A Hybrid Recommendation System Using Trust Scores in a Social Network
Various techniques for personalized recommendation have been studied. One of the most promising such techniques is collaborative filtering (CF). However, CF is unable to produce high quality recommendations when user rating data is lacking or insufficient. To address this “sparsity” problem of CF systems, this paper proposes a hybrid recommendation system. The proposed system improves recommendation quality by exploiting trust scores between users in a social network. The proposed system overcomes the weakness of CF for sparse user ratings databases and yields better performance than the conventional CF method.
KeywordsTrust scores Collaborative Filtering Recommendation System Social Network
Unable to display preview. Download preview PDF.
- 1.Jure, L., Daniel, H., Jon, K.: Predicting Positive and Negative Links in Online Social Networks. In: WWW 2010 (April 2010)Google Scholar
- 2.Lada, A., Eytan, A.: How to search a social network. In: HP Labs (2005)Google Scholar
- 3.Panagiotis, S., Eleftherios, T., Yannis, M.: Transitive Node Similarity for Link Prediction in Social Networks with Positive and Negative Links. In: RecSys 2010 (September 2010)Google Scholar
- 4.Paolo, M., Paolo, A.: Trust-aware Recommender Systems. In: RecSys 2007 (October 2007)Google Scholar
- 5.Zhang, Z., Wang, X.-M., Wang, Y.-X.: A P2P Global Trust Model based on Recommedation. In: Proceedings of the Fourth International Conference on Machine Learning and Cybernetics (August 2005)Google Scholar