Automates Process Formerly Accomplished by Human Editors
September 27, 2004 - CAMBRIDGE, Mass. - Eliyon Technologies, (www.eliyon.com), a next-generation search and automated content creation company, announced today that it has been awarded a U.S. Patent (No. 6,778,986), for a computer process that reads the content of a Web site, and performs a series of tests and evaluations to determine what "type" of site has been read. The newly patented process allows Eliyon to rapidly extract information from the site and use it to build profiles of business people and companies from the collected data. Eliyon has built a searchable database of more than 22 million profiles of business people from nearly 1.5 million companies.
"Eliyon has automated an editorial process that previously could only be accomplished by human editors, enabling our computers to create original content on a scale that previously was impossible, " said Jeremy Rothman-Shore, vice president of research and development for Eliyon Technologies. "Our new patent is proof of the uniqueness of our approach, which we have been able to translate to solid value for our customers."
The new patent covers the method and the apparatus for determining the type of a Web site. Unlike traditional search engines that are usually content to simply match strings of characters or words between a query and results, Eliyon's technology seeks to gain an understanding of the words in the query and the pages it reads on the Web. By understanding the type of Web site Eliyon's crawlers have visited, it determines the best way to extract the business-related information it seeks, for example using a very different procedure on a news site compared to an e-commerce site.
The system runs tests that provide quantitative statistical evaluation of candidate site types. Eliyon's computers evaluate site conditions including numbers of external and internal links, the site's morphology or structure and the distribution of multimedia elements in the site. The results of a mathematical combination of the test results cause Eliyon to conclude that a site is statistically likely to fit into a predetermined category, for example: