Toward a Language Modeling Approach for Consumer Review Spam Detection

Numerous reports have indicated the severity of fake reviews (i.e., spam) posted to various e-Commerce or opinion sharing Web sites. Nevertheless, very few studies have been conducted to examine the trustworthiness of online consumer reviews because of the lack of an effective computational methodology. Unlike other kinds of Web spam, untruthful reviews could just look like other legitimate reviews (i.e., ham), and so it is difficult to apply any features to distinguish the two classes. One main contribution of our research work is the development of a novel computational methodology to combat online review spam. Our experimental results confirm that the KL divergence and the probabilistic language modeling based computational model is effective for the detection of untruthful reviews. Empowered by the proposed computational methods, our empirical study found that around 2% of the consumer reviews posted to a large e-Commerce site is spam.