Research And Application Of The Method Of Business Component Identification With Graph Segmentation

With informatization process accelerating its pace, enterprise need flexible information system architecture to satisfy the changeable business demand, finally further to strength the ability to participate in international cooperation and market competition. Component-based information system can match the flexible changes of business demand in the way to assemble components; it becomes the best choice for enterprise.The successful implementation of component-based system must meet a precondition that having a stable component model. Business component identification is the first step of component model, it plays an important role. Currently, there are many identification methods of business component, but all methods are lack of guidance of standard methodology and practice application. With the research of above literature and investigation of enterprise application, this paper discusses the problem of "Information Isolation Island" has been impact on enterprise informatization process, and points that enterprise informatization should pay more attention on the whole application of information system. So through analysis of the characters and problems of available business component identification on business component modeling, this paper proposes a new method of business component identification with graph segmentation in order to improve the exactness and application of the process of business component modeling.In this paper the method of business component identification is centered on three aspects of business modeling,semantic abstraction and identification algorithm. The new identification method not only satisfies the quality of higher cohesion and lower coupling between business components, resulting in lower reuse cost, but also improves the reuse degree of business component. For the business modeling, it is introduced the notion of "traction" in Dynamic Essential Modeling of Organization (DEMO).It takes advantage of modeling benefits of domain analysis, objective decomposition and "traction" to result in the quality of large-granularity and general-purpose of business component; For semantic abstraction, the connection of business elements are extended to entity connection,process connection and entity-process connection, and the weighted graph is converted into business components resulting in accuracy of content of business component; For identification algorithm, it uses clustering identification with genetic algorithm, proposes the evaluation function in order to less communication information between business components for lower reuse cost.The last section in this paper, we applicate the business component identification method proposed in banking business of the securities card change, and evaluate the instance business result. The result indicates that the business component identification method proposed in the paper does improve the reuse degree and reduce the reuse cost of business component. The new identification method has practical use.