Research Of Data Sparsity Problem In Recommendation Systems Based On Collaborative Filtering

With the rapid development of information technology and ever-expanding of information resource, E-commerce also develops rapidly, recommendation systems are more and more widely applied in E-commerce websites. Recommendation system is an important component of personalized E-commerce service, which breaks traditional commercial operating pattern and plays a sales staff role in traditional commerce. It can increase the sales of merchandise and enhance the loyalty of customers. Now collaborative filtering technology is one of the most successful in recommendation systems, and it is applied widely. But with the customers increased and the factor of merchandises themselves limited, currently most of collaborative filtering algorithms usually have several major limitations, such as data sparsity, scalability, cold start and synonymity and so on. Almost every recommendation system, it is impossible for every customer rates on all the merchandises. In fact, the amount of the merchandises one customer bought is less than 1% of all, and the data set which customers rated on merchandises is sparse. In order to improve the quality of recommendation system, many researchers tried to analyze customers and merchandise information in different points.We introduced the basic knowledge of recommendation system detailedly in this paper, through expatiating the problems of collaborative filtering recommendation system educed the sparse problem. We summarized several methods to solve this problem. Then through comparing and analyzing between item-based collaborative filtering algorithm and user-based collaborative filtering algorithm, we proposed two improved methods for sparse problem from two different parts. One of the methods is improving the quality of algorithm, which combines the item genre and computes the similarities between the items that users preferred and target item. It can reduce the computing amount and get higher quality; another one is in order to reduce the sparse data, filling the unrated items through the method of Slope One, and getting the recommendation to target users. We theoretically detailed analyze the new methods and prove their feasibility. Then the experimental results that the new method is implemented with the benchmark experimental data set are given, the performance between the new methods and the old methods is compared and analyzed. The experiments show that the improved methods can alleviate the problem of sparsity effectively and improve the quality of recommendation.