Exploiting user's social network: A novel method to recommend most attractive and targeted service

The rapid development of the Internet and e-commerce has accumulated numerous data about products, services and customers, which contains meaningful information that recommender system can utilize to realize a more accurate recommendation. In daily life, it is nature for users to seek suggestion from a friend and the more professional and intimated the friend is, the more impact the suggestion has on the user's final decision. Inspired by this phenomenon, we develop a recommendation approach based on social trust network. In this paper, we take advantage of homogeneous effect in trust relations to calculate user similarities and build a personal weighted trust work. A trust propagation mechanism is proposed to attain trustworthy users and then we use a dynamic recommendation algorithm based on random walk model to achieve accurate recommendation. In the end, several comparing experiments are conducted to demonstrate the improvement of our recommendation method.