Interface Support for the Search Process

The user interface designer must make decisions about how to arrange
various kinds of information on the computer screen and how to structure
the possible sequences of interactions.
1. Interfaces for String Matching
 A common simple search need is that of the 'find' operation,
typically run over the contents of a document that is currently being
viewed.

 Usually, this function does not produce ranked output, nor allow
Boolean combinations of terms; the main operation is a simple string
match (without regular expression capabilities).

 The next degree of complexity is the 'find' function for searching

across small collections, such as the files on a personal computer's
hard disk, or the history list of a Web browser.

 A common problem arises even in these very simple interfaces. An

ambiguous state occurs in which the results for an earlier search are
shown while the user is entering a new query or modifying the
previous one. If the user types in new terms and but then does not
activate the search, the interface takes on a potentially misleading
state, since a user could erroneously assume that the old search hits
shown correspond to the newly typed-in query. One solution for this
problem is to clear the results list as soon as the user begins to type
in a new query.

An example of a simple interface for string matching, from

Netscape Communicator
 Another solution is to bring up a new window for every new query.
  Spelling errors are a major cause of void result sets. A spell-checking
function that suggests alternatives for query terms that have low
frequency in the collection might be useful at this stage.

2. Window Management
 When arranging information within windows, the designer must
choose between a monolithic display, in which all the windows are
laid out in predefined positions and are all simultaneously viewable,
tiled windows, and overlapping windows.

 The conventional command-line interfaces used by bibliographic

search engines are TTY based. When the system replies to a
command, the previous results screen’s contents are erased, forcing
the user to recall the previous one’s information.

 The primary design decisions in such a system relate to the

command or menu structure and the sequence in which the various
alternatives are presented.

 The windowing concept can be utilized in contemporary graphical

interfaces to split functionality into various concurrently visible
windows.

 Overlapping windows provide flexibility in arrangement, but can

quickly lead to a crowded, disorganized display.

 Henderson and Card built a system intended to make it easier for

users to move between 'multiple virtual workspaces'. The system
uses a 3D tiled window.

- By 'traveling' from one room to the next, users can change

from one work context to another. In each work context, the
application programs and data files that are associated with
that work context are visible and readily available for
reopening and perusal.

- The user should be able to leave a room dedicated to some

tasks, work on another task, and three days later return to the
first room and see all of the applications still in the same state
as before.
  Elastic windows are an extension to the workspace or rooms notion
to the organization of 2D tiled windows. By altering how much
screen space is taken up by the current role, the main goal is to
make switching between roles or tasks easier.

3. Example Systems

The following sections describe the information layout and

management approaches taken by several modern information access
interfaces.

a.The InfoGrid Layout

Left-Hand Side: This section is further subdivided into several areas:

 Structured Entry Forms: At the top, there are structured entry

forms where users can specify the properties of their query. This
area likely includes fields for inputting search terms and setting
query parameters.
 Iconic Controls: A column of iconic controls lines the left side of
the left-hand section. These icons likely represent various actions
or functions related to the information access process, such as
search, filtering, or sorting.
 Area for Retaining Documents: Along the bottom of the left-hand
section, there is an area designated for retaining documents of
interest. Users can likely save or bookmark documents here for
future reference.

Main Central Area: The central area is the primary space for viewing
retrieval results. These results can be displayed in various formats,
including thumbnail representations of the original documents or
organized clusters of documents, such as Scatter/Gather-style cluster
results. Users can interact with and select documents from this area.

Right-Hand Side: The majority of the right-hand section is dedicated to

viewing selected documents. The upper portion of this area typically
displays metadata associated with the document currently being
viewed. Metadata might include information like the document's title,
author, publication date, and source.

Graphical History: Below the document display area, there is an area

intended to show a graphical history of earlier interactions. This history
likely provides users with a visual record of their previous searches,
selections, and interactions within the interface

b. The SuperBook Layout

The SuperBook layout is similar to that of the InfoGrid but with some
notable differences and a history of iterative design. Here's an
overview of the SuperBook layout and its design evolution:

Main Layout Structure:

 Left-Hand Pane: Like InfoGrid, SuperBook has a left-hand pane. It

retains a table of contents-like display, providing an outline of the
content structure. This left-hand pane also indicates how many
documents containing search hits are present at each level of the
outline. This feature helps users understand the relevance of
different sections of the document hierarchy.
 Query Formulation: Below the table of contents view, SuperBook
provides an area for query formulation. Users can input their
search terms and specify query parameters here. This aspect allows
users to fine-tune their search.

 Terms Related to Query: Terms related to the user's query are

displayed in this window, assisting users in understanding the context
and potentially refining their search.

 Right-Hand Pane: The main right-hand pane is used for displaying

selected documents. Users can view the content of documents in
this area.

