TRPC: An Ensembled Online Prediction Mechanism for Trusted Recommender System

The most recent invasions in social networks make it inevitable to develop a network with high dependence and confidence to users. Even though recommender systems of today use advanced parallelism in web development, achieving trustworthiness in such a system has been a challenging task for several years. To overcome the existing sparsity, scalability and dynamism in new item/user issues, we propose a framework TRust Propagation and Clustering (TRPC), to build a trust network using the social distance between every pair of users and similarity measure of clustered users based on the users’ tastes and preference. Our proposed technique to predict the ratings of items by users involves three major steps which comprise both implicit and explicit social relationship and propagation mechanism. The second step involves clustering the trusted users and third step predicts the products/subjects between them based on the alike criteria. The proposed rating prediction promises a better eminent recommendation for all buyers and online users who gain access to the community. To validate the effectiveness of our work, we experimented with two real world datasets Epinions and Movie Lens. Key-Words: Trust propagation, Trusted path, Clustering, Social networks, Recommendation systems, Similarity metrics