Code-Smells Detection using Good and Bad Software Design Examples

Code-Smells refer to design situations that may adversely affect the maintenance of software. They make a system difficult to change, which may in turn introduce bugs. Codesmells are detected, in general, using quality metrics that formalize some symptoms based on detection rules. We propose in this work to consider the problem of code-smells detection as a multi-objective problem where examples of code-smells and well-designed code are used to generate detection rules. To this end, we use multi-objective genetic programming (MOGP) to find the best combination of metrics that maximizes the detection of code-smells examples and minimizes the detection of well-designed code examples. We evaluated our proposal on seven large open source systems and found that, on average, most of the different three code-smells types were detected with an average of 86% of precision and 91% of recall. Statistical analysis of our experiments over 51 runs shows that MOGP performed significantly better than state-ofthe-art code-smell detectors.