A REPUTATION BASED TECHNIQUE TO DETECT CLOAKED WEB SPAM

Abstract

Web spamming techniques are used by spammers in order to boost one's PageRank of the page and show themselves in top of the results. Spamming techniques can be classified as boosting techniques and hiding techniques. Search engines are facing problems mainly due to hiding techniques. Cloaking is a kind of hiding technique which is used to return different pages to the crawler and the user on their request to the cloaked web server. The cloaked web server always return same page to the user but it returns different pages to different crawlers because different search engines use different PageRanking techniques. Dynamic cloaking is a kind of cloaking technique in which web servers intermittently send cloaked and non cloaked pages to crawler on its request. There is no method to detect dynamic cloaking to the best of our knowledge. This dissertation discusses various techniques that exist to detecting cloaking and an exhaustive comparison between the existing techniques is done. In this dissertation a model to detect cloaking based on reputation has been proposed, implemented and evaluated. The proposed technique showed better space and temporal performance than the existing techniques.