Collaborative filtering algorithm integrating trust and preference of user's emotion

Taking user's emotion regulation as application background, this paper presents a collaborative filtering algorithm integrating trust and preference of user's emotion to meet user's emotional needs. Firstly, a user preference model based on ratings and trust is presented to address the scalability issue of user preference model in collaborative filtering. The proposed model uses the number of ratings to set two thresholds to extend the calculation strategy of user similarity weight to selectively assign the trust value and correlation to the rating value in the user preference model. Secondly, in the process of producing the candidate set of items, the emotional connotation of items is customized. The user preference for emotional connotation of item is introduced to make up for the neglect of user's emotions in collaborative recommendation. Experimental results show that the presented algorithm has good scalability and therefore improves user satisfaction.