earticle

영어

Collaborative filtering recommendation is one of the most effective recommending techniques, which provide customers with suggestions according to their interests. However, neighborhood based collaborative filtering methods confront great challenges of data sparsity and lack of accessorial information in the context of big data. To address these problems, we propose a hybrid model combining tag information and neighborhood based collaborative filtering. A folksonomy network model based on tag information is proposed to analyze the tag relevance between different items. And tag relevance is incorporated into rating prediction of neighborhood based collaborative filtering for improving the recommendation accuracy. Experiments on MovieLens and Netflix datasets are carried out to evaluate the performance of our method. The results show that our method outperforms other methods and can improve recommending quality effectively.