An exploration of link-based knowledge map in academic web space

The World Wide Web has become an important source of academic information. The linking feature of the Web has been used to study the structure of academic web, as well as the presence of academic and research institutes on the Web. In this paper, we propose an integrated model for exploring the subject macrostructure of a specific academic topic on the Web and automatically depicting the knowledge map that is closer to what a domain expert would expect. The model integrates a hyperlink-induced topic search (HITS)-based link network extending strategy and a semantic based clustering algorithm with the aid of co-link analysis and social network analysis (SNA) to discover subject-based communities in the academic web space. We selected to use websites as analytical units rather than web pages because of the subject stability of a website. Compared with traditional techniques in Webometrics and SNA that have been used for such analyses, our model has the advantages of working on open web space (capability to explore unknown web resources and identify important ones) and of automatically building an extendable and hierarchical web knowledge map. The experiment in the area of Information Retrieval shows the effectiveness of the integrated model in analyzing and portraying of subject clustering phenomenon in academic web space.