Research on Object-Oriented Design Defect Detection Method Based on Machine Learning

Design defects are one of the main reasons for the decline of software design quality. Effective detection of design defects plays an important role in improving software maintainability and scalability. On the basis of defining software design defects, according to C&K design metrics and heuristics, this paper extracts the relevant features of design defects. Based on classical machine learning methods, classifiers are trained for design defect, and candidate designs are classified by classifiers, so as to identify whether there is a design defect in the design. Experiments show that the method has high accuracy and recall rate in identifying design defects.