Research On Personalized Recommendation System Based On Rough Set Theory

With the development of the Internet,people can get more and more abundant information resources,these information resources in the convenience of people's lives,but also caused some problems.People usually need to expend a lot of energy and time to find what they want from the huge information database,so the "information overload" becomes more and more serious.How to glean the information that people really need from the clutter also needs to be addressed quickly.The recommendation system can solve the above problems.It can recommend to the user can achieve the corresponding requirements of the object,and then realize personalized service.However,there are still some problems in the development of recommender system.First,there are many ungraded items in the user-item scoring matrix because of the limited number and variety of items in real life,the sparsity of the user-item scoring matrix is caused by the sparsity of the user-item scoring matrix,and the problem of cold-start is caused by the lack of the user-item rating information when new items are added Third,the existing recommendation system often can not make accurate recommendations to users,there is a low recommendation quality.This paper describes the basic knowledge of personalized recommendation system and related theories,analyzes and compares the existing recommendation Algorithms,the sparsity,cold start and low recommendation quality of the existing recommendation algorithms are studied and improved methods are proposed.The main work of this paper is as follows:1.At present,the data set of recommender system is sparse.In view of this phenomenon,the rough set theory is used to extract the rules,through the user-item scoring matrix and the attributes of user and item,the corresponding decision table is constructed,and then the decision table is reduced to obtain the corresponding core values of each rule,and finally the decision rules of core values are obtained,to achieve all of its reduction,and then to score the ungraded items to predict the score and fill the original user-item score Matrix to calculate similarity in collaborative filtering,to alleviate the problem of data sparsity in user-item scoring matrix.Experiments show that the rule extraction based on rough set theory can effectively alleviate the sparsity of user-item scoring Matrix and improve the accuracy of the recommendation Algorithm.2.The article-based collaborative filtering algorithm only relies on the user-item scoring matrix when calculating the similarity,and does not consider the correlation between the items,they tend to give the item a lower rating,and because of these low quality ratings,the accuracy of the recommender system is low.In order to reduce the influence of user's subjective bias on the recommendation result,this paper introduces the concept of item classification.Combining the user-item scoring Matrix,the similarity between the two items is calculated to realize the personalized recommendation according to the preference of the target users.The experiment shows that the accuracy of the recommendation algorithm can be improved effectively by introducing the classification information of items to calculate the similarity of items.