Identifying Influential Users in Bipartite Networks and it’s Application in Recommender Systems

Please enable javascript in your browser.

Taheri, Mohammad | 2016

863 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 48808 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Movaghar, Ali
Abstract:
With the growth of spreading information across social networks, users face with many options. More specifically, try to find out information is so troublesome in this context.In response to this problem, recommender systems aim at providing suggestions of interest for end-users. Also, we see many social networks in real-life which have the potential to be used in a variety of recommender systems. In the bipartite networks of users and items weighted edges is formed between each user and each item in the training set. On the other hand, developing a recommender system that takes into account the social network of the user improves the accuracy of traditional recommender systems.The purpose of this project is to use the Hellinger distance amongst users to enhance existing recommender systems. At first, we propose a new metric to find important,influential, and more behavioral representative users in bipartite networks. Then, we attempt to improve the performance and efficiency of only-rating based recommender system by considering the achieved solution and using social recommender algorithms
Keywords:
Recommender System ; Social Networks ; Bipartite Networks ; Bipartite Networks ; Influential Users ; Behavioral Representative Users ; Social Recommendation Algorithms