Application Of Open Data-Mining Engine Based On Web Services In E-Commerce

Along with the popularization of Internet application, e-commerce, as the production of information era, has developed fast. Every day, an e-commerce system will create a lot of business data, how to use it to provide personal services for different customers, then improve their loyalty, which is much necessary in competitive e-commerce trade.Now, almost large e-commerce systems, like Amazon, eBay, dangdang, has used various data mining technology to provide personal services for their customers. But, the mining engines of these systems are all based on close architectures. These architectures always have some shortcomings in e-commerce mining applications, which have changeful demands and different users. Therefore, this paper focuses on how to build up an open data-mining engine. First, it starts with Service Oriented Architecture, including its concept model, virtue and realization-web services. Then this paper studies the architectures used in representative mining engines, and analyzes existing shortcomings of them. Against these shortcomings, this paper put forward a kind of architecture about an open data-mining engine based on web services. This architecture uses service-oriented idea and is combined with web services technology, which makes the engine separated from the e-commerce system and mining algorithm.In addition, this paper is based on the scientific researchproject-"New database technology based on web services", a sub project of national "863" high technology research development plan. This paper designs the data-mining subsystem for the project's application platform-"EGO365" e-commerce system, and the mining engine in subsystem uses the open data-mining engine based on web services. This paper also handtailor three kinds of data mining applications for the real business requirement of "EGO365": customer loyalty analysis, customer classification and commodity recommendation. These data mining applications provide personal services for the customers of "EGO365"straightly or non-straightly. Then this paper dissertates the realization process of data-mining engine's archetype, discusses the several factors that would affect the engine's performance and makes the suggestions on how to improve the performance.At last, on the basis of summarizing full paper and against the studied content, the article put forward some questions and looks into the distance.