Learning to Understand Information on the Internet: AnExample-Based Approach

摘要: The explosive growth of the Web has made intelligent software assistants increasingly necessary for ordinary computer users. Both traditional approaches—search engines, hierarchical indices—and agents require significant amounts human effort to keep up with Web. As an alternative, we investigate problem automatically learning interact information sources on Internet. We report ShopBot and ILA , two implemented that learn use such resources. learns how extract from online vendors using only minimal knowledge about product domains. Given home pages several stores, autonomously shop at those vendors. After its is complete, able speedily visit over a dozen stores CD vendors, information, summarize results user. translate Internet into own internal concepts. builds model source specifies translation between source‘s output ‘s world. capable leveraging small amount domain models many sources. show fast accurate, requiring number queries per source.