Abstract

In this report we consider a number of approaches to improving the efficacy of search algorithms for the World Wide Web and similar hyperlinked document collections. We observe that the exploitation of link topology has resulted in high precision search engines, and investigate possibilities for augmenting these techniques. We examine the use of metadata and structural observations as sources of information. Local (per page) and distributed (per site) techniques are used to exploit this information for the categorisation of web pages.

Citations