Helping users understand search results with categorized overviews

Try a search... and send us your feedback. Remember that this is research software. It may not work properly. (It may not even work at all.) Please send problem reports, comments, or suggestions to Bill Kules, wmk@cs.umd.edu. Thanks!

Query:

Project description

Categorized overviews of search results enable informative forms of interaction to support exploratory search. Meaningful and stable categories help searchers explore and understand large result sets.

Search engines are effective at generating long lists of search results. But the lack of overviews prevents searchers from effectively organizing and exploring their results. Categorizing search results using meaningful and stable classifications helps:

• Exploratory searchers - who have vague or evolving information needs
• Searchers who lack detailed knowledge about the subject of their search

In the following four screenshots, search results for the query "median" are coupled with a categorized overview based on thematic, geographic, and temporal categories. The number of results in each category provide a simple query preview. (Click an image to view it full-size.)

query median

Moving the pointer over the Science category pops up a list of its nonempty subcategories. It also highlights the visible results in the Science category in the result list:

science category

Moving the pointer over a result highlights the categories that it is a member of:

hightlight categories

Clicking on the Science category narrows the results to that category and updates the overview:

click science category

This screenshot shows results of the query "breast cancer" on government web sites. The top 200 search results have been organized into a hierarchy based on the federal government departments and agencies. Most of the results fall under the National Institutes of Health. The detailed result list has been filtered to NIH by clicking on the agency name. It is easy to which agencies have no pages in this set of search results (e.g. the Department of Education):

Search results shown with an expandable outliner overview

Principles of categorized search result visualization

We are developing a set of search result visualization principles, based on the premise that consistent, comprehensible visual displays built on meaningful and stable classifications will better support user understanding of search results.

1. Provide overviews of large sets of results (100-1000+)
2. Organize overviews around meaningful categories
3. Clarify and visualize category structure
4. Tightly couple category labels to result list
5. Ensure that the full category information is available
6. Support multiple types of categories and visual presentations
7. Use separate facets for each type of category
8. Arrange text for scanning/skimming
9. Visually encode quantitative attributes on a stable visual structure

Publications

Kules, B. (April 2006)
Supporting Exploratory Web Search with Meaningful and Stable Categorized Overviews
Ph.D. Dissertation from the Department of Computer Science

Kules, B., Kustanowitz, J., Shneiderman, B. (May 2006)
Categorizing Web Search Results into Meaningful and Stable Categories Using Fast-Feature Techniques
Proceedings of the 6th ACM/IEEE-CS joint conference on Digital libraries. 210-219.

White, R., Kules, B., Drucker, S., schraefel, m. (May 2006)
Supporting Exploratory Search
Communications of the ACM, 49(4), 36-39.

White, R., Kules, B., Bederson, B. (May 2006)
Exploratory Search Interfaces: Categorization, Clustering and Beyond
SIGIR Forum, volume 39, issue 2, December 2005

Kules, B., Shneiderman, B. (December 2005)
Using meaningful and stable categories to support exploratory web search: Two formative studies
Technical report HCIL-2005-31

Kules, B., Shneiderman, B. (January 2005)
Categorized Graphical Overviews for Web Search Results: An Exploratory Study Using U.S. Government Agencies as a Meaningful and Stable Structure
Proc. Third Annual Workshop on HCI Research in MIS, December 2004, Washington, DC

See also:
Distinguishing Forests from Trees in Search Engine Results
HCIL Research Highlight

Videos

[フレーム]

Downloadable MPG (18MB)

Participants

Bill Kules, Graduate Research Assistant
Jack Kustanowitz, Graduate Research Assistant
Ben Shneiderman, Professor, Computer Science

Support

This research is partially supported by an AOL Fellowship in Human-Computer Interaction and National Science Foundation Digital Government Initiative grant (EIA 0129978) "Towards a Statistical Knowledge Network."