Collaboorative Filtering Recommendation Service System Base On The User Interest And Trust

With the development of the Internet,a wider range of commodities are being sold through online platforms.Due to multiple destabilizing factors,effective online shopping becomes more and more difficult for people,and the shopping efficiency also decreases,which certainly restricts the development of e-commerce in China.The rapid growth of online information quantity and commodity categories poses a serious challenge to the recommendation system,and the problems of user interest orientation and product procurement risk evaluation in the user-based collaborative filtering recommendation are to be urgently solved.By improving the traditional user-based collaborative filtering algorithm,a user interest-based trusted shopping recommendation service system is formed.By adopting this algorithm,the user interest metadata are analyzed to form different interest categories and to produce various interest data sets;then,collaborative filtering algorithm training is carried out on these data sets to produce corresponding optimal models and predict the unscored items of the target users;finally,the recommended items are evaluated to obtain their credit risks and the items with high credit risks will be filtered out.Thus,the final recommended items are all reliable ones.The major contents are as follows:1)Design the architecture of Collaborative Filtering Recommendation Service System base on the user interest and trust.It composed of Preprocessing Subsystem,Recommendation Subsystem,Interaction Control Subsystem,Store Data Subsystem,Postprocessing System,Verify Subsystem.2)Designed the user interests tree.It can describe the user's good interests from space and time;Accomplished the algorithm for building and update the user's good interests tree.3)Improved now available Collaborative Filtering algorithm.Designed and Accomplished the Recommend Subsystem that improve increase efficiency and performance;Designed and Accomplished the Interaction Control Subsystem by input feature and output feature.4)Designed and Accomplished the recommend model training process base on the user interests tree.It cluster user's good data by the user interests tree and train model by Collaborative Filtering algorithm.5)It evaluated that how much user interested good risk by comprehensive evaluation technology.Designed and Accomplished trust risk filtering technology.6)Designed the test for the recommendation system and verified the recommend system.The experiments show that,the proposed recommendation system is efficient and stable,and the improved algorithm has higher accuracy and effectiveness.This system can filter high-risk good data.Test data date from Internet normal online shopping data.