Aspect Mining from a Modeling Perspective

Aspect mining aims at identifying, analyzing, and refactoring crosscutting concerns throughout a legacy system for the purpose of improving software modularization. Current research on aspect mining prevails at the implementation level as applied to source code. However, an aspect-oriented approach can be beneficial at various levels of abstraction and at different stages of the software lifecycle. This paper presents our investigation into raising the benefits of aspect mining to a higher level of abstraction through application of aspect mining algorithms to domain-specific models. A key contribution of the approach is a capability to identify crosscutting concerns early in development, which assists in modularizing a design through aspects before proceeding to the implementation level. Furthermore, our experience has led us to believe that aspects are easier to identify at the modeling level because much of the accidental complexities of implementation concerns are not present in the corresponding modeling abstractions.