Introducing Information

Retrieval
and Web Search
The literature searching process “is not an exact science but an art.”
Samuel Butler
Agenda

• Course schedule
• Introduction
• Some tricks to be more effective with Google
search
• Structured vs Unstructured Data
• Boolean retrieval
• Search Evaluation (Precision, recall, F-Measure)
• Introduction to ElK
• Reading: Chapter 1 - Boolean retrieval
Tricks to be more effective with Google Search

• Either this or That : buy cheap Android OR| iphone

• Using synonyms: Healthy ~ Food
• Searching within Websites: augmented reality site:
theverge.com
• Generic character (asterisk): Cinema the Number*
• When lots of word are missing: I wandered AROUND
(4) cloud
• Using a number range: GoPro hero $ 100 … 300
• Searching for a title or URL: Intitle
inUrl : sport

• Finding similar Websites: related: www.xxx.

Tricks to be more effective with Google search

• Unimportant search words: interesting books - buy

• Searching images : see the icon at the left of the search bar
• Searching Music
• Defining words and learning where they come
from? Define: xxxx etymology: mortage
• Finding a specific file xxx filetype:pdf
• Using google as spellcheck
• Tracking packages: enter tracking number (UPS, Fedex, etc.)
• Lunch an internal game: Atari Breakout
Information Retrieval

• Information Retrieval (IR) is finding material

(usually documents) of an unstructured nature
(usually text) that satisfies an information need
from within large collections (usually stored on
computers).

– These days we frequently think first of web search,

but there are many other cases:
• E-mail search
• Searching your laptop
• Corporate knowledge bases
• Scientific Papers
• Legal information retrieval

5
 Structured vs.
Unstructured Data
Unstructured (text) vs. structured (database) data

Mid-nineties

Today

7
Basic assumptions of Information Retrieval

• Collection: A set of documents

– Assume it is a static collection for the moment

• Goal: Retrieve documents with information

that is relevant to the user’s information need
and helps the user complete a task

8
The classic search model
User task Get rid of mice in a
politically correct way
Misconception?

Info need
Info about removing mice
without killing them
Misformulation?

Query
how trap mice alive Search

Search
engine

Query Results
Collection
refinement
How good are the retrieved docs?
▪ Precision : Fraction of retrieved docs that are
relevant to the user’s information need
▪ Recall : Fraction of relevant docs in collection
that are retrieved

irrelevant
Retrieved & Not Retrieved &
Entire document collection
irrelevant irrelevant

Relevant Retrieved

relevant
documents documents Retrieved
Retrieved && Not Retrieved &
relevant
Relevant But relevant

retrieved not retrieved

▪ More precise definitions and measurements to

follow later
10
Term-document incidence
matrices
Boolean Retrieval
Unstructured data in 1620
• Which plays of Shakespeare contain the words
Brutus AND Caesar but NOT Calpurnia?
• One could grep all of Shakespeare’s plays for
Brutus and Caesar, then strip out lines containing
Calpurnia?
• Why is that not the answer?
– Slow (for large corpora)
– NOT Calpurnia is non-trivial
– Other operations (e.g., find the word Romans near
countrymen) not feasible
– Ranked retrieval (best documents to return)
• Later lectures
Example : http://www.rhymezone.com/shakespeare/
12
 Term-document incidence matrices

Antony and Cleopatra Julius Caesar The Tempest Hamlet Othello Macbeth
Antony 1 1 0 0 0 1
Brutus 1 1 0 1 0 0
Caesar 1 1 0 1 1 1
Calpurnia 0 1 0 0 0 0
Cleopatra 1 0 0 0 0 0
mercy 1 0 1 1 1 1
worser 1 0 1 1 1 0

1 if play contains
Brutus AND Caesar BUT NOT Calpurnia
word, 0 otherwise
 Sec. 1.1
Incidence vectors

• So we have a 0/1 vector for each term.

