Multiple criteria fake reviews detection based on spammers' indicators within the belief function theory
Résumé
E-reputation becomes one of the most important keys of success for companies and brands. It is mainly based on the online reviews which significantly influence consumer purchase decisions. Therefore, in order to mislead and artificially manipulate costumers' perceptions about products or services, some dealers rely on spammers who post fake reviews to exaggerate the advantages of their products and defame rival's reputation. Hence, fake reviews detection becomes an essential task to protect online reviews, maintain readers' confidence and to ensure companies fair competition. In this way, we propose a new method based on both the reviews given to multiple evaluation criteria and the reviewers' behaviors to spot spam reviews. This approach deals with uncertainty in the different inputs thanks to the belief function theory. Our method shows its performance in fake reviews detection while testing with two large real-world review data-sets from Yelp.com.
Origine : Fichiers produits par l'(les) auteur(s)