Recipes

4/4/2018

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

For the most recent version of this document, visit our documentation website.

Relativity | Recipes - 1
Table of Contents
1 Adding translated text to documents 20
1.1 Recipe overview 20
1.2 Requirements 20
1.3 Directions 20

2 Adding documents to an existing Assisted Review project 21

2.1 Recipe Overview 21
2.2 Requirements 21
2.3 Directions 21
3 Advanced workflows with Relativity Legal Hold 22
3.1 Recipe Overview 22
3.2 Requirements 22
3.3 Directions 22

3.3.1 Custom fields 22

3.3.2 Custodian related information 25
3.3.3 Reporting 26
3.4 References 27
4 Cluster Visualization QC Workflow 27
4.1 Requirements 28

4.2 Directions 28
5 Coding the first item in a family group with the skip function 36
5.1 Recipe overview 36
5.2 Directions 36
6 Creating a read-only layout to lock produced documents 38
6.1 Recipe overview 38
6.2 Requirements 38

Relativity | Recipes - 2
 6.3 Directions 38
6.4 Considerations 42
7 Cluster Visualization QC Workflow 42
7.1 Requirements 42
7.2 Directions 42
8 Creating an application for managing attorney lists 51

8.1 Recipe overview 51

8.2 Requirements 51
8.3 Directions 51
9 Creating an optimized Relativity Analytics index (6.10-7.5) 52
9.1 Recipe overview 52
9.2 Requirements 53
9.3 Directions 53
9.3.1 Important Points 54

10 Creating an optimized Relativity Analytics index (9.5 and higher) 54

10.1 Requirements 54
10.2 Directions 54
10.2.1 Important Points 56
11 Creating commonly used dashboards in Relativity 56
11.1 Recipe overview 56

11.2 Requirements 56
11.3 Directions 57
11.3.1 Data Load QC 57
11.3.2 Database Stats 58
11.3.3 Document Characteristics 58
11.3.4 Review Prioritization 59
11.3.5 Review Progress/QC 60

Relativity | Recipes - 3
 11.3.6 Privilege Review/QC 61
11.3.7 Reviewer Stats 62
11.3.8 Search Term Review 63
11.3.9 Pre-Production QC 64
11.3.10 Post-Production Review 65
11.3.11 Email Threading 66

Language Identification 66
11.3.12 Tips 67
11.3.13 References 67
12 Creating commonly used Pivots 68
12.1 Recipe overview 68
12.2 Directions 68
13 Creating secured saved search folders for multiple groups 68
13.1 Recipe overview 68

13.2 Directions 69
14 Creating static searches using multiple-choice fields 69
14.1 Recipe overview 69
14.2 Requirements 70
14.3 Directions 70
15 Creating training resources tabs 70

15.1 Creating training resources tabs 70

15.1.1 Directions 70
15.2 Adding training resources to a workspace 71
15.2.1 Directions 72
16 Displaying family groups in a view 72
16.1 Recipe overview 72
16.2 Requirements 72

Relativity | Recipes - 4
 16.3 Directions 73
16.3.1 Information 73
16.3.2 Fields 73
16.3.3 Conditions 73
16.3.4 Sort 73
17 Displaying production fields in a layout 74

17.1 Recipe overview 74

17.2 Requirements 74
17.3 Directions 75
18 Email threading and near dupe - workflow alternatives 77
18.1 Recipe overview 77
18.2 Requirements 77
18.3 Directions 77
18.3.1 Running the Email Threading & the Near Dupe in a single SAS 77

18.3.2 Running only Email Threading in an SAS. 77

18.3.3 Running only Near Dupe in an SAS. 78
19 Email Threading Results - Identify threads with missing emails 78
19.1 Recipe overview 78
19.2 Requirements 78
19.3 Directions 78

19.4 References 80
20 Estimating diskspace utilization for databases and Analytics indexes 80
20.1 Requirements 80
20.2 Directions 80
20.2.1 Workspace sizes 81
20.2.2 Analytics index sizes 81
21 Exporting PDFs using document identifiers as file names 81

Relativity | Recipes - 5
 21.1 Recipe overview 81
21.2 Requirements 81
21.3 Directions 82
22 Exporting to HTML for expert witness review 82
22.1 Recipe overview 82
22.2 Requirements 82

22.3 Directions 82
22.3.1 Results 83
22.3.2 Offline coding form 84
23 Filtering to find empty fields 84
23.1 Recipe overview 84
23.2 Directions 84
24 Finding duplicates or near duplicates in a new set that match up to a previous set 85
24.1 Requirements 86

24.2 Directions 86
24.3 References 86
25 Finding emails exchanged within the same domain 86
25.1 Recipe overview 87
25.2 Requirements 87
25.3 Directions 87

26 How to calculate Precision and Recall without a control set 90

26.1 Recipe overview 90
26.2 Requirements 90
26.3 Precision and Recall Defined 90
26.3.1 Initial setup 92
26.3.2 View setup 92
26.3.3 Calculating Precision 92

Relativity | Recipes - 6
 26.3.4 Calculating Recall 96
26.3.5 Additional considerations 97
27 Identifying emails between two or more specific individuals 98
27.1 Requirements 98
27.2 Directions 98
28 Identifying standalone emails 103

28.1 Requirements 103

28.2 Directions 104
29 Importing productions in Relativity with natives 108
29.1 Overview 108
29.2 Requirements 108
29.3 Directions 108
30 Legal Hold checklist 110
30.1 Requirements 110

30.2 Directions 111

30.3 References 112
31 Leveraging processing metadata for quality control 112
31.1 Overview 112
31.1.1 Requirements 112
31.2 Directions 112

31.2.1 Important considerations 118

32 Loading external productions into Relativity 119
32.1 Requirements 120
32.2 Instructions 120
32.3 Notes 120
33 Maintaining previous Bates numbers between production sets 120
33.1 Recipe overview 120

Relativity | Recipes - 7
 33.2 Requirements 121
33.3 Directions 121
33.3.1 Option 1 121
33.3.2 Option 2 122
34 Mapping Processing Fields 122
34.1 Requirements 122

34.2 Directions 122

35 Mass importing into a file field 124
35.1 Overview 125
35.2 Requirements 125
35.3 Directions 125
36 Migrating cases from Summation 127
36.1 Recipe overview 127
36.2 Requirements 127

36.3 Directions 127

36.3.1 Migrating images 127
36.3.2 Migrating data 127
36.3.3 Migrating transcripts 128
36.3.4 Migrating productions 128
36.3.5 Migrating redactions 129

37 Multi-matter workspace setup in Relativity 129

37.1 Recipe overview 129
37.2 Requirements 130
37.3 Directions 130
37.3.1 User Groups/Security 130
37.3.2 Folder Setup 131
37.3.3 Field and choice setup 131

Relativity | Recipes - 8
 37.3.4 Views Setup 132
37.3.5 Layout Setup 132
37.3.6 Search indexes 133
37.3.7 Batches 134
38 Native Imaging - Dithering Algorithm Options 134
38.1 Recipe overview 134

38.2 Requirements 134

38.3 Directions 135
38.3.1 Dithering results for text-rich documents 135
38.3.2 Dithering results for image-rich documents 138
39 OCR Redacted Production Documents & Export Text 148
39.1 Requirements 148
39.2 Directions 149
40 Overlaying in the production object in Relativity 9.3 150

40.1 Requirements 151

40.2 Directions 152
40.3 References 154
41 Performing overturn analysis for Relativity Assisted Review 154
41.1 Recipe overview 154
41.2 Requirements 155

41.3 Directions 155

41.3.1 Correcting coding inconsistencies between true or conceptual duplicates 155
41.3.2 Identifying and correcting the most influential seed documents 155
41.3.3 Using the Overturn Analysis related items pane 155
42 Processing duplication workflow 156
42.1 Before you begin 156
42.1.1 Supported versions 156

Relativity | Recipes - 9
 42.1.2 Components 156
42.1.3 Considerations 157
42.2 Deploying and configuring the solution 157
42.3 Preparing the workspace 157
42.4 Running the solution 157
42.4.1 All Custodians script 158

42.4.2 All Source Locations script 159

42.4.3 Update Duplicate Status script 160
42.5 Viewing the results 162
42.5.1 All Custodians script 162
42.5.2 All Source Locations script 163
42.5.3 Update Duplicate Status script 163
43 Performing QC of Productions 164
43.1 Recipe overview 164

43.2 Requirements 164

43.3 Directions 164
43.3.1 Conversation before Production 165
43.3.2 Items to check in a production set 165
43.3.3 QC of production data 166
44 RAR checklist 166

44.1 Checklist 167

45 Relativity Assisted Review Reviewer Protocol (Sample Based and Active Learning) 168
45.1 Overview 168
45.2 Requirements 168
45.3 What is Assisted Review? 168
45.4 Selecting good example documents 168
45.4.1 Sufficient text 169

Relativity | Recipes - 10
 45.4.2 Email headers and repeated content 169
45.4.3 Images 169
45.4.4 Numbers 170
45.4.5 Families and the "Four Corners" test 172
45.5 Handling good language / bad example exceptions 173
45.6 Sample Based Learning 174

45.6.1 Applying the Use as Example field 174

45.6.2 Using the Excerpt text box 175
45.7 Coding Tips 176
46 Recording page-level information for a document 177
46.1 Recipe overview 177
46.2 Requirements 177
46.3 Directions 177
47 Regular expression searching - SSN and EIN 177

47.1 Recipe overview 177

47.2 Requirements 177
47.3 Directions 178
47.4 General considerations for all regular expression searches 181
48 Searching for symbols 181
48.1 Requirements 181

48.2 Directions 181

48.3 References 183
49 Relativity Structured Analytics – Best practices and workflows 183
49.1 Requirements 184
49.2 Directions 184
50 Relativity Structured Analytics - Email Threading 187
50.1 Requirements 187

Relativity | Recipes - 11
 50.2 Directions 187
50.2.1 Create saved searches 187
50.2.2 Create or configure an Analytics profile 188
50.2.3 Create a Structured Analytics set 189
50.2.4 Email threading results - reports 191
50.2.5 Setting up an Email Threading Saved Search 192

50.2.6 Setting up an Email Threading View 194

50.2.7 Email Threading Display 195
50.2.8 Working with email threading results 195
50.2.9 Special Considerations 196
50.2.10 Setting up an Email Threading View 197
50.2.11 Email Threading Display 198
50.2.12 Thread groups in the related items pane 199
50.2.13 Working with email threading results 200

50.3 Special considerations 204

50.4 References 204
51 Removing documents from a RAR example set 205
51.1 Requirements 205
51.2 Directions 205
51.3 References 205

52 Resolving common processing errors 206

52.1 Recipe overview 206
52.2 Requirements 206
52.3 Directions 206
53 Reviewer QC workflow ideas 210
53.1 Overview 210
53.2 Requirements 210

Relativity | Recipes - 12
 53.3 Directions 210
53.3.1 Create a Quality Control layout 210
53.3.2 Batch reviewed documents for QC review 212
53.3.3 Identify total documents put through QC and resulting overturns 213
53.4 Focus QC on Substantive Issues or Privilege 215
53.4.1 Conduct targeted substantive QC 215

53.4.2 Conduct Privilege QC 216

53.4.3 Check for family tagging conflicts 216
54 Sampling for repeated content 216
54.1 Requirements 216
54.2 Directions 216
54.2.1 Creating the sample 216
54.2.2 Save sample as list and list as saved search 217
54.2.3 Create and run repeated content identification as Structured Analytics set 218

54.2.4 Review results 218

54.2.5 Special considerations 218
55 Searching for custom object information on the Document object 219
55.1 Recipe overview 219
55.2 Directions 219
56 Searching for dates in Relativity 221

56.1 Types of date searches 222

56.2 Directions 222
56.2.1 Field filters 222
56.2.2 Saved search 223
56.2.3 Index search 223
56.3 Notes 224
57 Searching for documents with incorrectly loaded text 224

Relativity | Recipes - 13
 57.1 Recipe overview 224
57.2 Directions 224
58 Searching for handwritten documents 225
58.1 Overview 225
58.2 Requirements 225
58.3 Directions 225

59 Searching for single standalone emails 227

59.1 Requirements 227
59.2 Directions 227
60 Searching for terms while excluding email footers 228
60.1 Recipe overview 228
60.2 Requirements 228
60.3 Directions 228
61 Searching for unique identifiers 229

61.1 Overview 229

61.2 Requirements 229
61.3 Directions 229
61.3.1 Option A 230
61.3.2 Option B 230
62 Searching workflows 231

62.1 Requirements 231

62.2 Directions 231
62.2.1 Proximity search 232
62.2.2 Auto-Recognition 234
62.2.3 Searching for times 234
62.2.4 Filters 234
62.2.5 Troubleshooting workflow 235

Relativity | Recipes - 14
 63 Setting up in an upgraded 7.5 environment 236
63.1 Recipe overview 236
63.2 Requirements 236
63.3 Directions 236
64 Setting up CJK document workspaces in Relativity 237
64.1 Recipe overview 237

64.2 Requirements 237

64.3 Directions 238
64.3.1 Searching considerations 239
64.3.2 Analytics considerations 241
65 Sorting documents and keeping email families together 241
65.1 Recipe overview 241
65.2 Requirements 241
65.3 Directions 242

66 Starting a RAR project after review has begun 242

66.1 Requirements 242
66.2 Directions 242
66.3 References 243
67 Tally all email recipients 243
67.1 Recipe overview 243

67.2 Requirements 244

67.3 Directions 244
67.4 References 245
68 Tracking categorized documents to their example document 246
68.1 Recipe overview 246
68.2 Requirements 246
68.3 Directions 246

Relativity | Recipes - 15
 69 Transfer tags from other document review applications 247
69.1 Recipe overview 247
69.2 Directions 247
70 Threading review setup 247
70.1 Requirements 247
70.2 Directions 247

70.2.1 Sample view for Inclusive email with family 248

70.2.2 Sample view for Inclusive and non-Duplicate email with family 249
70.2.3 Sample view of the Related Items pane for an Email Thread Group 249
70.2.4 Sample view for Near Dupe Documents 250
70.2.5 Sample view of the Related Items pane for a Near Dupe Group 251
71 Understanding the statistical sampling formula 252
71.1 Recipe overview 252
71.2 Formula breakdown 253

72 Use case for Relativity Analytics - Clustering & Categorization 253

72.1 Common scenarios 253
72.2 Requirements 253
72.3 Directions 254
72.3.1 Scenario 1 254
72.3.2 Scenario 2 254

73 Using Analytics with CJK characters and non-English languages 254

73.1 Considerations for documents with CJK characters 255
73.1.1 Tokenization method 255
73.1.2 Workflow recommendations 255
73.2 Requirements 256
73.3 Recipe overview 256
73.4 Considerations for using Analytics with CJK characters 256

Relativity | Recipes - 16
 73.4.1 Considerations for creating Analytics indexes containing non-English languages 257
73.4.2 Considerations for using Structured Analytics with non-English languages 259
73.4.3 Checklist for using Analytics with non-English language and multiple language
workspaces 261
74 Using categorization to find privileged documents 262
74.1 Requirements 262
74.2 Directions 262

75 Using categorization to identify hot documents 263

75.1 Requirements 263
75.2 Directions 263
76 Using Cluster Visualization to jump start your RAR project 263
76.1 Requirements 263
76.2 Directions 264
77 Using custom objects to secure coding choices 267
77.1 Recipe overview 267

77.2 Directions 268

78 Using dtSearch to identify documents in other languages 268
78.1 Recipe overview 268
78.2 Requirements 268
78.3 Directions 269

79 Using Email Thread Visualization in your QC workflows 269

79.1 Requirements 270
79.2 Directions 270
80 Using Email Thread Visualization to leverage previously coded documents 271
80.1 Requirements 271
80.2 Directions 271
81 Using Near Duplicate analysis in review 272

Relativity | Recipes - 17
 81.1 Requirements 273
81.2 Overview 273
81.3 Directions 273
81.4 Workflow considerations 274
82 Using redactions from other applications 275
82.1 Recipe overview 275

82.2 Directions 275

83 Using reviewed documents to prioritize new documents via Analytics 275
83.1 Requirements 275
83.2 Directions 276
84 Using saved searches to complete conflict checks 276
84.1 Recipe overview 276
84.2 Directions 277
85 Using stop words and making some characters searchable in a dtSearch 279

85.1 Requirements 280

85.2 Directions 280
85.2.1 dtSearches and stop words 281
85.2.2 Stop words in languages other than English 282
85.2.3 Searching for Special Characters 282
86 Using tokens to customize stamps in a production 285

86.1 Recipe overview 285

86.2 Directions 285
87 Using the field tree to view folders 286
87.1 Recipe overview 286
87.2 Requirements 286
87.3 Directions 286
88 Using the Update Duplicate Status Script 288

Relativity | Recipes - 18
 88.1 Requirements 289
88.2 Directions 289
88.2.1 Publishing or loading additional documents 290
89 Verifying imaged natives 290
89.1 Recipe overview 290
89.2 Requirements 291

89.3 Directions 291

Relativity | Recipes - 19
1 Adding translated text to documents
If your documents require translation, this recipe describes one way of including both the original
text and the translated text in your Relativity workspace after you receive the translated documents
from the translation service.

1.1 Recipe overview

This recipe explains how to add translated text to your workspace, then search, view the translated
text alongside the original text, and potentially produce the translated text.

1.2 Requirements
n Relativity security access with the ability to edit/add fields
n Relativity Desktop Client

1.3 Directions
1. Create a field to hold the translated text using the following settings:
n Field type - Long Text
n Available in Viewer - Yes
n Included in Text index - Yes (not required, but helpful when searching with the keyword index)
2. After the documents have been translated and are ready for import, use the Relativity Desktop Client
to overlay the information into the newly created field. The translated text must exist in the load file
and not as separate text files.
3. Index the translated text. The newly added text may not be immediately searchable. The following
three options exist for indexing the text:
n Build a new dtSearch index with only the new translated text.
n Add this field to an existing dtSearch index, and then rebuild.
n Set the Include in Text Index attribute to Yes on the translated text field to add the content to
the keyword index.
4. Configure your workspace to display both the translated and original text:
a. Launch the standalone viewer.
b. Change the Extracted text drop-down to the translated text field you created in step 1.
c. Click the Sync button at the top of the standalone viewer to continuously update this window as
you navigate through documents.
5. If there is an agreement to share the translated text, export the content as you would any other field.

Relativity | Recipes - 20
2 Adding documents to an existing Assisted Review
project
2.1 Recipe Overview
Relativity allows you to add new documents to an Assisted Review project that is already in
progress. This recipe discusses the suggested steps to add documents to a live project.

2.2 Requirements
n Relativity Assisted Review
n Analytics Core
n Analytics

2.3 Directions
To apply a categorized value to new documents, you must complete a round. It is easier to get
immediate results if you have a round in process when beginning these steps. When you finish the
round, the new documents get categorized from the examples already in the system. Generally it
will be a QC round in progress, but any round type except Control Set will work. If you are not in
the middle of a round, begin one and code at least one document so a round can be finished.
1. Load the new documents into the workspace.
2. Ensure the new documents are included in your saved searches for the Analytics index.
3. Navigate to the Indexing & Analytics tab, select Search Indexes and then the Analytics
index.
4. From the Analytics Index console, rebuild the Analytics index to include the new documents.
A full rebuild of the Analytics index may or may not be needed. Consider the following:
n When adding new concepts or types of documents, a full rebuild is recommended.
n If the new documents are more of the existing types of documents and the system does
not need to learn anything new, an incremental population will probably suffice.
5. In the Assisted Review console, click the View Project Settings link. Ensure the saved search
used for Documents to be categorized includes the new documents.
6. Once manual review is completed for the current round and the index is finished rebuilding,
click Finish Round. The new documents will categorize alongside the old ones.
7. Return to the Project Home and verify that the total in the Docs in project column reflects
the additional documents.
8. Continue the Assisted Review project as normal.

Note: Any previously established control set in the project will no longer be in sync with the
document set. To correct this, make a new control set round to replace the previous one after
the existing round is completed.

Relativity | Recipes - 21
3 Advanced workflows with Relativity Legal Hold
When working in Relativity Legal Hold, there may be a need to customize it to fit your unique
workflows. In particular, we discuss tracking custodian information (like employee location),
tracking custodian response information in real time, and creating custom dashboards to visualize
and provide snapshots of your Hold data.

3.1 Recipe Overview

This recipe utilizes the flexibility provided by Relativity to integrate it into the Legal Hold
application, gaining the ability to track additional information or obtain information out of Relativity
in real time via Dashboards.

3.2 Requirements
n Relativity 9.4 or higher
n Relativity Legal Hold v4.1.437.2+

3.3 Directions
Below are some workflow suggestions one can employ to customize Relativity Legal hold to cater to
their unique needs and / or for advanced reporting functionality.

3.3.1 Custom fields

Create custom fields to track additional pieces of information that are unique to your Hold needs or
tie in well with your precedent of subsequent workflow. Below are a couple examples of these
fields.
Custom fields are on the Projects object to track information about the Department that the hold is
associated with. Below is an illustration of what the Custom field and the Custom Layout would look
like on a layout.

Relativity | Recipes - 22
On a similar note, custom fields are used on the Custodian object to track information like
Custodians Office location or company assigned devices and their information. Below is an
illustration of what those fields would look like on a layout.

Relativity | Recipes - 23
When creating the custom fields, layouts, or views, keep in mind the following considerations:
n You should not associate custom fields, layouts, or views with the Relativity Legal Hold application.
n You can enable the Open to Association setting to Yes.
n You can enable the Allow Group By and Allow Pivot settings to Yes.

Relativity | Recipes - 24
3.3.2 Custodian related information
The Custodian Status Tab can be unhidden and the view on the object made available. It’s useful to
report information about the Custodian’s interactions, such as:
n Number of reminders sent
n Numbers of escalations
n Date / Time the reminder was sent
n Date / Time the last escalation was sent
n Date / Time of the last scheduled escalation
n When questionnaire was completed and submitted
n When the Hold (Communication) Notice was sent
n When the Communication was viewed
n If the Hold notice was acknowledged
To create the Custodian Status Tab, fill out the settings as follows:
n Name - Custodian Status
n Link Type - Object
n Parent - Legal Hold
n Object Type - Custodian Status
To create a new View on the Custodian Status Object, fill out the setting as follows:

Relativity | Recipes - 25
 n Name - Custom View – Custodian Status
n Object Type - Custodian Status
n Fields - Select the fields of your choice from the following recommendations:
o Project
o Escalation Sent
o Reminders Sent
o Acknowledgment Status
o Questionnaire Status

n Sort Order - Custodian | Project

The Custodian Role Tab can also be unhidden and the view on the object made available. It's useful
to report information about the Custodian's interactions, such as:
n Custodian's role on a hold
n When the Custodian was released
To create the Custodian Role Tab, fill out the settings as follows:

n Name - Custodian Role

n Link Type - Object
n Parent - Legal Hold
n Object Type - Custodian Role

To create a new View on the Custodian Role Object, fill out the settings as follows:
n Name - Custom Role – Custodian Status
n Object Type - Custodian Role
n Fields - Select the fields of your choice from the following recommendations:
o Role
o Project
o Release Date
o Custodian:: Email

n Sort Order - Custodian | Project

3.3.3 Reporting
Dashboards provide a way to make information available in Relativity realtime, providing self-
service with very little training. Layering the Dashboards with custom views make it a powerful
reporting aid. Below are a few examples of custom Dashboards that address reporting needs.

3.3.3.1 Projects Object - Project Information Dashboard:

Name Group By Pivot On Type
Subject Matter Start Date Subject Matter Start Date <Grand Total> Line Chart
Project Status by External Counsel External Counsel Project Status Table
Department Office <Grand Total> Bar Chart (OR) Pie Chart
(Custom field)

Relativity | Recipes - 26
3.3.3.2 Custodian Object - Custodian Information Dashboard:
Name Group By Pivot On Type
Employee Status (Custom field) by Department <Grand Bar Chart (OR) Pie
Department Total> Chart
Employment Start Date Employment Start <Grand Line Chart
Date Total>
Custodian Office Office <Grand Bar Chart (OR) Pie
Total> Chart

3.3.3.3 Question Responses Object - Question Responses Breakdown:

Name Group By Pivot On Type
Answers by Project Project Answer Status Stacked Bar Chart
Custodian by Project Project Custodian Table

3.3.3.4 Mailbox Object - Question Responses Breakdown:

Name Group By Pivot On Type
Message Type Message Type <Grand Total> Pie Chart
Message Status by Project Project Message Status Stacked Bar Chart
Message Sent Date Sent Date <Grand Total> Line Chart

3.3.3.5 Custodian Status Object - Custodian Information Dashboard:

When the Custodian Status tab is available, we can use Pivots and Dashboards to visualize a lot of
this information and make it more digestible.

Name Group By Pivot On Type

Escalations by Custodian Custodian Escalations Sent Table
Resolutions by Custodian Custodian Resolved By Table
Reminders by Custodian Custodian Reminders Sent Table

3.4 References
Relativity Legal Hold

4 Cluster Visualization QC Workflow

Improve the efficiency and effectiveness of your quality control workflow by using cluster
visualization to help identify potential coding discrepancies. This recipe describes how you can use
cluster visualization to identify patterns and visually compare groups of conceptually similar
documents to uncover potential coding inconsistencies.

Relativity | Recipes - 27
4.1 Requirements
n An active Analytics index
n An existing cluster

4.2 Directions
This recipe involves a scenario where the case team has performed a first-level review for
responsiveness. Cluster visualization can help QC the responsiveness coding. The approach
described here could also be used to QC privilege coding or issue coding.
1. Ensure that the reviewed and coded documents are included in an existing cluster set.
2. From the Documents tab, select the cluster set in the cluster browser and click Visualize Cluster.

Relativity | Recipes - 28
 3. In the search panel, click Add Condition and select the responsiveness coding field. Create a condition
that will return documents coded as responsive. Click Apply, then click Run Search.

Relativity | Recipes - 29
 4. The system displays a heat map overlay on the cluster visualization that indicates the percentage of
documents coded as responsive in each cluster. The darker the shading, the higher the concentration
of responsive documents in the cluster.
5. Throughout this scenario, you can focus in on the documents in a cluster by right clicking the cluster
and clicking Select, then Apply. You can select multiple clusters by doing this for each cluster you want
to focus on.

Relativity | Recipes - 30
 6. The main cluster visualization view (the circle pack view) provides the following visual clues to aid in
your QC review:
a. Darker clusters – The darker shading indicates that a cluster contains a high percentage of doc-
uments coded as responsive, meaning that the documents in the cluster have been coded con-
sistently. Since documents within a cluster are conceptually similar, we would generally expect
them to be treated similarly, and this is what the darker shading tells us.

Relativity | Recipes - 31
 b. Lighter clusters – The lighter shading indicates that only a small percentage of documents in the
cluster have been coded as responsive. The documents in these lighter shaded clusters would
warrant further investigation to determine why only a small portion of conceptually similar doc-
uments were coded as responsive.

Relativity | Recipes - 32
 7. Continue your QC review by right clicking a darker shaded cluster (one with a high percentage of
responsive documents) and selecting View Nearby Clusters from the right click menu.

Relativity | Recipes - 33
 8. The Nearby Clusters visualization places the selected cluster in the middle of the screen and shows you
other clusters that are conceptually similar to it. The closer a cluster is to the center cluster, the more
conceptually similar it is.

Relativity | Recipes - 34
 Along with the circle pack, the clusters can be visualized on the dial by right clicking and selecting View
Dial.

9. The Nearby Clusters visualization reveals the following additional insights to help guide your QC review:
a. Darker clusters – We would expect clusters that are nearby, or conceptually similar to, the center
cluster to also have a darker shading, indicating that they too contain a high percentage of doc-
uments coded as responsive.
b. Lighter clusters - Clusters with a lighter shading would probably warrant additional investigation.
The system is telling us that these clusters are conceptually similar to the dark cluster in the
middle (which contains a high percentage of responsive documents). However, only a small por-
tion of the documents in these lighter clusters have been coded as responsive; we might want to
investigate why this is. A cluster that is completely white contains no responsive documents, yet
is conceptually similar our center cluster, and should also be investigated.

Relativity | Recipes - 35
 Note that with the Responsiveness filter set and active, only documents coded as responsive
are returned. If you would like to access all the documents in a selected cluster, simply clear
the box on the filter card, click Run Search, and all documents within a cluster will be
returned.

5 Coding the first item in a family group with the skip

function
When a document in a family group is marked as responsive, the rest of the group is typically
grouped as responsive. If the other family items are going to a second pass team for review, you can
quickly prioritize which groups you send to the second pass team to determine privilege and
importance

5.1 Recipe overview

This recipe explains how to use a saved search, a view, and the document skip function to determine
the next document available for review. By determining whether at least one item in a family has
been coded as responsive, you can remove any other family items from the review queue based on
that one coding decision, which is not propagated to the family group.

5.2 Directions
1. Create a new saved search and name it Responsiveness is Responsive and Family.
2. Choose the following search information and conditions:
n Set Includes to Include Family.
n Set Scope to Entire Workspace.

Relativity | Recipes - 36
 n Set Field to Responsiveness, Operator to any of these, and Value to Responsive.

3. Create a new view:

a. Name the view.
b. Give the view any order number.
c. Set Visible in Dropdown to Yes.
d. Set Object Type to Document.
e. Include the fields to display.
4. Select the following conditions:
a. For the first condition, set Field to Responsiveness and Operator to is not set.
b. Select AND.
c. For the second condition, set Field to (Saved Search), Operator to Document is not in, and
Value to Responsiveness is Responsive and family.

Relativity | Recipes - 37
 5. Use the Skip function with this view. After the first item in a family group is coded, even without
propagation set, the rest of the family group is removed from the review set without being coded. This
way you can manually code each document without the need for a first level review.

6 Creating a read-only layout to lock produced

documents
Rolling productions can potentially reverse the document coding that has already been produced. If
you still need to review documents, you can't lock down these coding fields. However, you can make
reviewed documents read-only to prevent changing coding on produced documents.

6.1 Recipe overview

This recipe describes how to add an object rule to your Document object that allows documents to
be viewed in a read-only layout.

6.2 Requirements
n Relativity security access
o Fields: Add
o Object rules: Add
o Ability to build layouts

6.3 Directions
Perform the following steps to create a read-only layout:
1. From the Layouts tab, create a new layout with the following settings to display the documents that
have been coded:
a. Object type: Document
b. Name: Previously Produced
c. Order: 10
d. Enable Copy From Previous: No
Alternatively, you can copy the original coding form and change only the fields that should not
be altered after production to read-only.

Relativity | Recipes - 38
 2. From the Fields tab, create a new field with the following settings to indicate which documents need to
be locked down:
a. Object type: Document
b. Name: Produced and locked
c. Field type: Single choice
d. Available in field tree: Yes

Relativity | Recipes - 39
 3. From the Choices tab, add a single choice field with the following settings:
a. Field: Produced and Locked (Document)
b. Name: Yes

Relativity | Recipes - 40
 4. From the Administration tab, select Object type and select the Document object.
5. In the Rules section, click New and create a new rule with the following settings:
a. Rule type: Default layout
b. Name: Production lock documents
c. Field: Produced and locked
d. Value: Yes
e. Action: Previously produced
f. Leave User can select another layout unchecked.

Relativity | Recipes - 41
 6. Once you've completed steps 1-5, a system admin must switch the produced documents choice in the
Produced and Locked field to Yes.
7. Examine the records. In the layout drop-down, there should only be the read-only version.

6.4 Considerations
n Make sure your reviewers don't have rights to edit layouts, as they can then switch read-only to No.
n This doesn't work on Mass Edit operations, and you must limit that right for users.

7 Cluster Visualization QC Workflow

Improve the efficiency and effectiveness of your quality control workflow by using cluster
visualization to help identify potential coding discrepancies. This recipe describes how you can use
cluster visualization to identify patterns and visually compare groups of conceptually similar
documents to uncover potential coding inconsistencies.

7.1 Requirements
n An active Analytics index
n An existing cluster

7.2 Directions
This recipe involves a scenario where the case team has performed a first-level review for
responsiveness. Cluster visualization can help QC the responsiveness coding. The approach
described here could also be used to QC privilege coding or issue coding.

Relativity | Recipes - 42
 1. Ensure that the reviewed and coded documents are included in an existing cluster set.
2. From the Documents tab, select the cluster set in the cluster browser and click Visualize Cluster.

Relativity | Recipes - 43
 3. In the search panel, click Add Condition and select the responsiveness coding field. Create a condition
that will return documents coded as responsive. Click Apply, then click Run Search.

Relativity | Recipes - 44
 4. The system displays a heat map overlay on the cluster visualization that indicates the percentage of
documents coded as responsive in each cluster. The darker the shading, the higher the concentration
of responsive documents in the cluster.
5. Throughout this scenario, you can focus in on the documents in a cluster by right clicking the cluster
and clicking Select, then Apply. You can select multiple clusters by doing this for each cluster you want
to focus on.

