一种基于混合相似度的协同过滤算法

ollaborative filtering is one of the most widely used method in personalized recommendations. The most important step of the method is obtaining the similarities among users by using ratings information so that system can predict user preferences. However, most similarity measures are not efficient enough in the face of cold start and data sparsity problem. To measure user similarity comprehensively and objectively, this paper introduces mixed similarity measure model. The model divides the user similarity into three parts to measure and then comprehensively analyzes the user\'s similarity in different periods. It also improves the accuracy of similarity by using user attribute similarity and item similarity as weighting factors. Experimentsshow that the proposed method relieves cold start and data sparsity issues and improves the prediction accuracy and quality.