Research On Multi-Factor Collaborative Filtering Recommendation Algorithm Based On PSO

The calculation methods of similarity in traditional collaborative filtering recommendation algorithm ignored the implicit meaning of numerical rating data and influence factors in the ratings. Aimed at the above problem, the paper construct different similarity influence model based on Multi-Factor which include rating D-value effects、rating value effects, item popularity effects,user habits affects and user activity effect in order to measure similar relationships among different users.and then,integrate all similarity influence and propose a Multi-Factor Collaborative Filtering Recommendation model. In addition,analysis the flaws of the above model and put forward a Multi-Factor Collaborative Filtering Recommendation Algorithm Based on PSO with the help of Particle swarm optimization algorithm(PSO).The algorithm make full use of the limited data resources, and alleviate the problem of sparse rating largely.We do the experiment on MovieLens and Netflix datasets and make Comparisons between our algorithm and the traditional collaborative filtering algorithm.The experimental results are well in agreement with the theoretical analysis. This work will contribute to the current recommendation system research.