Improvement And Application Research Of Collaborative Recommendation Algorithm

Information overload has seriously affected the efficiency of information acquisition.In order to get interesting information on the website,users need to spend a lot of time browsing with the uninteresting information.The recommendation systems help users to get the required information from the massive information,and enable the information manufactures stand out simultaneously.At present collaborative filtering technology is one of the effective techniques in the application of recommendation system.However there are still some problems in collaborative filtering technology,such as cold start,scalability and data sparseness.How to solve these problems has become a hot research topic.On the basis of the basic concepts and study status in the research of recommender systems,this thesis discusses the advantages and disadvantages of collaborative filtering algorithms.Comparing to the common recommendation techniques,then this thesis proposes forward two kinds of strategies to improve the present collaborative filtering algorithm.The purpose of the two improved algorithms is to improve the accuracy of recommendation in the case of sparse data.If the item category is fuzzy,the method based on user preference is appropriate.This algorithm introduces the concept of user preference,and determines the similarity of the two items by calculating the service attributes between different projects.According to the different users’ preference for the service attributes of the project,the algorithm calculates the correction factor of the service,so this algorithm can improve the accuracy of the recommendation.If the item category can be defined easily,it is more appropriate to use the collaborative filtering algorithm based on item category and attributes.With consideration of the items categories,this algorithm makes full use of user rating data and project information to recommend and improve the quality of the recommendation further.In this thesis,a movie recommendation system is constructed to verify the effectiveness of the improved algorithms and the quality of recommendation.The experimental results show that the improved proposed algorithms can effectively improve the accuracy of the recommendation results.