Introduction of synthetic and non-synthetic trust recommender models in collaborative filtering

Rapid expansion of the Internet makes competition between many websites and social networks more prevalent. Several users cannot choose their favorite options because of huge amount of information they will face. This causes information overhead problem. Recommender systems with collaborative filtering appeared in different area to solve this problem. In recent years, expansion of e-commerce websites and social networks gives form to credit mechanism which can be used to improve performance of collaborative filtering system and to eliminate their limitations. Most models of credit are based on scores which users give to items and also relation credits and popularity. In this model, if users scoring to items is little it will cause data dispersion. In this paper, different types of trust models in recommender systems will be studied. Also these systems are classified into synthetic and non-synthetic types. Among modern methods are hybrid personal trust and hybrid personal and group trust. Finally, all models are compared with each other and their advantages and disadvantages are clarified. Comparing with different methods shows that the final trust method has a higher recommendation quality than other collaborative filtering methods. It also increases accuracy of prediction superbly.