Combination of user’s judging power and similarity for collaborative recommendation algorithm

Abstract

As an effective way to solve information overload, the collaborative filtering (CF) algorithm has been widely used in the personalized recommendation. In order to improve the accuracy of recommendation, an improved collaborative recommendation algorithm is proposed. Firstly, evaluate the user’s judging power based on historical scoring; then, combine the user’s judging power and similarity to improve the traditional user-based CF algorithm. Experimental results show that the judging power is positively correlated with the recommendation abilities of users and also verify that the judging power extracts the depth information from historical scoring and factors to influence a user on adopting the recommendation results.