Relativity | Recipes - 45
 6. The main cluster visualization view (the circle pack view) provides the following visual clues to aid in
your QC review:
a. Darker clusters – The darker shading indicates that a cluster contains a high percentage of doc-
uments coded as responsive, meaning that the documents in the cluster have been coded con-
sistently. Since documents within a cluster are conceptually similar, we would generally expect
them to be treated similarly, and this is what the darker shading tells us.

Relativity | Recipes - 46
 b. Lighter clusters – The lighter shading indicates that only a small percentage of documents in the
cluster have been coded as responsive. The documents in these lighter shaded clusters would
warrant further investigation to determine why only a small portion of conceptually similar doc-
uments were coded as responsive.

Relativity | Recipes - 47
 7. Continue your QC review by right clicking a darker shaded cluster (one with a high percentage of
responsive documents) and selecting View Nearby Clusters from the right click menu.

Relativity | Recipes - 48
 8. The Nearby Clusters visualization places the selected cluster in the middle of the screen and shows you
other clusters that are conceptually similar to it. The closer a cluster is to the center cluster, the more
conceptually similar it is.

Relativity | Recipes - 49
 Along with the circle pack, the clusters can be visualized on the dial by right clicking and selecting View
Dial.

9. The Nearby Clusters visualization reveals the following additional insights to help guide your QC review:
a. Darker clusters – We would expect clusters that are nearby, or conceptually similar to, the center
cluster to also have a darker shading, indicating that they too contain a high percentage of doc-
uments coded as responsive.
b. Lighter clusters - Clusters with a lighter shading would probably warrant additional investigation.
The system is telling us that these clusters are conceptually similar to the dark cluster in the
middle (which contains a high percentage of responsive documents). However, only a small por-
tion of the documents in these lighter clusters have been coded as responsive; we might want to
investigate why this is. A cluster that is completely white contains no responsive documents, yet
is conceptually similar our center cluster, and should also be investigated.

Relativity | Recipes - 50
 Note that with the Responsiveness filter set and active, only documents coded as responsive
are returned. If you would like to access all the documents in a selected cluster, simply clear
the box on the filter card, click Run Search, and all documents within a cluster will be
returned.

8 Creating an application for managing attorney lists

At the start of a project, it’s typical for reviewers to receive a list of attorneys associated with a
case. To manage these lists as the review progresses, create a master list within Relativity , which
can be easily updated and maintained.

8.1 Recipe overview

This recipe shows you how to create a single object with multiple fields on it, allowing you to import
a CSV attorney list into your project.

8.2 Requirements
n Relativity Desktop Client (RDC)
n Microsoft Excel attorney list (this can be saved as a CSV file)

8.3 Directions
1. Create a new object in your workspace called Attorney. You may wish to place this new tab under an
Attorney List parent tab.
2. A Name field will automatically be created on the Attorney object. Change the Name field to Last
Name, First Name.
3. Click New Field to create the following single-choice fields on the Attorney object:
a. Alphabet

Note: This field is optional.

b. Law Firm or Employer

c. Client

Relativity | Recipes - 51
 4. Edit the Attorney tab view and layout to include the new fields.
5. In Excel, create or modify the attorney list you wish to import.
a. Name the columns in the spreadsheet the same as the fields created above.
b. The column that corresponds with the Alphabet field should be populated with the first letter of
each attorney’s last name. This will facilitate sorting and quick access to the list for your review-
ers.

6. Save the Excel file as a CSV file.

7. Launch and log in to the RDC.
8. Choose your workspace, and change the dropdown from the Document object to the Attorney object.
9. Choose your CSV file and import your list.
10. Enable filters to search on any combination of the newly created single-choice fields.
11. Add new attorneys as needed during review.
12. If necessary, add other fields and columns to match your workflow.

9 Creating an optimized Relativity Analytics index

(6.10-7.5)
Relativity Analytics uses only the documents you provide to create a search index. Because no
outside word lists are used, you must create saved searches to dictate which documents are used to
build the index. Generally, one index is built for each workspace. However, if you wish to limit
search results to certain document groups or have more than one language in the document set,
multiple indexes might give you better results.

9.1 Recipe overview

This recipe shows you how to use saved searches and filters to optimize a Relativity Analytics index.

Relativity | Recipes - 52
9.2 Requirements
n Relativity Analytics 6.10-7.5
n Existing decimal-type field

9.3 Directions
1. Run “Set Extracted Text Size” Script - The Set Extracted Text Size script is included with every instance
of Relativity . If it’s not already part of your workspace, select the script from the library and run it.
Select the field to hold the data, and click run. This will apply a number to your decimal field to hold the
text size in KB.
2. Saved Search Setup - Relativity Analytics indexes use saved searches to determine the document set
for the index. There are two default searches in the index setup. These searches will remove documents
without text, as well as documents with more than 2 MB of extracted text, for the training set. To optim-
ize the index, you will usually need to remove additional documents. The following types of documents
do not typically return desired results, and increase the size and build time of the index:
n Documents with OCR errors, such as “OCR not processed”
n Documents with poor-quality OCR, such as OCR on handwritten documents
n Compressed files, such as ZIP or RAR files
n System files
n Image files
n CAD drawings
n Maps
n Calendar items
n Excel files with minimal text (numbers and symbols are not conceptually searchable)
We only remove documents with more than 2 MB of text from the training set search. These
documents can have large amounts of text and many concepts, which make them difficult for the
system to index. For documents that have errors such as “OCR not processed,” we need to create our
own searches.
To optimize the index, we will create two saved searches—the training set and searchable set—instead
of using the defaults. The searchable set includes all documents that need to be searched, minus the
items listed above. The training set is a copy of the searchable set, but will limit the size of extracted text
to items less than 2 MB.
a. If you find other document groups that do not have helpful text, criteria can be added to these
searches later.
b. The only field to be returned in the saved searches will be the Extracted Text field.
c. Concept searching works by finding repeated word content across documents. This means that
metadata that repeats across documents, such as author and recipient, will bring emails
together conceptually because of that metadata, and not because of the actual content in the
body of the documents. For that reason, you should be mindful that email headers are con-
tained within the Extracted Text field—which brings us to the next section.
3. Filtering - Filters remove extraneous data that should not be conceptually indexed or searched. For
example, you can use filters to ensure that repeating header or footer information is not included in the
index to avoid the interference mentioned above. Note that filters don’t permanently remove data.
Rather, they inhibit data from being included in the indexes. There are five filter types:

Relativity | Recipes - 53
 n Email – Removes email header information throughout email-formatted text
n OCR – Uses groupings of three characters to determine if the text to be indexed contains valid
words; this cannot be edited
n Go Words – Removes bad OCR by comparing all words against a lengthy dictionary; this dic-
tionary file can be edited
n Repeated Content – Removes blocks of text, such as email footer information
n Java Regular Expressions – Identifies blocks of formatted text that can be removed from an index

9.3.1 Important Points

n Word order does not matter—only the co-occurrence of terms.
n Single words will not return valid results, except with keyword expansion.
n Analytics can search across languages. For example, running a concept search with French text can
work against an index that contains both French and English content. Analytics will find content with
French language within the larger English set. However, analytics will not translate French text into Eng-
lish.
n Analytics indexes are not available for searching when they are being incrementally updated or rebuilt.
A second inactive index can serve as a great backup, but will take up server space.

10 Creating an optimized Relativity Analytics index (9.5

and higher)
Published October 20, 2016
Relativity Analytics uses only the documents you provide to make a search index. Because no outside
word lists are used, you must create Saved Searches to dictate which documents are used to build
the index. Generally, one index is built for each workspace. However, if you wish to limit search
results to certain document groups or have more than one language in the document set, multiple
indexes might give you better results.

10.1 Requirements
n Relativity Analytics
n Existing decimal-type field

10.2 Directions
1. The key search is based on text size. Your processing tool might provide the text size. If not, run the Set
Extracted Text Size script. This updates a decimal field with the text size in KB.
2. Create a saved search of the data and (name it Analytics Data Search or something similar).
n Bring back all text less than 30 MB which translates to “Extracted Text Size is less than 30,000 and
Extracted Text Size is greater than 0. (This will return documents that have had this script run
and nothing too large that will encumber the system.)
n Ensure that the Extracted Text field is the only one returned.

Relativity | Recipes - 54
 3. Run Structured Analytics > Repeated Content Identification on your Analytics Data Search.
n Set the Minimum number of occurrences to 0.5% of the total population. Leave the rest of the
settings to the default.
n Scan the results of repeated content identification to ensure that the repeated text blocks cor-
respond to unwanted (non-authored) content.
4. Create a new Analytics profile:
n Set the Email header filter to Yes.
n Set OCR and Go Words to No.
n If you intend to run email threading, you can match up fields for Structured Analytics email
threading.
n Save the Analytics profile and link any repeated content filters which comprise non-authored con-
tent.
5. Create your Analytics index
n Use the Analytics Data Search created above for both training and searchable (unless it is for a
very large workspace).
n Select the Analytics profile created above.
n Set Optimize Training Set and Auto Remove Signatures to Yes.
6. Build your index.
1. The key search is based on text size. Your processing tool might provide the text size. If not, run the Set
Extracted Text Size script. This updates a decimal field with the text size in KB.
2. Create a saved search of the data and (name it Analytics Data Search or something similar).
n Bring back all text less than 30 MB which translates to “Extracted Text Size is less than 30,000 and
Extracted Text Size is greater than 0. (This will return documents that have had this script run
and nothing too large that will encumber the system.)
n Ensure that the Extracted Text field is the only one returned.
3. Run Structured Analytics > Repeated Content Identification on your Analytics Data Search.
n Set the Minimum number of occurrences to 0.5% of the total population. Leave the rest of the
settings to the default.
n Scan the results of repeated content identification to ensure that the repeated text blocks cor-
respond to unwanted (non-authored) content.
4. Create your Analytics index
n Use the Analytics Data Search created above for both training and searchable (unless it is for a
very large workspace).
n In the Advanced Settings, ensure that you set Optimize Training Set, Remove English Sig-
natures and Footers, and Email Header Filter to Yes.
n Link the appropriate Repeated Content filter results from Step 3 above.

Note: If you have a large amount of Repeated Content results, typically the results which are most
occuring and with the most words will suffice.

5. Build your index.

The following are explanations of the various settings selected above:

Relativity | Recipes - 55
 n Email Header Filter - removes email headers throughout email-formatted text.
n Repeated Content - removes blocks of text, such as email footer information. Instances of repeated
content may be detected and converted into filters automatically by a structured analytics set.
n Optimize Training Set - this option automatically excludes poor quality index training documents
based on an analysis of their text. Commonly excluded content includes extremely large documents
and files predominated by tables of numbers, long garbage strings of characters, and a preponderance
of symbols (rather than words).
n Auto-removal of email signatures and footers - this option automatically excludes email signatures
and footers when it finds them (although it is intentionally conservative to avoid over-exclusion. This
feature works for English signatures and footers only.
n Remove English Signatures and Footers - this option excludes email signatures and footers when it
finds them (although it is intentionally conservative to avoid over-exclusion). This feature works for Eng-
lish signatures and footers only.

10.2.1 Important Points

n Single word queries tend to not return valid results, except with keyword expansion.
n Analytics indexes are not available for searching when they are in the build phase of an incremental
update or if they are completely repopulated. A second inactive index can serve as a backup and will
allow for continued access during updates. But remember—this secondary index will take up server
space.
n Prior to population, remember to link the desired repeated content filters to the Analytics profile to be
used with your index.

11 Creating commonly used dashboards in Relativity

This recipe includes information on creating 12 different dashboards and the widgets saved on each
dashboard. These dashboards were presented during the session Creating Dashboards for your Case
Lifecycle at Relativity Fest 2016.

11.1 Recipe overview

This recipe walks through the fields to group by and pivot on to create some commonly used
dashboards in your case life cycle. You and your case team will be better equipped to explore and
prioritize your review, identify patterns and trends across custodians, and develop better review
strategies with real-time insight with these dashboards.

11.2 Requirements
Relativity 9.3 or higher

Relativity | Recipes - 56
11.3 Directions
Use the following dashboards as examples to generate ideas for dashboards you may want to create
in your cases:
n Data Load QC below
n Database Stats on the next page
n Document Characteristics on the next page
n Review Prioritization on page 59
n Review Progress/QC on page 60
n Privilege Review/QC on page 61
n Reviewer Stats on page 62
n Search Term Review on page 63
n Pre-Production QC on page 64
n Post-Production Review on page 65
n Email Threading on page 66
n Language Identification on page 66
n Tips on page 67

11.3.1 Data Load QC

To create the Data Load QC dashboard, group by and pivot on the following fields:
n Group by Custodian
n Group by Custodian and pivot on Sort Date (Year)

Relativity | Recipes - 57
11.3.2 Database Stats
To create the Database Stats dashboard, group by and pivot on the following fields:
n Group by Native Type
n Group by Extension
n Group by Sort Date
n Group by Key Term STR

11.3.3 Document Characteristics

To create the Document Characteristics dashboard, group by and pivot on the following fields:
n Group by Has Hidden Data
n Group by Unprocessable
n Group by Track Changes
n Group by Password Protected
n Group by File Type - Note that this field collapsed in the dashboard example below.

Note: You can collapse your widgets (Count of File Type highlighted below). If you save your dashboard
with a widget collapsed, it will remain collapsed.

Relativity | Recipes - 58
11.3.4 Review Prioritization
To create the Review Prioritization dashboard, group by and pivot on the following fields:
n Cluster Visualization Widget
n Group by Key Term STR
n Group by Custodian

Relativity | Recipes - 59
11.3.5 Review Progress/QC
To create the Review Progress/QC dashboard, group by and pivot on the following fields:
n Group by Responsive Designation
n Group by Privilege
n Group by Issues
n Group by Key Term STR
n Group by Custodian
n Group by File Extension

Relativity | Recipes - 60
11.3.6 Privilege Review/QC
To create the Privilege Review/QC dashboard, group by and pivot on the following fields:
n Group by Privilege
n Group by Privilege Designation
n Group by Privilege Terms STR
n Group by Custodian

Relativity | Recipes - 61
11.3.7 Reviewer Stats
To create the Reviewer Stats dashboard, group by and pivot on the following fields:
n Group by Reviewed By
n Group by Reviewed By and Pivot on Reviewed On
n Group by File Extension

Note: This was created using the Track Document Field Edits by Reviewer application, found in the
Relativity Community. It was set up to track the first edit on the Responsive Designation field. If you aren't
setting up this application in your template, you'll need to create this dashboard in your workspace, once
you’ve installed the application and created the fields.

Relativity | Recipes - 62
11.3.8 Search Term Review
To create the Search Term Review dashboard, group by and pivot on the following fields:
n Group by Responsive Designation
n Group by Key Term STR
n Group by Issue Coding STR
n Group by Issues
n Group by Privilege
n Group by Privilege Terms STR

Relativity | Recipes - 63
11.3.9 Pre-Production QC
To create the Pre-Production QC dashboard, group by and pivot on the following fields:
n Group by Custodian
n Group by Responsive Designation
n Group by Privilege
n Group by Key Term STR
n Group by Has Images
n Group by Issues

Relativity | Recipes - 64
11.3.10 Post-Production Review
To create the Post-Production Review dashboard, group by and pivot on the following fields:
n Group by Custodian and pivot on Production::Production Name
n Group by Production Type
n Group by Production::Production Name

Note: Currently the production fields are not enabled for group by/pivot on and the application is locked.
In future releases we plan to have the fields enabled for grouping and pivot in the production application.

Relativity | Recipes - 65
11.3.11 Email Threading
To create the Email Threading dashboard, group by and pivot on the following fields:
n Group by Inclusive Email
n Group by Inclusive Reason
n Group by Custodian
n Group by Date Sent (year)

Note: We recommend using your Email Threading view with this dashboard.

Language Identification
To create the Language Identification dashboard, group by and pivot on the following fields:
n Group by Primary Language
n Group by Docs_Languages::Language
n Group by Custodian

Note: We recommend using your Language Identification view with this dashboard.

Relativity | Recipes - 66
11.3.12 Tips
n Create a default Document List dashboard (lowest order number) with only the item/document list, no
additional widgets.
n You can only have one Cluster Visualization widget on a dashboard.
n You can create dashboards on other objects.
n Dashboards can be created in your template, so they carry over to your new workspaces. However,
some dashboards, such as those with search term report widgets or cluster visualization, will need to
be created in the individual workspace once you’ve created your clusters and/or STRs.

11.3.13 References
n Dashboards
n Dashboard and Pivot quick reference
n Advice Hub solution:Track document field edits by reviewer
n Dashboards
n Dashboard and Pivot quick reference
n Advice Hub solution:Track document field edits by reviewer
n Dashboards
n Dashboard and Pivot quick reference
n Advice Hub solution:Track document field edits by reviewer

Relativity | Recipes - 67
12 Creating commonly used Pivots
Relativity Pivot is a powerful feature that can help you visualize your data set in a digestible way.
Through interactive charts and graphs, Pivot allows you to gain insight into trends and patterns in
your case. The following suggestions describe how to make the most of Pivot.

12.1 Recipe overview

This recipe describes a number of Pivot charts that may be useful for analyzing data throughout
your project.

12.2 Directions
Set the following fields as suggested to create a number of useful Pivot charts.
Group by Pivot on Description
Custodian Sent date Review sent dates of emails for each custodian to verify a date range is not
missing from the collection.
Custodian File type Verify all documents were collected, including emails and loose files.
Custodian STR-Keywords Examine the count of keywords that occur by custodian.
Sent date Issues Examine the occurrence of issues over a time period.
Batch::Assigned Responsiveness View reviewer progress of responsive items across batches.
to
Cluster STR-Keywords Find out which clusters or document groups contain the most interesting
documents based on keywords.
Author Recipient Dive into internal emails, potentially privileged documents, or find per-
domains domains sonal emails received at work.
Author Issues or STR- Determine the most important domains where issues or keywords
domains Keywords appear.

13 Creating secured saved search folders for multiple

groups
When two groups need to add saved searches to your workspace ,without sharing them with each
other, you can secure the individual saved searches.

13.1 Recipe overview

This recipe describes how to create two secure folders to prevent two groups from seeing each
other’s saved searches.

Relativity | Recipes - 68
13.2 Directions
Set workspace security as follows:
Group Security Setting Comment
1 & 2 Search: View This ensures users in these groups can access the saved search browser, but can't
save searches outside of their assigned folders.
1 & 2 Search Folder:
View
1 & 2 Saved Search
Browser Access
1. Create two folders in the saved search browser, one for each group.
2. Secure the first search folder to remove Group 2, and then set permissions for Group 1 as follows:
Group Security Comment
Setting
1 Search: Edit This allows the user to save searches only in this folder. You can substitute Delete
and Add for the Edit permission if you want to allow the users to delete their own saved
searches or sub-folders.
1 Search
Folder: Edit
and Add
3. Repeat step 2 for the search folder for Group 2.

Note: If you need to make any modifications to either the folders or saved searches within the secured
folders, you must address these changes individually at the folder level and not in workspace permissions.

14 Creating static searches using multiple-choice fields

As databases update with additional documents, or as edits are made, saved searches provide
different results that include new documents. To maintain the same search results from before a
database update, you can create a field to store the results.

14.1 Recipe overview

This recipe describes how to create a multiple-choice field and populate it with records in a static
search. It also explains how to make this new field visible in the field tree to manage and review
your static searches.

Relativity | Recipes - 69
14.2 Requirements
n Mass updates
n Relativity security access with the following permissions:
o Fields: edit/add
o Choices: edit/add
o Document: edit

14.3 Directions
1. Create a new multiple choice field with the following settings to hold the items for each search.
n Object Type - Document
n Name - Static Searches
n Field Type- Multiple Choice
n Available In Field Tree- Yes
2. Add a choice to this field, and then name it after the appropriate search.
3. Build, and then run the desired search.
4. Mass update the Static Searches field with the choice you created in step 2.
5. Review the Field Tree and the Static Searches field to view your new search choices and results.

15 Creating training resources tabs

While in Relativity, users can access various resources by adding tabs that display training resources
so that users don't have to leave the Relativity environment and can quickly jump back into their
workspaces once they’ve found the information they need.
The first recipe shows you how to create tabs in your workspace that contain documentation,
webinars, and recipes. The second recipe helps you import a training resources application, which
creates a tab containing all of Relativity’s training and reference materials within the Relativity
interface.

15.1 Creating training resources tabs

15.1.1 Directions
1. From the home screen, click Admin Workspace Configuration > Tabs.
2. Click New Tab and create the Training Resources parent tab:
a. Name the tab Training Resources.
b. Set Order to 10.
c. Set Link Type to Parent.
3. Click New Tab and create the On-Demand Videos tab:
a. Name the tab On-Demand Videos.
b. Set Order to 10.
c. Set Link Type to External.

Relativity | Recipes - 70
 d. Set Parent to Training Resources.
e. Set Link to https://www.relativity.com/ediscovery-training/.
4. Click New Tab and create the Documentation tab:
a. Name the tab Documentation.
b. Set Order to 20.
c. Set Link Type to External.
d. Set Parent to Training Resources.
e. Set Link to the appropriate URL based on your Relativity version:
n RelativityOne: https://help.relativity.com/RelativityOne
n 9.5: https://help.relativity.com/9.5/
n 9.4: https://help.relativity.com/9.4/
n 9.3: https://help.relativity.com/9.3/
n 9.2: https://help.relativity.com/9.2/
n 9.1 – 9.0: https://help.relativity.com/9.0/
n 8.2: https://help.relativity.com/8.2/
n For earlier versions, locate the appropriate link on the documentation archives page:
https://help.relativity.com/9.3/Content/Resources/Shared/Documentation_archives.htm
5. Click New Tab and create the Recipes tab:
a. Name the tab Recipes.
b. Set Order to 30.
c. Set Link Type to External.
d. Set Parent to Training Resources.
e. Set Link to the appropriate URL based on your Relativity version:
n R1: https://help.relativity.com/RelativityOne/Content/Site_Resources/RecipesHome.htm
n 9.5: https://help.relativity.com/9.5/Content/Site_Resources/RecipesHome.htm
n 9.4: https://help.relativity.com/9.4/Content/Site_Resources/RecipesHome.htm
n 9.3: https://help.relativity.com/9.3/Content/Site_Resources/RecipesHome.htm
n 9.2: https://help.relativity.com/9.2/Content/Site_Resources/Home.htm
n 9.1 – 9.0: https://help.relativity.com/9.0/#../Subsystems/Default/Content/Site_Resources/Re-
cipesHome.htm
n 8.2: https://help.relativity.com/8.2/#../Subsystems/Default/Content/Site_Resources/Re-
cipesHome.htm
n For earlier versions, locate the appropriate link on the documentation archives page:
https://help.relativity.com/9.5/Content/Resources/Shared/Documentation_archives.htm

15.2 Adding training resources to a workspace

Note: You must have valid Relativity Community credentials in order to download any Community file linked to
from the documentation site. You'll need to enter those credentials on the Community login screen if you're
not already logged in. If you're already logged in to the Community at the time you click a link, the file is
automatically downloaded in the bottom left corner of your screen. If you get an error message stating "URL
No Longer Exists" after clicking a Community link, it may be due to a single sign-on error related to the SAML
Assertion Validator, and you should contact your IT department.

Relativity | Recipes - 71
 15.2.1 Directions
1. Visit the Relativity Community and search for RA_Training*.
2. Download the training application.
3. From the Applications tab in either the workspace or admin mode, install the application to your
desired workspace(s).
4. Once the application installs successfully, a new tab, Training Resources, appears in the work-
space(s) you selected. You will also see the sub-tabs for Tutorials, Videos, and Documentation.
Here, you can navigate to these tabs to access training resources without leaving Relativity.

16 Displaying family groups in a view

During review, it may be helpful to quickly see documents as they exist in their family groups.
Building a view to organize documents by family can support this workflow.

16.1 Recipe overview

This recipe describes how to create a view to display documents organized by family.

16.2 Requirements
n Workspace access with permission to the following tabs:
o Administration | Views

Relativity | Recipes - 72
16.3 Directions
Create a view with the following settings:

16.3.1 Information
View information:
n Object Type: Document
n Name: Documents
n Owner: Public
n Order: 10
Other:
n Group Definition: Family Group (Family related items field)
n Relativity Applications: none
n Query Hint: leave blank
n Visible In Dropdown: Yes

16.3.2 Fields
Suggested fields - Include Edit, File Icon, Control Number, Group Identifier, File Name (or Name
or Unified Title), Record Type, and any additional fields you want to place on the view

16.3.3 Conditions
Set conditions - None

16.3.4 Sort
It is necessary to sort based on the related item field. It might be Group Identifier or Family Group.
Then how the families are displayed is optional.
n Control Number
n Family Date - Use the Propagate Sent Date to Family Documents script to set values for this field.
Save the View.
Final View:

Relativity | Recipes - 73
 Note: The screen shot is not intended to show as indented (with dots), but to illustrate the view you see
when you use the indented list method to group the families together. Notice in the view that each family is
now divided by bolded lines.

17 Displaying production fields in a layout

At times, you may want to see the production field information for a document you are viewing in
Relativity. These steps will help you create a view and then add the Object list to a layout.

17.1 Recipe overview

The following instructions will provide a method for displaying your production related fields in a
document object layout.

17.2 Requirements
n Relativity 9.3 and higher
n Access to Views and Layouts, to include add and edit permissions

Relativity | Recipes - 74
17.3 Directions
To display your production related fields in a document object layout:
1. Go to the Views tab or your view drop down in the Folder Browser pane to create your new view.
2. Create the view on the production Information object.

3. Select the production fields you would like to display in the layout, such as Begin Bates, End Bates,
Begin Attachment, and End Attachment. In addition, you can select a condition, such as Begin Bates is
set and you can apply a sort order on the view, however a condition and sort order are not required.
4. Now that you have your view created, navigate to the layout in which you want to display the pro-
duction information object list.
5. Click the pencil icon ( ) to edit the layout, or go to the Layouts tab and select Build Layout on the
appropriate layout.
6. Select Add Object List from the Add Category drop down.

7. In the Object drop down, select the Production Information (Production) object. The View will default
to the new view you created on the Production Information object. If it does not, you can select it in

Relativity | Recipes - 75
 the View drop down.

8. Add the Object List for the Production Information Object.

9. Select Save and Close. The layout will update to reflect the Production field(s) you added to the Pro-
duction Information view you created.

Relativity | Recipes - 76
18 Email threading and near dupe - workflow
alternatives
Relativity workspaces often have a need for email threading and document near dupe detection.
System admins need to decide whether both functions should be addressed together or
independently.

18.1 Recipe overview

This recipe assists you in making an informed decision that is right for your workflow. This includes
suggested workflows for Email Threading and Near Dupe detection with examples of how to utilize
applicable Relativity functions.

18.2 Requirements
n Workspace & system admin rights

18.3 Directions
The following sections contain three workflow alternatives to help system admins who decide to run
a Structured Analytics Set (SAS) involving Email threading and/or Near Dupe detection on a data set.

18.3.1 Running the Email Threading & the Near Dupe in a single SAS
Use this workflow to identify email duplicate spares independent of the duplicate loose documents.
In other words, use this where the case demands that you handle the workflow for emails and loose
files separately. If both options are selected on a single set, only run near duplicate identification
against non-emails (attachments and loose files).
Once you run the SAS, you can create a document view or a batch of documents that contain the
documents flagged inclusive (Inclusive Email = Yes) and marked as non duplicate email (Email
Duplicate Space = No) for email documents.
You can then identify the loose documents as the documents flagged as Text Duplicate Principle
(Text Near Duplicate Principal = Yes).

18.3.2 Running only Email Threading in an SAS.

This workflow works well if you want to perform email threading and email duplicate detection on
an email only document set. For example:
n A case where emails are the only important documents. The case team wants to see only the most
inclusive, non-duplicate emails. Running a SAS with just email threading would be best.

Relativity | Recipes - 77
18.3.3 Running only Near Dupe in an SAS.
Use this workflow to identify duplicates whether they are emails or loose files. For example:
n Two parties in discovery produce documents. There is an agreement that each is to review documents
that were outside the set provided to the opposite side. In this case, the workflow requires you treat all
documents as loose documents and identify all near duplicate groups of single documents (just a Near
Duplicate Principal and no other documents in the group). You can then isolate all such documents via
a saved search, placed into a singular view or into review batches.
n A case team must identify duplicates across two or more groups of produced documents. Running the
Near Dupe process would be ideal. This allows the team to review only the necessary documents—the
Near Duplicates—while the other documents in the group are the duplicates.

19 Email Threading Results - Identify threads with

missing emails
You can identify which thread groups have items missing by looking at the Email Threading ID field.
The thread groups with missing items have a minus ( - ) sign in the threading ID.

19.1 Recipe overview

This recipe offers a workflow to use email threading results and identify threads with missing
emails.

19.2 Requirements
n Relativity 9.4 and above
n Structured Analytics

19.3 Directions
There are two ways to isolate the threads that have missing emails.
To use workflow suggestion number 1, complete the steps below:
1. Add the Email Threading ID field to the Email Threading view.
2. Filter for the minus sign to view the total threads with missing items.
3. Save those documents as a search.
4. Tag them as partial threads.
To use workflow suggestion number 2, create a widget to view the totals for partial versus complete
threads using the steps below:

Relativity | Recipes - 78
 1. Create a single choice field called Email Threads and allow pivot on and group by.
2. Create two choices on the new field:
1. Partial Thread
2. Complete Thread
3. Create a search for documents where the Email Thread group is set and return these fields:
1. Email Threads
2. Email Threading ID
4. Run the search.
5. Filter on Email Threading ID for the minus sign.
1. Tag documents returned as Partial Thread.
2. Then filter for Email Threads field is not set.

3. Tag the remaining documents as Complete Thread.

6. Create a widget.
1. Add Pivot (group by Email Threads, pivot and sort on grand totals)

Relativity | Recipes - 79
 2. Visualize the totals:

19.4 References
n Structured Analytics
n Structured Analytics
n Structured Analytics

20 Estimating diskspace utilization for databases and

Analytics indexes
Published July 19, 2013
When working with Relativity workspaces, system admins are often asked to estimate the diskspace
and the cost associated with a Relativity database and Analytics index. This recipe provides some
workflows and associated scripts that can help you calculate this for your workspaces.

20.1 Requirements
n Workspace with system admin-level access

20.2 Directions
The rule of thumb measurements and methods below help system admins estimate the size of their
Relativity database and associated Analytics indexes.

Relativity | Recipes - 80
It’s noteworthy that these metrics have a high standard deviation and tend to scale up or down
linearly. Every data set is different, so please note that this is only an average. Files will vary in size
considerably.

20.2.1 Workspace sizes

On average, every 100,000 records consume about 10 GB of database space. Database diskspace
usage is roughly one third of the total file size, including natives and images, for a workspace.
In addition to the above approximation, system admins can execute the Billing Statistics – Case Roll-
up script that is preinstalled in every Relativity environment. This script helps you calculate accurate
numbers on the size of your SQL database and the size of the files associated with the workspace.

Note: This script can only be run in Home mode.

20.2.2 Analytics index sizes

With default Analytics profile settings for an Analytics index, disk space usage is approximately 20
percent of the size of the corresponding SQL database’s MDF file.
To get an accurate measurement, system admins can use the Relativity Analytics Billable Estimate
script to determine the precise size of an Analytics index. This script runs on a workspace level
against a saved search of documents submitted to the Analytics index.

21 Exporting PDFs using document identifiers as file

names
Exporting documents in PDF format is a good way to create a portable subset of documents that you
can use when you can't access your Relativity environment, or when you need to share documents
with other parties. For ease of identification, you should create a cross-reference between the files
and the documents in Relativity, which you can do by naming the PDF files after the Document
Identifier field.

21.1 Recipe overview

This recipe describes how to export documents from Relativity in PDF format using the Relativity
Desktop Client and a saved search.

21.2 Requirements
n Workspace access with the following permissions:
o Search edit/add
o Corresponding search browser

Relativity | Recipes - 81
 n Relativity Desktop Client

21.3 Directions
1. Create a saved search with the documents to export.
2. Confirm all documents in the saved search have images created.
3. Use the Relativity Desktop Client to export the saved search.
a.  Select Tools | Export | Saved Search.
b.  Select the location for the export.
c.  Under the Image section, select the following settings:
n Data File Format - Select either Opticon or IPRO
n File Type - PDF
d. Under the Metadata section, select the following settings:
n Data File Format - Select the format that suits your needs.
n Data File Encoding - Select the encoding that suits your needs.
The export consists of two load files and PDF files named by the Document Identifier.

22 Exporting to HTML for expert witness review

This recipe describes how to share documents and their metadata with an expert witness without
giving the witness access to Relativity. Using the HTML export feature and a CSV file, you can review
documents and their metadata, as well as enter comments to later overlay into Relativity.

22.1 Recipe overview

This recipe walks you through the process of exporting documents to an HTML format that include
PDF, TIFF, or Native files.