Design Evolution:

 Early Overlapping Windows: In earlier versions of SuperBook,

overlapping windows were used. Users could create new text
boxes by selecting a rectangular area on the screen. These text
boxes had associated buttons that allowed users to jump to
occurrences of highlighted words in other documents or to the
table of contents. This design allowed for more flexibility but had
its challenges.
 Redesign for Efficiency: SuperBook underwent a redesign after
observing results from experimental studies. The redesign aimed to
provide users with fewer, well-chosen interaction paths, rather than
a more flexible but potentially confusing interface. Users were
found to be more efficient with this redesigned interface.

 Automating Common Actions: To streamline user interactions,

the system was redesigned to automate common actions. For
instance, if users wanted to view the overall frequency of a hit
and then navigate to the table of contents, the system was
redesigned to perform these actions automatically after a search.

 Adaptation for Smaller Displays: SuperBook also attempted a

redesign to make the interface more suitable for smaller displays.
This adaptation involved using small, overlapping windows to
accommodate limited screen space.

c. The DLITE

Interface Split
Functionality:

 Control of Search Process: DLITE separates its functionality into two

main parts. The first part focuses on controlling the search process.
Users interact with this part using a graphical, direct manipulation
display.

 Display of Results: The second part is dedicated to displaying search

results. DLITE uses a separate web browser window for this
purpose, where users can view detailed information about retrieved
documents, including bibliographic citations and full text. This
separation helps maintain a clear distinction between controlling
the search and viewing the results.

Graphical Representation:

 Graphical Objects: DLITE represents queries, sources, documents,

and groups of retrieved documents as graphical objects. For
instance, queries are created by filling out
 editable fields within a query constructor object, and they are
represented by small icons.

 Drag-and-Drop Interaction: Users can interact with these graphical

objects through drag-and-drop interactions. For example, a user
can drag a query icon and drop it onto iconic representations of
collections or search services. When a service is active, it responds
by creating an empty results set object and attaching the query to
it.

 Manipulation of Documents: The interface allows users to

manipulate documents within the result set. For example,
documents can be dragged out of the result set pool and dropped
into other services, such as a document summarizer or a language
translator.

Flexibility in Interaction:

 Flexible Architecture: DLITE provides a flexible interface

architecture that allows users to perform actions in a non-linear
order. Users are not bound by a rigid sequence of commands, and
they can explore different interaction paths.

 Guidance through Animation: To help users navigate this flexible

interface, DLITE makes extensive use of animation. For example, if a
user attempts an illegal operation, such as dropping a query into the
document summarizer icon, the system responds with animation,
moving the object left, right, and left again to mimic a "shake-the-
head-no" gesture. Animation is also used to convey the system's
state, such as showing the progress of search results retrieval.

Workspaces:

 Workspace Concept: DLITE incorporates the notion of a

"workspace" or "workcenter" for different kinds of tasks. Each
workspace can be customized with source icons representing
relevant sources and services for a specific task. For example, a
workspace for buying computer software might include icons for
reviewing software, searching for price information, and accessing
online credit services.

Reusable Queries and Retention:

 Query Reusability: DLITE allows users to store and reuse queries.

 This feature facilitates the retention of previously successful search
strategies, making it easier for users to revisit and refine their
searches over time.

d. The SketchTrieve

Interface Informal

Information Access:

 Informal Process: SketchTrieve is designed to depict information

access as an informal process where users can explore half-finished
ideas and partly explored paths. It recognizes that information
seeking is not always a structured or linear process.

Canvas or Workspace:
 Concept of a Canvas: SketchTrieve employs the concept of a
canvas or workspace where users can arrange and organize
retrieval results in a side-by-side manner. This canvas serves as a
space for retaining the context of previous interactions.

Graphical Representation:

 Graphical Objects: Information and search results are represented

as graphical objects within the canvas. These graphical objects can
include retrieved documents, search queries, annotations, and
connectors between them.

Flexible Manipulation:

 Flexible Operations: SketchTrieve allows users to perform

various operations on graphical objects and connectors. Users
can manipulate these elements to create, compare, and
recombine results as needed.

 Comparison and Recombination: The interface facilitates the

comparison of retrieval results by allowing users to arrange
documents side by side. Users can also recombine results by
connecting related elements via connectors.

Retention and Reusability:

 Retention of Context: The canvas serves as a space for retaining the

context of previous interactions. Users can save and bring back their
work, even if it's in a half-finished state, for later exploration and
refinement.

Handling Complex Queries:

 Dealing with Interrelated Queries: SketchTrieve addresses the

challenge of handling sets of related retrieval results, which may
involve minor modifications based on query expansion, relevance
feedback, and other modifications.

 Stacking or Grouping: One approach in SketchTrieve is to show sets

of related retrieval results as a stack of cards within a folder. This
enables users to extract subsets of cards for side-by-side viewing or
perform operations like difference comparisons.
The SketchTrieve interface encourages a more flexible and creative
approach to information access. By providing a canvas for users to
work with graphical objects and connectors, it supports the informal
nature of information seeking, making it possible to retain,
manipulate, and compare search results and queries in a way that
suits individual exploration styles.

