Neighbor Diversification-Based Collaborative Filtering for Improving Recommendation Lists

Recommendation systems are popular information filtering tools that help people find what they want. Accuracy is the most widely used metric for evaluating recommendation systems. Recently, many research works have focused on new measurements beyond the accuracy of recommendation systems. In this paper, we propose a neighbor diversification collaborative filtering algorithm to improve the recommendation lists. By using Movie lens dataset for empirical analysis, we investigated the influence of neighbor diversity to the recommendation accuracy, diversity, novelty and coverage. Intensive experimental results proved the efficiency of our proposed algorithm for improving recommendation lists.