Optimization of Multiple Kernels in Twin SVM for Decreasing Web Spam Page Detection Semantic Gap

Web pages are crawled and indexed by search engines for fast accessing data on the web. One of the challenges in the search engines is web spam pages. There are many approaches to web spam pages detection such as measurement of HTML code style similarity, pages linguistic pattern analysis and machine learning algorithm on page content features. One of the famous algorithms has been used in machine learning approach is Support Vector Machine (SVM) classifier. Unfortunately SVM could not achieve a reasonable accuracy in this scope. In order to classify non-linear data in a linear manner, the SVM needs to use the idea of the kernel, which leads to enhanced classification capabilities. A kernel, implicitly maps the data to a higher-dimensional space. Recently basic structure of SVM has been changed by new extensions called Twin SVM (TSVM) to increase robustness and classification accuracy using two separate hyperplanes. Because of using two separate hyperplanes in TSVM, it is better to use multiple kernels in it. Kernel functions are designed based on specific data sample. Therefore they cannot use for general purpose. In this paper we improved accuracy of web spam detection by using two nonlinear kernels into TSVM as an improved extension of SVM. These two kernels have been created based on genetic algorithm. The classifier ability to data separation has been increased by using two separated kernels for each class of data. Effectiveness of new proposed method has been experimented with two publicly used spam datasets called UK-2007 and UK-2006.