The Research On Collaborative Filtering Algorithm In Artificial Recommendation System

Collaborative Filtering Algorithm aimed at the main challenges of recommendation systems in E-Commerce, this dissertation explored and researched some key technologies of recommendation systems in E-Commerce.The main research results of this thesis are as follows:1.The dissertation analyzed three existing collaborative filtering modified algorithms, and found their problems:(1) DCRec algorithms could efficiently overcome the extreme sparsity of user rating data, but the accuracy of its recommendations will have a certain decline.(2) IRPRec algorithms provides better recommendation results than traditional CF algorithms. But the real-time algorithm is not high, can not be good to meet the recommended system performance requirements.(3) Item-clustering-based collaborative filtering algorithm could efficiently improve the real-time response speed of recommendation systems, but recommended only on the accuracy of intelligence with the traditional method is recommended.2.Through analyzing the problems of the modified algorithms, this dissertation researched current mainstream technique in personalized recommendation, collaborative filtering. On the base of introducing"concept hierarchy"and"community filtering"techniques, the dissertation modified collaborative filtering algorithm named"concept hierarchy"collaborative filtering algorithm. We do imitation experiment on improved algorithms proposed by the dissertation, through experiment's validation, the improved collaborative filtering algorithms based on concept hierarchy are better than traditional one in the aspect of veracity, integrality, diversity in recommendation, especially in sparse consumer rating datasets, the improved ones behave favorable recommendation performance.