TIMELINE BASED TRUST DISSEMINATION TOWARDS REVIEW SPAM DETECTION IN ONLINE REVIEWS

Spam is generated as information to pass multiple users at a time without concern to the recipient's knowledge. Spam has become a major concern in computer networking, due to its generalized message distribution, resulting in large network overhead and network blockage. Users are not aware of the spam delivery, storage and classification. Reliance on online reviews gives rise to the potential concern that wrongdoers may create false reviews to artificially promote or devalue products and services. In this work, we propose a time line based review-spamming detection based on the deviation between discrete observations on the events at different time periods in a continuous observation. In particular, we model the opinion vector dynamically about a review with its similarity among other reviews in database, then identified deviation of a particular review used to compute the trust scores for users, reviews, and targeted events respectively. Then these trust scores are effective indicators in spam detection to SVM classifier. Experiments on the semeval dataset show that the proposed detection scheme work better compared to feedback computational model.