A Hybrid Method Using Multidimensional Clustering-based Collaborative Filtering to Improve Recommendation Diversity

Abstract

This paper describes a hybrid recommendation approach for discovering individual users' potential preferences from multidimensional clustering view. The proposed approach aims to help users reach a decision to meet their diverse demands and provide the target user with highly idiosyncratic or more diverse recommendations. To this end, we propose a hybrid approach that incorporates multidimensional clustering into a collaborative filtering recommendation model to provide a quality recommendation. The proposed approach also provides a flexible solution for improving recommendation diversity and achieves a tradeoff between recommendation accuracy and diversity. The performance of proposed approach is evaluated using a public movie dataset and compared with two representative recommendation algorithms. The empirical results demonstrate that our proposed approach performs superiorly on increasing recommendation diversity while maintaining recommendation accuracy.