Analytical learning based on a meta-programming approach for the detection of object-oriented design defects

This study proposed a new defect-detection approach using a declarative meta-programming technique to analytical learning for object-oriented software. The extrapolating patterns are generated using analytical learning in which certain design defect characteristics can be understood through deductive learning. This study uses declarative meta-programming to represent the specific object-oriented components as logic rules with which design defects can finally be described. Using the two complementary techniques, the object-oriented software is transformed into the narrow related problem domain, in which design defect problems can be managed and simplified. The approach is validated by detecting design defects in certain open-source systems. The results obtained exhibit a superior precision to the conventional method. In application, the proposed strategy can be recognised as a flexible and automated system for detecting software design defects which many object-oriented software systems are able to use. © 2012 Asian Network for Scientific Information.