Model Transformation using Adaptive Systems

The key research focus of this paper is the combination of advantages from rule based and adaptive systems to produce a hybrid technique that is better able to handle transformations than either technique in its own right. The target problem for the techniques we are developing of reverse engineering is a significant problem when dealing with legacy systems but has great advantages over the significant costs of maintaining or reengineering the old code. The significant novelty of the system is the application of adaptive systems to the problem, these serve to reduce the complexity and quantities inherent in defining transformations rules for each individual case. Current reverse engineering approaches fail due to the difficulties of writing rules to recognize every possible pattern of code that maps to the higher level model.