An improvement of video recommender similarity measurement model

Collaborative recommender systems have succeeded in capturing the similarity between users and items based on ratings. However, they have rarely considered about the available information of the multimedia such as categories, delivery time and so on. Such information are valuable and feasible to solve rating bias problems in recommender systems. We found that user described their preferences directly to the item rating data is not comprehensive. In this paper, we design IBHF (Item-attribute Based Hybrid Filtering) based on movie features of the multimedia information. In the IBHF, we provide recommendation service by new Pearson method which is a similarity measure technique used to integrate movie attributes into the Item-based collaborative filtering (IBCF) framework in hopes of achieving better performance. The experiment prove that this method can make the items recommendation more ideal, and also provides a solution to solve the cold start problem to different recommendation items.