Electronic Commerce Recommendation System Based On Web Usage Mining

Along with the popularization and the application of Internet, Electronic Commerce is been popular in the world by its low cost, convenient, high speed and unlimited in space and time. Electronic Commerce provides more and more choices for the user, but at the same time its structure also becomes more complex. On the one hand, the users facing the massive commodities, one finding the needed commodity become more and more difficult; On the other hand, the business has also lost the relation with consumer in the massive electronic data. The personalized electronic commerce recommendation system provides one-to-one service according to the user behavior characteristic for the user, helps the user to find commodity which he needs, thus completes the purchase process smoothly. The business can enhance the sale of the electronic commerce system through the recommendation system, maintain the relation with customer, and promote the user's the loyalty and the degree of satisfaction.The electronic commerce recommendation system has widespread applied in each big retail sales website, but simultaneously it's also faces a series of challenges. Face the question of the current electronic commerce recommendation system, this article has conducted the beneficial research and the exploration to the electronic commerce recommendation algorithm as well as the electronic commerce recommendation system structure.The prime task includes:(1)Propose the customer choice question. Selects the customers who mostly purchase the commodity, taken as the goal customer, and also are finally can obtain the recommendation customer.(2)Improve the traditional association rule algorithm. Improved and realized the frequent item set production algorithm and the strong association rule production algorithm. And established and has realized the product association model in the improvement algorithm foundation.(3)Implement the multi-model recommendation system. This article proposes a kind of multi-model recommendation system and realizes it.