An Evidential Spammer Detection based on the Suspicious Behaviors' Indicators
Résumé
The e-reputation is the key factor for the success of different companies and organizations. It is mainly influenced by the online reviews that have an important impact on the company's development. In fact, they affect the buying decision of the customer. Due to this attraction, the spammers post deceptive reviews to deliberately mislead the potential customers. Thus, the spammer detection becomes crucial to control the fake reviews, to protect the e-commerce from the fraudsters' activities and to ensure an equitable online competition. In this way, we propose a novel method based on the K-nearest neighbor algorithm within the belief function theory to handle the uncertainty involved by the suspicious behaviors' indicators. Our method relies on several spammers indicators used as features to perform the distinguishing between innocent and spammer reviewers. To evaluate our method performance and robustness, we test our approach on two large real-world labeled datasets extracted from yelp.com.
Origine : Fichiers produits par l'(les) auteur(s)