Hybrid Trust-Driven Recommendation System for E-commerce Networks

Abstract

In traditional recommendation systems, the challenging issues in adopting similarity-based approaches are sparsity, cold-start users and trustworthiness. We present a new paradigm of recommendation system which can utilize information from social networks including user preferences, item`s general acceptance, and influence from friends. A probabilistic model, particularly for e-commerce networks, is developed in this paper to make personalized recommendations from such information. Our analysis reveals that similar friends have a tendency to select the same items and give similar ratings. We propose a trust-driven recommendation method known as HybridTrustWalker. First, a matrix factorization method is utilized to assess the degree of trust between users. Next, an extended random walk algorithm is proposed to obtain recommendation results. Experimental results show that our proposed system improves the prediction accuracy of recommendation systems, remedying the issues inherent in collaborative filtering to lower the user`s search effort by listing items of highest utility.