Spotting faked 5 stars ratings in E-Commerce using mouse dynamics

In the last decade, faked reviews have become a growing issue. While most studies have focused on text analysis to identify false reviews, the big companies are now switching to simpler review systems exclusively based on ratings (such as number of stars or like/dislike systems). In this paper, the possibility of detecting faked ratings from the analysis of mouse movements is explored for the first time. The participants were asked to evaluate twenty products on a five-star rating scale. For half of the products, they were encouraged to cheat to get an advantage. Results showed greater response times and wider mouse trajectories when users left false ratings then when they left true ones. Concerning the analysis of the ratings, results revealed an interaction between the type of rating (positive vs. negative) and deception (true vs. false rating). Moreover, the users’ attitudes toward the faked reviews has been investigated through a self-report questionnaire, pointing out that people are more prone to give neutral judgments rather than extremely positive or negative ratings when they are encouraged to cheat.