Intelligent information extraction and discovery for query and browsing operations

This presentation proposes an intelligent information discovery agent that may aid Internet users in performing efficient Web information retrieval and discovery through query formulation, information collection, information clustering and analysis. The underlying mechanism includes a fuzzy neural learning system for active learning that can be used in clustering operations. Due to the tremendous popularity and growth of heterogeneous information on the Internet, it may be difficult to index and categorize documents efficiently to speed up searching and browsing; this problem is made even more complex due to the broad spectrum of Internet users. Further, the internal representation of information retrieval systems may not always match the user´s intent (sometimes called the user concept), thus requiring several cycles of query, possibly with user feedback, to match the user´s desires with the retrieval results. Several approaches have been suggested to close the gap between the user´s intent and the actual retrieved documents taken from a dynamically increasing information database. These methods include intelligent browsing agents, meta-search engines that are interface agents to multiple search engines and Internet resources, and user concept matching for the Lotus Notes database. In this work, a novel intelligent query and browsing information framework is developed to aid searching, learning and discovering the user concept from multiple existing information retrieval systems. The framework is applied to accessing multiple Web search engines in parallel