Discovering web document associations for web site summarization

Abstract

Complex web information structures prevent search engines from providing satisfactory context-sensitive retrieval. We see that in order to overcome this obstacle, it is essential to use techniques that re- cover the web authors' intentions and superimpose them with the users' retrieval contexts in summarizing web sites. Therefore, in this paper, we present a framework for discovering implicit associations among web documents for effective web site summarization. In the proposed frame- work, associations of web documents are induced by the web structure embedding them, as well as the contents of the documents and users' interests. We analyze the semantics of document associations and describe an algorithm which capture these semantics for enumerating and ranking possible document associations. We then use these asociations in creating context-sensitive summaries of web neighborhoods.