Research In Difference Detecting And Merging On Domain-Specific Model

Domain modeling, as a domain-oriented software development method, can reflect the special needs of domain. For a particular domain, developers in specific domains are able to build a domain model promptly, basing on the domain modeling language created by domain experts. A great deal of engineering practice has shown that domain modeling is more efficient than unified modeling language which is based on MDA in development. With gradual improvement of the theory for domain modeling, the domain-oriented development methods are used by more and more organizations and companies, and the management of domain model has become an significant issue. Especially in the process of model creation, iterative type development for complex and massive software system often need different people to cooperate to construct the model in different stages. In order to capture and describe the system characteristics, it requires the ability to track and demonstrate the elements changes in the model, and,if necessary, it can merge these modeling results.This paper analyzes the domain-oriented development methods, including development process, domain modeling language, model integrated computing, development tools, then based on a domain-oriented method which is proposed by ISIS Vanderbilt University,we proposed a method which is able to detect and merge the model difference. The improved method replace the only identifier by the combination-identifier and structural similarity to determine the method of mapping model elements, and then the improved method divide the difference in domain model into effective differences and invalid differences base on the impact of model interpretation and code generation. In addition, the improved method divide effective difference into four aspect include creation, deletion, modification and movement, then we detect the model differences based on our proposed model element mapping algorithm. Based on the difference detection of domain model, we proposed a method which is able to detect the model conflict on the environment of people-development at the same time. The proposed method divide the conflict into deletion modification and so on. Additionally, the proposed method provide the conflict to users by category, so users can determine the reasons of conflict quickly and carry on the merger of model difference conveniently. At last, we designed and developed a system which is able to detect and merge the model differences, the system can verify the feasibility of the algorithm through signal flow domain model. The result show that the proposed method in this paper is feasible.