Construction And Research On Intellectualization Of Trade Management Electronic Commerce System

In recent years, implementing the architecture of the E-Commerce system facesmany technical challenges, including how the system can be improved inmaintainability, stability, scalability and operational performance. There are fewin-house developed E-Commerce systems in China which have feature-richfunctionality to provide intelligent business operation analyses based on data-mining.Most systems either can only provide simple statistic analyses based on OLAP or justuse relevant products purchased from abroad, both of which can not meet pressingdemands of the e-business industry. Therefore, this paper examines a new possibilityto develop an E-Commerce system based on J2EE framework and introduceintelligent business operation analyses with data mining to support business marketing.Then, the paper describes such development on the instances of the foreign trademanagement system adopted by LuTu.The system demonstrates a component-based design on a 3-tier architecture, usesdifferent design patterns such as MVC and value object, and applies data mining tocustom relationship management (CRM) analysis. The complete package containsfive key function modules. These subsystems are user management, businessmanagement, business operation analysis, export billing management and otherfundamental content managment. The author mainly develops the productmanagement and customer order management modules in the business managementsubsystem and the operation analysis subsystem, which altogether contain nearly 50web pages and more than 7,000 lines of source codes. The product management andcustomer product order management modules are further divided by MVC mode intosub-modules like user view, controller and business model, implemented by JSP+Struts and EJB components respectively. The implementation procedures includedatabase design, implementation of EJB and the value object mode in the model,implementation of JSP+Struts and the Dispatch Action scheme in the user view, andunit and integration tests. In the business operation analysis subsystem, author usesthe classification algorithm of information theory-based ID3 and C4.5 decision treegenerating algorithm to get user churn and value analyses, while applying K-meansclustering algorithm for user segmentation. The author also researches andimplements key Apriori algorithm and K-means clustering algorithm. The paperdiscusses their shortcomings and benefits, and proposes relevant improvements. After implementing the E-Commerce system of intelligent analyses, LuTudramatically lowers its expenses compared with the traditional foreign tradeoperational mode. For example, the cost of human resources reduces by 5-10%. Inaddition, after the company uses operation analysis subsystem to predict the customchurn and adopts some retention measures, the custom churn dropped from 8.991% to8.403%.