The Semantic Spider Web - The KnowItAll Project
(Page 3 of 4 )
Oren Etzioni, a computer scientist at the University of Washington, is leading an effort, known as KnowItAll, that takes the fundamental aspects of the Semantic Web and puts them to use through "domain independent systems that extract information from the Web in an autonomous, scalable manner." The system works by taking the most important facts from the content of each site on the Web. As of now, only obvious declarations of fact are capable of being understood.
Opine is an application from the KnowItAll project that cuts through online reviews and gathers the key facts relevant to the particular product. A restaurant, for instance, would be displayed with its location, a critique of the staff and food, etc., all in a short summary no more than a paragraph. Of course, users can click the hyperlinks to access more information. The application cannot be tested as of yet, as it is only a prototype, but it is well into the development stages.
TextRunner is another application from KnowItAll, similar to a search engine, that allows queries on general information like nutrition, history, and science. So far, it has mined information from over 100 million web pages, a mere one percent of the Web. A test search of George Washington yielded nine different categories arranged by most relevant and in a way that displays the results as a declaratory statement. Each statement also has the number of sources next to it, and if you click the number, the sources will be displayed as a list of separate, more specific statements.
(Screen shot of a TextRunner search results page)
Etzioni has already launched Farecast, which uses data mining techniques to find the cheapest airfares and hotel prices and then makes predictions about whether to make reservations or not, saving you the time of researching multiple sites and the risk of taking a wild guess. The test search I ran told me to wait at least seven days because there will likely be a significant drop in prices. And seeing as I ran my original searches on December 27, I'd say that's a good analysis.
Next: Getting a Clear Picture >>
More Web Hosting Articles Articles
More By Michael Lowry
