Chapter 14

Information Search and Visualization

Outline

I. New Terminology

• Information gathering, seeking, filtering (associated with visualization)
• Task objects are represented by interface objects. in structured relational databases, or libraries.
• Relations have items (tuples or records) with multiple attributes (fields).
• Attributes are atomic.
• Task actions are decomposed into browsing or searching.
• Task actions are represented by interface actions such as scrolling, zooming, joining or linking.
• Unstructured tasks include:

- Specific fact finding

- Extended fact finding

- Exploration of availability

- Open-ended browsing and problem analysis.

II. Searching in Textual Documents and Database Querying

Searching in textual documents

• Search engines make use of statistical ranking in their search.
• Most search engines provide a simple user interface but offer an advanced search

Database searches

• SQL has become a widespread standard for searching
• Requires training for use
• Type of queries:

- Natural language queries: limited to eliminating frequent terms or commands.

- Form-fillin queries: simplified query formulation (usually allow some Boolean combinations such as OR / AND)

• Five-phase Framework to coordinate design:

- Formulation: expressing the search

- Initiation of action: launching the search

- Review of results: reading messages and outcomes.

- Refinement: formulating the next step.

- Use: compiling or disseminating insight.

III. Multimedia Document Searches

• Image search: QBIC ( query by image content)
• Map search: searching by feature
• Design or diagram search: document-structure recognition
• Sound search: MIR (music-information retrieval) , uses audio input
• Video search: visual feature search (color, faces..) and textual features
• Animation search: searching of animation or morphing features

IV. Advanced Filtering and Search Interfaces

• Searching with complex Boolean queries: difficult to use
• Automatic filtering: user created profile
• Dynamic queries: uses sliders for numerical values, alphasliders for names, and buttons for small sets of categories
• Faceted metadata search: integrates category browsing with keyword searching
• Collaborative filtering: collective collaboration by users to combine their evaluations
• Multilingual searches: will allow users to select appropriate dictionaries and restrict keyword translation.
• Visual searches: uses specialized visual representations of possible values.

V. Information Visualization

a. 1D linear data

· linear, one dimensional

· include program source code / textual documents / dictionaries...

b. 2D map data

· include geographic maps / newspaper layouts...

· multiple-layer approach to deal with map data

c. 3D world data

· include objects that have volume and surfaces

d. Multidimensional data

· data contain items with n attributes in an n-dimensional space

· i.e. matrices that combine series of small bivariate representations

e. Temporal data

· Time series (i.e. electrocardiograms, weather data)

· items have a start and a finish time and items may overlap

f. Tree data

· each item has a link to a parent except the root.

· called hierarchies or tree structures

g. Network data

· items are linked to an arbitrary number of other items in a network

· complex in design

h. Overview task

· zoom feature, movable field-of-view box

i. Zoom task

j. Filter task

k. Details-on-demand task

l. Relate task

m. History task

n. Extract task

o. Challenges for information visualization

· import data

· combine visual representations with textual labels

· see related information

· view large volumes of data

· integrate data mining

· collaborate with others

· achieve universal usability