22.2 Requirements
n Workspace access with the following permissions:
o Saved search
o Print to TIFF
o Export to File mass action

n Relativity Desktop Client (RDC)

22.3 Directions
1. Create a saved search containing the documents you would like to review.
2. Image the documents.

Relativity | Recipes - 82
 Note: To create PDF files—which may be easier to review—make sure you have TIFF images
available.

3. In the RDC, select to export a saved search.

a. Select the search you’ve created for this project, or click the ellipsis to use production images.

Note: Fields are automatically brought over from the saved search.

b. Populate your destination settings according to your preferences.

i. Change the Metadata: Data File Format to HTML (.html).
ii. For file type, you can select PDF, TIFF, and/or Natives.

22.3.1 Results
The result is an HTML document that includes all of your selected fields and hyperlinks to the image
files.

Relativity | Recipes - 83
22.3.2 Offline coding form
If your expert witnesses need to make any comments while reviewing these records, create a CSV
file export from the saved search.
1. Edit your search to show the control number.
2. Using mass operations, export this list of control numbers to a CSV file.
3. Track their comments in this file.
4. Overlay this information into a new document field.
n Exporting with the RDC
n Saved search

23 Filtering to find empty fields

Occasionally, you may need to locate documents that are missing field content. You can use
advanced filtering options to find text fields that don't contain data.

23.1 Recipe overview

This recipe describes how to use a custom filter type to identify empty fields.

23.2 Directions
1. Under the Administration tab, select Fields.
2. Click Edit for the desired text field.
3. Change the Filter Type to CustomOnly.

4. Click Show Filters from the view.

Relativity | Recipes - 84
 5. Select (Advanced…) from the field filter’s drop-down to activate the CustomOnly filter.
6. Select is not set from the value drop-down and leave the text query field blank. Select is not like from
the value drop-down and leave the adjacent field blank.

7. Click Apply to see the fields that do not contain data.

24 Finding duplicates or near duplicates in a new set

that match up to a previous set
When a new set of documents arrives, it can be important to know how the documents match up to
the current set that have been reviewed. This recipe can be useful for the following scenarios:
n Finding the overlap between one document set and another.
n Finding the documents that match up to the coded items in your initial set.
n Comparing items to determine the coding process.

Relativity | Recipes - 85
24.1 Requirements
n Structured Analytics
n Saved searches

24.2 Directions
1. Define the old and new datasets by marking a field or using the prefix of the ID.
2. Create a Structured Analytics set of Near Duplicate items across old and new data.
3. Create a saved search of the tagged items in the old document set and include the condition that Tex-
tual Near Duplicate Group is set. Add the include Near Duplicate Items from the Relational Items drop
down to see both the old information and new data that contains near duplicates of the original data.
4. Create a second search with the first condition as (Saved Search) = First Search. The second condition
is to return all the new documents.
5. This search returns documents that are in the new set which match up as duplicates or near duplicates
of the requested items in the first set. You can include the Near Duplicates to this search to see both
the original and new set items.

24.3 References
n Structured Analtyics
n Searching
n Structured Analtyics
n Searching
n Structured Analtyics
n Searching
n Structured Analtyics
n Searching
n Structured Analtyics
n Searching
n Structured Analtyics - RelativityOne
n Searching - RelativityOne

25 Finding emails exchanged within the same domain

During the course of developing your case strategy, you might need to identify the amount of email
traffic within the same company or between two companies. This can help identify some of the
main players in the case along with communication patterns. Using Relativity, you can identify email
communications at the company domain level and analyze those communications using sender and
recipient information.

Relativity | Recipes - 86
25.1 Recipe overview
This recipe describes how to use a transform set to parse email domain information from your data
and use Pivot to analyze the volume of communications between or among companies.

25.2 Requirements
n Relativity security access:
o Field tab with add/edit permission
o Object type tab with add/edit permission
o Transform set tab with add/edit permission
o Transform add/edit permission
o Use Pivot/Chart permission

25.3 Directions
1. From the Administration tab, click the Object Type sub-tab.
2. Click New Object Type and create a new object type with the following settings:
a. Name: Domain
b. Leave all other fields as default
3. Create a new field with the following settings to hold the domain information:
n Object type: Document
n Name: Recipient Domains
n Field type: Multiple Object
n Associative Object Type: Domain
n Allow Group By: Yes
n Allow Pivot: Yes
n Leave all other fields as default

Note: The object field type must be Multiple Object for the transform set to run.

4. Create a new field with the name Author Domains.

5. Create a saved search that serves as the data source for your transform set.
a. From the Documents tab, click the Saved Searches browser icon to create a new search with the
following settings:
n Name: Domains Saved Search
n Conditions: Set conditions so that the saved search only returns documents for which
your Email From (the field that stores the author email address) or Email To (the field that
stores recipient email addresses) fields are populated. This allows the transform set to run
as efficiently as possible.
Email From::is set OR
o

Email To::is set
o

n Fields: Select only the Email From and Email To fields.

6. From the Transform Sets tab, click New Transform Set and create a transform set with the following set-
tings:

Relativity | Recipes - 87
 n Name: Domains
n Handler: Domain Parsing
n Data Source: Domains Saved Search (the saved search you set up in the previous step).
7. Click New next to the transform and enter the following:
n Transform set: Domains
n Name: Author Domains
n Source field: Field that stores the author email address (e.g., Email From or Email Author)
n Destination field: Author Domains field created above (where Object Type = Document)
8. Click New next to the transform again and enter the following:
n Transform set: Domains
n Name: Recipient Domains
n Source field: Field that stores the Recipient email address (e.g., Email To or Recipients)
n Destination field: Recipients Domains field created above (where Object Type = Document)
You can repeat this process with the CC and BCC field domains in the Recipient Domains field. You just
have to create a transform to do so. For this example, we concentrate on just the two fields we already
set up.
9. Click Start Transform: Full on the Transform Set Console. This runs through the workspace and
extracts domains from the email address field. When this process is finished, the status changes to
Completed.
Now you can use Pivot to determine the overlap between the two fields containing the author and
recipient domains.
10. From the Documents tab, click the Pivot icon.
a. Turn on Grid and List.
b. Group by Author domains and Pivot on Recipient domains.

11. Click Go.

If you scroll to the author domain enron.com and then go across to the recipient domain enron.com,
you can see in our example that 630 items are from and to people within the Enron domain. You can
also review those 630 documents from the list view beneath the grid.

Relativity | Recipes - 88
If you scroll to the author domain enron.com and then go across to the recipient domain enron.com,
you can see in our example that 68 items are from and to people within the Enron domain. You can
also review those 68 documents from the list view beneath the grid.

Relativity | Recipes - 89
26 How to calculate Precision and Recall without a
control set
26.1 Recipe overview
Relativity will automatically calculate Precision and Recall for Assisted Review projects that have a
control set. However, the control set is an optional round type, and not recommended for all
projects, specifically (though not limited to) situations where documents continue to be added or
removed throughout the process.
There are situations, however, where you might wish to calculate these values without a control
set, or alternatively other instances where you feel that your control set is not a reliable model of
your document universe.
Although this calculation is not performed for you automatically under these conditions, the
process is very simple, requires just a few fields on a view, some filtering, and very basic math.

26.2 Requirements
n Relativity Assisted Review
n Analytics
n View / Edit permission

26.3 Precision and Recall Defined

Precision is a measurement of system accuracy. Ask yourself: Of everything the system thought was
Responsive, how much of it truly was Responsive according to the human reviewers?
The equation:

Recall is the measurement of completeness. Ask yourself: Of everything Responsive that the system
was supposed to find, how much did it actually find?
The equation:

The following table illustrates the meaning of terms like “True Positive”:

Relativity | Recipes - 90
Consider the illustration below:

n The entire rectangle represents all of the documents of a control set, or for the purposes of
this exercise, a QC round which is being used as a control set.
n The circle contains all documents which the system has categorized as Responsive.
n Blue documents, regardless of their location, are truly Responsive, as decided by the human
reviewer.
n The blue documents inside the circle are True Positives.
n The blue documents outside the circle are False Negatives.

n Orange documents, regardless of the location, are truly Non-Responsive, as decided by the
human reviewer.
n The orange documents inside the circle are False Positives.
n The orange documents outside the circle are True Negatives.

n Under different circumstances, some of the documents outside the circle might also be
uncategorized. This workflow focuses solely on categorized documents, however.

Relativity | Recipes - 91
26.3.1 Initial setup
The purpose of a control set is to serve as a miniature model of your document universe where the
documents in the set are not used as seeds which influence categorization. They can however, be
categorized by seed documents from other rounds. This lack of direct influence allows the system
admin to gauge the accuracy and success of the project through the control set categorization
results.
However, it is possible to use a QC Round as an ad hoc control set. In order to do so, the following
conditions are required:

n The QC Round must be a random sample. This is true of all QC rounds, and this recipe
workflow is no exception. Remember, judgmental sampling is typically not appropriate for QC
rounds.
n Only sample from [Documents to be categorized] - Categorized, where [Documents to be
categorized] is the saved search in your project settings. This saved search is comprised of all
your categorized documents for both categories (such as Responsive and Not Responsive).
n Coding for the round should be complete, but the round itself should not be finished, and
post-round categorization should not be triggered. Once you finish the current round, the
documents in it become examples, and are no longer viable for this ad-hoc control set.

26.3.2 View setup

n Control Number
n RAR Sample Set
n Designation (this name may vary but should be your human reviewer designation field)
n Categories – XXX RAR Designation Cat. Set (XXX represents the project prefix)

26.3.3 Calculating Precision

There are two ways to calculate Precision.
Method 1:
The first method is extremely simple because Precision is the mathematical complement of your
Responsive Overturn Rate. In other words: 100% - Responsive Overturn % = Precision %.
To calculate Precision using this method:

1. Navigate to the Overturn Summary Report and scroll down to the section called Overturns
by Designation Category.
2. Identify the Responsive overturned docs percentage for the current round.

Relativity | Recipes - 92
 3. Subtract this value from 100% to calculate your Precision.

So, for the above example, 100% - 26.54% (Responsive Overturn) = 73.46% (Precision).

Method 2:

This method involves filters on the view which was set up earlier.
To calculate Precision using this method:

1. Navigate to your Assisted Review document view; ensure before you start that all filters have
been cleared.
2. Filter on the RAR Sample Set field for the current round. Again, coding should be complete
but the round should not have been Finished/Categorized.
3. Ask yourself the question: Of everything the system thought was Responsive, how much of it
truly was Responsive according to the human reviewers?
4. Filter for Responsive on the Categories – XXX RAR Designation Cat. Set field. This is
everything the system thought was Responsive. Make a note of the number of documents
returned. This will be the denominator for your calculation.
5. Filter for Responsive on the Designation field. Make a note of the number of documents
returned. This will be the numerator for your calculation.
6. Divide the numerator by the denominator. This is your Precision.
Example, Precision denominator (True Positive + False Positive): 456 documents:

Relativity | Recipes - 93
Relativity | Recipes - 94
Example, Precision numerator (True Positive): 335 documents:

Relativity | Recipes - 95
Example, Precision calculation: 335 ÷ 456 = 73.46%.

26.3.4 Calculating Recall

This method involves filters on the view which was set up earlier. To calculate Recall using this
method:

1. Navigate to your Assisted Review document view; ensure before you start that all filters have
been cleared.
2. Filter on the RAR Sample Set field for the current round. Again, coding should be complete
but the round should not have been Finished/Categorized.
3. Ask yourself the question: Of everything Responsive that the system was supposed to find,
how much did it actually find?
4. Filter for Responsive on the Designation field. This is everything the reviewers determined to
be Responsive this round. Make a note of the number of documents returned. This will be the
denominator for your calculation.
5. Filter for Responsive on the Categories – XXX RAR Designation Cat. Set field. Make a note of
the number of documents returned. This will be the numerator for your calculation.
6. Divide the numerator by the denominator. This is your Recall.
Example, Recall denominator (True Positive + False Negative): 630 Documents:

Example, Recall numerator (True Positive): 335 documents:

Relativity | Recipes - 96
Example, Recall calculation: 335 ÷ 630 = 53.17%.

26.3.5 Additional considerations

n Ensure that filters are applied in the proper order as failing to do so will give incorrect results
n Note that numerator is the same (True Positive) for both equations.
n It is also important to understand that these calculations apply only to your categorized
population, and do not include your uncategorized documents. These documents are often
removed from projects and/or reviewed separately, essentially negating the need to
determine their Precision and Recall values. Please note, however, this differs from the
automatic control set calculation, which does account for uncategorized documents.
n It is also possible to calculate Precision from a Responsive-only QC Round, as all of the
required information will still be available. Please note that a Recall calculation will not be
possible under these circumstances, however.

References
n Assisted Review
n Analytics
n Views
n Filters
n Assisted Review
n Analytics
n Views
n Filters
n Assisted Review

Relativity | Recipes - 97
 n Analytics
n Views
n Filters
n Assisted Review
n Analytics
n Views
n Filters
n Views
n Filters
n Assisted Review
n Analytics
n Views
n Filters
n Assisted Review
n Analytics
n Views
n Filters
n Assisted Review
n Analytics
n Views - RelativityOne
n Filters - RelativityOne
n Assisted Review - RelativityOne
n Analytics - RelativityOne

27 Identifying emails between two or more specific

individuals
This recipe describes how you can use dtSearch and search terms reports to identify emails in which
two or more individuals were involved in the correspondence. This includes emails between the
specific individuals, as well as instances in which the individuals are only in the recipient fields.

27.1 Requirements
n Relativity access

27.2 Directions
This recipe involves a scenario where the case team has a list of specific individuals and only wants
to review emails in which two or more of these individuals were involved in the email
correspondence.

Relativity | Recipes - 98
 1. Create a custom index containing only the correspondence fields (To, From, CC, BCC):
a. Create a saved search that returns only the correspondence fields (To, From, CC and BCC).

b. Create a custom dtSearch index using this new saved search as your searchable set.

2. Create a new Search Terms Report (STR).

a. Select the custom dtSearch Index that contains only the correspondence fields.
b. Select Report and tag for Type.
c. Ensure that Calculate Unique Hits is set to Yes.

Relativity | Recipes - 99
 d. Click Save.

e. Click Modify Terms, and then paste in the email addresses for the individuals you are trying to
identify.

Relativity | Recipes - 100

 f. Click Add Terms, and then click Done.
g. Click Run All Terms.
3. Create a saved search based on the two STR fields that were created (setting Calculate Unique Hits to
Yes creates a second STR field) when the STR completes. In the following example, the conditions are as
follows:
a. STR – Email Addresses_v1 – Unique “not these conditions” STR – Email Addresses_v1 Unique – is
set.
AND
b. STR – Email Addresses_v1 “these conditions” STR – Email Addresses_v1 – is set.
c. Ensure the STR fields are selected as fields returned in your saved search. You may also want to

Relativity | Recipes - 101

 include the To, From, CC, and BCC fields for reference.

You can now set up a workflow for reviewing these emails in which two or more individuals were
identified.

Relativity | Recipes - 102

28 Identifying standalone emails
This recipe describes how you can use saved searches after email threading is complete to identify
emails that have not been replied to or forwarded. You can then batch these documents together in
order to expedite the review process.

28.1 Requirements
n Existing email threading set

Relativity | Recipes - 103

28.2 Directions
This recipe involves a scenario in which the case team has performed email threading and would
now like to identify any standalone emails.
1. Ensure email threading is complete.
2. To find all emails that have not been replied to or forwarded, you need to return all emails that have
Indentation equal to 1, but where there is no Indentation greater than 1 within the thread group. This
requires you creating three saved searches:
a. Saved search 1 contains documents in which Indentation is equal to 1 and thread group is
included.

Relativity | Recipes - 104

 b. Saved search 2 contains documents in which Indentation is greater than 1 and thread group is
included.

Relativity | Recipes - 105

 c. Saved search 3 gives you the final results and is a nested search of saved searches 1 and 2. The
conditions for this search should be as follows:
n Document is in saved search “Indentation is 1 + Thread Group”
n Document is not in saved search “Indentation > 1 + Thread Group”
n Sort the results by Group Identifier to keep families together

Relativity | Recipes - 106

 The result of saved search 3 are the standalone emails which have not been replied to or
forwarded.

Relativity | Recipes - 107

29 Importing productions in Relativity with natives
When working with productions in Relativity, system admins may need to make documents available
as images and natives to expert witnesses. To do this, system admins will import the production
back into Relativity as images (along with the natives) when natives are available. This creates
duplicate files, which use more storage space.

29.1 Overview
This recipe shows you how to set up your workspaces to support the import of productions with
natives when natives are available. This approach conserves disk space by storing only one copy of
the images and the natives.

29.2 Requirements
n Workspace access
o Documents: Add, Edit, Delete
o Images: Add, Delete
n Relativity Desktop Client

29.3 Directions
1. When setting up and importing into the workspace, be sure that the files are available in a shared repos-
itory accessible from your Relativity environment, and import the data using the links to the repos-
itory.

Relativity | Recipes - 108

 2. Create a saved search for the production set.
3. Export the saved search. Make sure to:
n Include all fields necessary for your project.
n Clear the Copy Files from Repository box when exporting from the RDC.

4. Export your production set. Make sure to:

n Include all fields necessary for your project.
n Clear the Copy Files from Repository checkbox.
5. Create a new folder in your workspace. Secure the folder so only your expert witness group can access
it.
6. Import the production set you used in step 4 into the new Productions folder. Be sure to select the fol-
lowing File Repository Preferences setting:

Relativity | Recipes - 109

 7. Overlay the natives using the load file and the export from the saved search in step 3. Be sure to select
the following File Repository Preferences setting:

Note: You must perform the imports (Steps 6 and 7) on a machine that has access to the repository to
avoid import errors. The import of the natives (step 7) executes successfully only when natives have been
imported into the workspace (step 1) using the files that already reside in a valid location that Relativity can
access.

Importing through the Relativity Desktop Client - RelativityOne

30 Legal Hold checklist

When working with Relativity Legal hold, it's helpful to have a checklist of items you can address to
expedite the setup time in Relativity. This information helps you establish an efficient, repeatable
process.

30.1 Requirements
n Workspace access
n Relativity Legal Hold

Relativity | Recipes - 110

30.2 Directions
As you set up holds using Relativity Legal Hold, consider the following:
1. Determine whether you'll use a single workspace for all holds or separate workspaces for holds.
n Relativity Legal Hold supports having multiple legal holds in a single workspace. However, it's
beneficial as a system admin, to set up workspaces based on organizational groups (associated
with the hold), or similarity of workflow. This helps simplify security on holds.
n For example, you could have a workspace that contains all holds concerning HR-related matters,
and set up additional workspaces based on hold types. Alternatively, you could set up holds in a
workspace per organization or department.
2. Determine hold owners and other participants who need access to data.
n Create a Hold Owners group that has permissions to set up and configure a hold.
n Create a group for other participants who expect to receive (or have access to) hold reports.
3. Identify your custodians and the potential need for alert groups.
n For example, you can set up groups for IT, HR, Executives, Marketing, etc.
n A sample use case would be:
o Setting up an alert for the IT group when someone responds "Yes" to the question "Do
you use your home desktop for work purposes?" in a questionnaire.
o You can set up custom reminder notices for the Executives group.

4. Define your hold workflow with:

a. Notification & reminder templates (including criteria for defensible hold notifications):
i. Define the overview of the matter.
ii. Indicate explicitly (with examples) the types of information you want to retain.
iii. Specify data sources.
iv. Collect additional information on other possible sources via questionnaires.
v. Specify date ranges.
vi. Clarify the legal obligations to preserve information.
b. Communication & reminder frequencies:
i. Does one size fit all, or do I set them up depending on the group (for example, Exec-
utives get a reminder less frequently)?
ii. Do I need to set up a Global reminder (for example, a blanket email every 3
months)?
iii. Does one questionnaire fit the need for all groups, or do I need to craft separate
questionnaires for each group of recipients?

Note: Relativity maintains an audit log of all the actions associated with the legal hold; however, we
strongly recommend that you document all your activities leading up to the legal hold.

In addition, keep the following in mind as you establish a repeatable process to execute a legal hold:
n Factor in employee mobility and turnover.
n Consider checking for non-traditional data sources (for example, data in Cloud Services, mobile
devices).
n Define a process for immediate collection (to account for potential equipment failures).

Relativity | Recipes - 111

 n Record all activities and materials that leverage past preservation efforts (to be able to learn from those
efforts and replicate or improve).
Overall, remember that legal holds are an iterative process. To be in compliance, keep the
following best practices in mind:
n Issue timely written legal hold directives.
n Ensure custodians are clear about the whats and hows of compliance.
n Schedule timely follow-ups for confirmation and compliance.
n Provide periodic updates and reminders.
n Document every action and the basis for the decisions executed on a legal hold (for example, releasing
a custodian from a hold).

30.3 References
Relativity Legal Hold

31 Leveraging processing metadata for quality control

At some point, unprocessable files will likely appear in your processing projects. To correctly
address these files in your workspace, you will need to review the files and determine whether they
require further attention.

31.1 Overview
Files that are designated unprocessable don't actually have an error associated with them. This is
because these files weren't able to be processed in the first place, so they weren't able to register a
processing error. This recipe walks you through the process of identifying and reviewing
unprocessable files from processing.

31.1.1 Requirements
n Relativity processing
n Processing repository access

31.2 Directions
To identify and review unprocessable files, perform the following steps:
1. Create a view in your workspace called Unprocessable Documents on the document object with the fol-
lowing properties:
1. View Information:
1. Name - Unprocessable Documents
2. Order - 9999
3. Visible in Drop-down - Yes

Relativity | Recipes - 112

 2. Advanced Settings:
1. Object Type - Document
2. Visualization Type - Standard List

1. Information:
1. Object Type - Document
2. Name - Unprocessable Documents
3. Owner - Public
4. Order - 9999

2. In the view, include the processed metadata fields that provide additional descriptive information
about the files, such as the Document Extension, File Type, File Name (which includes the original doc-
ument extension), and File Size (especially useful for empty files) fields.

Relativity | Recipes - 113

Relativity | Recipes - 114
Relativity | Recipes - 115
 3. Next, add a condition to the view. The Unprocessable field uses a Yes/No field type. The yes/no value
indicates whether a file was able to be processed. If the file couldn't be processed, this field is set to Yes.

Note: The Unprocessable field is set to Yes on any file for which Relativity doesn’t have an Invariant plugin that
is capable of extracting text or imaging/OCR’ing that document type. For example, it’s not set for a corrupt file
for which Relativity can’t extract text, such as a corrupt Word document that logs an error during data
extraction. Unprocessable documents don't have errors associated with them because they never reach a
point at which they can register a processing error.

4. Finally, set your Sort to sort by Document Extension and File Type. This allows you to complete your
review while addressing similar file types together.

Relativity | Recipes - 116

 5. Once your view is created, you can create a dashboard. Set the Group By value to Document Extension
and the Pivot On value to File Type to examine the different file types associated with the document
extensions.

You can also reference your list view to identify records individually using the view.

Relativity | Recipes - 117

Once you've segregated the unprocessable files, you can begin your review.

31.2.1 Important considerations

n While these files are single records, you will want to watch for unsupported containers that could poten-
tially contain additional compressed files. Proprietary archives, like self-decrypting archives in .exe
format are becoming more prevalent.
n In the example shown here, there are also files created using a Macintosh machine. If data was col-
lected and processed from a non-Windows device, special attention should be made to files without
text extraction.
n Special considerations may be considered for file types not supported by the viewer. Installing the
appropriate software may be required on the reviewer workstations.
n If your processing profile did not include the Unprocessable field, you can reference the two reports
available on the Relativity Processing | Reports tab. These reports provide information about the file
types, by extension, with a section listing the unprocessable files.

Relativity | Recipes - 118

 o Discovered Files by Custodian - This report provides information on the file types discovered dur-
ing processing for the custodians associated with the selected processing sets. This report iden-
tifies the total processable and unprocessable file types discovered and categorizes the totals by
custodian. You can run this report on discovered or published processing sets.
o Discovered Files by File Type - This report provides information on the file types discovered dur-
ing processing for the custodians associated with the selected processing sets. This report iden-
tifies the total processable and unprocessable file types discovered and categorizes the totals by
file type. You can run this report on discovered or published processing sets. See Supported file
types for processing for a list of file types and extensions supported by Relativity for processing.
See Supported file types for processing for a list of file types and extensions supported by Relativ-
ity for processing. See Supported file types for processing for a list of file types and extensions
supported by Relativity for processing.See Supported file types for processing for a list of file
types and extensions supported by Relativity for processing.
Using the extensions in the report, you can create saved searches to identify the unprocessable files
using the document extensions:

32 Loading external productions into Relativity

You can import external productions into Relativity through the Relativity Desktop Client. This recipe
illustrates the basic workflow to set up a new production set and then load the load files (usually one
for metadata and one for the images) into the production set.
This is often useful when the case team asks that, after a production goes out the door, they can
access the documents just as they were in the production and that they be available for searching.

Relativity | Recipes - 119

This recipe also helps when you receive a production from the other side to help organize the
received data in the workspace.

32.1 Requirements
n Relativity Desktop Client (RDC)

32.2 Instructions
1. Build a new production set. This is an empty 'staging' production to load the received production into. 
2. Open the RDC.
3. Click Tools > Import > Production Load File… to:
a. Load the image load file (opt file) into the ‘staging’ production.
b. Use Append Only to ensure only new documents are created.
4. Click Tools > Import > Import Document Load File… to:
a. Load the .dat metadata file.
b. Map the appropriate metadata files and use the Overlay Only option on the Overwrite behavior
to the production documents that you want to load.

32.3 Notes
n Loading documents into a 'staging' production marks the production set as produced.
n When loading a production set into Relativity, the systems loads the newly created documents into the
root folder. It may be prudent to move these documents to a new folder to ensure proper foldering
and security of the documents.
n Using Append/Overlay during the metadata import creates new documents that did not have images,
but these documents aren't associated with the production. To associate these files with a production
set, create and load an image load file pointing to placeholder images and then overlay the metadata.
Using the Overlay Only option shows an error on any document not already loaded, and you can use
the error file as a list to identify documents that may not have had images.

33 Maintaining previous Bates numbers between

production sets
In some cases, you may need to produce documents with a second Bates number—while
maintaining their original Bates numbering—on the production image.

33.1 Recipe overview

This recipe shows you how to build production images that maintain their original Bates numbering.

Relativity | Recipes - 120

33.2 Requirements
n Workspace access
o Saved searches
o Production sets
n Relativity Desktop Client

33.3 Directions
There are two options for creating a new set of production images that retain the Bates numbering
from a previous production. Both options generate the same results.

33.3.1 Option 1
Load the documents to re-produced back into Relativity as new documents.
1. Create a saved search of the documents to produce again.
2. Export the documents’ images from the saved search.
Ensure you select the production version of the images in the Production Precedence section of the
Relativity Desktop Client.

3. Import the documents back into your original workspace as new documents. Use the Bates number as
the control number.

Relativity | Recipes - 121

 4. Create a production set with the documents you just loaded back into Relativity . The images contain
the Bates numbers from the original production, and from the new production.

33.3.2 Option 2
Produce the documents in a “staging” workspace.
1. Export the documents to re-produce from the original workspace.

Note: Ensure you select the correct production version in the Production Precedence section of the Relativity
Desktop Client.

2. Create a new “staging” workspace. Load the first version of the produced images into the staging work-
space.
3. Create a production set in the staging workspace, and then produce the documents. The images con-
tain the Bates numbers from the original production, and from the new production.
4. Export the re-produced documents from the staging workspace.

Note: Ensure you select the new production version in the Production Precedence section of the Relativity
Desktop Client.

5. Create a production set in the original workspace.

6. Overlay the re-produced documents as a production in the original workspace. This gives you a pro-
duction set containing the original documents in the original workspace—which contains images with
both sets of Bates numbers.

34 Mapping Processing Fields

This application loads the fields necessary for processing. These fields also store
metadata in your workspace. The application runs a script that maps identically named
Relativity processing fields to the newly loaded fields.

34.1 Requirements
You need to install the Relativity Processing application in the workspace
before you install the Mapping Workflow application:
n Fields
n Scripts

34.2 Directions
1. On the Relativity Applications tab, click New Relativity Application.
2. Select Import from File and click Browse.

Relativity | Recipes - 122

 3. Select the Mapping Workflow application and click Open.

4. Click Import.

Note: Relativity displays an error message after installation. This error is expected. Select Unlock
Processing Application and click Retry Import.

When the installation completes, Relativity displays a message indicating that import was
successful.
5. Navigate to the Script Library.
6. Execute the Map Processing Fields script provided with the application. After the script com-
pletes, you can view the processing fields mapped to their respective Relativity fields.

Relativity | Recipes - 123

35 Mass importing into a file field
A file type field allows users to link an outside document to a record. This field is only available in
custom objects. File fields offer the flexibility to store documents outside the scope of discovery in

Relativity | Recipes - 124

your workspace, such as CVs for your expert witnesses.

35.1 Overview
This recipe shows you how to upload multiple documents into a file field stored on a custom object.

35.2 Requirements
n Dynamic object with a file field
n Workspace access
o Relativity Desktop Client

35.3 Directions
After you create your custom object and add a new file field, you can use the RDC to import into this
object. This example uses an Expert Witness object.

1. Open the Relativity Desktop Client and your workspace.

2. Click the drop-down for the Documents object and select the object that stores the file field you’re
using.

Relativity | Recipes - 125

 3. Click Tools > Import, and then choose Expert Witness Load File.
4. Browse to your load file with the file field path.
5. Change Encoding and Characters to match the load file, and then click to the Field Maps tab.

Note: When you look at the list of available fields in your workspace, your file field does not appear. However,
the Native File Behavior is enabled.

6. Ensure the Load Native Files checkbox is selected.

7. Fill in the Native File Path Contained in Column field with the appropriate field from your load file.

8. Click Import > Import File.

9. Navigate back to your workspace. In the file field, you should see your records with a hyperlink to the
file.

Relativity | Recipes - 126

36 Migrating cases from Summation
At times, you may need to migrate case data from other software into Relativity. While a button for
transferring data from one database to another doesn't exist, there are some ways you can help
mitigate the process.

36.1 Recipe overview

This recipe highlights some best practices for migrating case information from Summation to
Relativity.

36.2 Requirements
n Workspace access
n Relativity Desktop Client

36.3 Directions
36.3.1 Migrating images
n Clients sometimes find it easier to use the image volumes and load files on their server rather than
exporting images from Summation. You can account for all the images by comparing the total images in
Summation with the total number of images contained in the volumes on the server.
n Use a conversion tool like iConvert to convert .dii files to .opt files.

Note: Use a text editor like TextPad to combine multiple .dii files into one. Doing so saves time during the
conversion and import processes.

36.3.2 Migrating data

n Often, not all data stored in Summation is contained in the original volume load files on the server. You
should perform an export from Summation that contains all metadata and work product fields.
n Fields that use lookup tables in Summation—such as Names, Issues, and Doctype—should be created
in Relativity as either single- or multiple-choice fields to match the previous fields.
n Summation folder information is stored in the Folder field. You can use that information in two ways:
o If you load the Folder field into a multiple-choice field, you see the folder structure in the tag
tree. Ensure that, when loading the data, you delimited to maintain the nested value to the “\”
character.

Relativity | Recipes - 127

 o If you want to reproduce the Summation folder structure in your Relativity folders, you need to
export the Folder field and import it into Relativity. During import, set the Folder Information
Column accordingly.

36.3.3 Migrating transcripts

n It may be easier to work with the original files received from the court reporting agency and import
them directly into Relativity . Once they’re in your workspace, use the Process Transcript mass action
to build the word index.
n Create a separate folder for your transcripts. You can build out a folder tree with each deponent’s
name as a sub-folder, or place them all in the root of the new transcript folder.

36.3.4 Migrating productions

You can export production images from Summation and imported into Relativity as a production set.
You need to create a production set in Relativity and import documents into the set through the
Relativity Desktop Client.

Relativity | Recipes - 128

36.3.5 Migrating redactions
To migrate redactions from Summation to Relativity, you need to burn the redactions into the
images and then import those images into Relativity .

37 Multi-matter workspace setup in Relativity

When working with a single workspace supporting multiple review matters, there are several ways
system admins can expedite setup and improve performance.

37.1 Recipe overview

This recipe highlights the items that system admins can reference while managing workspaces that
support multi-matter ESI for processing, review, and production.

Relativity | Recipes - 129

37.2 Requirements
n System admin access
o Add new users and groups permission

n Workspace access
o Edit, Delete permission for all items under the Admin tab

37.3 Directions
In this illustration, we’ll reference a single workspace that contains three different matters: A, B,
and C. Each matter is actively being reviewed by a combination of reviewers, some common to all
three matters and some who are unique to certain matters.
System admins should use the following best practices as they configure this type of workspace.