• To answer query: take the vectors for Brutus,
Caesar and Calpurnia (complemented) ➔
bitwise AND.
– 110100 AND
– 110111 AND
Antony and Cleopatra Julius Caesar The Tempest Hamlet Othello Macbeth

– 101111 = Antony
Brutus
1
1
1
1
0
0
0
1
0
0
1
0
Caesar 1 1 0 1 1 1

– 100100 Calpurnia
Cleopatra
0
1
1
0
0
0
0
0
0
0
0
0
mercy 1 0 1 1 1 1
worser 1 0 1 1 1 0

14
Answers to query

• Antony and Cleopatra, Act III, Scene ii

Agrippa [Aside to DOMITIUS ENOBARBUS]: Why, Enobarbus,
When Antony found Julius Caesar dead,
He cried almost to roaring; and he wept
When at Philippi he found Brutus slain.

• Hamlet, Act III, Scene ii

Lord Polonius: I did enact Julius Caesar I was killed i’ the
Capitol; Brutus killed me.

15
Bigger collections

• Consider N = 1 million documents, each with

about 1000 words.
• Avg 6 bytes/word including
spaces/punctuation
– 6GB of data in the documents.
• Say there are M = 500K distinct terms among
these.

16
Can’t build the matrix

• 500K x 1M matrix has half-a-trillion 0’s and 1’s.

• But it has no more than one billion 1’s. Why?

– matrix is extremely sparse.

• What’s a better representation?

– We only record the 1 positions.

17
Example
• D1: Mehdi habite à Rabat
• D2: Kenza habite à Fes
• D3: Mehdi et Kenza se sont rencontrés à Fes
terms D1 D2 D3 Mehdi AND Kenza
Mehdi 1 0 1 1 0 1

Habite 1 1 0 0 1 1

À 1 1 1
Rabat 1 0 1 0 0 1
Kenza 0 1 1 Mehdi OR Kenza
Fes 0 1 1 1 0 1
Et 0 0 1 0 1 1
Se sont 0 0 1
rencontrés 0 0 1 1
1 1
Why Boolean retrieval is not suitable

• Consider an average of 6 byte / word

• Consider an average of 300 words / page
• Consider we have to index a million pages (106)
• We need 1.5 * 109 byte ~ 1.5 Gb to store a
million pages documents (Data)
• To store the index (term-doc matrix) we need a
matrix with the lexicon as line and 106 column
• 3 * 105 * 106 = 3 1011 ~ 30 Gb
• Tdm is extremely sparse !!!
• Tdm is 30 Gb
Possible terms 250 * 106
=
Possible values 3 * 1011

0.083 possible value of 1

or 99.9 % of 0

To fix this issue we look at Inverted index

 The Inverted Index
The key data structure
underlying modern IR
Inverted index

• For each term t, we must store a list of all

documents that contain t.
– Identify each doc by a docID, a document serial
number
• Can we use fixed-size arrays for this?
Brutus 1 2 4 11 31 45 173 174
Caesar 1 2 4 5 6 16 57 132
Calpurnia 2 31 54 101

What happens if the word Caesar is added to document 14?

22
Inverted index

• We need variable-size postings lists

– On disk, a continuous run of postings is normal
and best
– In memory, can use linked lists or variable length
arrays Posting
• Some tradeoffs in size/ease of insertion
Brutus 1 2 4 11 31 45 173 174
Caesar 1 2 4 5 6 16 57 132
Calpurnia 2 31 54 101

Dictionary Postings
Sorted by docID (more later on why).
 Inverted index construction
Documents to Friends, Romans, countrymen.
be indexed

Tokenizer

Token stream Friends Romans Countrymen

Linguistic modules

Modified tokens friend roman countryman

Indexer friend 2 4
roman 1 2
Inverted index
countryman 13 16
Initial stages of text processing

• Tokenization
– Cut character sequence into word tokens
• Deal with “John’s”, a state-of-the-art solution
• Normalization
– Map text and query term to same form
• You want U.S.A. and USA to match
• Stemming
– We may wish different forms of a root to match
• authorize, authorization
• Stop words
– We may omit very common words (or not)
• the, a, to, of
Indexer steps: Token sequence

• Sequence of (Modified token, Document ID) pairs.

Doc 1 Doc 2

I did enact Julius So let it be with

Caesar I was killed Caesar. The noble
i’ the Capitol; Brutus hath told you
Brutus killed me. Caesar was ambitious
Indexer steps: Sort

• Sort by terms
– And then docID

Core indexing step

 Indexer steps: Dictionary & Postings

• Multiple term entries

in a single document
are merged.
• Split into Dictionary
and Postings
• Doc. frequency
information is added.

