Collaborative Filtering Recommendation Algorithm Based on Multifactorial Clustering

Abstract

Recommendation systems applications have been applied to many fields and they are urgent need to better exploit. Collaborative filtering is a widely used recommendation technique. In collaborative filtering recommender systems, users are requested to provide ratings to the items they have purchased. By analysis of the ratings, systems can recommend items that are likely to be of interest to the users. However, it is not perfect yet. A few problems need to be solved such as sparse data and low recommended accuracy. In this paper, a novel collaborative filtering recommendation algorithm based on multifactorial clustering is presented. Firstly, two similarity measurements are defined according to user attributes and item category. Then, the linear combination of the two similarity values is applied to construct the user adjacency matrix. Secondly, in order to narrow down the search range of the nearest neighbors for the target users, we use Kmeans and differential clustering algorithm to group the users. By this doing, we can apply user correlation indices to predict scores within the group. Finally, the predicted score is weighted based on the scoring and attributes. That is to say, the impact of multiple factors is fully considered, such as rating information and attributes. The proposed method is evaluated using the Movielens dataset. Diversity experimental results demonstrate that the proposed method has outstanding performance in prediction accuracy and recommendation comprehensive performance.