Ranking Academic Web Pages and Research Documents Based on Importance.

      Аннотация: Traditional Ranking Techniques such as Page Rank, Hits etc. rely on the structure of the hyperlink where web spammers can easily exploit the hyperlink structure for their profits to add more links to their webpages and the same links can easily be taken back once the page rank is calculated. Thus Browse Rank gets popularity in which ranking is based on the browsing behavior of the web user. Instead of considering number of inlinks and outlinks, the number of hits a web page can be a competing factor. The higher the number of hits, the higher the page value. But Browse Rank gives equal value for all hit. It treats all the web users equally. In our work we consider academic and research web users and we propose a framework for ranking academic web pages and research documents in which we evaluate web users, score them based on their academic contributions in terms of various publications they published. We also identify their area of interest based on their publications. Finally we rank academic web pages and scientific articles by considering web user’s score.

Ключевые слова: personalized search, Page Ranking, SVM, classification, User behaviour, Markov model, Weighted Ranking