Why frequency?
Will discuss later.
Where do we pay in storage?

Lists of
docIDs

Terms
and
counts
IR system
implementation
• How do we
index efficiently?
• How much
storage do we
need?
Pointers 29
Query processing with an
inverted index
The index we just built

• How do we process a query? Our focus

– Later - what kinds of queries can we process?

31
Query processing: AND

• Consider processing the query:

Brutus AND Caesar
– Locate Brutus in the Dictionary;
• Retrieve its postings.
– Locate Caesar in the Dictionary;
• Retrieve its postings.
– “Merge” the two postings (intersect the document sets):
2 4 8 16 32 64 128 Brutus
1 2 3 5 8 13 21 34 Caesar

32
 Sec. 1.3
The merge

• Walk through the two postings

simultaneously, in time linear in the total
number of postings entries

2 4 8 16 32 64 128 Brutus
1 2 3 5 8 13 21 34 Caesar

If the list lengths are x and y, the merge takes O(x+y)

operations.
Crucial: postings sorted by docID.
33
Intersecting two postings lists (a “merge” algorithm)

34
The Boolean Retrieval
Model
& Extended Boolean
Models
Boolean queries: Exact match

• The Boolean retrieval model is being able to ask a

query that is a Boolean expression:
– Boolean Queries are queries using AND, OR and NOT
to join query terms
• Views each document as a set of words
• Is precise: document matches condition or not.
– Perhaps the simplest model to build an IR system on
• Primary commercial retrieval tool for 3 decades.
• Many search systems you still use are Boolean:
– Email, library catalog, Mac OS X Spotlight

36
Example: WestLaw http://www.westlaw.com/

• Largest commercial (paying subscribers)

legal search service (started 1975; ranking
added 1992; new federated search added
2010)
• Tens of terabytes of data; ~700,000 users
• Majority of users still use boolean queries
• Example query:
– What is the statute of limitations in cases
involving the federal tort claims act?
– LIMIT! /3 STATUTE ACTION /S FEDERAL /2
TORT /3 CLAIM
• /3 = within 3 words, /S = in same sentence

37
Exercise

• Try the search feature at

http://www.rhymezone.com/shakespeare/
• Write down five search features you think it
could do better

38
Phrase queries and
positional indexes
Phrase queries

• We want to be able to answer queries such as

“stanford university” – as a phrase
• Thus the sentence “I went to university at
Stanford” is not a match.
– The concept of phrase queries has proven easily
understood by users; one of the few “advanced
search” ideas that works
– Many more queries are implicit phrase queries
• For this, it no longer suffices to store only
<term : docs> entries
A first attempt: Biword indexes

• Index every consecutive pair of terms in the text

as a phrase
• For example the text “Friends, Romans,
Countrymen” would generate the biwords
– friends romans
– romans countrymen
• Each of these biwords is now a dictionary term
• Two-word phrase query-processing is now
immediate.
 Longer phrase queries

• Longer phrases can be processed by breaking them

down
• stanford university palo alto can be broken into the
Boolean query on biwords:
stanford university AND university palo AND palo alto

Without the docs, we cannot verify that the docs

matching the above Boolean query do contain the
phrase.
Can have false positives!
 Sec. 2.4.1
Issues for biword indexes

• False positives, as noted before

• Index blowup due to bigger dictionary
– Infeasible for more than biwords, big even for them

• Biword indexes are not the standard solution

(for all biwords) but can be part of a compound
strategy
 Sec. 2.4.2
Solution 2: Positional indexes

• In the postings, store, for each term the

position(s) in which tokens of it appear:

<term, number of docs containing term;

doc1: position1, position2 … ;
doc2: position1, position2 … ;
etc.>
Positional index example

<be: 993427;
1: 7, 18, 33, 72, 86, 231; Which of docs 1,2,4,5
could contain “to be
2: 3, 149; or not to be”?
4: 17, 191, 291, 430, 434;
5: 363, 367, …>

