Research On Larger-granulary Model Transformation Issues Based On Semantic Preservation In Model Driven Architecture Framework

With the rapid development of the market, the competition of the enterprises becomes fiercer and fiercer. In order to survive in a changing marketplace, the enterprises are forced to adjust their business constantly. At the same time, the changing business information, namely software requirement, put forward higher request to the development of the enterprise management software. Therefore, the solution that how to quickly adapt to the constantly changing software requirements has become an important challenge in the field of software development.To address this challenge, the development of methods to realize enterprise management software has acquired outstanding achievements through Model-Driven Architecture. MDA is an architecture composed of different levels of abstraction models and the conversions of these models. The development of enterprise management software based on MDA is a process driv en by the model transformation, where the business model is taken as the carrier. Through the model transformation, it can reveal the change of the business requirements in the form of the change of model performance and pass these changes to the software model until to the software system, to improve the enterprise management software to quickly respond to the changing market demand. This paper mainly researches the key technologies of model transformation under the framework of MDA. Based on the traditional transformation rules, we introduce the idea of data mining to construct the large granularity transformation rules. Through the rules of large granularity transformation conversion, it achieves the coarse-grained model transformation to improve the efficiency and accuracy of model transformation.The main work of this paper includes the following aspects:(1) To address the problems of the complex forms of expression and the difficulties of semantic proof in model of semantic transformation rules, it foc uses on the conversion rules in abstract syntax and concrete syntax, including atomic conversion rules and complex(large granularity) transformation conversion rules. Concentrating on the existing conversion rules, it also analyses the structure of transformation rules and semantic relationships between them and studies the organization form of the conversion rules. To support the reusing of the transition rules, it investigates the compound operation for conversion rules and provide s a theoretical basis for the construction of complex(large granularity) transformation conversion rules.(2) Aiming at the difficulties of extracting semantic information in the construction process of model transformation rules, it analyses the method to construct the existing conversion rules, and the technology of the Rough Set Theory to extract semantics, plus it makes a close study of the method to construct the transformation rules based on rough set. According to the change of model elements concerned in the model transformation, the evolution of modeling language and other reasons, it analyses several situations where constructing transformation rules are affected, and studies an incremental updating technique of conversion rules. In order to evaluate the conversion rules and the measure of complexity metrics, the matrics for conversion rules are also investigated, which provide a basis for selection of transformation rules.(3) To address the problems that only the atomic transition rules of model transformation may lead to a failure of(not supported) model elements conversion or determine the conversion rules repeatedly, Study the method to use the composite conversion rules to achieve the model conversion. First of all, from viewpoint of the fact that enterprises have the similar business models and software models includes similar business models and similar software models, this paper analyses the form of these similar patterns in transformation rules, and proposes the concept of transformation model. In addition, the method of constructing the composite conversion rules was studied. Aiming at the composite conversion rules in the model transformation, it also studies the detection method of the convertible sub models corresponding to the composite conversion rules, and analyses the semantic relations between the set of the convertible sub models and source models, in addition to the maintaining method to classify the convertible sub models keeping the semantic consistency, which provides the theory basis for the semantic consistency of model transformation.(4) Aiming at the conversion solutions to use the rules of multi-granularity transformation conversion for model transformation(classification of the convertible sub models), most of the conversion solutions need to be optimized and verified.Firstly, it needs to analysis the transformation rules which have participated in the model transformation and their execution sequence, and study on the selection method of the transformation rules based on t he optimal clustering analysis. Then, according to the selection of the conversion rules, it studies the verification method of sequence transformation rules which k eep the semantic consistency.(5) Based on the theory, methods and algorithms proposed in this thesis design and develop the MDA-oriented tools and the model transformation tools, using the conversion between logistics business process models as application cases to verify the theory proposed and to validate the method mentioned herein from the practical point.