A Review on Trust-based Recommender Systems

By growing the e-commerce sites, a new challenge is information overload. The problem refers to huge information about items, users and activities, which make following of the information flow in real world impossible. Recommender systems help users to find interested items in huge databases in e-commerce sites faster and easier. Variety techniques have been proposed for performing recommendation, including collaborative filtering, contentbased, demographic filtering and hybrid methods. Although collaborative filtering is the most successful technology for recommender systems, it suffers from several inherent issues such as data sparsity, cold start, accuracy and malicious attacks. Trust-based approaches use trustworthiness as a factor to solve traditional problems and improve recommendation results. Based on previous researches, trust may be global or local, explicit or implicit, and be measured based on friendship, membership, social activity or other methods. In this paper we discuss about different trust aspects and categories of trust-based approaches. We will also review by detail on the most important trust-based approaches and will discuss about them.