37.3.1 User Groups/Security

Set up groups with reviewers:
n Common to the three matters, there are four potential combinations:
o A,B,C
o A,B
o B,C
o A,C

n Unique to matters, there are three groups:

o A
o B
o C
The following table represents the different user groups and how they are distributed:

As illustrated, you'll have a minimum of seven unique user groups for the three matters in this
workspace. The minimum number of unique user groups a workspace requires (ignoring tiered
permissions within a user group, which will be described in the following example) is determined by the
following formula:
(2^m) – 1, where m is the number of matters being set up in the workspace
System admins must analyze which users belong in which groups and consider the permission levels
required for each matter and each user. For example, if we add a user to UG-4 and UG-3, and the
permission levels for the matters are identical, that particular user may be for UG-7.
However, if the permission levels at UG-5 and UG-3 are different—perhaps the user has permissions to
code in Matter A and Matter B, but not the documents in Matter C—then tiered user group access is

Relativity | Recipes - 130

 required. In that case, UG-4 and UG-3A — which, unlike UG-3B, do not include permissions to code
documents — may be more suitable.
A best practice for creating folders is to continue to set up the documents in a workspace by custodian.
There are two advantages to this:
o For end users, the workspace will look similar to the typical single matter workspaces, and
navigation will be familiar.
o If some custodians are not part of a certain matter, those custodian folders can be locked at the
folder level for the appropriate user groups, thus restricting document access to those users.

37.3.2 Folder Setup

37.3.3 Field and choice setup

For quick identification of documents in a multi-matter workspace, system admins can use a
required multi-choice field to tag documents appropriately. For example, it could be titled Matter ID
and include the matter names as choices. Use this field to identify documents not tagged for any of
the matters, as well as for creating batches.
Add fields and choices to the document object. However, if you anticipate adding more fields —
such as for review or production— outside of the standard Designation and Issue fields, there are
two ways to approach this.
n Continue to add the system admin fields as a part of the document object.
n We recommend adding the system admin fields to a separate object, with one object set for each
matter. Refer to the Building Custom Objects Workbook.
Adding additional fields will affect performance as the database grows, but you can make the
database more scalable for new fields as your document count or the needs of each of the matter
grow by adding them on a separate object.

Relativity | Recipes - 131

With the second option, the new fields created on a separate object will need to be set up so that
they will be available in the views on the Document object. This can be accomplished by opening the
field to association, as highlighted by the red box below.

37.3.4 Views Setup

If fields are set up on a separate object, those fields and their associated choices are available in a
view under the Documents object, as described in the Field & Choice Setup section of the Building
Custom Objects Workbook. Once the fields are available in a view, you can use the view to filter for
documents. Below is a view with fields from the Document object (boxed in blue) and the Matter-1
object (boxed in green).

37.3.5 Layout Setup

Fields set up on a separate object can be added to the document layout with an Associative Object
List, highlighted in the red box below. Once the object has been included, it will appear in the coding
pane — shown in the blue box — along with views that have associative object fields and their
choices.

Relativity | Recipes - 132

Once the object has been included in the coding pane, the field will appear in the layout as shown
below. Note that the New, Link, and Unlink buttons can be turned off depending on the permission
level for the user group to the object.

You can link multiple associative objects to the layout being built if you set up separate objects for
each matter.

37.3.6 Search indexes

Depending on the needs of your workspace, you can create separate dtSearch indexes for
documents in each of the matters.
This may not be recommended in all multi-matter workspace setups, but in such cases it is
recommended to have at least two tiered indexes. Refer to the following table for more details.
Search Index Tiers Use Drawback
Across all documents (and all n For linear review and n Slower Index Builds and
matters) in the workspace batching slower query response
n Supports review QC for times
responsiveness to matters.
Across a subset of documents n Supports quicker index
responsive to each matter builds and faster querying

Relativity | Recipes - 133

Search Index Tiers Use Drawback
n Securable to user groups
Keyword search automatically indexes all documents within the workspace, so if your search access
is matter-based, access to keyword search needs to be disabled for appropriate user groups.

37.3.7 Batches
The first step is to identify the documents included in the different matters within the workspace.
Batching can be done as part of the linear review workflow, or a Relativity Analytics-based
categorization workflow. In a categorization workflow, begin by sampling documents the same way
you would perform issue coding using categorization.
As described in the Fields Setup section of the Building Custom Objects workbook, batching can
leverage the Matter ID field for further review or to tag the documents during initial review.
All other system admin setup functions — such as production sets, markup sets, fields, persistent
highlight sets, and choices on the document object — are managed the same way they would be in a
single matter workspace. However, we recommend the following best practices:
n Use intuitive naming conventions. For example:
o SEC productions on Matter A could be MA – Production SEC 20110229
o EPA markup sets on Matter B could be MB – EPA Priv Set
n System admins need to be particularly aware of permissions provided to each user group. We
recommend setting up tiered access for each user group combination based on who needs access to
the views, layouts, markup sets, production sets, etc., as described above.
n Since batching is based on saved searches, further review, batching, and dtSearch indexes can be based
on search results responsive to the matters in question.

38 Native Imaging - Dithering Algorithm Options

Relativity native imaging offers multiple dithering algorithms you can apply towards imaging your
documents.

38.1 Recipe overview

This recipe presents the various dithering algorithms available within Relativity native imaging. This
recipe provides you with additional information and use cases for each algorithm, including the size
and clarity of images you create.

38.2 Requirements
n Workspace and system admin access

Relativity | Recipes - 134

38.3 Directions
System admins can select from one of eight different dithering algorithms when they choose to
image with native imaging:
n Clustered 6x6
n Clustered 8x8
n Clustered 16x16
n Dispersed 4x4
n Dispersed 8x8
n Dispersed 16x16
n Floyd & Steinberg
n Threshold

38.3.1 Dithering results for text-rich documents

Below are the testing results with the different algorithms for text-rich documents. All results are
300DPI B&W only.

Relativity | Recipes - 135

Image Dithering Type Size Image Sample Magnified (150% scale)
Clustered 6x6 27 KB

Clustered 8x8 25 KB

Clustered 16x16 23 KB

Floyd & Steinberg 24 KB

Dispersed 4x4 24 KB

Relativity | Recipes - 136

Image Dithering Type Size Image Sample Magnified (150% scale)
Dispersed 8x8 24 KB

Dispersed 16x16 24 KB

Threshold 0 2 KB

Threshold 127 22 KB

Threshold 215 22 KB

Relativity | Recipes - 137

Image Dithering Type Size Image Sample Magnified (150% scale)
Threshold 255 22 KB

38.3.2 Dithering results for image-rich documents

Below are the testing results with the different algorithms for image-rich documents. All results
300DPI B&W only.
Image Dithering Type Size Image Sample Magnified (150% scale)
Clustered 6x6 45 KB

Relativity | Recipes - 138

Image Dithering Type Size Image Sample Magnified (150% scale)
Clustered 8x8 39 KB

Relativity | Recipes - 139

Image Dithering Type Size Image Sample Magnified (150% scale)
Clustered 16x16 33 KB

Relativity | Recipes - 140

Image Dithering Type Size Image Sample Magnified (150% scale)
Floyd & Steinberg 70 KB

Relativity | Recipes - 141

Image Dithering Type Size Image Sample Magnified (150% scale)
Dispersed 4x4 64 KB

Relativity | Recipes - 142

Image Dithering Type Size Image Sample Magnified (150% scale)
Dispersed 8x8 52 KB

Relativity | Recipes - 143

Image Dithering Type Size Image Sample Magnified (150% scale)
Dispersed 16x16 69 KB

Relativity | Recipes - 144

Image Dithering Type Size Image Sample Magnified (150% scale)
Threshold 0 2 KB

Relativity | Recipes - 145

Image Dithering Type Size Image Sample Magnified (150% scale)
Threshold 127 79 KB

Relativity | Recipes - 146

Image Dithering Type Size Image Sample Magnified (150% scale)
Threshold 215 27 KB

Relativity | Recipes - 147

Image Dithering Type Size Image Sample Magnified (150% scale)
Threshold 255 20 KB

39 OCR Redacted Production Documents & Export Text

Redactions are a ubiquitous aspect of document review. The challenge is to prepare redacted
documents for production in a timely and efficient manner. This recipe assists in completing OCR on
redacted documents, and exporting text, while minimizing unnecessary work.

39.1 Requirements
n Workspace access
o OCR Profile, OCR Set, Production Sets
o Create access on Saved Searches

Relativity | Recipes - 148

 o Create Field access
n Relativity Desktop Client

39.2 Directions
First, OCR a production. You should OCR your production after you run the production, but before
you perform the final export. Perform the following steps to OCR documents with redactions:
1. Create an OCR profile with your preferred settings. You must have an OCR profile before you can save
an OCR Set.
2. Create a long text field to hold your OCR results. You can't OCR directly into the Extracted Text field.
3. Create an OCR Set.
4. Point the OCR Set to completed Production. If you only want to OCR redacted documents for the pro-
duction, select Yes next to Only OCR Production Documents Containing Redactions.
5. Click Save on the OCR Set. The OCR Set Console appears.
6. Click OCR Documents from the OCR Set Console to kick off the OCR job.
When the job completes, the text becomes ready to export.
Now you can export a production using text precedence. To only produce non-redacted text when
exporting a production, you must use the Text Precedence feature. When you export a production
through the RDC, don't add the Extracted Text field to the exported field list. This results in
producing redacted text. In order to preserve the integrity of your redactions, don't add the
Extracted Text field to the selected columns tab.

Relativity | Recipes - 149

When you switch to the Destination Files tab, you can set Text Precedence in the Metadata section.
To set Text Precedence:
1. Set the File Format, Encoding, and Export Files options to meet your needs.
2. Click the ellipsis next to Text Precedence. The Pick Text Precedence window appears.
n Based on the screenshot below, the RDC first checks to see if there is OCR'd text available in the
OCR Destination Field. If this field isn't set, the RDC exports the Extracted Text.
3. Export your production set documents.

40 Overlaying in the production object in Relativity 9.3

Prior to Relativity 9.3, production fields were stored on the Document object. Now in 9.3 and higher,
they're populated and stored on an object called Production Information. These fields include:
n Begin Bates
n End Bates

Relativity | Recipes - 150

 n Begin Attachment
n End Attachment
n Has Redactions
n Data Source
n Production
n Production Name
n Production Type
n Sort Order
After you run a production in 9.3 or higher, Relativity automatically populates the fields listed
previously on the Production Information object. If any modifications need to be made to the fields
on this object, you can run an overlay using the Relativity Desktop Client.
The Production field is a unique value that contains the Artifact ID value of the Production Set
combined with the Artifact ID value of the document. This value ensures that this field is truly a
unique document. Here is an example of the format of the production field:
[Artifact ID of Production Set}_{Artifact ID of Document} or XXXXXXX_XXXXXXX

Let’s envision a scenario in which you ran a production and did not specify a relational field. You
currently do not have any Attachment values in the Begin and End Attachment fields. This would
require you to overlay new Begin Attachment and End Attachment values into the Production
Information object. Here is an example of what this load file should look like:

40.1 Requirements
n Relativity 9.3

Relativity | Recipes - 151

40.2 Directions
1. Select the Production Information object on the object selection.

2. Select Overlay on the Overwrite behavior selection.

3. To map the fields, the Name [Identifier] field must be mapped to the Production field. Map the
remainder of the fields to overlay the values into the corresponding fields on the Production Inform-
ation object. In this scenario, we select the Begin and End Attachments fields. Select the Name [Iden-
tifier] field as the Overlay Identifier.

Relativity | Recipes - 152

 Running this overlay using the Production field as the Name [Identifier] ensures that the New
Begin/End Attachment values are loaded properly to the Workspace.
The overlay creates the following updates to the Begin Attachment and End Attachment fields.

Relativity | Recipes - 153

40.3 References

41 Performing overturn analysis for Relativity Assisted

Review
During a computer-assisted review, a case team moves through several rounds of coding to train
the system on the document collection and validate the computer’s results. A meta round isn’t a
formal round, but a between-rounds workflow used to make any necessary adjustments to the
project to prepare for the next round. A meta round usually consists of identifying, analyzing, and
correcting (re-coding) documents which have a significant and adverse effect on project results.
We recommend that you make these adjustments prior to finishing a round and categorizing
documents. This allows the system to make use of the corrections performed during the meta round
and apply them to the next true round.

41.1 Recipe overview

This recipe teaches Relativity project managers to identify high-ranking overturns and highly
influential seed documents that are responsible for a large number of overturns.

Relativity | Recipes - 154

41.2 Requirements
n Admin rights or access to the console
n Analytics index (active with queries enabled)

41.3 Directions
41.3.1 Correcting coding inconsistencies between true or conceptual duplic-
ates
n Each seed-overturn pair has a rank (or score) which indicates the degree of conceptual similarity they
share. The maximum possible score is 100, which means the two documents are conceptual duplicates.
Conceptual duplicates are documents which may or may not have identical text, but do contain the
same conceptual content according to the Analytics index. While it is possible that conceptual duplic-
ates may also be exact textual duplicates (i.e., documents with the same MD5 hash value), this should
not be assumed from a score of 100.
n We recommend that you use the Overturn Documents report to locate these documents by filtering
on the round and sorting by descending rank. A good best practice is to re-evaluate each seed-overturn
pair having a rank of 95 and higher to see which document was coded correctly, as well as whether each
is a suitable example.

41.3.2 Identifying and correcting the most influential seed documents

n When viewing overturn reports at the end of a round, the same few documents can be responsible for
many overturns. If those seed documents were incorrectly coded, they can greatly inflate the overturn
rate for the entire project. Finding and correcting these situations is an essential component of meta
round protocol.
n The quickest way to find the most influential documents is by using Pivot on the Overturned Docu-
ments report. Simply choose Seed document in the Group by drop-down and leave the <Total Only>
drop-down as is.

41.3.3 Using the Overturn Analysis related items pane

n Once a document has been targeted for re-evaluation during a meta round, you can navigate directly
to it using the hyperlinks in the Overturned Documents report. Once you reach the core reviewer inter-
face, open the Overturn Analysis related items pane by clicking the RAR icon in the bottom right corner.
n Clicking the file icon next to the document's control number opens the target document in a separate
viewer window. This allows a reviewer to compare the two documents side by side to assist in the
decision-making process.
n Analytics - RelativityOne

Relativity | Recipes - 155

42 Processing duplication workflow
The Processing Duplication Workflow is a Relativity application that identifies master and duplicate
documents, all custodians, and all source files for documents. It also provides capabilities to identify
unique, master and duplicate files based on a relational field.
This page contains the following sections: 
n Before you begin below
n Deploying and configuring the solution on the next page
n Preparing the workspace on the next page
n Running the solution on the next page
n Viewing the results on page 162

42.1 Before you begin

42.1.1 Supported versions
This solution is supported in Relativity 9.0 – 9.5 and RelativityOne.

Note: Some versions of this application may not be eligible for support by Relativity Client Services. For
more information, see the Version support policy on the Relativity documentation site.

Click either of the following links to access the solution files.

Note: You must have valid Relativity Community credentials in order to download any Community file
linked to from the documentation site. You'll need to enter those credentials on the Community login
screen if you're not already logged in. If you're already logged in to the Community at the time you click a
link, the file is automatically downloaded in the bottom left corner of your screen. If you get an error
message stating "URL No Longer Exists" after clicking a Community link, it may be due to a single sign-on
error related to the SAML Assertion Validator, and you should contact your IT department.

Solution version Supported Relativity version

7.0 9.4 - 9.5
6.0 9.0 - 9.3

42.1.2 Components
This custom solution consists of the following components:
n Relativity application
n Relativity scripts that run at the workspace level within a script group

Relativity | Recipes - 156

42.1.3 Considerations
n This script should only be run by a system admin. If you are not a system admin, we recommend you
do not run this script.
n The solution may require you to create fields in your environment. Please see Update Duplicate Status
script for detailed information about the fields that need to be created.
n This solution is especially useful when you're using Relativity's ECA and Investigation feature, which was
introduced in Relativity 9.4. For more information, see ECA and Investigation.
n You have the option of tagging a document as Responsive and then propagating that value to the doc-
ument’s family members. For more information, see Applying propagation to documents.

42.2 Deploying and configuring the solution

To deploy and configure the solution, add it to the Application Library as a Relativity application by
following these steps:
1. Log in to Relativity.
2. Click the user drop-down menu in the upper-right corner of Relativity, and then click Home.
3. Click on the Applications & Scripts tab, and then click the Application Library tab.
4. Click on the Upload Application button.
5. Next to Application File, click Choose File.
6. Navigate to and select the Processing Duplication Workflow.rap file included in the solution, and then
click Open.
7. Click the Save button.

42.3 Preparing the workspace

After you add the solution to the Application Library, you're ready to install and configure it to a
workspace by following these steps:
1. Click the user drop-down menu in the upper-right corner of Relativity, and then click Home.
2. Click the Applications and Scripts tab, and then click the Application Library tab.
3. Click the name of the Processing Duplication Workflow application.
4. Next to Workspaces Installed, click Install.
5. Next to Workspaces, click .
6. Select the check box next to the workspace that you want to install the solution in. To install the solu-
tion in more than one workspace, select the check box next to each additional workspace that you
want.
7. Click OK.
8. Click Save.
9. After refreshing the page, the Workspaces Installed category displays the progress of the installation.

42.4 Running the solution

After you install and configure the solution to a workspace, a Processing Duplication Workflow tab is
added to the workspace. Within the Processing Duplication Workflow tab is a tab called Processing

Relativity | Recipes - 157

Scripts which contains three Relativity scripts within a script group:
1. All Custodians script
2. All Source Locations script
3. Update Duplicate Status script

42.4.1 All Custodians script

The All Custodians script populates a field with the names of all custodians who own a document. To
run the All Custodians script: 
1. Click the Processing Duplication Workflow tab, and then click the Processing scripts tab.
2. The Processing script page appears. Select the radio button next to All Custodians.

3. Complete the following fields: 

n Saved Search - Required. Select the saved search which will contain the group of documents the
script will be executed against.

Relativity | Recipes - 158

 n Relational Identifier - Required. Select the relational field which defines groups of duplicate doc-
uments.
n Custodian - Required. Select the Custodian object that contains your custodian information.
n Custodian Field - Required. Select the field that contains the custodian name.
n Destination (output) - Required. Select the long text field to store the semi-colon delimited list of
custodians.
4. Click Run.

42.4.2 All Source Locations script

The All Source Locations script identifies all source locations for a duplicate document. To run the
All Source Locations script:
1. Click the Processing Duplication Workflow tab, and then click the Processing scripts tab.
2. The Processing script page appears. Select the radio button next to All Source Locations.

Relativity | Recipes - 159

 3. Complete the following fields: 
n Saved Search - Required. Select the saved search which will contain the group of documents the
script will be executed against.
n Relational Identifier - Required. Select the relational field which defines groups of duplicate doc-
uments.
n Source Field - Required. Select the long text field that contains the source location for doc-
uments.
n Destination Field (output) - Required. Select the long text field to store the semi-colon delimited
list of source locations.
4. Click Run.

42.4.3 Update Duplicate Status script

Before running the Update Duplicate Status script, you need to create fields in your environment by
doing the following:
1. Navigate to the workspace in which you are running the solution
2. Create a new single choice field on the Document object with three choices (names must be exact, but
are not case-sensitive): Unique, Master, or Duplicate.
3. Make sure a whole number field is created for the Custodian object If utilizing the order of the cus-
todians associated with the documents.
4. Make sure each custodian has a unique order number. Do not use numbers less than 1.
5. Verify that a Custodian single-object field exists on the Document object. Create a Custodian field if one
does not already exist.
6. Set up a saved search that returns parent level emails and parent level loose documents. The duplicate
status field will be set on the parent level documents, and then family members will be included with
the parents, as described below.
n If using Relativity Processing, parent documents can be identified by including a search condition
where the Level field equals 1.
To run the Update Duplicate Status script:
1. Click the Processing Duplication Workflow tab, and then click the Processing scripts tab.
2. The Processing script page appears. Select the radio button next to Update Duplicate Status.

Relativity | Recipes - 160

 3. Complete the following fields:

n Saved Search - use the saved search described above.

n Duplicate Status Field - use the duplicate status field created above.
n Relational Identifier - select a file duplicate hash field. If using Relativity Processing to load data,
the Processing duplicate hash field can be used.
n Duplicate Sort Order - leave this blank unless you have set up a Custodian field to maintain a cus-
tom priority sort order. When this field is blank, the system sorts on document artifact ID, so the
first document loaded in the workspace becomes the Master document when duplicates are
identified. See Viewing the Results for more information on how this field operates.
After the script runs, the duplicate status will be populated for top level, parent documents. At this
point, you have a couple of options for including family members:
n
Option 1 - Create a search that returns all documents where the duplicate status is set to Master or
Unique, and include family.
n Option 2 - Copy the parent document duplicate status value to child documents and then create a
search that returns all documents where the duplicate status is set to Master or Unique. You would

Relativity | Recipes - 161

 not need to include family in this search.
o One way to copy the duplicate status values is to deploy the ‘Propagate Coding Post-import’
solution and run the script, Propagate Coding Post-Import. This script will copy the duplicate
status value from the parent to each child document so that all documents in a family will have
the same duplicate status value. For more information, see Propagating to one field on demand.
o An alternative way to copy the duplicate status values to is to construct searches to return each

status, include family, and then use a mass edit to update the child documents with the status
value of the parent.
o After copying the parent duplicate status value to children, create a search that returns all non-

duplicates. Family does not need to be included in this search, since all family members will now
have a duplicate status value.
n Make the non-duplicate documents identified in the above steps available to reviewers.

42.5 Viewing the results

The results of running depend on which script was executed.

42.5.1 All Custodians script

After you run the All Custodians script, the results appear as a report on the script page.

The following table lists and describes the columns in the report.

Column Description
Document Identifier The value of the document
identifier on the Document
object.
Custodian The value of the Custodian
associated with the document.

Relativity | Recipes - 162

Column Description
Relational Identifier The value of the relational field
chosen to define groups of
related documents.
All Custodians per Relational Group The semi-colon delimited list of
all custodians that are associated
with the document in the
relational group.

42.5.2 All Source Locations script

After you run the All Source Locations script, the results appear as a report on the script page.

The following table lists and describes the columns in the report: 

Column Description
Document Identifier The value of the document identifier on the Document
object.
Folder Path The value of the source field file location associated with
the document.
Relational Identifier The value of the relational field chosen to define groups of
related documents.
All Source Locations The semi-colon delimited list of all source locations
associated with the document.

42.5.3 Update Duplicate Status script

When the script is executed, the selected Duplicate Status Field is cleared for all documents in the
workspace. Once the field is cleared, the Duplicate Status Field is populated with the following
values for any documents included in the selected saved search:
n Unique – Documents where there is only one document in the saved search with the same relational
identifier
n Master – If Duplicate Sort Order is selected, documents where there is more than one document in the
saved search with the same relational identifier and the lowest order of the associated custodian.  If
multiple documents in the same group shared the same custodian, then the lowest document artifact
ID. If Duplicate Sort Order is not selected, documents where there is more than one document in the
saved search with the same relational identifier which have the lowest document artifact ID in the rela-
tional group.
n Duplicate – If Duplicate Sort Order is selected, documents where there is more than one document in
the saved search with the same relational identifier, which do not have the lowest ordered custodian in
the relational group. If Duplicate Sort Order is not selected, documents where there is more than one
document in the saved search with the same relational identifier which do not have the lowest artifact
ID in the relational group.

Relativity | Recipes - 163

 n Not Set – Documents will remain "Not set" in the selected Duplicate Status Field if the selected Rela-
tional Identifier is not set.
Any documents not included in the selected saved search will be excluded from the logic to
calculate duplicate status and will not be populated in the selected Duplicate Status Field.
When the script is complete, you will receive an "Update Complete" message on the script page.
Documents in the saved search are updated.

43 Performing QC of Productions
Creating and exporting a production successfully can be complicated. It's important to perform a QC
of your productions to ensure accuracy.

43.1 Recipe overview

This recipe points out all possible areas of conflict and errors to consider when creating and
exporting a production.

43.2 Requirements
n Workspace Access
o Saved Searches
o Production sets
n Relativity Desktop Client
n This recipe is applicable to all versions of Relativity.

43.3 Directions
When you're preparing to create a production, it's beneficial to verify some standard checklist items
to prevent incorrect items mistakenly included in a production. Many of these items are intended to
check for conflicts between items in a group and to show where unwanted items are brought in
while segmenting desired items and including a group.
This recipe assist you in preparing a plan on how to move forward when locating unwanted data.
This recipe also assist you in identifying who reviews this data and when a review of the data should
occur.
See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:
See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:

Relativity | Recipes - 164

See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:
See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:
See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:
See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:
See Using saved searches to complete conflict checks for conflict searching when considering the
following in the production set:

43.3.1 Conversation before Production

Consider the following questions before you create a production:
n What field are reviewers using to code responsiveness of documents?
n What field are reviewers using to denote privilege?
n What markup set does the production use?
n Which fields do you want to produce?
n Which fields do you want to produce if you're producing redactions?
n Is there a production set tag?
n What order are you producing items in?
n What is the bates prefix?
n How many digits are in the bates number?
n What language do you want to use on the placeholders? (if applicable)
n What type of production is it? (images, natives, both)
n Do you want to export tiffs or PDFs?

43.3.2 Items to check in a production set

Review the production set for the following document codings:
n Not responsive
n Privileged
n Redacted, not tagged redact (make sure to search on proper markup set)
n Tagged redact, not redacted (make sure to search on proper markup set)
n Previously produced
n Privilege screen terms, not marked privileged
n Imaged and to be produced natively
n To be imaged, not imaged
n Corrupt/unprocessable in production set
n Password protected documents in production set
n Duplicate/email thread/near duplicate of privilege document
n Not reviewed/question/not sure for responsiveness or privilege
n Outside of relevant date range

Relativity | Recipes - 165

43.3.3 QC of production data
Review load files for:
n Ensure the number of rows in DAT file is correct.
n Ensure the fields in the DAT file are correct.
n Ensure the native and text paths are correct.
n Verify the number of images in image load files.
n Verify the number of natively produced documents.
n Ensure the bates and sort order are correct.
n Ensure the header/field names in DAT file are correct.
Sample and Check Review images for:
n The correct first bates number.
n The correct confidentiality endorsement.
n Redactions are in the appropriate location.
n Any other special endorsements.
n The correct number of images from image load files.
n The image type (black and white, color).
n No thumbs.db file is present.
Review text files for:
n Ensure the text of the first document matches image of the first document.
n The OCR text for redacted documents.
n Ensure empty text files only exist for documents with no text in database.
n Ensure no thumbs.db file is present.
n Ensure file names contain the bates number.
Sample and Review native files for:
n The correct number of native files.
n The proper extensions/document types produced natively.
n No thumbs.db file is present.
n Ensure file names contain the bates number.

44 RAR checklist
Before you begin a RAR project, think about whether the project is a good fit for assisted review.
Assisted review centers on the concept of training the system, so it learns how to interpret
uncategorized documents. The system learns best from documents that are good examples. To be
good examples, documents should have rich text with lots of concepts, not just numbers.
Also consider what constitutes a responsive document. If, for instance, responsiveness hinges on a
name or a date, that's not enough for RAR, because there are no concepts to learn, only absolutes.
Successfully completing a RAR project requires you to spend a little time at the beginning of the
process to determine if RAR is the best way to proceed.

Relativity | Recipes - 166

Every RAR project has specific needs, goals, and deliverables. Customize the following checklist to fit
the needs of each project. The checklist is also useful as a guide to the most commonly required
items for RAR.

44.1 Checklist
1. Ensure the document set you plan to use is a good population for Assisted Review. Ensure the pop-
ulation contains the following:
n Minimum 50k records with text
n Concept rich files (not primarily numbers)
n Issue or privilege coding is in a separate field or not part of RAR workflow
2. Make sure you set your timeline and goals. The stakeholders should discuss goals and timelines prior
to beginning a RAR project. This helps establish clear deliverables. Decide and determine the following:
n Determine level of Precision, Recall, and F1
n Decide on a manual review plan (i.e., all docs categorized as Responsive; privilege screen only)
n Put a production plan in place
3. Ensure your administrative prep is complete. Use the following to ensure you complete all admin-
istrative tasks before reviewers look at documents. This decreases the chance of delays during the pro-
ject.
n Install the Assisted Review application
n Create a layout and fields for the project
n Build a layout
n Create a RAR designation field
n Create choices
n Create the use as example field
n Create a text excerpt field
n Build a Reviewers’ view
n Complete the index build
n Eliminate documents without enough text
n Run the extracted text size script
n Create saved searches (training set and searchable set)
a. Optional – Use the following if the project requires categorizing for issues:
nKey issues field (single choice)
n Key issues excerpt field (long text)
n Cluster all documents in RAR universe
n Pull UNCLUSTERED documents from project
b. Optional - Use the following to further cull documents:
nEmail threading
n Key word culling
n STR
4. Ensure you prepare your reviewers. Reviewer preparation is key to success. A RAR project isn't like
other document coding. Use all the tools available to train all reviewers in RAR protocols.
n Show the RAR for End Users webinar to reviewers
n Distribute and discuss Assisted Review Reviewer Protocol

Relativity | Recipes - 167

 5. Complete the RAR project. Planning in advance ensures a successful wrap up. Ensuring all tasks are
complete is important for the client’s satisfaction as well as defensibility.
n Meet project goals
n Precision/recall
n Achieve stabilization
n Manual review under way
n Complete production

45 Relativity Assisted Review Reviewer Protocol

(Sample Based and Active Learning)
45.1 Overview
This recipe is intended as a reference for reviewers in a Relativity Assisted Review project. We’ve
included some best practices and special considerations that are unique to a computer-assisted
review.

Note: This recipe is applicable to both Sample Based and Active Learning projects. To see information only
applicable to Sample Based Learning, see the Sample Based Learning section below.

45.2 Requirements
Analytics Server

45.3 What is Assisted Review?

Computer-assisted review is a workflow that captures and analyzes reviewers’ coding decisions and
amplifies them across a data set. Based on user input, the system learns how to categorize
conceptually similar documents in the document universe.
Relativity Assisted Review utilizes a specific type of text analytics called latent semantic indexing
(LSI). It is important to remember that LSI uses only text in its decision-making; numbers, symbols,
and images are not considered during the machine learning process.

45.4 Selecting good example documents

Because all machine learning is derived from text, it is important to note that some documents may
be highly responsive but undesirable as example documents for an Assisted Review project.

Relativity | Recipes - 168

45.4.1 Sufficient text
In order for a document be considered a good example for machine learning, it must contain a
sufficient quantity of text to train the system. Assisted Review’s text analytics engine learns from
concepts, rather than individual words or short phrases. It’s best to think in terms of sentences or
paragraphs rather than a few words when deciding if a document has sufficient conceptual
language. For that reason, very short documents that contain only a few words or phrases are
typically not good examples.

45.4.2 Email headers and repeated content

Email headers, confidentiality footers, and other types of repeated content are typically filtered out
prior to review. They should not be considered when determining whether a document is a good
example for a computer-assisted review.
Consider the following document:

In this example, the system will only learn from the text that is not framed by a red box. Even if the
subject line and short sentence fragment that remain are responsive, there is not enough text to
warrant this document’s inclusion as an example.

45.4.3 Images
Consider the following example from the Enron data set. This document, which appears to contain
useful text, is actually a JPEG image of a paper document:

Relativity | Recipes - 169

A reviewer might read the above language and find the document to be responsive. However, when
we switch to the extracted text of the document we see the following:

Because the system only works with a document’s extracted text, all of the responsive text located
in the image will be unavailable for machine learning. Consequently, this document, while highly
responsive, turns out to be a poor example document.

45.4.4 Numbers
As mentioned above, numbers are not considered in the machine learning process. It follows that
spreadsheets consisting largely of numbers, while potentially responsive, do not make good example
documents.
Consider the two examples below. Both are spreadsheets, but only the second would make a good
example document.

Relativity | Recipes - 170

Relativity | Recipes - 171
45.4.5 Families and the "Four Corners" test
A document is only a good example if there is text on the document’s face—within the four corners
of the document—that makes it responsive.
The following scenarios violate the Four Corners Test, and will not offer good example documents:
n The document is a family member of another document which is responsive.
n The document comes from a custodian whose documents are presumptively responsive.
n The document was created within a date range which is presumptively responsive.
n The document comes from a location or repository where documents are typically responsive.
This issue is especially prevalent with regard to document families, so additional emphasis is
warranted:
A reviewer should never include a document as an example based on the content of a family
member.

Relativity | Recipes - 172

For example, consider the following email. Note that it mentions an attachment. If the attachment is
responsive, the reviewer might be tempted to include the email as a responsive example, too. Doing
so would violate the Four Corners Test. Again, if there is no sufficient, responsive language on the
document’s face, it should not be used as an example.

