The Institute of Physics Publishing (IOPP; http://www.iop.org), a U.K.-based scholarly society publisher with a record of innovation in electronic publishing, has selected Vivísimo (http://vivisimo.com), a Pittsburgh-based advanced search engine company, to enhance its Electronic Journals service. Vivísimo runs a highly acclaimed Web metasearch engine, but it also markets its software to publishers and other data-hungry environments.According to Lloyd Fletcher, manager of enterprise application development at IOPP, the particular aspect of Vivísimo's technology that attracted IOPP was its search results clustering. He said: "Search results clustering is the way of the future for information retrieval, and Vivísimo is a huge leap forward in that direction. By experimenting with Vivísimo's approach, we aim to demonstrate how clustering can add tremendous value for the tens of thousands of physicists who use our online journals." IOPP plans to make the enhanced access available for searching to both subscribers and nonsubscribers of its Electronic Journals (http://www.iop.org/EJ) service. Searchers will receive results with clusters organized into key terms.
Raul Valdes-Perez, president and co-founder of Vivísimo, said: "Users who search for scientific information ignore most of what they find because of the sheer tedium of plodding through search results that are displayed one-dimensionally. The Vivísimo Clustering Engine will allow physicists to see much further into the physics literature, just as telescopes allowed astronomers to see farther in the heavens. I predict that, just as with the invention of telescopes, users will more easily discover knowledge they didn't even know existed."
Vivísimo receives high praise from leading critics of the search engine field. Gary Price, co-author of The Invisible Web and Webmastery columnist for Searcher magazine, is a "big fan" and "uses it all the time" in its metasearch Web search engine version. He particularly finds the clustering useful. He said: "Unlike Northern Light folders, which are pre-mapped to a fixed taxonomy structured by librarians, Vivísimo creates its clusters on the fly using terms taken from the brief descriptions or snippets in the search results."
Because IOPP deals with a highly focused subject area and a field that has an elaborate jargon of its own, Vivísimo will face different challenges in handling the company's publications than it would with broad Web metasearch issues. However, Vivísimo already has a version of PubMed (MEDLINE) available on its site to showcase its search engine. Price pointed out that even if the new service "required a lot of tweaking, there's a lot of brain power at Vivísimo to do it." The Vivísimo service was founded in June 2000 by Carnegie Mellon University researchers and receives support from the Commonwealth of Pennsylvania's Innovation Works, a regional development company, as well as Carnegie Mellon.
The Institute of Physics, which has a membership of 35,000, promotes the advancement and dissemination of knowledge in pure and applied physics. It launched its Web site in 1995. Fletcher commented that Vivísimo looked particularly attractive to the organization because IOPP has no classification or taxonomy as applied to its own publications, making a document-clustering tool like Vivísimo's especially useful. However, he indicated that if IOPP had had a taxonomy, Vivisimo could have incorporated it. Since the IOPP Web site carries some outside databases, such as the Institute of Electrical Engineers' INSPEC file, I asked if IOPP would possibly incorporate Vivísimo into its external files if the "experiment" proved successful. Fletcher indicated it might.
When IOPP tested the clustering feature with in-house experts, Fletcher reported that the Vivísimo treatment was judged 90-percent right in its clustering. The feature also helps in discovering new areas of subject development, avoiding the "mummy's curse," in which human catalogers have to recognize a term before approving it for usage and then leaving the earlier material using the term un-indexed and unretrievable by that term as an authorized descriptor or metatag. With fully automated, "real-time" subject groupings, such as Vivísimo offers, this problem could be avoided.
To see what the enhancements will look like when fully implemented, check out a test run on a year's worth of articles from the Journal of Physics: Condensed Matter (http://vivisimo.com/docs/jpcm.html). The initial launch of the enhancement covering all of IOPP's 29 journals on its site should take place sometime in mid-April.