Study On Collaborative Filtering Recommendation Based On Users' Interest Clustering

Recently, personalized recommender systems, especially collaborative filtering recommender systems, have achieved widespread successes on the Web. The tremendous growth in the amount of available information and the kinds of commodities to Web sites poses some key challenges for recommender systems, so the problems of cold-start and sparsity in collaborative filtering recommendation, and the user interests'orientation problems of user-based collaborative filtering algorithm are in dire need to be solved.By analyzing problems of user-based collaborative filtering algorithm, an improved user interest clustering based collaborative filtering algorithm was proposed. This new algorithm takes synthetically into account the influence of the sparsity of user-item rating matrix and the class of the user's interest. In the nearest neighbors query, firstly classify the items and then cluster the users by the user-item rating matrix to find out the target user's nearest neighbor user set. At last, by computing the similarity of the target user and the user in the nearest neighbors set to find out the nearest neighbors set.Individual user interest determines users' visitation in a certain extent. To address the problem of extreme sparseness of user rating data, a novel collaborative filtering algorithm based on users' interest clustering is proposed. The method predicts item ratings that users have not rated by the similarity based on users'interest clustering, then uses a new similarity measure to find the target users' item recommendation set.The experimental results show that the improved algorithm, i.e. the collaborative filtering recommendation algorithm based on users' interest clustering is more accurate and efficient comparing with the traditional user-based collaborative filtering recommendation algorithm.