45.5 Handling good language / bad example exceptions

Quite frequently, a reviewer may encounter a document containing highly responsive language
which should be coded as non-responsive for an external reason, such as a carve-out agreement
between the litigating parties. For example, consider the draft contract fragment below:

Relativity | Recipes - 173

Let’s assume that the parties have agreed to produce only final, executed contracts. This contract is
a draft, which would make it non-responsive. However, this document contains a great deal of
extremely responsive language, and to submit it as a non-responsive example would teach the
system incorrectly. Consequently, the correct action here is to mark it as non-responsive, but not
submit it as an example.

45.6 Sample Based Learning

45.6.1 Applying the Use as Example field
Users select or reject documents as examples for the system via the Use as Example field. During
categorization, Assisted Review will check the Use as Example field checkbox for all documents that
are part of a project sample set.
If a document appears to be a good example for machine learning, simply code the designation field
as you normally would and leave the Use as Example field checked.
If a document is known to be responsive but is not a good example, code the document as
responsive and uncheck the Use as Example button.

Relativity | Recipes - 174

45.6.2 Using the Excerpt text box
When a document is mostly non-responsive but contains examples of responsive text, use the
Excerpt Text box. Highlight the responsive text, right click, and choose Add to Excerpt Text. Code
the document as responsive.

Relativity pastes the selection into the Excerpt Text field on the Assisted Review coding layout. You
can perform this action multiple times per document. Each time you select text and click Add to
Excerpt Text, the text is automatically appended to the Excerpt Text field.

Relativity | Recipes - 175

However, it is important to remember that good example language is measured in concepts
(sentences and paragraphs) and not single words or short phrases. Do not use the Excerpt Text
option for short keywords or phrases.

45.7 Coding Tips

n Consistency is crucial.
n Consult fellow reviewers on difficult coding decisions to ensure unanimity.
n Don’t touch.
n Never add choices to the Designation tag.
n If you are unsure about a document or have a technical difficulty, consult with the project man-
ager to identify workflow solutions.
n Double check.
n Always check the extracted text of a document to be sure it matches the content in other views.
Whenever possible, review from the Extracted Text viewer.
n When in doubt, ask.
n If there is an aspect of the Assisted Review workflow that is confusing, do not guess. Ask a sys-
tem admin or project manager about the proper course of action.
Optimal use of Assisted Review mandates careful adherence to the recommended workflow. Always
consult with a system admin or review manager when confusion or problems arise. For additional
support for Assisted Review, please contact [email protected].
n Assisted Review - RelativityOne

Relativity | Recipes - 176

46 Recording page-level information for a document
Occasionally, documents may not be paginated correctly or may require changes to pagination for
use as exhibits. This workflow allows you to note these page breaks.

46.1 Recipe overview

This recipe uses an object to store information about a document to call out page-level information.

46.2 Requirements
n Workspace access
n Object creation permission

46.3 Directions
Perform the following steps:
1. Create a new object type called Work Orders.
2. Create fields in the Work Orders type for coding values, such as Page Number, Bates Number, or
Description.
3. Add the newly created fields to a document coding layout.
4. Set the Work Orders object as an associative object on the layout. The Work Order tab can hold all the
individual Bates Numbers of the work orders.
5. You can also create a document view to identify the documents with work orders.

47 Regular expression searching - SSN and EIN

During a review, you may want to search documents for social security numbers and/or employer
identification numbers. Customizing a dtSearch index and using regular expressions can assist in
locating these numbers.

47.1 Recipe overview

This recipe shows you how to customize a dtSearch index and use a regular expression to return a
set of documents containing social security numbers or employer identification numbers.

47.2 Requirements
n Workspace access
o Search index – Edit/Add and corresponding tab

Relativity | Recipes - 177

47.3 Directions
1. Create a dtSearch index.
2. Change the alphabet file to include the dash as a searchable character.
a. Enter the following value under [Letters] // Original letter, lower case, upper case,
unaccented:
i. [Space] [-][Space] [-][Space] [-][Space] [-]

Note: You must have a leading space. You can't have a trailing space.

ii. Place a second instance of the string for lower case.

Relativity | Recipes - 178

 b. Delete the hyphen character - from the [Hyphens] section. For other symbols you would like to
make searchable, delete the appropriate symbols from the [Spaces] or [Ignore] section.

Relativity | Recipes - 179

 3. Perform a full build of the dtSearch index.

4. Run a search using the following regular expression for social security numbers:
"##((?!000)(?!666)([0-8]\d{2}))\-((?!00)\d{2})\-((?!0000)\d{4})"
"##((?!000)(?!666)([0-6]\d{2}|7[0-2][0-9]|73[0-3]|7[5-6][0-9]|77[0-1]))(\s|\-)((?!00)\d{2})(\s|\-)
((?!0000)\d{4})"

Relativity | Recipes - 180

 5. Run a search using the following regular expression for employer identification numbers:
"##\d{2}-\d{7}"

Note: The quotes and # characters are necessary for the dtSearch index to recognize the regular
expression. Quotes can't be Smart Quotes; they must be ASCII quotes.

47.4 General considerations for all regular expression searches

Each regular expression in a search request must match a single word in the index. This results in
the following limitations:
n A regular expression can't contain a space or any character that's indexed as causing a word break.
n It can only reference searchable letters.
n The terms are subject to the maximum length of a word, which is 32 by default. You can change this
limit using the Options.MaxWordLength configuration Instance setting table value.
n The case should match the case in the index (lower case unless the index is case-sensitive).

Note: dtSearch uses TR1 regular expressions as implemented in Visual C++, which Microsoft's Developer
Network documents.

48 Searching for symbols

This recipe demonstrates how to make symbols searchable in a dtSearch index. It also demonstrates
how to use regular expressions to search for certain symbols reserved as search operators, such as
the % sign.

48.1 Requirements
n Workspace access
n Search index – Edit/Add and corresponding tab

48.2 Directions
1. Create a dtSearch index.
2. Change the alphabet file to include the % sign as a searchable character:
a. Enter the following under [Letters] // Original letter, lower case, upper case, unaccented:
i. [Space] [%] [Space] [%] [Space] [%] [Space] [%]

Note: You must have a leading space. You can't have a trailing space.

Relativity | Recipes - 181

 ii. Place a second instance of the % % % % string in the lower [Letters] section immediately
after _ _ _ _:

b. Delete the % sign from the [Ignore] section.

c. Repeat these steps for any other symbols that you need to make searchable. Delete the appro-
priate symbols from the [Spaces] section.
3. Perform a full build of the dtSearch index.
You can now search for terms containing the % sign using a regular expression. For example, if you
need to search for documents that contain the term 75%, you would enter the following in your
search box (ensuring you select the proper dtSearch index):

Relativity | Recipes - 182

\W is the regular expression metacharacter that stands for any symbol. A dtSearch for "##75\W"
hits on 75%, as well as 75# and 75! if you had made # and ! searchable. "##" signals to Relativity to
treat the string as a regular expression. Let’s say you need to search for documents that contain any
variety of percentage values, where the strings are least 2 numbers and at most 3, such as: 100%,
37%, 214%, 95%. You could use the following regular expression:

The quotes and # characters are necessary for the dtSearch index to recognize the regular
expression. Quotes can't be Smart Quotes; they must be ASCII quotes.

Note: For most symbols, once you have made them searchable in the alphabet file, you can type them
directly into the dtSearch box without using regular expressions. Thus to search for 30!, enter 30! into the
dtSearch box after you have made ! a searchable character. You only need to employ \W for certain
symbols such as the % sign.

48.3 References
n Searching with Regular Expressions
n Searching with Regular Expressions
n Searching with Regular Expressions
n Searching with Regular Expressions
n Searching with Regular Expressions
n Searching with Regular Expressions
n Regular Expressions for Beginners webinar

49 Relativity Structured Analytics – Best practices and

workflows
Published September 30, 2013
This recipe provides some best practices around the setup and use of Structured Analytics.

Relativity | Recipes - 183

49.1 Requirements
n Workspace access and system admin rights
n Relativity Structured Analytics application

49.2 Directions
Consider the following guidelines when using Structured Analytics in Relativity:
n There are advantages to running each of the Structured Analytics Set operations individually. For
example, if you need to make changes to a particular operation, it is best to make the changes and
rerun the set for that one operation, rather than re-running a set with multiple operations selected.
n The ideal groups of Structured Analytics operations to perform as separate sets are:
o Email threading and Near Duplicate Identification
o Language Identification
o Repeated Content Identification

n We do not recommend clicking Clean Up Set following your first Full Analysis if you expect to run an
Incremental Analysis in the future. Keeping the results of the first full run on the server will expedite the
incremental run.
n When setting up email threading, the Identify emails in document set search items should also be in
the Select document set to analyze search, but the documents to analyze should also include attach-
ments.
n When running Near Duplicate Identification, system admins have two options:
o Run with email threading – the Near Duplicate Identification process identifies all textual duplic-
ates within your document set which are not emails, while the email threading operation iden-
tifies all the email duplicates.
o Run it without email threading (either by itself or with another Structured Analytics operation)

– the Near Duplicate Identification process includes all email and non-email textual duplicates.
n When performing Repeated Content Identification, you should change the default settings if you’re
working with large data sets. See Sampling for repeated content for more information.
Once you run the Structured Analytics Set, the following workflows can assist with an efficient
review process.
n System admins can set up a review to include only the inclusive email threads. Listed below are some
settings for a Saved Search that can be used to batch documents:
o Search name and settings (you can choose whether to include family members)

Relativity | Recipes - 184

 o Search conditions

o Search fields

o Sort order

n Similar to identifying email threads, you can set up a review for all near duplicate items by setting up a
saved search to identify such documents.

Relativity | Recipes - 185

 o Search name and settings (you can choose whether to include Near Dupe Groups)

o Search conditions

o Search fields

Relativity | Recipes - 186

 o Sort order

n As a system admin, you can continue to manually identify repeated content items and include them as
a part of the Repeated Content Filters tab. Alternatively, you can use the Repeated Content Iden-
tification feature and link to an Analytics profile.
n To facilitate review in multiple languages, system admins can use the Language Identification feature to
batch out documents based on the primary language identified, or use the Pivot feature to identify and
filter on the languages the system identified.

50 Relativity Structured Analytics - Email Threading

Published January 22, 2015
Relativity Analytics Email Threading reduces the time of email review by arranging entire email
conversations in sequence. Structured Analytics identifies inclusive emails, the most complete
message in a thread. This allows reviewers to bypass other emails in a thread that contain only part
of the inclusive content. Reviewing non-inclusive emails is unnecessary.
This recipe provides the necessary setup instructions to properly view and utilize Email Threading
results and reports.

50.1 Requirements
n Workspace access and system admin rights
n Relativity Structured Analytics application

50.2 Directions
50.2.1 Create saved searches
First, create the following saved searches when setting up a Structured Analytics Email Threading
set.
n Saved Search #1 = Parent emails in document set: Create a saved search that returns only
the parent emails and family. The search conditions you set will depend on the metadata
fields available in your workspace.

Relativity | Recipes - 187

 o "Email From" field is set
o Parent Document ID is not set
o Include Family
First, create the following saved searches when setting up a Structured Analytics Email Threading
set.
n Saved Search #1 = Parent emails in document set: Create a saved search that returns only
the parent emails. The search conditions you set will depend on the metadata fields available
in your workspace.
o "Email From" field is set
o Parent Document ID is not set
n Saved Search #2 = Parent Emails and Family: Create a saved search that returns all doc-
uments across which you intend to run email threading. You can set the saved search so that
it returns all the documents in your workspace or limit the results according to any particular
conditions you may want to set. Ensure that you include both the parent emails and any
attachments in your search results. You can do this by setting the drop-down to Include Fam-
ily.
o (Saved Search) : Document is in [Parent Emails]
o Include Family

Note: For RelativityOne, documents with more than 30 MB of extracted text will automatically be excluded
from the structured analytics set and you no longer need to manually exclude them using an additional
search.

n Saved Search #3 = Document set to analyze: Using search #2 above, exclude documents

based on extracted text size. Create a third saved search with the following conditions:
o (Saved Search) : Document is in [Parent Emails and Family]
o Extracted Text Size is less than 30720 KB

50.2.2 Create or configure an Analytics profile

Next, create or configure an Analytics Profile for your Email Threading set. Perform the following
steps:
1. From the Indexing and Analytics tab, select Analytics Profiles. A list of existing Analytics pro-
files displays.
2. To create a new profile, click New Analytics Profile. Or, you can edit an existing profile. The
Analytics Profile layout appears.

Note: Rather than modify the Default profile, you should always create a new profile or edit an
existing profile you have created.

3. Email Threading Settings - map the email threading fields on the Analytics Profile Layout to
the appropriate workspace fields containing the following specified metadata.

Relativity | Recipes - 188

 n Email Header Fields:
n Email from field – sender’s email address.
n Email to field – recipients’ email addresses.
n Email cc field – carbon copied (Cc) recipients’ email addresses.
n Email bcc field – blind carbon copied (Bcc) recipients email addresses.
n Email subject field – email subject line.
n Email date sent field – email sent date. Typically a date field, this field needs to
contain both the date and time.

Note: When setting up an Email Threading Structured Analytics set, if you choose to
not use email header fields, then you will not need to map any of the above email
header fields. In this scenario, email threading relies on extracted text, and the Parent
Document ID and Attachment Name fields.

n Email Metadata Fields

Note: The following fields are all optional.

n Attachment name field – file name of the document attachment. This only needs
to be populated for child documents and not the parent email. This field is used
to display an attachment’s file name and icon in the Email Threading Display visu-
alization field. Relativity Processing creates the Attachment name field. You may
use a File name field instead.

n Conversation ID field – Microsoft Conversation index number generated by

Microsoft Outlook to identify email conversations.
n Parent document ID field – a field that indicates the Parent Email ID for an
attachment. The parent document ID field should be set for all attachments
(including attachments that are emails), but should not be set for parent emails.
o If you do not have the parent document ID data, you can run the Populate
Parent ID to Child script to populate this field. For more information on run-
ning this script, see Relativity Script Library.

Note: In order to properly thread and visualize attachments, the Parent document ID
and Attachment name fields must be mapped.

5. Advanced Options – these settings affect the building of the conceptual Analytics Index, but
will not have any effect on your Structured Analytics Email Threading set. Thus, you may
leave the default settings as is in this section.

50.2.3 Create a Structured Analytics set

You are now ready to create a Structured Analytics Email Threading Set.

Relativity | Recipes - 189

 1. From the Indexing and Analytics tab, select Structured Analytics Set. A list of existing Struc-
tured Analytics sets displays.
2. To create a new set, click New Structured Analytics Set. The Structured Analytics set layout
appears.
3. Set Structured Analytics Set Information:
n Enter a name for your Structured Analytics set, and then select Email threading as the
operation to perform.
n In Select document set to analyze, select the saved search you created that will return
all documents across which you wish to run email threading.
n In Analytics server selection, select Automatic.
Enter your email address in the "email addresses" text box if you wish to receive a notification
when the email threading analysis is completed.
4. Set Email Threading Settings:
n In Select profile for field mappings, select the Analytics profile that you had set up
earlier.
n In Use email header fields, select Yes in order to use any email header fields in your
workspace for the threading analysis. Setting this option to No will not send any email
header fields for analysis, forcing Structured Analytics to rely only on the extracted
text field of an email for threading. If you choose No, you will not need to map the
email header fields in your Analytics profile.
n Click Save. The Structured Analytics Set Console appears.
n In Identify parent emails in document set, select the saved search you created that
will return only parent emails from the document set across which you wish to run
email threading.
n In Select profile for field mappings, select the Analytics profile that you had set up
earlier.
n In Use email header fields, select Yes in order to use any email header fields in your
workspace for the threading analysis. Setting this option to No will not send any email
header fields for analysis, forcing Structured Analytics to rely only on the extracted
text field of an email for threading. If you choose No, you will not need to map the
email header fields in your Analytics profile.
n Click Save. The Structured Analytics Set Console appears.
5. Run Email Threading Analysis:

n Click Run Full Analysis to begin running your Email Threading analysis.

Relativity | Recipes - 190

 n
You can monitor the progress of the email threading analysis in a separate window by
clicking on the pop-out icon in the upper right corner of the progress pane.

50.2.4 Email threading results - reports

After running an email threading operation, we recommend reviewing the available reports. On the
Structured Analytics Set console, click View Email Threading Summary to open the report. This
report contains a graphical summary and tables that list a breakdown of all the emails analyzed.
See Viewing the Email Threading Summary.See Viewing the Email Threading Summary.See Viewing
the Email Threading Summary.See Viewing the Email Threading Summary.See Viewing the Email
Threading Summary.See Viewing the Email Threading Summary.See Viewing the Email Threading
Summary.

Relativity | Recipes - 191

The Email Threading Summary pie chart provides a graphical representation of the percentages of
inclusive emails, inclusive duplicate emails, and non-inclusive emails.

50.2.5 Setting up an Email Threading Saved Search

To view the results of the email threading operation, we recommend creating an Email Threading
view in the Documents tab. But first, create a saved search that will be the basis of the Email
Threading view.
For more information on creating saved searches, see Saved Search.

Relativity | Recipes - 192

 1. Fill out the appropriate information for Name, Owner, and Scope.
2. Email Threading Saved Search - Set Conditions
To only return email thread groups (or possibly documents that have been part of an email threading
set) set the following conditions:
n Email Thread Group: Is set
n Add Related Items: + Family Group
o Add relation field that groups emails and their attachments together. Otherwise, attach-
ments will not be returned in the Email Threading saved search or the view to be created
in the next section.

3. Email Threading Saved Search - Set Fields

Add the following output fields to your Email Threading view:
n Email Threading Display
n Email Thread Group
n Email Threading ID
n Inclusive Email
n Inclusive Reason
n Email Duplicate Spare
n Control Number
4. Email threading Saved Search - Set Sort Order
Configure each field below to sort in ascending or descending order, depending on the case team's
preference:
n Email Thread Group - initial nine characters of the Email Threading ID. This value is the same
for all members of an Email Thread Group which are all of the replies, forwards, and attach-
ments following an initial sent email. This field is relational.
n Indentation - a whole number field indicating the node level of the document.
n Email Threading ID - ID value that indicates which conversation an email belongs to and where
in that conversation it occurred. The first nine characters indicate the Email Thread Group, (all
emails with a common root), and the subsequent five-character components show each level in
the thread leading up to an email's top segment. For example, an email sent from A to B, and

Relativity | Recipes - 193

 then replied to by B, has the initial nine characters for the message from A to B, plus one five-
character set for the reply.

50.2.6 Setting up an Email Threading View

To view the results of the email threading operation, we recommend creating an Email Threading
view in the Documents tab. The saved search you just created will be the basis of this view.
For more information on creating views, see Views.
1. Navigate to the Documents tab in your workspace.
2. Click New View in the views drop-down to create a view.
n Email threading View - Basic Information
o View Information - Fill out the appropriate information for Name, Order, Visible in Drop-
down.
o Advanced Settings - select the following Advanced Settings options when creating the
Email Threading view:
Object Type - Document
l

l Visualization Type - Indented List

l Group Definition - Email Thread Group

l Indentation Method - Numeric

l Indentation Definition Field - Indentation

l Display Field - Email Threading Display (Choose Control Number if you wish to see

indentation dots in the view)

n Email Threading View - Set Fields- Add the following output fields to your Email Threading view:
o Email Threading Display  
o Email Threading ID
o Inclusive Email
o Inclusive Reason
o Email Duplicate Spare
o Control Number

n Email Threading View - Set Conditions - Set the conditions of the view such that it returns the
saved search created in the previous step. In the screen shot below, the saved search was
named Email Threading Results:
o (Saved Search) = Document is in = Saved Search created in previous step

n Email Threading View - Set Sort Order - Configure each field below to sort inascending or des-
cending order, depending on the case team's preference:

Relativity | Recipes - 194

 o Email Thread Group - initial nine characters of the Email Threading ID. This value is the
same for all members of an Email Thread Group which are all of the replies, forwards, and
attachments following an initial sent email. This field is relational.
o Indentation - a whole number field indicating the node level of the document.
o Email Threading ID - ID value that indicates which conversation an email belongs to and
where in that conversation it occurred. The first nine characters indicate the Email Thread
Group, (all emails with a common root), and the subsequent five-character components
show each level in the thread leading up to an email's top segment. For example, an email
sent from A to B, and then replied to by B, has the initial nine characters for the message
from A to B, plus one five-character set for the reply.
Your email threading view will appear similar to the following. The blue line between rows
separates distinct threads of email messages, and is activated as a result of setting the
view's visualization type to Indented list.

Note: If you wish to see the latest email in a thread first, set Indentation to sort in descending
order.

50.2.7 Email Threading Display

Note: See Email threading results for more information on the Email threading Display field and how to use
it on the Thread Groups related items pane in the Viewer.

50.2.8 Working with email threading results

After running an email threading operation and setting up your email threading view, use the
following sample workflow to narrow down your documents to review.

Relativity | Recipes - 195

50.2.8.1 Identifying unique documents to review
By reviewing only inclusive emails and skipping non-inclusives and duplicates, your review process
would be most efficient. The Analytics engine derives the email threads and determines which
subset of each conversation constitutes the minimal inclusive set. Non-inclusive emails are
redundant because all non-inclusive content is also contained in inclusive emails. The inclusiveness
analysis ensures that even small changes in content will not be missed by reviewers.
To avoid redundant review, you can look at the Email Duplicate Spare field in conjunction with the
Inclusive Email field. The Email Duplicate Spare field will be set to Yes for all but one duplicate in
each group of duplicative emails. A suggested workflow is to review only the inclusive, non-duplicate
spare emails and their attachments, while skipping the non-inclusive and duplicate spare emails.
These fields may be used in Saved Searches and views to easily locate the desired documents, as
follows.
To identify and view a list of only unique documents using your email threading results and email
threading view, perform the following steps:
1. Set up a saved search that will only return inclusive, non-duplicate spare emails and their attachments.
2. Follow the overall steps described above in the section, Setting up an Email Threading Saved Search.
But for the conditions section, apply the following settings instead:
n Inclusive, Non-Duplicate Spares Saved Search - Set Conditions
o Email Thread Group: is set, AND
o Inclusive Email: is YES, AND
o Email Duplicate Spare: is No
o Add Related Items: + Family Group
l Add the relational field that groups emails and their attachments together. Other-
wise, attachments will not be returned in the saved search or the view to be cre-
ated in the next step.
3. Set up a view for reviewers to access only the inclusive, non-duplicate spare emails and their attach-
ments.
4. Follow the overall steps described above in the section, Setting up an Email Threading View. Set the con-
ditions of the view such that it returns the Inclusive, Non-Duplicate Spares saved search created in this
section.

50.2.9 Special Considerations

Consider the following guidelines when using Structured Analytics in Relativity:
n There are advantages to running each of the Structured Analytics Set operations individually. For
example, if you need to make changes to a particular operation, it is best to make the changes and
rerun the set for that one operation, rather than rerunning a set with multiple operations selected.
n The ideal groups of Structured Analytics operations to perform as separate sets are:
o Email threading and near duplicate identification
o Language Identification
o Repeated Content Identification

Relativity | Recipes - 196

50.2.10 Setting up an Email Threading View
To view the results of the email threading operation, we recommend creating an Email Threading
view in the Documents tab.
See Setting up an email threading viewSetting up an email threading view in the Admin guide.
For more information on creating views, see Views.
n Navigate to the Documents tab in your workspace.
n Click to create a view.
1. Email Threading View - Basic Information
n
View Information:
o Fill out the appropriate information for Name, Order, Visible in Dropdown.

n Advanced Settings:
Select the following Advanced Settings options when creating the Email Threading
view:
Object Type - Document
o

Visualization Type - Indented List

o
o Group Definition - Email Thread Group
o Indentation Method - Numeric
o Indentation Definition Field - Indentation
o Display Field - Email Threading Display (Choose Control Number if you wish to see

indentation dots in the view.)

2. Email Threading View - Set Fields
Add the following output fields to your Email Threading view:
nEmail Threading Display
n Email Thread Group
n Email Threading ID
n Inclusive Email
n Inclusive Reason
n Email Duplicate Spare
n Control Number
3. Email Threading View - Set Conditions
To only return email thread groups (or possibly documents that have been part of an email
threading set) set the following condition:
n Email Thread Group : is set
4. Email Threading View - View Sorting
Configure each field below to sort in ascending order:
n Email Thread Group - initial nine characters of the Email Threading ID. This value is
the same for all members of an Email Thread Group which are all of the replies, for-
wards, and attachments following an initial sent email. This field is relational.
n Indentation - a whole number field indicating the node level of the document.

Relativity | Recipes - 197

 n Email Threading ID - ID value that indicates which conversation an email belongs to
and where in that conversation it occurred. The first nine characters indicate the
Email Thread Group, (all emails with a common root), and the subsequent five-char-
acter components show each level in the thread leading up to an email's top segment.
For example, an email sent from A to B, and then replied to by B, has the initial nine
characters for the message from A to B, plus one five-character set for the reply.
Your email threading view will appear similar to the following. The blue line between rows
separates distinct threads of email messages, and is activated as a result of setting the view's
visualization type to Indented list.

Note: If you wish to see the latest email in a thread first, set Indentation to sort in descending order.

50.2.11 Email Threading Display

Note: See Email threading results for more information on the Email Threading Display field and how to use
it on the Thread Groups related items pane in the Viewer.

The Email Threading Display field provides the following visual information about email threading:
1. Indentation Bubbles
The indentation bubbles indicate each email message's indentation level within the thread.
For example, the first email in the chain would be "1," an email responding to the first email
would be "2," and an email responding to the second email would have a "3." Email
indentation bubbles go up to 99. For messages with an indentation level over 99, the bubbles
display "99+."

Relativity | Recipes - 198

 The color of the indentation bubble indicates inclusiveness. Inclusive email messages contain
the most prior message content and have a "Yes" value in the Inclusive Email field.
Emails that are both inclusive and non-duplicative appear blue. Non-inclusive or duplicate
spare messages appear gray. In the screen shot below, Emails #1 and #2 are non-inclusive.
Email #3 is inclusive and not a duplicate spare.

2. Message and file type icons

The Email Threading Display field includes the following file type icons:

n Reply - original file name begins with RE:

n Forward - original file name begins with FW:
n Other - original file name does NOT begin with either RE: or FW:
n Unknown - file type cannot be found.
n Email contains attachments - file is an email containing attachments. Attachments
are documents included in your emails in document set saved search. The following con-
ditions apply to the icon display in the Relativity workspace:
o The document type determines the attachment's file type icon.
o The attachment’s file name that appears is based on the value in the attachment
name field.

50.2.12 Thread groups in the related items pane

To improve the review efficiency, the Email Threading Display field automatically appears when
viewing the Thread Group in the related items pane. You can also add the Email Threading Display
field to any other view in Relativity.

Relativity | Recipes - 199

In the related items pane, click the Thread Group icon to display all messages in the same
thread group as the selected document. You can also click the Email Duplicates icon to show all
of the messages identified as duplicates of the selected document, or click the Near Dupe Groups
icon to show all items that are textual near duplicates of the selected document.
After deploying the Analytics application, you can create fields that propagate coding choices to
these Structured Analytics related groups. See Applying propagation to documents for more
information.

50.2.13 Working with email threading results

After running an email threading operation and setting up your email threading view, use the
following sample workflow to narrow down your documents to review.

Relativity | Recipes - 200

50.2.13.1 Identifying unique documents to review
By reviewing only inclusive emails and skipping duplicates, your review process would be most
efficient. The Analytics engine derives the email threads and determines which subset of each
conversation constitutes the minimal inclusive set. Non-inclusive emails are redundant because all
non-inclusive content is also contained in inclusive emails. The inclusiveness analysis ensures that
even small changes in content will not be missed by reviewers.
To avoid redundant review, you should look at the Email Duplicate Spare field in conjunction with
the Inclusive Email field. The Email Duplicate Spare field will be set to Yes for all but one duplicate
in each group of duplicative emails.
A suggested workflow is to review the inclusives and their attachments, while skipping the
duplicate spares and their attachments. These fields may be used in Saved Searches and views to
easily locate the desired documents, as follows.
To identify and view a list of only unique documents using your email threading results and email
threading view, perform the following steps:
1. Click the Documents tab in your workspace.
2. Select your new email threading view from the drop-down menu on the view bar.
3. In the drop-down to the right of In This Folder & Subfolders, select Include Family.
4. Click located next to the Search... drop-down menu to toggle on search conditions.
5. Add the following search conditions with an AND operator:
n Field: Email Duplicate Spare
Operator: isValue: No
n Field: Inclusive Email
Operator: isValue: Yes

6. Click Search.
To save this search, click the Save as search icon . See Saving searches on the Documents tab for
additional information.

50.2.13.2 Email Thread Visualization

You can also review your results using email thread visualization.

1. To launch the email thread visualization pane, click the email thread visualization ( ) icon in the
lower right corner of the displayed document in the viewer. The pane expands and shows the email
thread group for the selected document in the viewer.
2. From the visualization, the currently opened document is outlined in gray and displays as larger than
the unselected documents in the thread.

Relativity | Recipes - 201

 3. A legend will appear in an opened state showing the meaning behind the different icons. You can col-
lapse the legend if you want to see more of the visualization and it will remain closed until you log out
of Relativity.

4. If you hover over an email icon, a tooltip window displays who the email is from and the sent date. It
also lists any attachments or duplicate spares for the selected email. You can click on the attachment
name or the name of the duplicate spare to open it in the viewer.

Relativity | Recipes - 202

 5. Coding highlight allows you to see how emails in the current thread are coded for a particular yes/no or
single-choice field. Coding discrepancies that could exist in a coded email thread are visually apparent,
making it very easy to make corrections or see where mistakes were made during review. Click on the
Highlight Field switch to enable coding highlighting.

6. Select the field you want to highlight in the email thread visualization pane from the Coding Highlight
field drop-down (e.g. Confidential Designation). The emails in the thread are now highlighted with dif-
ferent colors corresponding to the available choices for the field.

Relativity | Recipes - 203

 7. The email thread visualization tool can be used to code entire branches of emails quickly and visually.
You can mass edit from thread group related items pane or the visualization and it’s reflected in real
time. For more information, see Email Thread Visualization.

50.3 Special considerations

Consider the following guidelines when using Structured Analytics in Relativity:
n There are advantages to running each of the Structured Analytics Set operations individually.
For example, if you need to make changes to a particular operation, it is best to make the
changes and rerun the set for that one operation, rather than rerunning a set with multiple
operations selected.
n The ideal groups of Structured Analytics operations to perform as separate sets are:
o Email threading and near duplicate identification
o Language Identification
o Repeated Content Identification

50.4 References
Email Threading
Email Threading
Email Threading
Email Threading
Email Threading

Email Threading
Email Threading

Relativity | Recipes - 204

51 Removing documents from a RAR example set
When working in Relativity, you may need to remove certain documents from a RAR project if
they're not good examples for training the RAR project. This recipe explains how to remove the set
of example documents provided to a RAR project at any point during the RAR project.

51.1 Requirements
n Relativity access
n An active RAR project

51.2 Directions
This workflow assumes that you are between rounds in a RAR project (i.e., you've finished the
current round and are ready to begin the subsequent round). Perform the following steps to remove
documents from a RAR project after the start of the project.
1. Identify the documents that are not good example documents for the RAR project per the Reviewer
Protocol specifications.
a. Create a saved search against your RAR project universe with the following conditions:
n (RAR Use As Example = Yes) AND
n ((Extracted Text Size <=0.75KB) OR
n ((Extracted Text Size >=2048KB) AND (RAR Excerpt Text = Not Set)) OR
n (File Type = compressed files, RAR, ZIP, system files, image files, CAD drawings, maps, and
calendar items, number-rich documents))
2. Once you've identified bad example documents, flag the documents by setting the RAR Use as Example
field to No.
3. Proceed with the regular RAR workflow. Create new training or QC rounds per the needs of your pro-
ject/reports (you can create a round with a single document to force categorization).
Once you complete the round in step 3, all the documents that were flagged RAR Use as example =
No (in step 2) will no longer be used by the categorization engine to make suggestions on the
documents in the RAR project universe.

51.3 References
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review

Relativity | Recipes - 205

 n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics - RelativityOne
n Relativity Assisted Review - RelativityOne

52 Resolving common processing errors

Password-protected and corrupt files may appear in your processing projects. To add these files to
your workspace, you need to remove the password restriction or repair the corrupt files.

52.1 Recipe overview

This recipe describes how to repair files and remove password protection from them. You can then
add the files to your workspace during processing.

52.2 Requirements
n Relativity Processing
n Processing repository access

52.3 Directions
1. Select the Processing > Processing Sets tab.
2. Select the processing set with the errors that you want to resolve.

3. Click View ErrorsView Job Errors on the console for the current processing set. You can now view the
errors in the processing set on the Processing > Document ErrorsErrors tab.

Relativity | Recipes - 206

 4. Click on the name of the file with errors that you want to resolve. You can now view the error details,
which lists two different file paths:
n Document file location - This path includes the actual file ending with the .doc file extension.
You can click this link to open the document.
n
Document folder location - This path indicates the path to the folder that contains the
document with the error. You can click this link to open the folder.

