Personalized Recommendation Based on Information Flow in Data Intensive Services

Abstract

With the development of data intensive service like online social networks, there is more and more user-generated content (UGC), the large amount of information make it increasingly difficult for the users to select their needed information quickly and efficiently. It is becoming a common problem that how to help users find the information they are most interested in, and how to make the different types of information or ideas spread quickly. To solve this problem, we propose a personalized recommendation method based on the information flow, i.e., we regard the personalized recommendation problem as a potential interest graph construction problem, and structural hole and opinion leader discovery problem in the interest graph. We present a structural holes discovery algorithm based on potential interest graph to find structural hole, and an opinion leader discovery algorithm based on node credibility. The experimental results show that our proposed method is effective in providing personalized recommendation in data intensive services.