Research On Model Driven Architecture Based Model Transformation Technologies In Software Engineering

Model-Driven Architecture (MDA) is for the purpose of enhancing the efficiency of software development. MDA, component-oriented and service-oriented software architecture, and middleware based distributed development environment complement one another, and become recipes to the development of large-scale software in heterogeneous environment. MDA is the research focus in software engineering. Model-centric development is what MDA aims at. Comparing with traditional code-centric development, model-centric development provides higher level abstraction, which is much closer to applications. Generally, software developed using model-centric development works on heterogeneous platforms. Therefore, model-centric development can be effectively applied in development of network based large-scale complex information systems.Model transformation is the core technique of MDA. So far, a number of domestic and foreign organizations have launched their works around applying model transformation in software development, and have made certain valuable progresses. Firstly, several model transformation languages have been developed and used to feed variant application demands of model transformation. Secondly, rules for transformation between metamodels have been highly concerned. Taking these rules as the specifications, transformation codes now can be automatically generated. Finally, many works have been focused on code generating techniques for variant application domains. These techniques help to improve the efficiency of development of domain-specific software. Nevertheless, model transformation is a new topic, there are several drawbacks when applying it on developing Web based applications. It is difficult to reuse model transformation codes. Code framework after transformation needs to consummate.Our works took software development for Web based applications, such as e-commerce, as the background, to overcome the drawbacks of model transformation technique. We presented formal definitions for some related concepts in the first place. Then taking these definitions as the foundation, we focused our works on model transformation for object oriented design pattern, model weaving, and metadata mapping. Finally we designed and implemented a Web Application Development Platform (WADP), verified the validity of our techniques and approaches. Our contributions are listed as follows:1. We proposed a set of formal definitions to model transformation related concepts including model, metamodel, metametamodel, model transformation, model weaving, and model conformance, basing on MDA Guide, MOF and QVT specifications. And then, we constructed a model transformation formal framework. Hence we concreted a theory basis for researching on model transformation techniques.2. Design pattern describes a repeatedly presenting issue during software designing, as well the solution to it. Applying design pattern enables developers to reuse it to solve a specified designing issue. Design pattern is commonly used in software system modeling to enhance design quality and reusage. Applying and maintaining design pattern in models is a duplicating work. In order to automatically apply design patterns, such as bridge pattern and abstract factory pattern, in models, we carried out a design pattern oriented model transformation approach. By analyzing rules and features of design pattern oriented model transformation, we designed a set of atom mappings and pattern mappings, as well the combination of them, the model transformation code template. Such a template can generate XSLT based model transformation code, and enable automatically applying design pattern in software models.3. The heterogeneity of different model transformation techniques makes it difficult to reuse model transformation code. We provided a QVT-based model transformation weaving framework named QMTW. QMTW describes transformation rules at a higher abstraction layer and supports up-to-date model transformation specification from OMG. Transformation rules described using QMTW can generate codes conforming to variant model transformation languages, and eliminate the heterogeneity of model transformation techniques partly.4. Metadata defines data of some kind of data specification. Metadata includes storage pattern, meaning, and related constraints of data. General metadata mapping adulterates semantic information between source data and target data, and structural information of target data, so automatically generating mapping codes is difficult. We provided a metadata mapping oriented model transformation approach. It separates metadata mapping rules into structural rules and semantics rules. Structural rules are defined by target metadata and semantics rules are described by a weaving framework QMTW. Relevant mapping codes can be generated by model weaving tools. Thus we have done automatically generation of metadata mapping codes.5. Basing on model weaving and metadata mapping, we designed and implemented a Web Application Development Platform (WADP). WADP has the infrastructures such as modeling tools, model storage, and model templates library, for Web applications. Furthermore, it has dynamic replaceable model transformer and code generator. Business models described by users can automatically generate prototypes and back-end databases for Web applications. Using this platform, we have developed five systems such as Shenzhen Bijin Club E-commerce Platform and .infor cooperating electronic government affair supporting platform. Our practice shows that WADP improves the efficiency of development and maintenance of Web applications.