Fixing class design inconsistencies using self regulating particle swarm optimization

作者：George Renu; Samuel Philip

来源：Information and Software Technology, 2018, 99: 81-92.

DOI：10.1016/j.infsof.2018.03.005

• Class diagram
• Activity diagram
• Consistency
• Self regulating particle swarm optimization
• Artificial intelligence

Context: The practice of using Unified Modeling Language models during software development and the chances of occurrence of model inconsistencies during software design are increasing. Hence detection of intra-model design inconsistencies is significant in the development of quality software.
Objective: The existing approaches of detecting class attribute inconsistencies rely on human decision making. Manual detection of inconsistencies is exhaustive, time consuming and sometimes incomplete. Therefore, we propose an automated and novel approach to perform consistency check of class attributes using artificial intelligence.
Method: Inconsistency in attribute definition and specification is detected and fixed with self regulating particle swarm optimization (SRPSO) algorithm that uses a fitness function to optimize the consistency of attributes in class diagram and activity diagrams. SRPSO is preferred since the best particle is not influenced by its or others experience and uses its direction as the best direction and the remaining particles use self and social knowledge to update their velocity and position.
Result: The use of artificial intelligence technique for detection and fixing of inconsistencies during the software design phase ensures design completeness through generation of models with consistent attribute definitions and a significant improvement in software quality prediction, accurate code generation, meeting time deadlines, and software production and maintenance cost is achieved.
Conclusion: Ensuring consistency and completeness of models is an inevitable aspect in software design and development. The proposed approach automates the process of inconsistency detection and correction in class attribute definition and specification using SRPSO algorithm during the design phase of software development.

• 出版日期2018-7