Detecting Software Bad Smells from Software Design Patterns using Machine Learning Algorithms

摘要

Design patterns and smells are two approaches that assure design quality. Code smells are symptoms of the design issues which can further hinder the maintenance of the software system. Software Design Patterns represent solutions to various design problems. The code having good design patterns is expected to have less bad smells but their inadequate use can lead to more design smells. Therefore, a code having design patterns can have design smells. This study finding some design pattern and smell pairs which co-exist in the code. This study uses J48 supervised machine learning algorithm to detect the desired relation. The model presented in this paper successfully finds the desired results.