Analyzing Content-based Heuristics for Persian Web Spam Detection

The rapid growth of web spam in the World Wide Web has motivated researchers to propose algorithms for combating web spam. Despite using these techniques, the search engines do not perform well in detecting Persian spam websites. In this paper, we analyze the effectiveness of many previously proposed content-based features on detecting Persian spam websites, and also present a number of new content-based features. As another approach, we explain and examine our Bag-Of-Spam-Words (BOSW) method to do web spam detection. In this method, we represent each document as a vector of specific words selected from a spam corpus. Finally, we apply a number of feature selection methods and use various kinds of classification algorithms to classify the Persian websites. For this purpose, we have created a dataset of Persian hosts. Our results show that using the BOSW method with the SVM classifier has the best performance in detecting Persian spam websites.

کلیدواژه ها:

Persian web spam ، web spam detection ، cotent-based features