Relativity | Recipes - 207

 5. Perform one of the following tasks to resolve the specific type of error:
n Password protection error - If password protection caused the error, use the password bank to
resolve the error.
n Corrupt file - If a corrupt file is corrupt caused the error, click the Document folder location
path. Locate the file in the folder. Next, open it with a tool used to repair corrupt files. After

Relativity | Recipes - 208

 repairing the file, save the new file in the same file location. Ensure that the new file has the same
file name as the corrupt one.

Note: When you repair the file, you modify it, therefore changing the meta data. Consider making a copy of
the file before making any changes to preserve the original meta data.

1. Retry the discovery phase after you resolve the file errors:
a. Use the mass operations window in the lower-left corner to check the files you fixed.
b. Select Checked and Retry.
c. Click Go.

2. Click Ok to retry the errors.

3. Monitor the retry progress via the progress window.

Relativity | Recipes - 209

53 Reviewer QC workflow ideas
53.1 Overview
The quality and accuracy of the review team’s work is a very important component of any project.
This recipe outlines some suggested workflows to perform Quality Control measures on first level
reviewers’ work. The requirements for QC and the extent of QC conducted is always dictated by
the needs of each project, but some form of QC will almost always be necessary. This recipe
discusses building and using a separate QC layout, setting up a QC sample set to gain insight into
the team’s accuracy, selecting and applying key substantive and privileged terms to QC measures,
and checking for tagging conflicts.

53.2 Requirements
n Workspace access
o Use Sampling, Pivot, Tally/Sum/Average
o Delete and Add Fields, Choices, Layouts, Views
o Delete and Add all Search functions

53.3 Directions
General QC workflow
1. Create the QC Layout
2. Batch documents for the QC team to review
3. Identify total documents put through QC and resulting overturns

53.3.1 Create a Quality Control layout

Once the first level team completes its review on a given set of documents, the QC team can get
started. The best workflow is to be sure first level review is complete on the set of documents
before placing the documents in any QC queue. We suggest setting up a separate layout for the QC
team’s exclusive use.
To set up the layout:
1. Include all fields the first level reviewers coded, such as any Responsiveness determination
fields, Privilege, Confidentiality and Issue Designation.
2. Add a “Second Level QC” field to track the number of documents put through QC.
3. Add choices for the QC team to make its decisions. The QC team may record any overturns on
second level as well. Name the choices appropriately for your matter.
4. Secure the layout so that only the QC team can view and code it.
Below are two examples of QC layouts. Both allow users to QC the first level review and make any
necessary coding changes.

Relativity | Recipes - 210

Or, configure the QC Layout to track the QC team’s overturns. If the QC team disagrees with the
first level reviewer’s decision, they can mark the changes by category, e.g “Responsiveness
Change” or “Privilege Change” as in the following example.

Relativity | Recipes - 211

53.3.2 Batch reviewed documents for QC review
1. Create a saved search of all documents that have been through first level review (Responsive
Designation is set).
2. Create a QC Batch Set using the reviewed documents search as the Batch Data Source.
Enable Auto Batch to continuously populate new batches for the QC team.

Relativity | Recipes - 212

53.3.3 Identify total documents put through QC and resulting overturns
After the QC team finishes its second level review, use a saved search and Pivot to calculate
overturns and team accuracy statistics.
n The saved search conditions should include:
o Second Level QC is set and
o Second Level QC Overturns is set(Include any specific first level review batches or
folders as well
n Run the search
n Pivot on Second Level QC Overturns and view the results in a chart:

To include the statistics in your reporting, save results by exporting to a file from the Pivot view:

Relativity | Recipes - 213

Use Statistical Sampling
Create a QC Sample set of all reviewed documents
The Relativity sampling features can be used to determine overall accuracy rates for individual
reviewers, groups, or the entire review team. Here are the steps to create the sample set:
1. Set the view to all documents.
2. Filter on documents where Responsiveness is set.

3. Set up a sample set:

n Activate the sampling menu.

n Choose either the Fixed Size, Percentage or Statistical sampling type.

4. Save as a list to keep the set static.

Relativity | Recipes - 214

 5. Save the list as a search.

6. Create review batches using the sample set search as the batch data source.
7. Assign to the QC team for review using the procedure described above.

Note: To learn more about using sampling and how best to apply it to your review, see Sampling and
Understanding the Statistical Sampling Formula.

53.4 Focus QC on Substantive Issues or Privilege

53.4.1 Conduct targeted substantive QC
A suggested method to make sure the reviewers are catching all responsive documents is to
conduct some QC based on the substantive issues in the litigation. Targeted QC review focuses only
on documents that contain relevant words or phrases, but where the first level reviewer marked
the document Non-Responsive. Use a list of relevant terms to define search parameters. You may
be able to obtain a list from the attorneys or case team handling the litigation. Or, use the terms
that were used to set up the search terms reports.
Following are examples of how to set up targeted QC:
n Use key terms from a Search Terms Report:
o On the Field Tree, choose an STR with Unique hits.

o Filter to show only those documents marked Non-Responsive.

o Create a saved search of the returned documents.
o Create review batches for the QC team or create a sample set for review.
OR
n Create a saved search to find documents where Responsiveness may have been missed.
o Enter the relevant text as the search terms.
o Set the search conditions to return only documents marked Non-Responsive.

Relativity | Recipes - 215

 Note: When creating STRs, choose “Report and Tag” to automatically list the tags and search terms in the
Field Tree.

53.4.2 Conduct Privilege QC

If privilege is an issue in the review, some measure of QC focused on either privileged terms,
attorneys’ names, or others who might create privilege is likely needed. Use the methods outlined
above for substantive QC to conduct the Privilege QC.

For targeted Privilege QC

1. Enter the list of names or firms who may create privilege as the search terms.
2. Set the search conditions as Privilege is not set.

53.4.3 Check for family tagging conflicts

n Often, you will need to check for inconsistent coding in family groups. We recommend con-
structing a set of saved searches to ensure that all members of each family are coded con-
sistently for Responsiveness and for Privilege.
n The Advice team’s recipe, Searching Workflows, includes a step-by-step process to accom-
plish a tagging conflicts check.

54 Sampling for repeated content

Relativity Analytics Indexes benefit greatly from targeted removal of boilerplate text, especially
email confidentiality footers. Structured Analytics offers a Repeated Content Identification
operation, but on large collections it can be slow and resource intensive.
This recipe provides information on using random samples to achieve repeated content
identification in a much quicker fashion and with significantly lower resource utilization than
running on the full collection - all without sacrificing the quality of the results.

54.1 Requirements
n Relativity 8.1 or higher
n Relativity Structured Analytics application
n Document set larger than 100,000 documents (no need to sample if you're already smaller than that)

54.2 Directions
54.2.1 Creating the sample
1. First, create your random sample of the target documents.
2. Navigate to the Documents tab and restrict yourself to the documents you want to focus on. It might
be everything in the workspace, the searchable documents from your index, or a set limited by file
type, email inclusives, or some other subset.

Relativity | Recipes - 216

 3. Once you're looking at the document set you want to analyze, create a random sample by clicking
4. In the Sampling dialog, set the following:

n Type: Fixed
n Size: 100,000
5. Click Sample. You should now see only 100,000 documents listed on the page.

54.2.2 Save sample as list and list as saved search

At this point, you're ready to create the list. From there you'll create the saved search that you'll
reference in your Repeated Content Identification run.

Note: In Relativity 8.1-9.2, you create lists with an icon rather than a mass action. Other than that minor
workflow change, this recipe works for releases prior to 9.3. See the Lists topic in the Admin guide for more
information.

To create the list as a mass operation:

1. From the bottom of the page, click all 100,000 and then click Save as list.
2. When prompted, name the list "100k random sample".
3. Navigate to the Lists tab.
4. Click 100k random sample, then click Create Search from List. You should now have the 100k random
sample search under the List Searches folder in your saved search browser.

Relativity | Recipes - 217

54.2.3 Create and run repeated content identification as Structured Analytics
set
To create a Structured Analytics Repeated Content Identification Set:
1. From the Indexing and Analytics tab, select Structured Analytics Set. A list of existing Structured Ana-
lytics sets appears.
2. To create a new set, click New Structured Analytics Set. The Structured Analytics Set layout appears.
3. In the Structured Analytics Set Information layout, set the following:
n Structured analytics set name: Enter a name for your Structured Analytics set, such as
"Repeated Content Identification on sample".
n Select operation: Repeated content identification
n Select document set to analyze: select your saved search, 100k random sample.
n Analytics server selection: select Automatic.
n Enter your email address: enter an email address if you wish to receive a notification when the
email threading analysis completes.
4. In the Repeated Content Identification layout, consider the following:
n You should modify the Repeated content settings slightly to suit your needs. We recommend
the defaults with the exception of the Minimum number of occurrences. If you are using a ran-
dom sample, we recommend you change this setting to 0.004 (i.e. 0.40%) multiplied by the num-
ber of documents you are submitting. For example, with 100,000 documents, set the Minimum
number of occurrences to 400. If you are not using a sample (for instance, if you have a small col-
lection with fewer than 100,000 documents), then we advise setting it to 0.005 times the number
of documents you’re submitting. See Special considerations below for more information.

Note: You may over time decide that you prefer setting this value higher or lower, but the above
recommendation is consistent with our experience with most customers. Contact
[email protected] if you would like to discuss in greater detail.

5. Click Save. The Structured Analytics Set Console appears.

6. To run repeated content identification analysis, click Run Full Analysis. You can monitor the progress
of the operation in a separate window by clicking the export icon in the upper right corner of the pro-
gress pane.

54.2.4 Review results

After the operation completes, review the resultant filters to ensure that they are indeed boilerplate
and not authored content. Accomplish this task by using filters along with the Ready to index field.

54.2.5 Special considerations

As mentioned above, the minimum number of occurrences setting should be configured in
proportion to the number of documents. While 400 is generally appropriate for 100,000 document
samples, larger or smaller sets necessitate proportional modification of that number.
To retrieve more filters, the minimum number of occurrences can be reduced. However, we don't
recommend setting it lower than 100 on a 100,000 document sample, as the results can become

Relativity | Recipes - 218

more subject to sampling error. Consider running the operation across a judgmental sample
instead. For example, just parent emails, or just Word and PDF documents.

55 Searching for custom object information on the

Document object
To search for custom object information on the Document object create a multiple object field that
links both objects together.

55.1 Recipe overview

You can search for the custodians' information from the document object using a custom custodian
object. The workspace should also contain documents associated with certain custodians.

55.2 Directions
To search for custom object information on the Document object, perform the following steps:
1. Create a multiple object that links the Document object to the custom Custodian object. In this
example, Custodian Obj Linker is the name of the multiple object .

2. Set Open to Associations to Yes on the Custodian object fields you want to make searchable on the

Relativity | Recipes - 219

 Document object (in this example, the Start Date and End Date).

When creating a new Saved Search, or toggling search conditions on, you can search across your
Custodian object Start Date and End Date fields. Look for the field name that starts with the name of
your multiple object. In this example, it is Custodian Obj Linker.

You can also filter on the Custodian object Start Date and End Date fields by adding them to a view
on your Document object.

Relativity | Recipes - 220

You can now filter on these date fields on this Document view.

56 Searching for dates in Relativity

This recipe provides steps to search for dates within Relativity using a filter, a saved search, or a
dtSearch.

Relativity | Recipes - 221

56.1 Types of date searches
You can search for dates in Relativity using three different methods:
n Field filters
o Search for single date or single date range
o Requires metadata
n Saved search
o Search for multiple date or multiple date ranges
o Requires metadata
n dtSearch auto-recognize
o Search for dates within document text and in multiple formats
o Does not require metadata

56.2 Directions
56.2.1 Field filters
To search for a date using a field filter, perform the following steps:
1. Ensure the date field you want to search for is in the current view. Add it to the view if necessary.
2. Turn on filters, and then enter a valid search string in the text box filter.
The following table lists examples of valid date and number searches, as well as the expected result
set.
Valid search strings Return items where...
>= 7/24/2008 [FIELD VALUE] >= '7/24/2008'
<= 7/24/2008 [FIELD VALUE] <'7/25/2008'
= 7/24/2008 ([FIELD VALUE] > '7/23/2008') AND ([FIELD VALUE] <'7/25/2008)
>= 07/27/2008 1:23 PM [FIELD VALUE] >= '07/27/2008 1:23 PM'
<= 07/27/2008 1:23 PM [FIELD VALUE] <= '07/27/2008 1:23 PM'
= 07/27/2008 1:23 pm [FIELD VALUE] = '07/27/2008 1:23 PM'
7/24/2008 BETWEEN 8/24/2008 ([FIELD VALUE] >= '7/24/2008') AND ([FIELD VALUE] <= '8/24/2008')
7/24/2008 1:23 PM BETWEEN 8/24/2008 ([FIELD VALUE] >= '7/24/2008 1:23 PM') AND ([FIELD VALUE] <=
3:45 PM '8/24/2008 3:45 PM')
07/27/2008 ([FIELD VALUE] >= '07/27/2008') AND ([FIELD VALUE] < '7/28/2008')
>= 100 [FIELD VALUE] >= '100'
<= 100 [FIELD VALUE] <= '100'
= 100 [FIELD VALUE] = '100'
The following table includes examples of invalid data and number search strings.
Invalid search strings Invalid reason
> 7/24/2008 You must use the equal sign with the greater than operator (as in >=).

Relativity | Recipes - 222

Invalid search strings Invalid reason
< 7/24/2008 You must use the equal sign with the less than operator (as in <=).
>= 0/24/2008 The search string includes the value 0 for the month.
= 0/24/2008 The search string includes the value 0 for the month.
0/24/2008 BETWEEN 8/24/2008 The search string includes the value 0 for the month in the starting date.
7/24/2008 BETWEEN 0/24/2008 The search string includes the value 0 for the month in the ending date.

56.2.2 Saved search

To search for a date in Relativity using a saved search, perform the following steps:
1. Create a saved search.
2. Select the date field you want to search for in the Conditions section.
You can set additional conditions to search for non-consecutive dates, including non-consecutive
date ranges.

56.2.3 Index search

Some cases require you to search for dates contained within the text of a document. Because dates
can appear in various formats, the auto-recognize feature in dtSearch is useful.
To use this feature, build a new dtSearch index with Auto-recognize date, email, and credit card
numbers set to Yes. Once the index is complete and active, you can search for dates within the text
of a document.
The auto-recognize feature searches for strings that appear to be dates. It uses English-language
months, including common abbreviations, and numerical formats. For example, dtSearch recognizes
the following date formats:
n January 15, 2006
n 15 Jan 06
n 2006/01/15
n 1/15/06
n 1-15-06
n The fifteenth of January, two thousand six
It doesn't recognize DD/MM/YYYY format.
Note the following date and date range search strings:

Relativity | Recipes - 223

 n To search for a date, enter a date expression between the parentheses in the string "date()"; for
example, "date(jan 10 2006)"
n To search for range of dates, enter a date range between the parentheses in the string "date()"; for
example, "date(jan 10 2006 to jan 20 2006)"
n To search for a range of dates near the word "apple," enter "date(jan 10 2006 to jan 20 2006) w/10
apple"
n dtSearch doesn't support unterminated date. To search for any date after or before a particular date,
enter a bounded range with a maximum or minimum value for the bounds. The maximum value for a
year is 2900, and the minimum value is 1000. For example, "DateField contains date(jan 10 2006 to jan 1
2900)"

56.3 Notes
n By default, date fields use the filter type of text box. However, you can change this to custom in the
field information.
n If you need more complexity than the text box filter type can provide, you can set up a saved search.

57 Searching for documents with incorrectly loaded

text
If data is loaded with the wrong encoding selected, you must locate those documents and ensure
they load into Relativity properly.

57.1 Recipe overview

This recipe shows you how to search for documents that loaded into Relativity with encoding set
incorrectly.

57.2 Directions
1. Create a new dtSearch index, and then remove all stop words.
2. Create a search terms report with all of the stop words using that index.
3. Create a saved search with the following conditions:
n the STR field is not set AND
n the Extracted Text field is set
The results yield all documents with extracted text, but don't contain any stop words. Documents
that fall into this category are the documents with the wrong encoding.

Note: Documents in other languages may also appear in your results.

Relativity | Recipes - 224

58 Searching for handwritten documents
This recipe provides a workflow that can help identify handwritten documents.

58.1 Overview
Relativity doesn't automatically identify handwritten documents. This recipe demonstrates how you
can use custom search indexes and search terms reports to execute searches that can identify
potential handwritten documents or documents that contain poor quality OCR.

58.2 Requirements
n Workspace access
n Edit and delete permissions on search indexes
n Edit and delete permissions on searches

58.3 Directions
Note: This recipe assumes that the documents contain available text or were previously OCR-ed to capture
all possible text.

To search for handwritten documents, perform the following steps:

1. Create a dtSearch index called dtSearch-NoNoiseWords.
a. Set the Searchable Set to the subset of documents you want to search for handwriting, and
ensure the extracted text or OCR field returns in the search.
b. Copy and paste the list of noise words into a Notepad file, and then remove all the noise words
from this index.

c. Save the index, and then perform a full build.

Relativity | Recipes - 225

 2. Create a search terms report (STR) with the following settings:
n Name- STR-NoiseWord
n Index- dtSearch-NoNoiseWords
n Searchable Set- Same set you previously identified
n Type- Report and Tag
n Calculate Unique Hits- No
b. Click Modify terms, and then enter the list of noise words that you previously copied and pas-
ted.

c. Click Run All Terms.

3. Create a search with the following settings to identify documents that may contain handwritten text
once the STR completes.
a. Name - Potential Handwritten Docs
b. Conditions:
n Field- STR-NoiseWord
n Operator- Not These Conditions
n Value- STR-NoiseWord is Set

Note: Make sure the searchable set for this search is the same as the searchable set used above.

The results of the search narrow down the universe of documents to the documents that potentially
contain handwritten text or poor quality OCR.

Relativity | Recipes - 226

59 Searching for single standalone emails
To assist in organizing document review, this recipe shows you how to separate standalone emails.
Standalone emails are emails with no further conversation other than the original email.

59.1 Requirements
n Structured Analytics - Email Threading
n Saved Search
n Workspace Access
o Threading fields
o Saved Search

59.2 Directions
1. Set up a saved search to return multiple emails with multiple threads.
a. Use the Email Threading ID field. This field indicates how many threads are in a conversation. The
search in the below example asks for items that contain second level emails.
b. Include the Thread group to pull in the rest of the conversation.

2. Create another saved search to return all records where Email Threading is set, but excludes the search
in step 1.

Relativity | Recipes - 227

60 Searching for terms while excluding email footers
While searching for a term, you may receive false positive hits because that term appears in
standard email confidentiality footers. For example, if you search for documents that contain the
term “Confidential,” the term returns extraneous results. These hits can interfere with relevant
results.
With some simple search syntax, you can search for a term, and exclude it only when it appears in a
standard email footer.

60.1 Recipe overview

This recipe shows you how to use the NOT operator with a proximity search to search for a specific
term. By using this search you can exclude hits in a repeated email disclaimer.

60.2 Requirements
n dtSearch index

60.3 Directions
Using your dtSearch index, construct a search for your term using the NOT and w/n operators to
exclude the term only when it's part of a repeated email disclaimer. For example, to find documents
that contain the term “Confidential” use the following syntax:

Relativity | Recipes - 228

 Example 1 for one footer:
Confidential NOT w/1 "The message, together with any attachment, may contain
confidential and privileged information"
Example 2 for more than one footer:
Confidential NOT w/1 "The message, together with any attachment, may contain
confidential and privileged information" or "Privileged and Confidential information
provided in this message is not for use"

Using this syntax returns documents that contain the term “Confidential” and excludes documents
where “Confidential” only appears in the email footer. If a document contains “Confidential” in the
body of the email and in the footer, it returns as a hit.
Your proximity is equal to the number of words that appear between the first word in the disclaimer
and your search term. In this example, “Confidential” is one word from the start of the email footer
string.

61 Searching for unique identifiers

You can run searches for a list of unique identifiers, like control numbers or Bates numbers, within
Relativity.

61.1 Overview
This recipe shows you how to set up custom search indexes and then use simple search operators to
identify a list of documents.

61.2 Requirements
n Workspace access
o Edit and delete permissions on search indexes
o Edit and delete permissions on searches

61.3 Directions
This recipe assumes that the documents you're attempting to identify contain a unique identifier in
a fixed-length text or long text field.
The examples in this recipe show you how to search for the control number of the document.
However, you can use the same workflow process for other unique identifiers like Bates numbers.

Relativity | Recipes - 229

61.3.1 Option A
Use this option when you have fewer than a few hundred documents to identify.
Create a search with the following conditions to identify documents with the control number:
n Name- Control Number List
n Conditions
o Field- Control Number
o Operator- Is
o Value
l ABC001
l ABC002
l ABC003

Ensure that you separate the unique identifier values by a line break. This may require you to
reformat the string of control numbers using Notepad++ or Textpad.

61.3.2 Option B
Use this option when you have more than a few hundred documents to identify.
1. Create a saved search called ControlNumber - for Index with the subset of documents you want to
search against.
2. Ensure that the only field the saved search returns is the Control Number.
3. Create a new dtSearch index called dtSearch-ControlNumber.
4. Perform a full build on this new index using the saved search from step 1.
5. Create a search with the following settings to identify documents with the control number once the
index is complete:
n Name- ControlNumber List
n Search Index
n Search With- dtSearch-ControlNumber
n Search String- ABC001 OR ABC002 OR ABC003

Ensure that you separate the unique identifiers with an “OR” operator. This may require you to
reformat the string of control numbers using Notepad++ or Textpad).
The results of the search help identify the documents by their control numbers.

Relativity | Recipes - 230

You can also paste this list of identifiers into a search terms report, which provides an OR search for
all of the identifiers. Select the Report and Tag option to write this information to a field.
The following table helps you determine the use case for each workflow option.

Option A Option B
Ideal when searching for fewer than a few hundred Ideal when searching for more than a few hundred
documents. documents.
Keyword indexes any length of text, but limits the search dtSearch doesn't index (by design) control
for control numbers to 4000 characters in length (overall numbers longer than 32 characters.
length of all characters you search).
Ensure you use the IS operator when searching. You can If your control number contains periods (.) or
also use the CONTAINS operator if, for example, you are hyphens (-), you must index those characters prior
trying to bring in all family members that begin with the to building the index, or enclose the control
same prefix. number in quotes.

62 Searching workflows
Relativity provides great flexibility to help you identify documents through searches so that you can
ready them for further review and analysis. This recipe highlights setting up multiple search
indexes, applying various search techniques, and using filters properly when executing searches.

62.1 Requirements
n Workspace access
o Edit, Delete access on Search Indexes
o Edit, Delete access on Searches

62.2 Directions
Relativity offers two types of search indexes, Basic Keyword Search and Specialized dtSearch.
The Basic Keyword Search Index populates automatically, is available out of the box, and supports
basic Boolean operations. These operations include AND, OR, NOT, and the wildcard (*) operator.
The Specialized dtSearch Index is custom built and must be set up with an index to query against. In
addition to basic Boolean operations, it also supports the following functionality:
n Wildcard operator for a single digit (=)
n Stemming operations (~)
n Fuzzy search operation (%)
n Range Searching on Numeric fields (~~)
n Customize data set to be indexed via custom saved search
n Proximity Searching

Relativity | Recipes - 231

 n Customized Noise Word List
n Auto Recognition of email addresses, credit card numbers, and dates.

The following includes best practices for the use of advanced operators and workflow options:

62.2.1 Proximity search

Proximity search uses operators to search certain terms in proximity to other terms in a document.
Use the following recommendations when employing proximity searching:
n To use directional proximity searching use "pre /x " operators. To use non-directional proximity search-
ing use "w /x" operators.
n To determine the beginning and end of a document, use reserved word with either ("xfirstword") or
("xlastword"). Use these operators to search for metadata like email addresses or footers within a doc-
ument.
n To determine proximity. The distance between terms is important so you know whether you receive
true hits or false positives. The following image illustrates the proper syntax for proximity searches:

The following image shows all correct and incorrect combinations of proximity searches that result
in successful hits when using proximity operators:

Relativity | Recipes - 232

Sometimes additional factors affect the distance between terms. The following image demonstrates
a search for a string of words in proximity to another word. The search in the following image takes
the following into account:
n Noise words count as words when calculating proximity.
n Punctuation counts as whitespace when using default settings.
n Relativity treats line breaks and consecutive space characters as single spaces.
n dtSearch default stop words and connector words like AND, OR, BETWEEN, and NOT count as words
when calculating proximity.

In this case, we continue to calculate the distance of the string ("confidential and/or privileged
information:) from the word ("message") as shown above. However, when using connector words,
system admins should create an index that removes stop words from the noise word list.

Relativity reserves the following noise words and characters, which continue to behave as operators,
as well as being noise words: and, or, not, to, contains, xfirstword, xlastword, ", ( ), *, ?, %, @, ~, #,
&, :, =.
Once that index is available for query you can either place the search string in quotes or apply
stemming to the connector words to override their function as a connector.

Relativity | Recipes - 233

62.2.2 Auto-Recognition
Auto-Recognition identifies email addresses, dates, and credit card numbers in the data set you
want to index. You can turn this feature on and off when you build your index. The search returns
the results regardless of the data format. Auto-Recognition adds some time to your index build, but,
depending on your case, the benefits can be significant.

62.2.3 Searching for times

If you want to search for times in the body of documents, perform a full-text search with your
dtSearch index for a specific time. Keep in mind that some characters cause a word break, such as
the colon and period. Searching for 12:15 p.m. results in searching for four words: 12, 15, p, and m.

62.2.4 Filters
In addition to the Search Indexes, you can also use filters to search on metadata fields and narrow
down the review set of documents.

The following list includes metadata fields, their corresponding field types, and the filter type
available in Relativity:
n File Type - Single-Choice, List Filter
n Custodian - Single-Choice, Pop-Up Filter
n Date Sent - Date, Textbox Filter
n Email To - Long Text, Textbox Filter
n Email Subject - Fixed Length Text, Custom Filter
n Designation - Single Choice, Multi-Choice List
The following table shows different types of filters you can set up with available fields in Relativity.

Relativity | Recipes - 234

62.2.5 Troubleshooting workflow
To troubleshoot and test searches, use a white board approach to map out searches before you run
them. As part of the process, try to take into account all possible variations, so you can see which
documents Relativity returns and to gain a better understanding of the search. For instance, if you
want to find email family groups with inconsistent coding, numerous possibilities for searches exist.
The following illustration shows a mapping of nine email family groups, each with an email and two
attachments.
Construct searches (1 and 2) and mark the corresponding documents, and their family members,
with hits for each of the searches. At this point, when you look at the board, you can see that the
four email family groups have one thing in common. The documents are responsive to both
searches. So, to close it out, construct a third search that pulls back documents common to both
searches.

Relativity | Recipes - 235

63 Setting up in an upgraded 7.5 environment
To get the most out of Relativity for Relativity 7.5, there are a few instructions to follow as you
transition to the new application after upgrading your environment.

63.1 Recipe overview

This recipe provides considerations to keep in mind as you take advantage of for Relativity 7.5 in
your pre-existing workspaces.

63.2 Requirements
n Relativity 7.5
oUpgraded Relativity environment from an earlier version
n Workspace access with the following roles:
o System Administrator
o Script Admin

63.3 Directions
1. Make sure the Distributive Transaction Coordinator (DTC) is turned on. See the Upgrade Guide for
details.

Relativity | Recipes - 236

 2. Once the Relativity environment has been upgraded:
a. On existing workspaces with an earlier version of ):
i.  Rename the tab for this application " 7.4,” to make it easily identifiable.
ii. projects progressing on the earlier version of can continue as usual. The earlier version of
is not affected by the environment upgrade.
iii.  When you start a new project in an existing workspace, install the application for Relativ-
ity 7.5.
b. On existing workspaces (without any application):
nWhen you start a new project in an existing workspace, install the application for Relativ-
ity 7.5.
c. On New Workspaces (setup in Relativity v7.5):
n When you start a new project in an existing workspace, install the application for Relativ-
ity 7.5.
3. Repeat steps 2a-2c for each workspace in your environment.

64 Setting up CJK document workspaces in Relativity

When working with documents that contain Asian character sets, specific settings help expedite your
setup time in Relativity. This information can help you build a clean workspace with more accurate
fields to support your Asian language document sets.

64.1 Recipe overview

This recipe highlights items to be aware of as you work with workspaces that contain CJK character
sets.

64.2 Requirements
n Workspace access with the following permissions:
o Field: edit/delete
o Searches: edit/delete
n Relativity Desktop Client

Relativity | Recipes - 237

64.3 Directions
1. In your workspace, set the Unicode field to Yes (this is the default in Relativity 7.5 and higher).

2. Set up fields such as Custodian, email fields, File Name, and Extracted Text as Unicode-compliant to cap-
ture CJK characters.
3. Your processing vendor should have provided you with data files that are Unicode/CJK character-com-
pliant. If this is the case, import the data in a Unicode-compliant format. Be sure to include any extrac-
ted text.

Relativity | Recipes - 238

 Setting up the fields as Unicode-compliant, and leaving the default settings in the dtSearch index
(dtSearch index alphabet list: CJKRanges = 2e80-ac00 ac00-d7af f900-faff fe30-fe4f), allows you to index
characters in CJK languages where each character is treated as a separate word.

64.3.1 Searching considerations

Consider the following with SQL, dtSearch, and CJK documents. Essentially each character is its own
word or token so proximity searching works.

64.3.1.1 Keyword Search

n Within a Relativity workspace, a system admin can select the language to use in the SQL Full Text Lan-
guage. SQL Full Text Language determines the correct stemming and word-break characters used in the
full text index.

n Once you configure your SQL full-text language to the correct language (Japanese in this example), you
can perform keyword searching and filtering in that language.

64.3.1.2 dtSearch
Index
Once you set the fields to be Unicode compliant, leaving the defaults in the dtSearch index setup
(dtSearch index alphabet list: CJKRanges = 2e80-ac00 ac00-d7af f900-faff fe30-fe4f) allows you to
index characters in CJK languages where each character is treated as a separate word, enabling
word breaking with CJK.

Relativity | Recipes - 239

Word/Character Search
As explained previously, you can store or convert Chinese, Japanese, and Korean text to Unicode so
that you can use dtSearch to search for words in these languages just as you search for words in
other languages. However, while dtSearch can search for literal word matches (or wildcard or fuzzy
matches), there are some limitations on the support in dtSearch for Chinese, Japanese, and Korean
text. Those limitations include:
n Some documents store text in a way that does not separate the words with spaces. Instead, all of the
text in a document runs together, and a language-specific dictionary is needed to find word breaks.
dtSearch does not have the ability to identify word breaks in these documents because it doesn't
include any language-specific dictionaries. To make this type of text searchable, enable an option in
dtSearch to automatically insert word breaks around Chinese, Japanese, and Korean characters. Once
you enable this option, Relativity treats each character as a single word for indexing and searching pur-
poses.
n This feature is turned on by default in Relativity dtSearch. Because each term is searchable, we recom-
mend searching for multiple characters to assist in retrieving more accurate results. When looking for
multiple characters, use the proximity connectors to assist in finding desired results.
n The same text can be presented in different ways depending on the context. dtSearch searches for a
word as it is provided in the search request and does not generate additional grammatical or script vari-
ations for words in Chinese, Japanese, and Korean.
n The dtSearch Engine has an API that you can use to integrate with dictionary-based language analyzers
from companies such as Basis Technologies. But while the dtSearch standalone desktop does allow for
integration, the instance within the Relativity environment does not support integration.
For non-Western languages (such as Chinese, Japanese, Hindi, and so forth) there are additional
considerations and workarounds that may provide assistance in locating search hits.

64.3.1.3 Mandarin considerations

n Mandarin characters are treated as words. Therefore the Mandarin characters do not need extra space
because each character can be compared against a list. If there is a match, then it's highlighted.
n Numbers are not treated like Mandarin characters. For example, when you use the number 54, you
need everything until there is a space. Add a space to the subject so the term is 54 所[space], then the
rest of the line, it will highlight.

n The term 54 所 does not match the subject. The subject has the additional characters, hence no
match.

Relativity | Recipes - 240

 n The Mandarin characters do not need the space because each character, such as 所, is treated
uniquely, therefore, it highlights.
n Keyword Search – Once you configure your SQL full-text language to the correct language (Japanese in
this example), you can perform keyword searching and filtering in that language.
n dtSearch – Once you set up the fields to be Unicode compliant, leaving the defaults in the dtSearch
Index setup (dtSearch index alphabet list: CJKRanges = 2e80-ac00 ac00-d7af f900-faff fe30-fe4f), you can
index characters in CJK languages where each character is treated as a separate word, enabling word
breaking with CJK.

64.3.2 Analytics considerations

n Use the structured data analytics language identification operation (or PLI language technique) to
identify the languages of the documents in your workspace.
n Training your Analytics index in one language as opposed to multiple languages generally produces a
better quality index. However, if many of your documents contain multiple languages, this can change.
n Be sure to set up your stop words in the same language as your index.

