Unit-based web site search.

Abstract

Nowadays, web sites may contain large amount of information and searching within web sites has become an important means to access information. Unfortunately, existing site search engines often fail to return web pages that best match user information needs a they ignore site-specific knowledge of the indexed web sites. Meanwhile, web unit mining is a new classification approach that explores conceptual knowledge to classify web pages from a web site. It models a web site with a set of concepts and discovers instances of those concepts, known as web units. Web units together provide the semantic structure of a web site, which could be used to enhance site search.