Study On Recommendation System Of E-commerce Content Based On Data Mining

With the development of e-commerce, personalized content recommendation system is particularly important for e-commerce site. This paper introduces the relevant concepts of e-commerce personalized recommendation and the research status of China and foreign countries. It analyzes and studies a variety of recommendation algorithms and their advantages and disadvantages. It put forward corresponding improvement ideas based on data mining methods for traditional collaborative filtering method of data sparseness, cold start and singular problems found. It uses the User Rating Matrix for the data sparseness problem, making implicit data conversion data to display. It designs a clustering-based collaborative filtering recommendation system for old start issues, and improving the K-means clustering algorithm depends on the initial value of K. It brings in the average user similarity. It describes the improved algorithm in details. This paper introduces a method of correlation analysis for the strange discovery problem. It takes advantage of association rules identifying the frequent item-sets from the data sets, and then uses these rules to create a process that describes the relationship of goods, making recommendation of the cross-category become true.Finally, it describes in detail. The overall framework of e-commerce recommendation system and the design of functional modules, as well as the personalized content recommendation module functions. And the content of the study is applied to the electronic commerce system. By making comprehensive assessment test on the actual Web site page views and purchase amount, we get the results that the improved collaborative filtering algorithms can solve the problems faced by the traditional algorithm effectively, improve the site’s page views and purchase amount. It further confirms the personalized recommendation system applying to e-commerce sites has practical significance.