| With the popularization of Internet and the rapid development of e-commerce sites, the problem of commodity information overload becomes more and more serious.How can we help the viewer quickly and efficiently find the needed goods when they face to a large number of commodity information has become the urgent problem of the development of e-commerce sites. The emergence of e-commerce recommendation system provides a way to solve these problems, however, there are still some problems in the use of existing e-commerce recommendation system, for example, some e-commendation recommendation systems have low recommended efficiency and can not meet the user’s individual needs. Therefore, the research on e-commerce recommendation system and recommended technology has great practical value.The research of e-commerce recommendation system is always focused on the recommended technology, because the selection of the recommended technology is directly related to the merits of the quality of recommendation. The association rules recommended techniques is more popular in current research at home and abroad for the research of e-commerce recommendation technology. However, in practical application, the association rules recommended techniques also exist some problems, such as, it is difficult to discover association rules, it is more difficult to find a strong association rules between commodities in the case of sparse data, and algorithm will produce a large number of candidate itemsets in the process of implementation, and so on. These issues must find a solution.In response to these issues, this paper presents an improved association rules recommended technology. That is, by using the method of combining the concept hierarchy tree and FP growth algorithm to mine association rule. This algorithm can solve the problem of data sparsity and large quantities of merchandise concepts, and the process of calculation does not produce a large number of candidate itemsets, at the same time, this algorithm has the advantage of mining time and can effectively overcome the disadvantage of Apriori algorithm. In order to verify the accuracy and efficiency of the improved algorithm, the authors make this improved algorithm run on.NET platform in using C#language, then through the data experiment to analysis the efficiency of mining. Data experimental results show that:Firstly, the improved algorithm also has the correctness and has a higher efficiency of mining compared with Apriori algorithm and pure FP growth algorithm. Secondly, the improved algorithm can also find out the implied valuable relationship between the different levels of commodity, and can provide a richer, more universal significance of knowledge to personalized recommendation model, then can meet the needs of more users.Finally, this paper design a personalized e-commerce recommendation model based on the improved FP growth algorithm. This model takes a clothing shopping website as application background. The main content includes the architecture design of the recommended model, the main function of the recommended model, the recommended model workflow analysis, analysis of the various functional sub-module of the recommended model as well as the back-end database design. The model can mine the user’s interest preferences by analyzing the user’s purchase history, and then provide users with accurate, real-time personalized recommendation service in the constantly updated. |
