Collaborative Ifltering Algorithm Analysis And Research In E-commerce Recommendation System

With the rapid development of e-commerce and Internet information technology,more and more users to purchase the goods and services they wanted through theE-commerce platform. In order to meet the needs of different users, the quantity andtypes of goods that E-commerce site provides to users has increasing rapidly.Although users have more opportunities to purchase their favorite products, but usersoften get lost in the numerous product list, so the user is difficult to accurately findtheir favorite products. E-commerce recommender system have ability to collect userpurchase information and track the user’s needs change, analysis of user preferences,and then recommended for the user they might be interested and users satisfied.E-commerce recommendation system has been widely used in various types ofe-commerce platform, has received extensive attention because of its gooddevelopment and prospects.In this article，I briefly discusses the e-commerce，then analysis why do aE-commerce website have to use the personalized recommendation system and what’sbenefits can be brought for a site。Then this thesis has researched the basic frameworkof E-commerce recommendation system and a variety of recommendation technology.on this foundation, focusing on collaborative filtering methods which is the earliestand the most widely used e-commerce recommendation system. For data sparsityproblem in collaborative filtering techniques, the classic user-based collaborativefiltering algorithm suffer from similarity calculation accuracy, this thesis propose aCART-based collaborative filtering algorithm. Firstly, to calculate the sparse user-item rating matrix’s unranked element by classification and regression tree, then usethe user-based collaborative filtering algorithm to make prediction in a dense matrix.The experiments showed that the CART-based collaborative filtering algorithms havebetter performance, increase the accuracy of the recommendation.