Shrinking the Internet
Visual software tool speeds up Web searches
BLOOMINGTON, Ind. (Tuesday, October 15, 2002) -- Google leave you a little flat? Search results pages feel too one-dimensional for your taste? It may soon be a lot easier to find that informational needle in the Internet haystack.
A new approach to showing Internet search engine results, developed by Indiana University information scientists, doesn't overload the user with pages of text, but instead summarizes possible topics of interest with graphical panache.
"Our goal was to develop intelligent tools that seek out information for the user in a way that's easier for them to digest," said information scientist and informatics expert Javed Mostafa, who oversaw the project and wrote some of the algorithms that drive the new interface. "Past studies have shown that people are much more likely to process pictures faster than long lists of textual information. And they have a better memory for pictures."
For example, searching for pop icon Britney Spears' name would result in a selection of topics rather than specific pages, with similar topics sharing a color or shade, floating near each other in a sea of other possible topics of interest, Mostafa said. "History" and "biography" might be colored blue and found on the far right side of a results screen, with topics not flattering toward the singer grouping themselves elsewhere. Clicking on a particular topic would then lead the Web browser to a new screen with links to pages most in tune with the selected topic.
Mostafa, the Victor H. Yngve Associate Professor of Information Science and associate professor of informatics at IU, helped create an online working example of the new approach, viewable at http://ella.slis.indiana.edu/~junzhang/dlib/IV.html. Topics shown in the example were gleaned from the Web site of the electronic publication D-Lib Magazine, which this month features an article about Mostafa's project, at http://www.dlib.org/dlib/october02/zhang/10zhang.html.
To get similar topics like "teacher" and "educate" to appear close to each other on the graphically oriented results screen, Mostafa and his team developed programs to extract key terms from thousands of common English terms that appear in documents and to calculate similarity values among these terms.
While the tool is not yet extendable to the entire World Wide Web, Mostafa said he would like to improve on already excellent search tools that will make finding information on the Web a lot easier.
"Currently, you can put in a search term or two with a popular search engine such as Google," Mostafa said. "You get a list. You can go forward or backward. You can click on an item in the list, and that's it. That's the end of the search. Visualization allows you to start with a list of key topics that are covered in the selection of hits, making it more likely you'll actually find what you're looking for."
Mostafa has faculty appointments in the School of Library and Information Science, the School of Informatics, and the Cognitive Science Program. He is the director of the Laboratory of Applied Informatics Research located at IU's Bloomington campus (http://lair.indiana.edu).
Research assistants Junliang Zhang and Himansu Tripathy helped implement some of the algorithms driving the software and are co-authors of the D-Lib paper. The project was funded by a grant from the National Science Foundation Digital Libraries Program (http://www.dli2.nsf.gov).
To talk to Mostafa, contact David Bricker at 812-856-9035 or firstname.lastname@example.org.
Office of Communications Marketing
530 E. Kirkwood Ave., Suite 201
Bloomington, IN 47408
"Information Retrieval by Semantic Analysis and Visualization of the Concept Space of D-Lib(R) Magazine," D-Lib Magazine, October 2002, Vol. 8 No. 10
Related SLIS News stories:
IU Home Pages
Google got you down? Visualize this
Posted October 17, 2002