Predict user interest with respect to global interest popularity
Wenhao Zhu, Kangkang Niu, Guannan Hu, Jiaoxiong Xia · 2014
With the development of personalized recommendation, the method of user interest prediction has been a hot research topic. Usually, predict methods use individual related parameters such as user ratings to infer possible user interests. A potential problem with these methods is that the credibility of the user ratings is rarely questioned or considered during the process of prediction. However, as a common knowledge of social science, people can be affected by group actions. For example, individual ratings can be affected by public opinion. In this paper, we propose an approach to predict user interest with respect to popularity facts. It operates in two aspects. First, interest item popularity gives a weight for each rating value to promote popular items. On the other hand, the prediction calculation is adjusted on the basis of individual item popularity and global item popularity. The experiment results show that this method can improve the accuracy of interest prediction.