Recommender systems are one of the recent inventions to deal with information overload problem and provide users with personalized recommendations that may be of their interests. Collaborative filtering is the most popular and widely used technique to build recommender systems and has been successfully employed in many applications. However, collaborative filtering suffers from several inherent issues that affect the recommendation accuracy such as: data sparsity and cold start problems caused by the lack of user ratings, so the recommendation results are often unsatisfactory. To address these problems, we propose a recommendation method called “MFGLT” that enhance the recommendation accuracy of collaborative filtering method using trust-based social networks by leveraging different user's situations (as a trustor and as a trustee) in these networks to model user preferences. Specifically, we propose model-based method that uses matrix factorization technique and exploit both local social context represented by modeling explicit user interactions and implicit user interactions with other users, and also the global social context represented by the user reputation in the whole social network for making recommendations. Experimental results based on real-world dataset demonstrate that our approach gives better performance than the other trust-aware recommendation approaches, in terms of prediction accuracy.
Journal of University of Babylon for Pure and Applied Sciences.
2019.
Improve the Performance of Advice Systems Based on Cooperative Liquidation Using Trust Relationships
Students
Teachers
Graduates
Faculties
