An item based collaborative filtering system combined with genetic algorithms using rating behavior

Abstract

With the sharp increment of information on the Internet, many technologies have been proposed to solve the problem of information explosion in people’s life. Collaborative Filtering (CF) recommendation system is one of the most popular and efficient ways of solutions, especially item based CF systems. While traditional item based CF recommendation algorithms either ignore the diversity of different users’ rating behavior or do not deal with it efficiently. In this paper, we present a novel similarity function using the average rating for each user instead of the overall average rating for all users. In order to find the optimal similarity function, we use genetic algorithm (GA) to optimize the weight vectors associated to the similarity function. A series of comparison experiments are conducted to demonstrate the effectiveness in terms of the quality of prediction of the proposed method.