Study Of The Architecture Design And Recommended Algorithm In E-Commerce System

E-commerce is a new method of trade that emerges and develops on the international market. E-commerce Web site build easier, but making it more effective is difficult. How to attract customers and enhance customer loyalty is the key to the competition of enterprises. System architecture which is made to support enterprise decision-making can provide data for personalized service, and personalized recommendation algorithm can enhance customer loyalty. Therefore, system architecture of E-commerce and personalized recommendation algorithm have become a hot theme in E-commerce system.This article firstly introduces corporate information factory (CIF), Architecture needs of e-commerce and the interface between enterprise information factory and e-commerce and so on. How to efficiently combine CIF and Web environment to supply data for personalized recommendation are mainly described.Combining the CIF and the Web environment, a stable, flexible interface can be established. Then some data can be supported to Web environment. Aimed at e-commerce enterprises that have many websites, an architecture based on distributional Web ODS is proposed. A unique view concentrating all data information and an abstract, integrated high-performance information management platform can be provided. Web ODS can interact with personalized engine based on the information of customer characteristics, to start a series of events.Collaborative filtering is the most widely used recommendation technology in the personalized recommendation system. However, existing collaborative filtering algorithms do not consider the relationship between the every year specific event and the user purchase behavior. For this reason, a collaborative filtering algorithm considering the year schedule event is proposed. It introduces the time weight function, make recommendation of the commodity which purchased by the consumer of close neighbor and which's purchased time is close to current user access time and belong to the same period to be higher. It reflects the correlation between event and recommended algorithm.In order to verify the feasibility of e-commerce architecture based on distributional Web ODS and time weighted collaborative filtering algorithm with annual schedule events considered, uses OWB to construct Web ODS based on simulated electronic bookstore, and achieve presented algorithm in the form of Java code. Both achieve the desired results.