| With the development of the Internet,the amount of data in the Recommender System are growing rapidly.Recommendation system can provide personalized recommendation for users,so it has been widely used.Collaborative filtering is one of the most effective algorithms,but it also has some problems,such as,scalability.Based on the collaborative filtering method,this paper puts forward some improving methods.The main works of this paper are as follows:(1)The commonly used similarity calculation method does not take into account the impact of the number of common scores.In order to solve this problem,this paper presents an improved formula for calculating similarity,which takes into account the number of common scores.Compared to the traditional method,it can measure similarity between items better.(2)In order to improve the scalability of the collaborative filtering,this paper proposes an improved collaborative filtering recommendation method based on fuzzy clustering.In the offline phase,items will be divided into different clusters using fuzzy clustering technology.Then calculate the similarity and neighbor search only in the cluster,which greatly reduces the search range of similarity computation and neighbor search stage.(3)when there is a new score,traditional collaborative filtering methods need to update the similarity in the entire project space,this method consumes large time and resource and affect the scalability of the system.To solve this problem,this paper proposes a collaborative filtering recommendation method based on incremental update,when there is a new score,the method only calculate the changing similarity between items,and then make recommendations according to the new similarity.(4)By doing some experiments on MovieLens and Yahoo! Movies data sets,the experiment results show that the improved algorithm can improve the scalability of the recommendation system,while maintaining good accuracy. |
