Web spam has extremely undermine the ranking structure of Web pages (or information retrieval) in search engines. The Web spam manipulates Web pages maliciously by different spamming echniques such as Boosting and Hiding Web spamming techniques. The spammer sole intention for undermine the ranking structure is to obtain a higher rank than they deserve for their Web pages which results negative impacts on search results. Modifying the search results ranking by spammers has raised the complication level of Web searching and time consumption for Web users to fetch significant information. In contemplation of enhance the quality of search engines results, the main focus of this research study is to develop an adaptive and effective method for identifying and classifying the web spam or search engine spam into pre-determined categories.

The overall goal of this thesis is to develop a transfer learning based method for detect Web spam or search engine spam which will be able to reduce training label insufficiency problem or domain adaptation problem to detect the unseen spam introduced from different distribution or feature space. Moreover, this method should be effective or accurate for detect and classifying the spam as one of search engine spam category. This thesis also provide the study on learning relevant important features for Web spam and their effect on performance of propose method.