Article


Article Code : 139507081215313470(DOI : 10.7508/jist.2016.03.008)

Article Title : The Surfer Model with a Hybrid Approach to Ranking the Web Pages

Journal Number : 15 Summer 2016

Visited : 1258

Files : 643 KB


List of Authors

  Full Name Email Grade Degree Corresponding Author
1 Javad Paksima Paksima@pnu.ac.ir Faculty Member PhD
2 homa khajeh khajeh121@yahoo.com Graduate M.Sc

Abstract

Users who seek results pertaining to their queries are at the first place. To meet users’ needs, thousands of webpages must be ranked. This requires an efficient algorithm to place the relevant webpages at first ranks. Regarding information retrieval, it is highly important to design a ranking algorithm to provide the results pertaining to user’s query due to the great deal of information on the World Wide Web. In this paper, a ranking method is proposed with a hybrid approach, which considers the content and connections of pages. The proposed model is a smart surfer that passes or hops from the current page to one of the externally linked pages with respect to their content. A probability, which is obtained using the learning automata along with content and links to pages, is used to select a webpage to hop. For a transition to another page, the content of pages linked to it are used. As the surfer moves about the pages, the PageRank score of a page is recursively calculated. Two standard datasets named TD2003 and TD2004 were used to evaluate and investigate the proposed method. They are the subsets of dataset LETOR3. The results indicated the superior performance of the proposed approach over other methods introduced in this area.