Web Email Spam Detection

Spam on the Web is an important issue for search engines. It reduces the quality of search results and frustrate genuine people trying to find content. You wouldn't even believe how much it costs economically just to have a high-ranking position in search engines. That brings so much advertisement value and a lot of traffic directed towards the site. This research work presents a sophisticated spam detection system consisting of link-based and language- model (LM)-based features applied to identify and filter web spam. The model combines quantitative data and qualitative link properties, including link reliability and semantic coherence between linked pages. It uses language modelling to analyse content with hyperlink context so that semantically unrelated pages will be flagged for potential web spam detection. Finally, it uses cosine similarity to match the content of a suspected page against trusted no spam pages to improve precision. This is achieved by joining structural and contextual analysis. Thus, this approach is a promising way of going decisive in detecting web spam so that a cleaner search engine result would lead towards an improved on browsing experience.