Research On Methods Of Model-Driven Software Component Classification And Retrieval

Reusable software development can effectively improve the quality and the efficiency of software development. With the developing of reusable software research, software component repository is regard as an important basic tool of reusable software development. The industrial domain and academia pay more and more attention to software component repository. Therefore, the software components classification and retrieval is one of research emphases. At present there are several methods of software components classification and retrieval. They manage software components by their own point of view. However, these methods regard software components as isolated entities. The relationships between components and their running environment are not considered when the components are classified and retrieval. A model-driving software component classification and retrieval method is represented in this dissertation to resolve this problem. This method manages software components by the point of application systems view. By this method the relationships between software components and their application frameworks are considered when classifying software components. In this method, the retrieval conditions of software components is got form the application system design model.This dissertation introduces several representational software components classification and retrieval methods and summarizes the problems in existing methods and the solution approaches.A component based software system model (CBSSM) and an application framework model (CFM) is represented. CBSSM and CFM are models based on Petri Net and to serve for software components classification and retrieval. CBSSM is a tree layered model, application processing model, component abstract model, performance analysis model. During CBSSM modeling, software components are abstracted form functions requirement of application system by the behavior relativity of Petri Net and the non-functionality of the application system is analyzed and validated. CFM model simplifies CBSSM model. CFMmodel describes the structure of framework, relationships between software components, relationships between software component and frameworks.A software component classification and retrieval method is presented based on CBSSM model and CFM model. CBSSM model provides retrieval conditions of the software components. CFM model provides relationships between software components. Therefore, model is regarded as the core and basic of the model-driving software components classification and retrieval method.Relationship based software components classification is represented. The characteristic of faceted software components classification is analyzed. Based on these characteristics, the SOM clustering is improved to solve the problems that the clustering result is disturbed by input sequence and the structure of SOM must be predefined. Based on improved SOM clustering, an automatic component classification is provided to free faceted classification from term space. SOM clustering has two disadvantages such that their topology construction need be defined in advance and the cluster result is disturbed by order of learning samples. So the training process of SOM clustering is improved to increase the accuracy of software component classification.The method of Establishing suffix index tree for software components is represented. The suffix index tree describe the terms of all facets. By CFM model, the possibility that different components are applied in the same framework is estimated. The framework information for every component is added to the index tree. So the software components found at the same time can be deployed in the same application system. This method improves the retrieval efficiency and provides more service to the development of application system.