• For phrase queries, we use a merge

algorithm recursively at the document level
• But we now need to deal with more than
just equality
 Sec. 2.4.2
Processing a phrase query

• Extract inverted index entries for each distinct

term: to, be, or, not.
• Merge their doc:position lists to enumerate all
positions with “to be or not to be”.
– to:
• 2:1,17,74,222,551; 4:8,16,190,429,433; 7:13,23,191; ...
– be:
• 1:17,19; 4:17,191,291,430,434; 5:14,19,101; ...
• Same general method for proximity searches
 Sec. 2.4.2
Proximity queries

• LIMIT! /3 STATUTE /3 FEDERAL /2 TORT

– Again, here, /k means “within k words of”.
• Clearly, positional indexes can be used for
such queries; biword indexes cannot.
• Exercise: Adapt the linear merge of postings to
handle proximity queries. Can you make it
work for any value of k?
– This is a little tricky to do correctly and efficiently
– See Figure 2.12 of IIR
Positional index size

• A positional index expands postings storage

substantially
– Even though indices can be compressed
• Nevertheless, a positional index is now
standardly used because of the power and
usefulness of phrase and proximity queries …
whether used explicitly or implicitly in a
ranking retrieval system.
Positional index size
• Need an entry for each occurrence, not just
once per document
• Index size depends on average document size Why?

– Average web page has <1000 terms

– SEC filings, books, even some epic poems … easily
100,000 terms
• Consider a term with frequency 0.1%
Document size Postings Positional postings
1000 1 1
100,000 1 100
 Rules of thumb

• A positional index is 2–4 as large as a non-

positional index

• Positional index size 35–50% of volume of

original text

– Caveat: all of this holds for “English-like” languages

Combination schemes

• These two approaches can be profitably combined

– For particular phrases (“Michael Jackson”, “Britney
Spears”) it is inefficient to keep on merging positional
postings lists
• Even more so for phrases like “The Who”
• Williams et al. (2004) evaluate a more sophisticated
mixed indexing scheme
– A typical web query mixture was executed in ¼ of the
time of using just a positional index
– It required 26% more space than having a positional
index alone
IR vs. databases: Structured vs unstructured data

• Structured data tends to refer to information

in “tables”

Employee Manager Salary

Smith Jones 50000
Chang Smith 60000
Ivy Smith 50000

Typically allows numerical range and exact match

(for text) queries, e.g.,
Salary < 60000 AND Manager = Smith.
52
Unstructured data

• Typically refers to free text

• Allows
– Keyword queries including operators
– More sophisticated “concept” queries e.g.,
• find all web pages dealing with drug abuse
• Classic model for searching text documents

53
Semi-structured data

• In fact almost no data is “unstructured”

• E.g., this slide has distinctly identified zones such
as the Title and Bullets
• … to say nothing of linguistic structure
• Facilitates “semi-structured” search such as
– Title contains data AND Bullets contain search
• Or even
– Title is about Object Oriented Programming AND
Author something like stro*rup
– where * is the wild-card operator

54
Performance Measures

Precision, Recall and F-Measure

Example

• Let’s discuss the performance of a classification

(Predict Category)
• Example: we aim at classifying cars
• Confusion Matrix

P N
(Predicted) (Predicted)
P True Positive False Negative
(Actual)
N False Positive True Negative
(Actual)

Mercedes Not Mercedes

Confusion Matrix
P N
(Predicted) (Predicted)
P True Positive False Negative
(Actual)
N False Positive True Negative
(Actual)

• True Positive: predict Mercedes car is really a Mercedes

• False Positive: Predict Non-Mercedes car as Mercedes
• False Negative: Predict Mercedes car as non Mercedes
• True Negative: Predict non-Mercedes car as Non-Mercedes
Performance Measure: Simple Accuracy
P N
(Predicted) (Predicted)
P True Positive False Negative
(Actual)
N False Positive True Negative
(Actual)

𝑁𝑜 𝑆𝑎𝑚𝑝𝑙𝑒𝑠 𝑃𝑟𝑒𝑑𝑖𝑐𝑒𝑑 𝐶𝑜𝑟𝑟𝑒𝑐𝑡𝑙𝑦

• Simple Accuracy =
𝑇𝑜𝑡𝑎𝑙 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑆𝑎𝑚𝑝𝑙𝑒𝑠
• What is wrong with this measure ?
Performance Measures

