Collaborative filtering recommendation model with user similarity filling

Filtering recommendation system is always key and hot point in electronic commerce research; to obtain recommendation result with high accuracy, performance, universality and strong adaptation, improve recommended efficiency and veracity of collaborative filtering recommendation system and provide more personalized recommendation service for users, a kind of collaborative filtering recommendation algorithm integrating user similarity and rating attribute has been designed in the thesis. Firstly, attribute dimensionality of users and corresponding value have been collected, and then rating information on interest of users for the project has been collected to enhance partition degree of user similarity; then user attribute is used to balance similarity among users; at last, multiple Datasets are used to carry out simulation test. Result of the simulation test shows that user depending-on method is adopted in the thesis, which can substantially raise quality of recommendation, and the recommendation can meet practical requirements of users and of practical value for application.