The three speakers this morning, Paul Pedersen from MarkLogic, Raul Valdes-Perez from Vivisimo, and Ben Bederson from the Human-Computer Interaction Lab at the University of Maryland, are explaining the technology they employ in their companies and the research they’re doing in aid of the search process. Paul discussed taxonomies, navigation, and the difference between search and query. Frankly, I missed the distinction until I realized that, to him, search, which he characterized as fuzzy, returning some relevant information, non-extensible, read-only, and using "pidgin" query language, equated more to Web search than the online searching I grew up with. Query, to him, was more in the computer science realm, where an exact query against a database returns every matching element and is fully programmable. That’s more like what I expect from Dialog, but I don’t expect it from Google or Yahoo!
Raul claims the main problem facing us today in not information overload, it’s information overlook. People, when faced with too much data simply ignore most of it. His answer to information overlook is, no surprise here, clustering results from multiple search engines and sources. Obviously, he’s showing how Clusty works. One of his query examples is bonabos, an animal I had never heard of before. Now he’s searching cockroaches. OK, I have heard of those.
Visualization is the topic of the third talk. It lets you find patterns and explore information. There are lots of pitfalls, such as meaningless spatial relationships. He advocates a zoomable user interface. He’s got a demo of a visual search engine for children’s books designed for kids to browse. There is an advanced search for adults that’s more textual, but the adults like the visual interface without the words. In another project, PaperLens, Ben’s showing relationships among conference papers. Now there’s an idea I’d like to see expanded beyond a lab experiment. It would be a great product for the real world.
Editor, ONLINE: The Leading Magazine for Information Professionals