PHS

Abstract

Information gathering on the Web is based on hyperlink traversal. Especially, topic-based information search and extraction is an important technique to support many new services on the Web. In this paper, we propose a webpage traversal algorithm called a Priority-based Hyperlink Search(PHS) which uses a set of well-defined rules to determine the traversal direction over webpages based on the priority ranks of respective hyperlinks. The prioritization rules are modeled based on human user``s webpage navigation behaviors of hyperlink traversal. During the traversal, all the hyperlinks in the present webpage are given priorities according to the priority rules so that the overall traversal flow moves toward the webpages with higher probabilities of containing matching information to the topic of search interest. To determine the priority of hyperlink, the PHS divide the webpage into basic information units using similarity thesaurus. The algorithm terminates by dynamically measuring the statistical matching trend on each traversal. This allows application designers to design services with a faster and more accurate information gathering. Web application designers (or application users) can choose the levels of information collection for the purpose of their applications. We carried out a performance evaluation and analysis on the proposed algorithm in comparison with other algorithms using a test document with an answer set and a query set.