Ant Colony Optimization for Matching Class Diagrams

Identifying the optimal match between two software models is a preliminary for several model management scenarios. This includes model retrieval, consolidation, and evolution. However, the task has exponential time complexity. Ant Colony Optimization is gaining popularity for providing reasonable solutions for different discrete optimization problems. This paper proposes an Ant Colony algorithm for matching UML class diagrams, with their similarity quantified based on their names, attributes, operations, and structural information. Using a case study of ten pairs of class diagrams, the performance of the Ant Colony Optimization algorithm is empirically tested and compared to that of the basic genetic algorithm, in terms of solution accuracy and execution time. The results indicate the superiority of the Ant Colony algorithm over the genetic algorithm, for the three accuracy measures: accuracy, precision, and recall.