Note: Refer to the Using Analytics with CJK characters recipe for more detailed considerations.

65 Sorting documents and keeping email families

together
You may want to sort your document collection by date; however, it's difficult to keep emails and
their attachments together. This is because family items can contain different dates or date values
in different fields. To sort documents chronologically and keep email families together, you need a
date field that is consistent for emails and their attachments.

65.1 Recipe overview

This recipe shows you how to create an additional date field. It also shows you how to populate it
with the sent date from the email to the attachments using a script. This creates a consistent date
field for emails and their attachments, which makes the sorting process easier.

65.2 Requirements
n Relational field identifying email families.
n Relativity security access
o Field- edit/Add
o Document- edit
o Permissions- access to Scripts tab
o Groups- system Admin and Script Admin (if you didn't add the script to the workspace.)

Relativity | Recipes - 241

65.3 Directions
1. Create a Date field to hold the date information. Call this field Parent Date, Family Date, etc.
2. Create a Yes/No field to flag documents when the script affects them. If you add additional data to the
workspace at a later date, and need to run the script again, this flag ensures the script runs only on the
newly added documents. Call this field Date Propagated.
3. Run the Propagate Sent Date to Family Documents script. If you don't see this script listed under the
Scripts tab, add it to the workspace from the Relativity Script Library. Once you run this script, Relativ-
ity presents you with the following inputs:
a. Date Field – Original- enter the field from which you want the date propagated, such as Sent
Date.
b. Date Field – Sort- enter the field to which you want the date propagated (the field you created in
step 1).
c. Family Identifier- enter the relational field that defines the email families.
d. Date Propagated - enter the Yes/No field you created in step 2.
e. Document Sort Field - select the field for which to order documents by instead of using Artifact
ID to order documents in each family. This accounts for some document families where the low-
est Artifact ID isn't the parent.
The script takes the date value from the field you select in step 3(a) from the email, and writes it to
the field you select in step 3(b) for the email and attachments. Once the script completes, you can
use the field you create in step 1 to sort your documents and keep email family groups together.

66 Starting a RAR project after review has begun

While working in Relativity, you may need to start a RAR project after review has already begun on
the documents in your workspace. This recipe explains how to use Assisted Review on an existing
data set that has been partly reviewed.

66.1 Requirements
n An active Analytics index
n Assisted Review Application

66.2 Directions
This recipe assumes that the following:
n The Assisted Review application is installed.
n An optimized Analytics index has been built and includes the project documents.
n The project is a good fit per the RAR Checklist.
n Text for the documents is available or the documents have been OCR-ed to capture all possible text.
n Documents/a subset of documents have been reviewed.
Once you've established these prerequisites, perform the following steps to integrate an Assisted
Review workflow midway into review:

Relativity | Recipes - 242

 1. Create an Assisted Review project. Keep in mind the “Documents to be categorized” saved search must
include all the documents on which you'd like the engine to make its coding decisions.
2. Create a saved search to identify the documents that have been already reviewed and coded.
3. Identify the documents from step 2 that were coded per the Reviewer Protocol specifications to ensure
they are good training examples. Once identified, tag the documents on the following fields, both of
which you specified when you set up the project in step 1:
a. RAR Designation field
b. RAR Use as Example field
4. Create a saved search for all documents tagged on the RAR Designation field that were good examples
(from step 3).
5. Create a new round in the Assisted Review project for pre-coded seeds to include the documents that
have already been coded during manual review that you identified as good examples (use the saved
search created in step 4).

Note: The saved search must exclude documents marked No for use as example, as RAR uses all coded
documents in a pre-coded seed saved search as training examples.

6. Proceed with the regular RAR workflow. Finish the pre-coded seed round and proceed to subsequent
rounds (training or QC rounds per the needs of your project / reports).

66.3 References
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics
n Relativity Assisted Review
n Analytics - RelativityOne
n Relativity Assisted Review - RelativityOne

67 Tally all email recipients

This recipe describes a workflow to count the total number of recipients to an email, based on
recipient metadata.

67.1 Recipe overview

To achieve this workflow, the user will consolidate the Email To, Email CC, and Email BCC fields into
one field, separated by semicolons. Then, the user will run the Delimiter Count By Saved Search

Relativity | Recipes - 243

script to count the number of semicolons in the consolidated field.

67.2 Requirements
n Correctly populated email metadata fields
n The Delimiter Count by Saved Search scrip from NSerio, located on their website

67.3 Directions
To tally all email recipients, use the following steps:
1. Create a field to store the consolidated values of Email To, Email CC, and Email BCC.
1. Field Type: Long Text
2. Field Name: Email All Recipients
2. Create a field to store the number of delimeters.
1. Field Type: Whole Number
2. Field Name: Email Recipient Count
3. Create a saved search of all the documents for which you'd like to tally the number of recipients.
1. (Optional) You can pull these results in a saved search by marking the Email To, Email CC, or
Email BCC as set.
1. Return these fields:
n Email All Recipients
n Email Recipient Count
n Email To
n Email CC
n Email BCC
4. Using the filter pane, filter on the Email To field for Is Set.
5. Use the Replace mass action on all documents returned by the filter.
1. Field to Update: Email All Recipients
2. Action: Replace Entire Field
3. Update With: Field
4. Field: Email To
6. Use the Replace mass action on all documents returned by the same filter as step 4.
1. Field to Update: Email All Recipients
2. Action: Append to End
3. Update With: Text
4. Text: ; (semicolon, no spaces)
7. Clear your previous filter.
8. Using the filter pane, filter on the Email CC field for Is Set.
9. Use the Replace mass action on all documents returned by the filter.
1. Field to Update: Email All Recipients
2. Action: Append to End
3. Update With: Field
4. Field: Email CC

Relativity | Recipes - 244

 10. Use the Replace mass action on all documents returned by the same filter as step 4.
1. Field to Update: Email All Recipients
2. Action: Append to End
3. Update With: Text
4. Text: ; (semicolon, no spaces)
11. Clear your previous filter.
12. Using the filter pane, filter on the Email BCC field for Is Set.
13. Use the Replace mass action on all documents returned by the filter.
1. Field to Update: Email All Recipients
2. Action: Append to End
3. Update With: Field
4. Field: Email BCC
14. Use the Replace mass action on all documents returned by the same filter as step 4.
1. Field to Update: Email All Recipients
2. Action: Append to End
3. Update With: Text
4. Text: ; (semicolon, no spaces)
15. Install the Delimiter Count by Saved Search script from NSerio, located on their website. Directions and
documentation about the script are available from NSerio.
16. Run the Delimiter Count by Saved Search script with the following settings:
1. Saved Search: Search created in step 3.
2. Source Field: Email All Recipients
3. Field Delimiter: ; (semicolon, no spaces)
4. Count Destination Field: Email Recipient Count
17. When the script completes, it updates the Email Recipient Count field for all documents returned in
your search to reflect the total number of recipients across their Email To, Email CC, and Email BCC
fields.

67.4 References
n Fields
n Mass replace
n Saved search
n Scripts
n Fields
n Mass replace
n Saved search
n Scripts
n Fields
n Mass replace
n Saved search
n Scripts
n Fields
n Mass Replace

Relativity | Recipes - 245

 n Saved search
n Scripts

68 Tracking categorized documents to their example

document
During a categorization project, it's sometimes important to know which example document was
used by the Relativity Analytics index to categorize the rest of your documents. This information can
help you improve your example set and yield a more accurate categorization set.

68.1 Recipe overview

This recipe shows you how to create a tab containing a view that displays both your categorized
documents and the example document used by the Analytics index to make its categorization
decision.

68.2 Requirements
n Applicable to all Relativity versions
n Relativity security access
o Permissions: Access to Tabs option under the Administration tab
o Tab: Edit/Add
o View: Edit

68.3 Directions
1. Your categorization set must have a Maximum Categories per Document value greater than 1.
2. Under Tabs, create a new tab to display the Analytics categorization results.
a. Name the new tab and select an Order.
b. Set Link Type to Object.
c. In the Parent field, you may wish to place the new tab under the Analytics tab.
d. Set Object Type to Analytics Categorization Result.
e. Leave Is Default as No.
3. Navigate to the newly created tab.
4. Create a new view with the object type of Analytics Categorization Result.
5. Ensure the following fields are added to the view:
a. Document – [Name of Categorization Set]
b. Category
c. Category Example
These fields show you the document that was categorized, the name of the category, and the
example document that was used by the Analytics index to make its categorization decision.

Relativity | Recipes - 246

69 Transfer tags from other document review
applications
Migrating data from other document review applications to Relativity is a straightforward process
through the exporting and importing of load files. However, some preparation is necessary to
transfer tag information from the applications.

69.1 Recipe overview

This recipe describes how to transfer tags to the field tree in your Relativity workspace.

69.2 Directions
1. In the application database, write the tag(s) information to a field. Use a semicolon to separate the indi-
vidual tag or folder naming.
2. Export the data from the application, using default delimiters. Ensure that at least the Document Iden-
tifier and Tag fields are in the export.
3. Create a field in your Relativity workspace to hold the tag information.
a. Set the Field type to Multiple choice.
b. Ensure the field is available in the field tree.
4. Import the tag data into the field created in step 3. Ensure that the multi-value delimiter is set to semi-
colon in the Relativity Desktop Client.
5. Navigate to the field tree and observe your tags.

70 Threading review setup

This recipe suggests some ideal views to make available to your reviewers to assist with the
document review process.

70.1 Requirements
n Workspace access
n Relativity Structured Analytics application

70.2 Directions
The following sections contain sample views that you can create along with workflow use cases for
each. The use cases provide details about the results of the Email Threading Structured Analytics
sets. These views can be very useful for your reviewers. See the table that appears after the
following views for an explanation of each of the fields.

Relativity | Recipes - 247

70.2.1 Sample view for Inclusive email with family
You can use this view to identify all the inclusive emails that need to be reviewed. This view is
particularly useful in the following situations:
n You're running threading analysis on a Custodial level.
n Duplicate emails matter from a production perspective.

This view provides context for review by displaying family members like attachments to an email
when going doc to doc.
In conjunction with this view, the Related Items pane for a Thread Group view can help reviewers
find context for the email that is being reviewed.

Relativity | Recipes - 248

70.2.2 Sample view for Inclusive and non-Duplicate email with family
You can modify the conditions on the preceding view to return only inclusive, non-duplicate emails,
plus family. You can use this view to identify all the inclusive emails that need to be reviewed. This
view is useful when you're running threading analysis across the entire dataset, and the case team
has agreed to produce just the inclusive non-duplicate emails.
This view provides context for reviewing by displaying family members like attachments to an email
when going doc to doc.
In conjunction with this view, the Related Items pane for a Thread Group view can help reviewers
find context for the email that they're reviewing.

70.2.3 Sample view of the Related Items pane for an Email Thread Group
This default view is available in the Related Items pane. It allows reviewers to identify the context of
an email that is being reviewed. You can use this view with both sample views identified previously.

Relativity | Recipes - 249

70.2.4 Sample view for Near Dupe Documents
This view displays all near duplicate documents that are grouped together and identifies the
percentage of similarity to the Principal documents.

Note: From a review standpoint, it's important for the reviewer/case team to be aware that the document
with the most amount of text is identified as the Principal document.

Relativity | Recipes - 250

70.2.5 Sample view of the Related Items pane for a Near Dupe Group
This default view is available in the Related Items pane. It allows reviewers to identify the context of
an email that is being reviewed. You can use this view in conjunction with the Sample view for Near
Dupe Documents.

Label Field(s) / Logical Description

Groups
A Control Number Unique Identifier for the document
B Email Threading Dis- Custom (system) field created following Email Thread analysis that provides a
play visual representation of the Indentation of the email (Label-F), Subject Line, and
Attachments to the email.
The indentation bubbles for Email threading display have changed slightly in the
Relativity 9.4 October product update, please refer to the Email Threading Display
section in to see how this will display for your version of 9.4.
C Email Thread Custom (system) field created following Email Thread analysis that maintains the
Group relationship of all the emails in an email thread.
D Inclusive Email / Custom (system) field created following Email Thread analysis that identifies if the
Inclusive Reason email is Inclusive and the reason for inclusiveness (typically tied to contents of the
message and/or presence of an attachment)
n Inclusive email = Yes: The email contains all the contents of the prior
emails.
n Inclusive Reason = Message: The contents of the email classify the email as
Inclusive.
n Inclusive Reason = Attachment:  The email contains an attachment that
needs to be reviewed.

Relativity | Recipes - 251

Label Field(s) / Logical Description
Groups
E Email Duplicate n Custom (system) field created following Email Thread analysis that iden-
Spare / Email tifies whether the document is a duplicate email in the collection analyzed
Duplicate ID and provides a unique identifier to that duplicate email.
n Email Duplicate Spare = No: Not a Duplicate email

F Indentation This Custom (system) field created following Email Thread analysis defines the
depth of the email in the email chain. The example here represents that the email
was third in the chain.
G N/A An example email along with associated attachment.
H N/A An example Inclusive email (based on the message contents).
I N/A An example Inclusive, but duplicate email (based on the message contents).
J N/A An example Inclusive email (based on the message contents & presence of an
attachment).
K Textual Near Dupe Custom (system) field populated following Near Dupe analysis that maintains the
Group relationship of textually near duplicate documents.
L Textual Near Dupe Custom (system) field created following Near Dupe analysis that identifies
Similarity / Textual whether the Principal document (the one with the most text) and the Similarity
Near Duplicate Prin- score provides the score of all the associated Near Duplicate documents in that
cipal Near Dupe Group.
M Email Threading Icon available in the Related Items pane in the document viewer that, when
Relational Field clicked, identifies all the documents related to that Email Thread group.
Icon
N Near Dupe Rela- Icon available in the Related Items pane in the document viewer that, when
tional Field Icon clicked, identifies all the documents related to that Near Dupe group.
O Relativity Compare This text comparison icon, when clicked, allows you to view the differences in text
Icon between two documents (useful for comparing near dupe documents.)

71 Understanding the statistical sampling formula

Relativity uses human decisions on a sample set of documents to learn the coding parameters of
your review project. One way to select your sample set is to pull a random group of documents from
your data set. When this option is selected, calculates the number of documents to pull with a
statistically valid formula.

71.1 Recipe overview

This recipe breaks down the formula used to calculate the number of documents in a random
sample set. You can reference this formula to ensure your random sample set is statistically valid.

Relativity | Recipes - 252

71.2 Formula breakdown
The following formula uses the total number of documents as well as the confidence level and
margin of error you select for your project to determine a statistically valid size for your sample set.

n Z = Z value, which is based on the confidence level (e.g., 1.96 for 95% confidence level)
n p = reflects the chance of a defect (i.e., overturn), expressed as a decimal; Assisted Review con-
servatively sets this at .5
n c = confidence interval — or margin of error — expressed as a decimal (e.g., .04 = ± 4%)

72 Use case for Relativity Analytics - Clustering &

Categorization
Published September 27, 2013
Relativity Analytics features — including clustering and categorization — allow users to expedite a
review by organizing, identifying, and prioritizing documents. This recipe highlights techniques for
expediting review using Relativity Analytics, as well as tips for utilizing the Analytics index to identify
other hot documents.

72.1 Common scenarios

Common scenarios for using clustering and categorization include the following:
n You are working with a large number of documents. This data may not be coded at all, or it may con-
tain documents that were coded with a multiple choice field for issues.
n You have limited time to complete the review.
n You have a limited number of reviewers.

72.2 Requirements
n Relativity Analytics index
n Workspace access
o Document – Edit

Relativity | Recipes - 253

72.3 Directions
72.3.1 Scenario 1
You have no prior knowledge of the data set. Clustering in Relativity groups conceptually similar
documents without the need for example documents or user input.
Perform the following steps:
1. Use clustering to automatically organize documents into groups of related data.
2. Batch and assign documents to reviewers based on the clusters.
3. Bulk review groups of clustered documents.
Once the above operation is complete, you may find clusters of documents that are clearly
irrelevant to your case, such as spam emails. Instead of reviewing hundreds or even thousands of
junk emails one at a time, reviewers and system admins can eliminate impertinent documents with
minimal time, effort, or subject matter expertise.

72.3.2 Scenario 2
You need to find key documents in an opposing production. Relativity’s categorization functionality
identifies and groups similar documents together based on a set of example documents manually
identified by the user. When you receive documents in an opposing production, and hot documents
have been identified in your documents, you can pass these document values to the new production
items by making them examples for the Analytics engine.
Perform the following steps:
1. Create a categorization set and use the Issue Designation multiple choice field as the Categories and
Examples source.
2. Use the Synchronize feature for your new categorization set.
3. Click Categorize All to categorize the opposing counsel's documents with your categorization set and
to identify opposing counsel's documents which are similar to your hot documents.
Setting the available Categories and Examples source option to use your multiple choice designation
field enables the Synchronize feature for categorization. The Synchronize feature automatically
creates categories for all choices associated with the specified field and designates example records
for all documents with this field coded. With the example document records identified in your data
set, categorization identifies and organizes similar documents in the opposing counsel's data set.

73 Using Analytics with CJK characters and non-English

languages
Published April 28, 2014

Relativity | Recipes - 254

Relativity Analytics works with Chinese, Japanese, and Korean characters. The system uses a process
called tokenization to break a stream of text into a list of tokens, which are words, phrases, symbols,
or other meaningful elements distinguishable within the text. The list then becomes input for further
processing, such as searching.
This recipe describes how Relativity Analytics handles CJK languages and provides some special
considerations for leveraging text analytics in a workspace with CJK documents.

Note: Note: This document draws from various Content Analyst resources with their permission.

73.1 Considerations for documents with CJK characters

Consider the following when working with documents that contain CJK characters.

73.1.1 Tokenization method

CJK languages use long strings of characters, and what constitutes a word can be context-specific
and subject to debate. Given that words (or "features") are one of the primary building components
behind CAAT's concept indexes, a process for splitting CJK text into words is crucial.
To support this functionality, a number of algorithms run during document ingestion and filtering
which split the CJK text into words. Once these algorithms run, you can build high quality indexes in
these languages. This saves time and money by dramatically reducing the effort and expense of
reviewing and understanding unstructured data.

73.1.2 Workflow recommendations

When building Analytics indexes for documents with multiple languages, we recommend that you
create a separate index for each language. Consider a case with a small set of Chinese documents
mingled with English documents. If you build one index for the entire population, searches against
the Chinese documents will always return the other Chinese documents. That’s because, regardless
of whether the content is truly similar, those documents have the most characters in common.
Building separate language indexes improves precision because the engine can focus on the words
for conceptual similarity rather than just linguistic characters.
Published January 19, 2016
Relativity Analytics works with Chinese, Japanese, and Korean characters (otherwise known as CJK
characters in relation to computers). CJK languages have writing systems based entirely or partly on
Chinese characters. These languages have some special characteristics that need to be handled
properly in regards to computing and searching methodologies. Some of these characteristics are
also common in other languages, such as Thai or Arabic.
The characteristics include:
n Non-delimiting of words with spaces
n Lack of word wrapping and line breaking

Relativity | Recipes - 255

 n Existence of thousands of complex characters
n Existence of multiple, conflicting versions of character sets
n Ability to edit characters by authors
n Horizontal and vertical layout frequently used
The system uses a process called tokenization to break a stream of text into a list of tokens, which
are words, phrases, symbols, or other meaningful elements distinguishable within the text. The list
then becomes input for further processing, such as searching.

73.2 Requirements
n Relativity 9.0 or higher
n Workspace access and system admin rights

73.3 Recipe overview

This recipe describes how Relativity Analytics handles CJK languages and provides some special
considerations for leveraging text Analytics in a workspace with CJK documents to create Analytics
indexes and to take advantage of Structured Analytics features.

Note: This document draws from various Content Analyst (CAAT®) resources with their permission.

73.4 Considerations for using Analytics with CJK characters

The Analytics engine is powered by individual terms, rather than sentences or phrases. Therefore,
challenges with various sentence structures are not a concern with the Analytics engine. Regardless
of the language, the text must be split into words (i.e., tokenized) in order to be indexed by
Relativity Analytics. CJK languages use long strings of characters, and what constitutes a word can
be context-specific and subject to debate. Given that words (or "features") are one of the primary
building components behind CAAT's concept indexes, a process for splitting CJK text into words is
crucial. To support this functionality, a number of algorithms run during document ingestion and
filtering which split the CJK text into words. Once these algorithms run, you can build high quality
indexes in these languages. This saves time and money by dramatically reducing the effort and
expense of reviewing and understanding unstructured data.
Relativity Analytics uses the CAAT® software for conceptual indexing. CAAT® uses the
getWordInstance method of the Java breakiterator class to handle all language tokenization, except
Chinese. The globalization libraries are provided through the International Components for Unicode
(ICU). Services provided by ICU are widely used by companies like Google, Apple, IBM, and others.

Note: For more information on the ICU, see http://site.icu-project.org/.

There are two approaches to word-breaking (tokenization) for all Java languages (other than
Chinese):
1. Rule-based Iteration: In this approach, spaces and punctuation define word boundaries.

Relativity | Recipes - 256

 2. Dictionary-based Iteration: In this approach, there are ambiguous word boundaries (no space sep-
aration, spaces in the middle of words, long compound words, etc.). This approach is used primarily for
supporting non-Latin based languages. Java only applies the dictionary-based approach when neces-
sary.
Chinese poses some unique challenges for Analytics. First of all, almost all characters can be one-
character words by themselves. Moreover, they can join other characters to form multi-character
words. This leads to a large amount of segmentation ambiguities. Second, compounding is the
predominant word-formation device in modern Chinese. It is often difficult to tell whether a low-
frequency compound is a word or phrase, and the lexicon can never exhaustively collect all low-
frequency compounds. Third, the same pool of characters is also used in constructing proper names.
Identifying proper names will be a problem, too. Finally, some specific morphological structures
such as reduplication and A-not-A construction also need to be taken into consideration.
To handle these characteristics in Chinese, a different approach is taken. CAAT® applies a Maximal
Matching algorithm in order to properly tokenize Chinese. This dictionary-based word segmentation
alternative approach is taken due to previous limitations with the Java class.

73.4.1 Considerations for creating Analytics indexes containing non-English

languages
When creating an Analytics index, it is recommended, if possible, that the index only contains one
language in its documents. Training the index in one language as opposed to multiple languages
produces a better quality index. Additionally, if the data set is overwhelmingly one language (e.g.,
English) with only a small amount of another language (e.g., French), then the French documents
would likely be outweighed by the English documents. The conceptual analysis of the French
documents would not be as deep, and they might all be clustered together due to sharing so many
common terms that are not found in the English documents.
Typically, there is a language that is used for the majority of documents in a workspace (i.e., the
Primary language). Other languages are referred to as secondary languages. If a secondary
language is less than 1K documents, an Analytics index might not be feasible for such a small data
set. In this case, you may either include the secondary language documents in the index, or instead
use Structured Analytics to analyze these documents.
If the index contains many multi-language documents, it may be impossible to split the languages. In
this case, consider the following Analytics index settings below, especially Concept Stop Words, to
ensure that it is a good quality index. You may apply stop words from both languages to this index.

73.4.1.1 Analytics index settings

Supports non-English English support only
Concept stop words Email header filter
Regular expression filter Automatically remove signatures and footers
Repeated content filter OCR filter
Optimize training set Go Words filter

Relativity | Recipes - 257

73.4.1.2 Concept stop words considerations
Concept stop words are very common terms that are filtered from the Analytics index in order to
improve quality. By default, the Analytics profile contains a Concept Stop Words field of only English
stop words. These are customizable by the user in Relativity. The English stop words should be
replaced by the appropriate stop words for the language being indexed. Analytics profiles are
copied over when a workspace is used as a template; it is recommended that you create multiple
profiles in the workspace template for each commonly indexed language.
kCura does not publish stop word lists. The following are three places where you can find stop word
lists:
n Ranks NL (http://www.ranks.nl/stopwords) - 40 languages are available.
n Microsoft SQL Server Stop Lists - 33 languages are available. Perform a query as follows:

SELECT LCID, Name FROM sys.syslanguages

SELECT * FROM sys.fulltext_system_stopwords WHERE language_id = ####

n Take the default English stop words and use a translator tool to translate to the desired language.

73.4.1.3 Regular expression filter considerations

The Regular Expression filtering allows for a high level of customization. Creating these filters
requires a working knowledge of regular expression methodology. This filter uses the regular
expression syntax supported by the java.util.regex.Pattern Java class, which is very similar to Perl
regular expressions.
Due to the English-only support by the Email Header filter, it is recommended to create a regular
expression filter for the indexed language to suppress email headers. For example, consider the
following German email header:

Von: John Smith

Gesendet: Mittwoch, 26 August 2015 17:05
An: Simon, Jane
Cc: Johnson, Ed
Betreff: Hallo

The Email Header filter would not find this text. Instead, use a Regular Expression filter as follows:

(?i)(Von.*?Gesendet.*?An.*?Cc:.*?Betreff.*?\r\n)

Note: The regular expression flag (?i) forces a case insensitive match. This is necessary because matching for
what to filter out from Structured Analytics analysis is case-sensitive.

73.4.1.4 Repeated content filters considerations

It is highly recommended to use Repeated Content filters on the Analytics index. The Auto-remove
signatures and footers feature only supports English, so the Repeated Content filter is needed to

Relativity | Recipes - 258

filter out the unwanted text. Repeated Content identification supports all Java-compatible languages
and may be used to identify the filters. As always, it is recommended to evaluate the contents and
impact of the filters before applying them to an Analytics profile.

73.4.1.5 Optimize training set considerations

The Optimize training set feature produces a better quality Analytics index by excluding documents
from the training set that are poor quality for index training. This function takes into account many
different characteristics, such as number of unique terms, punctuation, whitespace, size of text, etc.
This feature is applied after the word-breaking has been applied to the documents.

Note: The Optimize training set feature supports all Java-compatible languages and should be enabled.

73.4.2 Considerations for using Structured Analytics with non-English lan-

guages
The following table lists the four Structured Analytics features along with their language support:
Function Supports
Language identification 173 languages
Textual Near Duplicate identification All Java-compatible languages
Repeated content identification All Java-compatible languages
Email threading English Email Headers only

73.4.2.1 Language identification considerations

Language identification examines the extracted text of each document to determine the primary
language and up to two secondary languages present. This allows you to see how many languages
are present in your collection, and the percentages of each language by document. You can then
easily separate documents by language and batch out files to native speakers for review.
For multi-language documents, it returns the top three languages found and their approximate
percentages of the total text bytes (e.g. 80% English and 20% French out of 1000 bytes of text means
about 800 bytes of English and 200 bytes of French). Language identification does not use a
conceptual index for this operation; a Structured Analytics set is used. The Analytics engine is
trained on the prose of each supported language. The operation analyzes each document for the
following qualities to determine whether it contains a known language:
n Character set (e.g., Thai and Greek are particularly distinctive)
n Letters and the presence or absence of accent marks
n Spelling of words (e.g., words that end in “-ing” are likely English)
Language identification is a Naïve Bayesian classifier, using one of three different token algorithms: 
n For Unicode scripts such as Greek and Thai that map one-to-one to detected languages, the script
defines the result.
n For CJK languages, single letters (rather than multi-letter combinations) are scored.

Relativity | Recipes - 259

 n For all other languages, language Identification ignores single letters and instead uses sequences of
four letters (quadgrams).
It also ignores punctuation and HTML tags. Language identification is done exclusively on lowercase
Unicode letters and marks, after expanding HTML entities &xyz; and after deleting digits,
punctuation, and <tags>. For each letter sequence, the scoring uses the 3-6 most likely languages
and their quantized log probabilities.
The analysis does not use a word list or dictionary. Instead, the engine examines the writing to
determine the language. The training corpus is manually constructed from chosen web pages for
each language, then augmented by careful automated scraping of over 100M additional web pages.
Though the algorithm is designed to work best on sets of at least 200 characters (about two
sentences), testing has shown that it does perform well with small fragments, such as Tweets.

Note: Language identification in Relativity 9+ supports 173 languages. Language identification considers all
Unicode characters and understands which characters are associated with each of the supported
languages. For example, Japanese has many different character sets - Kanji, Katagana, Hirigana, all of which
are supported. See Supported languages matrix in the System Guides for the full list of supported
languages.

73.4.2.2 Textual Near Duplicate identification considerations

Textual Near Duplicate identification finds documents with highly similar text and places them into
relational groups. You may then use near duplicate groups in searching or filtering.

Note: This feature supports all Java-compatible languages.

73.4.2.3 Repeated content identification considerations

Repeated content identification finds blocks of text that are repetitive throughout the documents
and automatically creates filters. You can choose which filters you’d like to link to the Analytics
profile in order to optimize the Analytics index.

Note: This supports all Java-compatible languages and is highly recommended to use with non-English
cases.

73.4.2.4 Email threading considerations

Email threading is designed to find English email headers.
n When email headers themselves are in English, then the Analytics engine will thread them even if the
contents are not in English (e.g., a Japanese email subject).
n When the headers themselves are not in English (this commonly happens when a non-English speaker
hits "reply" in his or her email client and the email client then includes the headers in the embedded
message), the Analytics engine will not be able to parse them out for email threading.
Processing engines tend to insert English-language headers on top of extracted email body text
when they process container files such as PST, OST, or NSF. These headers, such as "To," "From,"

Relativity | Recipes - 260

"Subject," etc., take their contents from specific fields in the container file. But the email body text
does not, strictly speaking, contain the headers.

When the Analytics engine parses emails, it looks for English-language cues to determine when it is
or isn't in an email header. In particular, it's looking for words like "To, From, CC, Subject" in the
case of the traditional headers. And it's looking for format such as "on <date>, <author> wrote:" for
single-line replies. There are other variations, but the email header logic is English-centric.
Email threading will be affected as follows by non-English email headers:
n Groups of emails which should be in a single thread will split into multiple threads. This is due to not
matching up the foreign-language-surrounded segment with its English-language version (either when
the email itself is in the collection, or when the email was replied to by both an English and a non-Eng-
lish email client).
n There will be fewer segments than desired in the Email Thread Group of a document which contains for-
eign email headers.
n If emails contain mixed languages in header fields, for example some field names are in English and
some are in a different language, your Indentation field is lower than expected because Analytics
doesn't identify the emails to be emails.

Note: The Analytics engine does not offer capability to effectively deal with email headers which are in a
non-English language.

73.4.3 Checklist for using Analytics with non-English language and multiple
language workspaces
The following table can be used as a general ordered checklist for using Analytics successfully in
workspaces with a non-English primary language or with multiple languages:
Process description Non-English work- Multi-language work-
spaces spaces
Run Textual Near Duplicate identification
a a
Run language identification
a a
Determine if there will be separate indexes (i.e., at least 1K
documents of a given language)
a
Find appropriate stop word list
a a
Create one new Analytics profile with appropriate stop words
a
Create two new Analytics profiles with appropriate stop
words (if separate indexes)
a
Create saved searches
a a
Run Repeated Content identification against the Searchable
Set search
a

Relativity | Recipes - 261

Process description Non-English work- Multi-language work-
spaces spaces
Run two different Repeated Content identification sets
against each Searchable Set search (separately)
a
Choose Repeated Content filters to apply
a a
Apply Repeated Content filters to the Analytics profile
a
Apply Repeated Content filters to each appropriate Analytics
profile
a
Create index(es) with Optimize Training Set set to Yes and
Auto-Remove Signatures set to No
a a

74 Using categorization to find privileged documents

Published February 22, 2012
Relativity's categorization functionality can help you locate potentially privileged documents that
may have been overlooked during review. This recipe describes how to use the Synchronize feature
in Relativity Analytics to categorize existing privileged documents and use them as examples to
locate similar documents that may also be privileged.

74.1 Requirements
n An active Analytics index
n Categorization data source (saved search)
n Edit/Add access to Analytics Categorization Set

74.2 Directions
1. Create a new Analytics Categorization Set (found under the Analytics tab) with the following settings:
a. Set the Documents to Be Categorized and Analytics Index.
b. Set Privilege as the Categories and Examples Source.
c. Save your Categorization Set.
2. Click Synchronize in the Categorization Set console.
3. Click Categorize All Documents.
4. Once categorization is complete, a field is created that contains the categorization results. It will be
named Categories – <your Categorization Set name>. To begin screening for privileged documents, sort
on the CSR – <your Categorization Set name>::Category Rank field, and prioritize the higher-ranking
documents.

Note: Categorization has the potential to return many false-positive results. You can focus on more viable
results by raising the Minimum Coherence Score.

Relativity | Recipes - 262

75 Using categorization to identify hot documents
Published April 18, 2012
Using Relativity Analytics and methods for finding initial key documents, you can easily find other
potentially hot documents.
This recipe describes how to use Analytics to find hot documents based on records that have already
been marked as hot.