• Let’s classify 10,000 cars

• Our classifier shows
– 9,990 Non-Mercedes
– 10 Mercedes

9,990
• Accuracy = = 99.9 %
10,000
• This accuracy makes no sense and is misleading
Performance Measures

• Precision: of the car classified Mercedes, how

many are actually Mercedes ?
• Recall: of the cars that are actually Mercedes,
how many are classified as Mercedes?
•
Performance Measures
P N
(Predicted) (Predicted)
P True Positive False Negative
(Actual)
N False Positive True Negative
(Actual)

• Precision: of the car classified Mercedes, how

many are actually Mercedes ?
1. Number of cars classified Mercedes = TP + FP
2. Number of cars actually Mercedes (when
classified as Mercedes) = TP 𝑇𝑃
Precision =
𝑇𝑃+𝐹𝑃
Performance Measures
P N
(Predicted) (Predicted)
P True Positive False Negative
(Actual)
N False Positive True Negative
(Actual)

• Recall: of the cars that are actually Mercedes,

how many are classified as Mercedes?
1. Number of cars actually Mercedes = TP + FN
2. Number of cars classified Mercedes (when
actually Mercedes) = TP 𝑇𝑃
Recall =
𝑇𝑃+𝐹𝑁
Performance Measures : Example cars

𝑇𝑃 𝑇𝑃
Precision = Recall =
𝑇𝑃+𝐹𝑃 𝑇𝑃+𝐹𝑁

P N
(Predicted) (Predicted)
P
Precision = 0
0
(𝑢𝑛𝑑𝑒𝑓𝑖𝑛𝑒𝑑)
(Actual)
True Positive False Negative

0 10
0 N
Recall = =0 (Actual)
False Positive True Negative
0+10
0 9,990
9,990
• However, Accuracy = = 99.9 % !!!
10,000
 Performance Measures : an Other Example

𝑇𝑃 𝑇𝑃
Precision = Recall =
𝑇𝑃+𝐹𝑃 𝑇𝑃+𝐹𝑁

P N
(Predicted) (Predicted)

Precision = 10+9,990
10
= 0.001 P
(Actual)
True Positive False Negative

10 0
10 N
Recall = = 1.0 (Actual)
False Positive True Negative
10+0
9,990 0
𝑇𝑃+𝑇𝑁
• However, Accuracy = = 0.1 %
𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁
!!!
 Comparing Performance of two Systems

• System 1 • System 2
– Precision = 70%
– Recall = 60%
? – Precision = 80%
– Recall = 50%

• Witch system is better ?

• We need a single measure to compare the performance of the
two systems
1
F𝜷 = 1 1
𝛽× + 1−𝛽 ×
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑅𝑒𝑐𝑎𝑙𝑙

• 𝛽 𝑖𝑠 𝑎 𝑓𝑎𝑐𝑡𝑜𝑟 𝑡ℎ𝑎𝑡 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 𝑡ℎ𝑒 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑐𝑒 𝑜𝑓 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛

• Greater 𝛽, Greater importance of Precision
Comparing Performance of two Systems
1
F𝛽 = 1 1
𝛽× + 1−𝛽 ×
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑅𝑒𝑐𝑎𝑙𝑙

• Cancer classification requires Precision, hence

high 𝛽.
• Making sure Mercedes Car are classified such
as requires Higher Recall
• System 1 • System 2
– 𝛽 = 0.5
– F𝛽 = 0.6461
> – 𝛽 = 0.5
– F𝛽 = 0.6153

System 1 performs greater than System 2

Comparing Performance of two Systems
1
F𝛽 = 1 1
𝛽× + 1−𝛽 ×
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑅𝑒𝑐𝑎𝑙𝑙

• Cancer classification requires Precision, hence

high 𝛽. → 𝑒. 𝑔., 𝜷 = 0.95

• System 1 • System 2
– 𝛽 = 0.95
– F𝛽 = 0.6942
< – 𝛽 = 0.95
– F𝛽 = 0.7766

System 2 has a better performance than System 1