Efficient and Trustworthy Review/Opinion Spam Detection

Abstract

The most common mode for consumers to express their level of satisfaction with their purchases is through online ratings, which we can refer as Online Review System. Network analysis has recently gained a lot of attention because of the arrival and the increasing attractiveness of social sites, such as blogs, social networking applications, micro blogging, or customer review sites. The reviews are used by potential customers to find opinions of existing users before purchasing the products. Online review systems plays an important part in affecting consumers' actions and decision making, and therefore attracting many spammers to insert fake feedback or reviews in order to manipulate review content and ratings. Malicious users misuse the review website and post untrustworthy, low quality, or sometimes fake opinions, which are referred as Spam Reviews. In this study, we aim at providing an efficient method to identify spam reviews and to filter out the spam content with the dataset of gsmarena.com. Experiments on the dataset collected from gsmarena.com show that the proposed system achieves higher accuracy than the standard na?ve bayes.