75.1 Requirements
n Active Analytics index
n Analytics categorization set
n Edit/Add access to Analytics categorization set

75.2 Directions
1. Identify hot documents to serve as example documents, using any of these methods:
a. Use key search terms to find hot documents.
b. Use key documents from your review to serve as examples against opposing production.
c. Use the Find Similar Documents functionality to begin a review of main document items.
d. Cluster and analyze documents from each cluster.
2. Mark your hot documents as examples, using any of these methods:
a. Use a layout, and then apply values to those records.
b. Right-click, and then apply the document as an example.
c. Import a list of document identifiers into the Analytics Example Object.
3. Categorize the examples against the database, and check for more hot documents.

76 Using Cluster Visualization to jump start your RAR

project
This recipe describes how to using cluster visualization can help you locate training documents for
your Assisted Review project. This recipe assumes that the case team has identified and tagged a
few responsive or hot documents based on an initial assessment or investigation. Using cluster
visualization, additional documents can be found that are conceptually similar to this initial set of
documents. These conceptually similar documents will then be used to create a training round in a
RAR project.

76.1 Requirements
n An active Analytics index
n An existing RAR project

Relativity | Recipes - 263

 n A cluster set (using default settings) that includes the RAR project documents

76.2 Directions
1. In the Cluster browser, select the cluster set that includes the documents in your RAR project, and then
click Visualize Cluster.

2. In the filter panel, create a filter condition that returns the previously identified responsive or hot doc-
uments, and then click Run Search.

3. The system applies a heat map overlay to indicate which clusters contain the coded documents.
a. Set the cluster visualization depth slider to 2 to see the sub-clusters that contain matching doc-
uments.

Relativity | Recipes - 264

 b. Optional: Click Highlight Matches in the legend to better visualize the relevant clusters (this
feature became available in the 9.3.256.4 product update).

Relativity | Recipes - 265

 4. Select the clusters that contain matching documents by holding down the control key (command key
on Mac), and then clicking on each of the shaded clusters. Click Apply.
5. The filter is currently set to return only the previously coded documents. To return all the documents in
the selected clusters, disable the filter by clearing the check box in the filter card and clicking Run
Search.

6. Now that all the documents in the selected clusters are displayed in the document list, save these doc-
uments as a search using the Save as Search functionality. Save the search in the search folder for your
RAR project so it can be accessed when creating a new round. In the following example, the search is
saved in the folder for the Responsive Review RAR project.

Relativity | Recipes - 266

 Note: You can also use the Save as List feature to save the documents as a list, then create a search from the
list and move the search into the RAR project search folder.

7. Next, go to your RAR project and create a new training round:

a. On the Project Home page, click Start Round.
b. Select Training for the round type, and in the saved search for sampling field, select the search
created above.
c. Under the Sampling Methodology section, adjust the settings to control the number of doc-
uments that you want the system to randomly select from the saved search. For example, you
could set the sampling type to fixed sample size, and then enter the number of documents you
want in the round.
d. Click Go.
You have now created a training round that focuses on documents that are conceptually similar to
the previously identified important documents.

77 Using custom objects to secure coding choices

You may need to secure coding choices in a field so that you can control which groups are able to
see them.

77.1 Recipe overview

This recipe shows you how to create a custom object and connect it to a document object via a field
to secure coding values.

Relativity | Recipes - 267

77.2 Directions
1. Create a new object in your workspace, and name it Coding Choices.
2. A Name field is automatically created on the Coding Choices object. Change the Name field to Issue
Choices.
3. Click New Field to create a field for the issue type.
a. Name the field Issue Type.
b. Set Object Type to Coding Choices.
c. Set the Field Type to Single Choice.
d. Click Save and New.
4. Create another field to hold the issue selection on the Document object.
a. Name the field Issues.
b. Set the Object Type to Document.
c. Set the Field Type to Multiple Object.
d. Set the Associative Object Type to Coding Choices.
5. Click the Coding Choices tab.
6. Edit the view to include the Issue Type field.
7. Import items using the Relativity Desktop Client, or use the New Coding Choices button to add items.
This recipe allows you to add choices manually.
8. Edit the Coding Choices layout to include the Issue Type field.
9. Add issues to the Issue Types.
10. Add items to the Coding Choices list and secure them as necessary.

78 Using dtSearch to identify documents in other

languages
A Relativity Analytics index running Primary Language Identification (PLI) can be very helpful to
determine languages. However, you can encounter a language not included in your index, or when a
simpler dtSearch workflow is sufficient for identifying documents in other languages.

78.1 Recipe overview

This recipe describes how to use stop words as search terms to yield dtSearch results for a specific
language.

78.2 Requirements
n Workspace access
o dtSearch creation rights
o Saved search creation permissions
n A list of stop words for the desired language

Relativity | Recipes - 268

78.3 Directions
The concept behind this process is simple. A document in any language is likely to contain a stop
word in that language. Consequently, you can take any language’s stop words and convert them into
go words.
1. Create a dtSearch index of your extracted text. You may wish to double-check that your Extracted Text
field was enabled for Unicode characters, if applicable.
2. Obtain or create a list of stop words for the language you wish to locate in your data set.
3. Construct a search using the dtSearch index, using the stop words as search terms separated by OR.

79 Using Email Thread Visualization in your QC

workflows
Use Email Thread Visualization as part of your QC workflow to visually identify and correct potential
privilege or designation coding conflicts.

Relativity | Recipes - 269

79.1 Requirements
n Relativity 9.5 or higher
n An Email Threading set

79.2 Directions
1. Set up a privilege conflict search that returns documents that are coded not privileged or have not
been reviewed for privilege within an email thread that contains privileged documents. See the recipe
Using saved searches to complete conflict checks for additional information.
2. Load a document returned by the above search into the document viewer.
3. In the viewer, click on the Show/Hide Email Thread Visualization icon ( ) to open the Email Thread
Visualization panel. 
4. In the Display Options section, enable Highlight Field, and select your privilege coding field. Note that
the field list includes single choice and yes/no fields.
5. Click LEGEND to display the field highlighting legend.
6. In the example screen shot below, the color coded field highlighting allows you to quickly see where an
email has been coded as Not Privileged (yellow highlighting), and where the Privileged field is not set
(gray shading).
7. Click on any square node to view the email in the viewer.
8. The interactive visual indicators allow you to quickly focus in on and correct potential inconsistencies,
and provide context as to where the emails are within the conversation thread.

Note: Reviewers can also have the Email Thread Visualization console open with Highlight Field enabled
while conducting their initial review. Doing this provides reviewers with real-time visual feedback regarding
coding decisions, allowing them to work more consistently and efficiently.

Relativity | Recipes - 270

80 Using Email Thread Visualization to leverage
previously coded documents
When new documents are loaded into a workspace, leverage previous coding decisions using Email
Thread Visualization to help prioritize your review of the newly loaded documents. Email Thread
Visualization allows you to visually focus in on newly loaded documents that are likely to be relevant
because they are part of a thread that contains relevant documents.

80.1 Requirements
n Relativity 9.5 and higher
n An Email Threading set

80.2 Directions
1. Ensure documents are tagged with a load identifier so that newly loaded documents can be readily
identified. 
2. Load new documents into the workspace and tag those documents with a load identifier value.
3. Build an email threading set that includes existing documents plus newly loaded documents.
4. Create saved search 1 to return documents that have been coded responsive and include email thread
group.
5. Create saved search 2 to return documents in search 1 that have a load identifier indicating they are
newly loaded documents.
6. Search 2 returns the newly loaded documents that should be prioritized, since these documents are in
email threads containing documents that were previously coded as responsive.
7. Run search 2 and open a document returned by the search in the document viewer.
8. In the viewer, click on the Show/Hide Email Thread Visualization icon ( ) to open the Email Thread
Visualization panel.
9. In the Display Options section, enable Highlight Field, and select your data load indicator field. Note
that the field list includes single choice and yes/no fields.
10. Click LEGEND to display the field highlighting legend.

Relativity | Recipes - 271

 11. In the screen shot below, the newly loaded documents are easily identified by the yellow highlighting.

12. Click DISPLAY OPTIONS and select the responsiveness coding field.
13. Click LEGEND.
14. As can be seen by the field highlighting in the screen shot below, all emails in the thread have been
coded responsive except for the newly loaded documents, which are not set (shaded gray).

15. Email Thread Visualization provides context as to where the newly loaded emails are within the con-
versation thread, and allows reviewers to focus in on these documents by clicking on the gray shaded
nodes to review and code the documents.

81 Using Near Duplicate analysis in review

Relativity can identify textually similar documents to assist in and speed up the review process. Near
duplicate analysis is best suited for grouping documents which can then be batched for review based

Relativity | Recipes - 272

on the similarity, or used to create new document sets for further analysis. The goal is for reviewers
to have the ability to see similar documents at the same time based on their textual similarity.

81.1 Requirements
n Analytics server setup

81.2 Overview
After running a Near Duplicate analysis, system admins should view the Textual Near Duplicate
Summary on the set’s Structured Analytics console, which breaks down the number of textual near
duplicate groups that have been identified, along with averages of percentage of similarity and of
the number of documents per similar document group.
Textual Near Duplicate Identification sorts the documents by size, from largest to smallest. This is
the order in which they are processed. The most visible optimization and organizing notion is the
principal document. The principal document is the largest document in a similar group and is the
document that all others are compared to when determining whether they are near duplicates. If
the current document is a close enough match to the principal document, as defined by the
Minimum Similarity Percentage, it is placed in that group. If no current groups are matches, the
current document becomes a new principal document. When the process is complete, only principal
documents that have one or more near duplicates are shown in groups.

When running the process, a Minimum Similarity Percentage is assigned. This parameter indicates
how similar a document must be to a principal document to be placed into that principal's group.

81.3 Directions
System admins should create a Textual Near Duplicates view for the review team.
1. In the Near Duplicate Identification view, add the following output fields:
nTextual Near Duplicate Principal - identifies the principal document with a "Yes"
value. The principal is the largest document in the duplicate group. It acts as an anchor
document to which all other documents in the near duplicate group are compared.
n Textual Near Duplicate Similarity - the percent value of similarity between the near
duplicate document and its principal document.
n Textual Near Duplicate Group - the identifier for a given group of textual near duplic-
ate documents.
2. Add a condition to only show documents where the Textual Near Duplicate Group field is set.
3. Set the following sort orders on the Near Duplicate Identification view to list the textual near
duplicate principals with the highest percentage of textual near duplicate similarity at the
top:
n Textual Near Duplicate Group - Ascending
n
Textual Near Duplicate Principal - Descending
n Textual Near Duplicate Similarity - Descending

Relativity | Recipes - 273

All documents should be reviewed in this process. Use the grouping and similarity to speed up the
review process. The Relativity Compare function can compare two documents to assess their
similarities and differences.
Reviewers will be able to view documents that are extremely similar but not identical to each
other. For example, the case team may need to ensure a series of very similar reports are coded
the same way. Another possible use is to help locate additional privileged documents that might
have been missed during first pass review. In situations like these, it is common to use a view that
displays textual near duplicates. Prior to the view’s creation, a system admin will run a near
duplicate analysis during which documents with similar text patterns are placed together into
relational groups. Exact syntax and word order are heavily considered during this analysis.
Here are the two most common fields you’ll likely encounter on a textual near duplicate view:
n Textual Near Duplicate Principal - identifies the principal document with a “Yes” value. The
principal is the largest document (as measured by amount of text) in the duplicate group. It
acts as an anchor document to which all other documents in the near duplicate group are
compared.
n
Textual Near Duplicate Similarity - the percent value of similarity between the near
duplicate documents in a given group and their principal document. Each group of textually
similar documents will contain a principal, which is typically the document in the group that
contains the most text. All documents in a near duplicate group will be assigned a score that
indicates how similar each document is to its principal.
Consider the following example. The first document, AZIPPER_0011323, is the group’s principal, as
indicated by the “Yes” value in the Textual Near Duplicate Principal field. It also has a score of 100.
All principals will have a score of 100, as they are by definition 100% similar to themselves,
however, not all documents with a score of 100 are necessarily principals. The documents
underneath are part of the principal’s relational group. The second and third documents are
identical to the principal. We know this because they are 100% similar to it, as shown in the Textual
Near Duplicate Similarity field on the far right of the view. The last three documents are very
closely similar to the principal but are not exact duplicates; their scores indicate they are each 99%
similar to the principal.

81.4 Workflow considerations

Textual near duplicate groups have a relational field that can be used to code several documents at
once. Documents contained in the near duplicate group are textually similar, but similarity is
usually not enough to treat near-duplicates as identical documents for the purposes of review. As
such:
n It is not recommended to propagate coding on near duplicates, unless other analysis or evid-
ence points to their similarity to justify such a step.
n
It is not recommended to delete or otherwise shelve near duplicates and opt to focus solely
on the principal document of each group.
n
Document attachments may also differ on near duplicates, furthering the necessity to review
or analyze all documents in the group.

Relativity | Recipes - 274

82 Using redactions from other applications
When transferring data to Relativity from other systems, redactions are often part of the original
data. No one wants to lose this work, but there is no way to export and import redactions between
databases. The following solution helps retain these redactions.

82.1 Recipe overview

This recipe shows you how to create a production from another application with the redactions
burned into the images.

82.2 Directions
1. Export data from the original database.
2. Create a production, burning in redactions.
3. Export the production with single-page TIFFs and an OPT file.
4. Export either native files, multi-page TIFFs, or PDFs from the original application with the path in a DAT
file.
5. Import the data into the new database.

Note: Load native files/multi-page image files as native files with metadata.

6. Import the production from your original application as single-page images into Relativity . Use redac-
ted versions of images for production unless changes need to be made. A user can then delete the
image and create it again from the clean native version.

83 Using reviewed documents to prioritize new

documents via Analytics
Published April 3, 2013
As a case develops, you may need to add more documents to your workspace several times
throughout your review. When this happens, the documents you have already coded can help
prioritize your review of the new documents.
This recipe shows you how to use Relativity's near duplicate functionality to leverage previously
coded documents to determine the potential values of new documents in your workspace.

83.1 Requirements
n Structured Analytics for Near Duplicates including new and reviewed documents
n Workspace access
o Saved searches
o Structured Analytics

Relativity | Recipes - 275

83.2 Directions
1. After new documents are loaded into your workspace, update your Structured Analytics Saved Search
and Run Full Analysis.
2. Using a saved search, identify the documents already coded as responsive and add Include Near Dupe
Groups. Based on those results, you can create a second saved search to identify potentially respons-
ive documents.
a. Your first saved search should contain responsive and Near Dupes.
n Name: Responsive and Near Duplicates
n Includes: Include Near Dupe Groups
n Conditions: Designation is Responsive
b. A second saved search will contain potentially responsive documents.
n Name: Potentially Responsive Documents
n Conditions:
o Documents are in the Responsive and Near Duplicates search
o Designation is not set
3. To leverage non-responsive documents, follow the same protocol.
a. The first saved search will contain non-responsive and Near Duplicate documents.
n Name: Not Responsive and Near Duplicates
n Includes: Include Near Dupe Groups
n Condition: Designation is Not Responsive
b. The second saved search will contain potentially non-responsive documents.
n Name: Potentially Not Responsive Documents
n Conditions:
o Documents are in the Not Responsive and Near Duplicates saved search
o Designation is not set

84 Using saved searches to complete conflict checks

When producing documents, check for conflicts between family members in a group and for family
members that didn't receive a coding value. You can combine saved searches to perform this
conflict check.

84.1 Recipe overview

When producing documents or looking for discrepancies in datasets, here is a string of searches that
can be constructed to find those conflicts. The key is when you bring in the related items. This recipe
looks at family group differences. When a search is executed, the related group inclusion occurs
after all other conditions have been met. This brings items into the search that might not fit all the
conditions, but these are the differences we want.

Relativity | Recipes - 276

 Note: To read a technical breakdown of how Relativity builds and executes complex searches such as this
conflict check, see Searching behind the scenes.

84.2 Directions
To create saved searches for your conflict check:
1. Create a saved search to find what we do want: responsive items and their family items.
a. Set Name to Responsive and Family.
b. Include related group and Family.
c. Set Scope to Entire Workspace.
d. Set Conditions to Responsive, Operator to any of these, and Value to Responsive under Condi-
tions.
e. Add any Fields you find useful, but this is not the search that will display at the end.

Relativity | Recipes - 277

 f. Click Save.

2. Create a saved search to find items without coding or anything that is NOT set to Responsive.
a. Set Name to Conflicts.
b. Set Scope to Entire Workspace.
c. Create a Logic Group and add:
a. Set Condition to Responsive, and Operator to is not set.
b. Add an additional Condition, and select OR.
c. Set second condition to Responsive any of these, and Value to Not Responsive under
Conditions.
d. Add Condition of (Saved Search) document is in Responsive and Family.

Relativity | Recipes - 278

 e. Click Save.

This search will find items that are not coded or are not responsive that are family members for
Responsive items. This can be used in a variety of ways to find differences between populations.
Change the related items group from Family to Near Duplicates to find overlap between groups of data
that are near duplicates.

85 Using stop words and making some characters

searchable in a dtSearch
Relativity ignores words that don't act as meaningful criteria when you create dtSearch and
keyword queries. Ignored words are known as stop or noise words. Search indexes automatically

Relativity | Recipes - 279

include the default list of stop words. However, you can edit this list in the dtSearch list to suit your
needs. This recipe includes an overview of stop words and steps to create custom lists.

Using the dtSearch alphabet file, system admins can take originally ignored words and make them
searchable. However, dtSearch contains some characters and punctuation characters that you can't
make searchable.

85.1 Requirements
n Workspace access
o Edit, delete access on search indexes
o Edit, delete access on searches

85.2 Directions
Relativity references the default list of stop words each time you create a new index. System admins
can't edit stop words in keyword searches. The default stop word list consists of punctuation marks,
single letters and numbers, and the following words:

Relativity | Recipes - 280

 Note: Relativity ignores stop words. However, Relativity doesn't ignore their position in the search phrase
set. So, if you execute the query “apple w/6 pear,” the search returns the phrase “apple tree is far from the
pear” even though it contains the stop words “is,” “from” and “the.”

85.2.1 dtSearches and stop words

The default list of stop words is the same in a dtSearch as in a keyword search. The primary
difference is that you can customize the dtSearch index list. For example, if the word "never" is
important to your litigation, remove it from the stop words list, so that your search results always
return that word.
To create a custom stop word list, perform the following:
1. Create a new dtSearch index, and then name it "dtSearch - updated stop words."
2. Select your extracted text search for the Searchable set.
3. Delete the word "never" from the Noise Words list.

Relativity | Recipes - 281

 4. Save the list, and then perform a full build on your new index.

85.2.2 Stop words in languages other than English

You can set up stop words to search documents in other languages. If the workspace primarily
contains of documents in a different language, see this page for an overview of suggested stop
words for use in nineteen additional languages.

Note: The following are restricted from dtSearch in Relativity. These are reserved noise words and
characters which will continue to behave as operators as well as being noise words:

Relativity restricts the following from dtSearch. The following noise words and characters continue
to behave as operators: and, or, not, to, contains, xfirstword, xlastword, ", ( )< *, ?, %, @, ~, #, &, :,
=.

85.2.3 Searching for Special Characters

The following descriptions are for characters in the ASCII 33-127 range.

85.2.3.1 The Alphabet file

dtSearch defines letters as characters to index. For most characters, this also means they are
searchable. However, you can never make the characters discussed in the previous section

Relativity | Recipes - 282

searchable. By default, this includes all alphabetic characters (a-z and A-Z) and all digits (0-9).

Note: dtSearch is case insensitive.

85.2.3.2 Spaces

Note: You must have valid Relativity Community credentials in order to download any Community file
linked to from the documentation site. You'll need to enter those credentials on the Community login
screen if you're not already logged in. If you're already logged in to the Community at the time you click a
link, the file is automatically downloaded in the bottom left corner of your screen. If you get an error
message stating "URL No Longer Exists" after clicking a Community link, it may be due to a single sign-on
error related to the SAML Assertion Validator, and you should contact your IT department.

dtSearch defines a space character as a character that causes a word break. Relativity doesn't index
these characters, so you can't search them. By default, dtSearch treats the following characters as
spaces:
\09\0a\0c\0d !@"#$&'()*+,./:;<=>?[\5c]^`{|}~
Values listed as \## are Unicode characters. Their definitions are:
n \09 - horizontal tab
n \0a - line feed
n \0c - form feed
n \0d - carriage return
n \5c - backslash (\)

85.2.3.3 Ignored
dtSearch defines an ignored character as a character that's ignored when processing text. You can't
search these characters. By default, dtSearch ignores the following characters:
\08 %
\08 is the backspace character in Unicode.

85.2.3.4 Hyphen
dtSearch defines hyphens as characters that receive special processing in dtSearch. By default,
dtSearch only classifies the - character as a hyphen.

85.2.3.5 Restricted characters

Either a limitation in dtSearch or the implementation of the dtSearch API restricts characters from
being searchable.

85.2.3.6 Always Restricted

dtSearch always restricts the following characters, regardless of implementation:
" ( ) * ?

Relativity | Recipes - 283

85.2.3.7 Using reserved characters in the Alphabet list
dtSearch treats reserved characters as operators regardless of what is set in the Alphabet file.
Remember how those operators act when interpreting whether a solution works.
For example, will adding the % sign to the alphabet, and removing it from the Ignore List return
"apple%"?
The % is the fuzzy operator, meaning, meaning any one character (or no character) can be in this
spot and return results. This is very similar to how * (wildcard) or ? (wildcard for any single
character) work. dtSearch reserves the % sign as an operator. Thus, even though adding % to the
alphabet and removing it from the Ignore list enables % to be a searchable character, % can't be
successfully used as a search term. However, you can employ the following Regular Expression to
bring back apple%:
"##apple\W"
The word apple% here is brought back because the regular expression metacharacter \W matches
for any non-alphanumeric character, which includes the % sign.

85.2.3.8 Searching for &

85.2.3.9 Searching for reserved characters such as &
Perform the following steps to search for the & character:
1. Make & a searchable character in your dtSearch index.

2. Delete & from the [Spaces] string, so dtSearch doesn't interpret it as a space (word break) in the index.

3. Perform a full build of the index.

You can now search for the & character in your workspace.

Relativity | Recipes - 284

86 Using tokens to customize stamps in a production
Occasionally, you might want your Bates numbers to include a confidentiality footer, or other
information in the same stamp location. Using tokens — codes that represent metadata for records
in the database — you can arrange multiple fields together in a free text box.

86.1 Recipe overview

This recipe describes how to use tokens for stamping on a production in a free text field instead of
pulling down a particular field in the database.

86.2 Directions
Using production sets you can stamp documents with various pieces of information. Typically, there
is one field selected from the drop-down for each stamp location (e.g., the right footer or left
header). However, if you want to produce with more than one piece of metadata information, you
can to use tokens to customize your stamps.
You can use tokens for any data field and you can use them in combination with static text and/or
other tokens. The syntax to use a token is {!Field:ArtifactID!}. You can find the ArtifactID for a field
by adding the ArtifactID field to your Fields view.
An example of combining static text and a token would be: Custodian: {!Field:1036657!}. In this
example, Custodian is the static text and acts as a label for the token, the custodian field.
An example of combining multiple tokens would be: {!Field:1036657!} {!Field:1037086!}
{!Field:1037082!}. In this example, you are branding a string of three fields together on an
image.You can use hard returns or spaces between the tokens. You can also add separators, such as
carriage returns, hyphens, or commas, between the fields.

Relativity | Recipes - 285

87 Using the field tree to view folders
Occasionally, system admins need to use the folder browser view for security and still want to see
the visual arrangement of folders as the document was originally stored. Using the Field Tree, we
can mimic the folder browser.

87.1 Recipe overview

This recipe shows you how to create a multi-choice field and open it to the Field Tree. Folder choices
may be created when users perform their initial data load, or later on as an import overlay. Users
may move a document from one folder choice to another (or display the document in multiple
folder choices) by creating a document layout and editing as desired.

87.2 Requirements
n Folder paths less than 200 characters

87.3 Directions
1. Create a new field named Folders and open to the Field Tree.
2. Use a load file with slashes between folder names.
3. Import into the multiple choice field. This loads into the field with semi-colons separating items.

The appearance of folders and folder choices on the field tree look very similar with the exception
of the icon.

Relativity | Recipes - 286

The image on the left is Folder Browser. The image on the right displays folder choices in the Field
Tree.
To change the location of a document within a folder choice, or to display the document in more
than one folder choice, create a layout on the document level, as displayed below.
The layout below only has the folders field available. Use the checkboxes just like any other multiple
choice selection, adding or removing documents from choices to simulate movement between
folders.

Relativity | Recipes - 287

88 Using the Update Duplicate Status Script
If global deduplication has not been applied to documents prior to loading them into a workspace,
you can use the Update Duplicate Status script, which is included in the Processing Duplication
Workflow solution, to identify duplicate documents.
This recipe provides additional guidance for using the Update Duplicate Status script. The script tags
documents as either master, unique or duplicate. Using these tags, you can filter out duplicates, and
pass only master and unique documents, together with their family members, on to the review
team.

Relativity | Recipes - 288

88.1 Requirements
n Relativity 9.1 or higher
n The Processing Duplication Workflow solution has been deployed in the workspace.

88.2 Directions
1. Publish or load documents into a workspace.
2. Set up a saved search that returns parent level emails and parent level loose documents. The duplicate
status field will be set on the parent level documents, and then family members will be included with
the parents, as described below.
n If using Relativity Processing, parent documents can be identified by including a search condition
where the Level field equals 1.
3. Create a single choice Duplicate Status field with 3 choices: Unique, Master, and Duplicate.
4. Run the Update Duplicate Status script:
n Use the saved search described above.
n Use the duplicate status field created above.
n For the Relational Identifier, select a file duplicate hash field. If using Relativity Processing to load
data, the Processing duplicate hash field can be used.
n Leave Duplicate Sort Order blank, unless you have set up a Custodian field to maintain a custom
priority sort order. When this field is blank, the system sorts on document artifact ID, so the first
document loaded in the workspace becomes the Master document when duplicates are iden-
tified.

n After the script runs, the duplicate status will be populated for top level, parent documents.
n At this point, you have a couple of options for including family members:
n Option 1 - Create a search that returns all documents where the duplicate status is set to
Master or Unique, and include family.

Relativity | Recipes - 289

 n Options 2 - Copy the parent document duplicate status value to child documents and
then create a search that returns all documents where the duplicate status is set to
Master or Unique. You would not need to include family in this search.
oOne way to copy the duplicate status values is to deploy the ‘Propagate Coding
Post-import’ solution and run the script, Propagate Coding Post-Import. This script
will copy the duplicate status value from the parent to each child document so that
all documents in a family will have the same duplicate status value. For more
information, see Propagating to one field on demand.
o An alternative way to copy the duplicate status values to is to construct searches

to return each status, include family, and then use a mass edit to update the child
documents with the status value of the parent.
o After copying the parent duplicate status value to children, create a search that

returns all non-duplicates. Family does not need to be included in this search, since
all family members will now have a duplicate status value.
n Make the non-duplicate documents identified in the above steps available to reviewers.

88.2.1 Publishing or loading additional documents

After publishing or loading additional documents into the workspace, follow the steps described
above to identify Master, Unique and Duplicate documents and to make non-duplicate documents
available to reviewers.
When you are publishing or loading additional documents, keep the following in mind:
n Run the Update Duplicate Status script against all parent level documents in the workspace for which
you want to identify duplicates - both the newly published/loaded documents and the existing doc-
uments. Running the script clears the duplicate status field and then re-populates it.
n The duplicate status for an existing document could change from Unique to Master if a duplicate of an
existing, unique document is subsequently published/loaded into the workspace. This change in status
will not impact the workflow described above, as all non-duplicate records (both Master and Unique)
are being identified for review.

89 Verifying imaged natives

After imaging documents, it is important to verify that they are formatted as you expected. You may
need to re-image certain documents with different settings, which can easily be accomplished by
creating Imaging Profiles.

89.1 Recipe overview

This recipe shows you how to QC your images using sampling, thumbnails, the viewer, and a QC
layout.

Relativity | Recipes - 290

 Note: This recipe is only relevant to versions of Relativity before 8.1. Relativity 8.1 introduced a QC Review
feature that accomplishes the same goals. See QC Review for more details.

89.2 Requirements
n Relativity 7.3 or higher
n Relativity Web Client (ensure you have the latest version)
n Workspace Access
o Add fields
o Add layouts
o Script Admin
o View image
o Viewer

89.3 Directions
1. Create the fields to be used with the Relativity Sample Script, which are added to the Image QC (IQC) lay-
out. Use IQC_ or another designation in front of the field names to assist with filtering for IQC-related
items. In addition, you may want to secure these fields so that reviewers do not see them. Some sug-
gested fields are:
n  IQC_Sample
o  Field Type - Yes/No (this will be used for the Sample Script)

n IQC_Color
o  Field Type - Single Choice
o Choices 
l Add
l Remove
n IQC_Excel Limit Pages
o  Field Type - Yes/No
n IQC_Excel Page Size
o Field Type - Single Choice
o Choices
l Letter (8.5 x 11)
l Letter (11 x 8.5)
l Tabloid (11 x 17)
l Ledger (17 x 11)
n IQC_Excel Print Area
o Field Type - Single Choice
o Choices
l Original
l Ignore

Relativity | Recipes - 291

 n IQC_Email Resize Tables
o Field Type - Yes/No
n IQC_Word/Excel Show Track Changes
o Field Type - Yes/No
n IQC_Word Include
o Field Type - Multiple Choice
o Choices
l Comments
l Fields Codes
l Hidden Text

2. After you create your IQC fields, build a layout that has all of these fields available. Use the IQC_ naming
convention to maintain organization. Add your fields according to your needs.
3. Create a view that contains all your IQC_fields and the document type, if you have it. You will filter on
these fields to re-image documents, if necessary.
4. Run the Sample Script to assemble a random sampling of your imaged documents to be inspected.
This step is not necessary if you choose to inspect all your documents.
n There are two options:
o Fixed number
o Percentage of sample universe
n When running the script, use the newly created IQC_Sample field as your Sample Indicator field.
n After running the script, create a Saved Search for your Sample Set to QC that consists only of
the condition where the value for IQC_Sample is YES. Name this search “IQC_Sample_Universe.”
Please note that if you did not use the Sample Script, you will not have to create this search.
5. To verify images, launch the thumbnail viewer and the standalone viewer. Thumbnails are located at
the lower-right corner of the viewer.

a. If you do not see the option for thumbnails, update your viewer to the latest version. You can
download a new version of the viewer from the Workspace Details tab.
b. Launch the standalone viewer and use the Viewer Mode, located in the upper-right corner. Using
both of these will allow you to compare the image to the native. If using multiple monitors,
move this to your second monitor.

Relativity | Recipes - 292

 c. Additionally, display the IQC Layout. While examining images, make the necessary coding
changes on the layout. These fields will be used to filter for documents that may need to be re-
imaged.
d. You can also delete individual pages from a document by selecting the page’s thumbnail and
clicking Delete in the lower-right corner. You can select multiple pages by holding down the
control key.
6. Under Imaging Profiles, build a new IQC Imaging Profile. This will be used to re-image documents. You
can turn specific features on or off from this profile in order to preserve the default or original profile.
a. Choose the IQC view that contains your document type and the IQC fields. Filter the view for spe-
cific document types and then for changes required to re-image the document.
b. Example workflow for reimaging specific features: You may need to add tracked changes to a doc-
ument that you find with tracked changes turned on. First, filter for the IQC_Word/Excel Show
Track Changes field to be YES, and then filter for document type WORD. Examine the rest of the
IQC_Word fields for any additional features that may need to be added.
c. Three groups of documents are returned:
n Group 1: Add Track Changes only
n Group 2: Add Track Changes and Comments
n Group 3: Add Track Changes, Comments, and Field Codes
d. Select Group 1, where you only need to add Track Changes. In your IQC Imaging Profile, change
the Show Track Changes option to YES. Save your profile and return to the documents tab. Filter
for ONLY documents needing only Track Changes (IQC_ Word Include is null), use the Image
mass function, and choose your IQC Image Profile.
e. Continue this process for the remaining QC items, always double-checking the changes made to
the IQC Imaging Profile.

Relativity | Recipes - 293

Proprietary Rights
This documentation (“Documentation”) and the software to which it relates (“Software”) belongs
to kCura LLC and/or kCura’s third party software vendors. kCura grants written license agreements
which contain restrictions. All parties accessing the Documentation or Software must: respect
proprietary rights of kCura and third parties; comply with your organization’s license agreement,
including but not limited to license restrictions on use, copying, modifications, reverse engineering,
and derivative products; and refrain from any misuse or misappropriation of this Documentation or
Software in whole or in part. The Software and Documentation is protected by the Copyright Act of
1976, as amended, and the Software code is protected by the Illinois Trade Secrets Act. Violations
can involve substantial civil liabilities, exemplary damages, and criminal penalties, including fines
and possible imprisonment.
© 2018. kCura LLC. All rights reserved. Relativity® and kCura® are registered trademarks of
kCura LLC.

Relativity | Recipes - 294