UW News

November 20, 2008

Tapping librarians, researchers aim beyond Google toward credibility

Googling might indeed be like searching a box of chocolates, but a new research effort aims to show where the best goodies are.

Reference Extract is a new international effort to build more credible Web searches based on vetting by librarians around the world. Michael Eisenberg, a professor and former dean at the UW School of Information, is a lead researcher on the project.

The planning phase is funded by a $100,000 grant from the John D. and Catherine T. MacArthur Foundation. The information school at Syracuse University and the Online Computer Library Center, the largest library cooperative in the world, are partners on the project as well. OCLC shares resources among 69,000 libraries in 112 countries and territories.

“Sometimes, the simplest ideas are the most powerful,” said Eisenberg. “The best search engines are great for basic searches, but sometimes Web site results lack credibility. If a librarian recommends a site, however, you can be more sure.”

Search results will be weighted toward sites most often used by librarians at institutions such as the Library of Congress, the UW, the State Library of Maryland and more than 2,000 other libraries around the world.

Instead of overtly asking librarians for suggestions, however, researchers will mine their work unobtrusively, thereby gathering hundreds of thousands, if not millions of Web sites.

If the project goes as planned, librarians will serve users on a whole new scale, said R. David Lankes, an associate professor at Syracuse and director of its Information Institute.

Asked why the librarians’ work will be tapped rather than that of topic experts, Eisenberg said librarians are more likely to focus on unbiased sources. “They may be in the best position to judge credibility itself,” Eisenberg said.

The research will use the same sort of frequency decisions — the number of times a Web site is linked — as those used by Google, but also the context in which a site has been recommended. For example, was it for specialists or the general public? “We’ll try a lot of different things,” said Eisenberg. “Remember that Google tweaks its algorithms on a daily basis.”

Reference Extract follows the work of the Credibility Commons, another joint effort between UW, Syracuse and the MacArthur Foundation, which examined the scope of Web site credibility.

People interested in learning more about Reference Extract and providing input may visit its Web site.