A Data Mining Prototype SOA Perspective

With the development of information technology,Business Intelligence plays an important role in decision-making and management of an organization.Data within an enterprise and its parterners have a very significant impact on enterprise strategy plan & decision management,which are regarded as important scientific envidence and technical support.Hence,how to build a BI system with the features of openness, multiplexing and flexibility has become a hot subject in both Academia and Industry,so as to meet requirements for organization decision-making and management,even to integrate cross-industrial BI systems into an organization's IT infrastructure.For the past few years,service becomes the standard for resource encapsulation and sharing under open network environment.Especially,as Web Service is spreaded and applied,Service Oriented Architecture is gradually evaluated and highly reconginized.Meanwhile,Decision-makers want not only to acquire analysis information relevant to their business from current information systems,but also more eager to obtain predictive knowledge to assist business activities.At present,based on Data Warehouse, Maturity and development make Data Mining mainstream technology for corporation decision-making and predictive analysis.This thesis dictates new mind and way to building enterprise service bus based data mining oriented business intelligence system.The architecture of new Business Intelligence system features not only high autonomy and flexibility but also concepts of share and openness absorbing enterprise service bus principle,so that an unified, convenient,fast,and scalable technical supporting platform is provided for all kinds of corporation dicision making.Main components of SOA based data mining oriented Business Intelligence system contains enterprise service bus,data warehouse,OLAP server,data access service,data mining service,data mining access language service,metadata management and visualization as well as frontend application.At the same time,this thesis adopts mainstream data mining standards,in order to ensure the conformance to industrial products.As a result,data mining technologies and tools in this new Business Intelligence system are reliable and advanced.Mainstream technologies are adopted to integrate one solution to core data mining components in this new BI system.A prototype is created and a frontend application customized to validate this new method's practicability and extension.