Searches for New Information over World Wide Web

摘要

We address the problem of specifying Web searches based on user's search intent and preferences when the search target is not a well-known concept in English. A concept not well known in English is either a newly developed concept expressed in English or is a word or phrase that represents a foreign language concept in English. In such a case a Weighted Semantic-Taxonomy Tree (see [1]) may not be immediately constructed from the user search target. In addition, another tool, a Web taxonomy agent (see [2]) that was relied on in [1], may produce no results when applied to word/concept that is not well known in English. This paper extends the direction of work done in [1] that has proposed an agent-based system, called WebSifter II, that was a result of several years of research for development of a personalizable search engine [3, 4]. The work in [1] and experiments that it has described with WebSifter II has led us to investigate an additional mechanism to extend WebSifter II by providing for it an alternative form of obtaining information about user intent for Web search in order to better satisfy a user need for finding new information on the Web.