The Study Of Consistency Issues In Model Transformations Based On Graph Theory

The core idea of MDA is to abstract the business logic to get modeld called PIMs(Platform Independent Model), which is a complete description of system functions independent of concrete technologies. Then develop different transformation rules for different implementation techniques, and transform these PIMs into PSMs (PSM, Platform Specific Model) related to implementation platform. And finally, perform the code generation from PSMs. MDA deals with the complexity of software development by high abstraction level, reduces the impact of demands changing on the implementation by the seperation of business logic and specific techniques, and improves the efficiency and quality by model reuse and automatic model transformation. Meanwhile, MDA uses UML as a standard modeling language, and defined a whole series of standards, from modeling to model transformation and then to code generation. These standards can enhance the compatibility between different systems developed based on MDA.Model transformation is the core technology of MDA . Its correctness is a determinant of the software quality development, and therefore the correctness of model transformations has become one of the most important study areas in MDA . Currently, the study mainly focus on two aspects: syntactic level and semantic level. The form, syntactic correctness verification, has mature solutions yet. However, the lack of semantic description in modeling language, leads to so many difficulties in the study of semantic correctness verification. The existing solutions either deal with a particular area, or may be too cumbersome and complex to realize. There has not been a more common and feasible solution.The main content of this paper is to study the semantics of modeling languages, and try to extract the formal description of semantic properties that need to be preserved in model transformations. Then propose a model transformation framework for consistency checking and perform the verification with the aid of graph theories to guarantee the correctness of model transformations. This solution can be used in the process of model transformations and code generation with some universal significance.In general, the main contributions of this paper are listed:1. Summarize the correctnesson criterion of model transformations and existing verification methods, and then propose a verification method of semantic property preservation to ensure the consistenty between models before and after model transformations. A conflict detection algorithm is given for the conflict detection between transformation rules and semantic properties preservation. so that the correctness of model transformation can be verified.2. Proposes a graph representation of models and object-oriented codes, and then uses the graph transform theory to express model transformation rules, and semantic properties preservation. By this way, graph theory can be used in this model transformation framework for consistency checking to get a better realization.3. By extending the critical techniques, the existing graph tools can be used to realize this framework, and a simple example of model-driven software development is given to show the feasibility and validity of this framework.