4. Examples of Poor Use of Overlapping

Windows Overlapping Windows Display

(Bibliographic System):

Problem: In the case of the bibliographic system described, an

overlapping windows display was used. This approach can lead to
problems when users need to remember the context as each new
window obscures the previous one.
Solution: A later redesign of the system adopted a monolithic
interface, which is a more structured and user-friendly approach.
Monolithic interfaces provide a clear layout with
different sections for various functions, making it easier for users to
access and manipulate information.

Literal Transformation (LEXIS-NEXIS Interface):

Problem: The LEXIS-NEXIS interface suffered from a literal

transformation from a TTY interface to a graphical interface. It did not
take advantage of the graphical environment's capabilities and failed
to provide an efficient user experience.

Issues Highlighted:
 Lack of window context retention when switching functions.
 Poorly-formatted ASCII rendering of metadata.
 Inefficient use of available screen space for displaying full-text
documents.

Solution: To address these issues, the LEXIS-NEXIS interface could

benefit from the following improvements:

 Implement a more dynamic and responsive interface that

retains context when switching functions.
 Utilize graphical capabilities for rendering metadata,
improving readability and presentation.
 Optimize the display of full-text documents to make better use
of available screen space and provide a more user-friendly
reading experience.

5. Retaining Search History

Showing Available Choices and Search History:

 Information seeking strategies and behaviors include showing

available choices at any given point, documenting past moves,
tactics, and longer-term strategies.
 Users should be able to annotate the choices made and
information found during a search session.
 Search sessions should be bundleable, allowing users to save
and access individual portions of a session.
Incorporating Personal Preferences:

 There is interest in incorporating personal preference and usage

information into query formulation and search results utilization.
History Mechanisms in User Interfaces:

 History mechanisms, such as command lists, have been

available in many search systems in the past.
Graphical History Presentation:

 Graphical history presentation mechanisms have been introduced,

allowing tracking of commands and results visually.
Hierarchical Structure for Access History:
 Some interfaces use a hierarchical structure, like a "slide sorter"
view, to save the state in miniature form.
 The VISAGE system for information visualization is an example of an
interface that uses this approach.
VISAGE Interface:

 VISAGE, while primarily designed for creating presentations, can

also be useful for retaining search action history.
 It allows users to visualize individual states, making it easier to
recognize and navigate through search history.
In summary, retaining search history is crucial for supporting various
information seeking strategies and behaviors. While traditional history
mechanisms exist, newer graphical and hierarchical approaches, like
those in the VISAGE system, provide more effective ways to document
and access search history in information access interfaces. These
mechanisms enhance the usability and efficiency of such interfaces by
helping users keep track of their actions and decisions during
information retrieval tasks.

6. Integrating Scanning, Selection, and

Querying Challenges in Information

Access Interfaces:

 Many information access interfaces do not effectively support

strategies or sequences of operations.
 Operations like using the output of one query as input for
another query in a later search session are often not well-
supported.
Integration of Scanning and Querying:

 Users prefer interfaces that integrate scanning and query

specification.
 However, overly unrestricted interactions can lead to errors and
wasteful behavior, suggesting that interactions need some level
of guidance.
Examples of Integrated Interfaces:

 The new web version of the Melvyl system allows users to take the
output of one query and modify it for re-execution.
 Workspace-based systems like DLITE and Rooms enable the
 storage and reuse of previous states but may not integrate
scanning, selection, and querying effectively.
The Cat-a-Cone Interface:

 Cat-a-Cone integrates querying and browsing of large category

hierarchies and text collections.
 It uses 3D+animation interface components to support browsing and
searching.
 Key components include the separation of category
hierarchy from document representation and the
assignment of first-class status to text content.
 Users can start searches by typing category names or free-text
queries.
 Retrieved documents are presented in a 3D+animation book
format, allowing users to interact with them.
 The interface combines metadata, starting points, scanning,
querying, and relevance feedback.
Evaluation and User Preferences:
 User preferences for visible hierarchical organization, zooming,
and easy navigation support the design decisions in the Cat-a-
Cone.
 The interface provides hierarchical representation of term
meanings, simultaneous viewing of different levels of description,
and the ability to jump between categories easily.
 Cat-a-Cone also allows documents to be placed at intersections of
multiple categories and explicitly links document contents with
category representation.

In summary, the integration of scanning, selection, and querying is a

challenge in information access interfaces. While many interfaces do
not effectively support these operations, some, like the Cat-a-Cone
interface, aim to provide a fluid and flexible interaction between
browsing and search, as well as between categories and documents.
Such interfaces offer advantages like hierarchical representation,
simultaneous viewing of categories, and ease of navigation,
contributing to improved user experiences during information
retrieval tasks.