LAUSR

Code smells instigate due to the consistent adoption of bad programming and implementation styles during the evolution of the software which adversely affects the software quality. They are essentially focused and prioritized for their effective removal based on their severity. The study proposed a hybrid approach for inspecting the severity based on the code smell intensity in Kotlin language and comparing the code smells which are found equivalent in Java language. The research work is examined on five common code smells that are complex method, large class long method, long parameter list, string literal duplication, and too many methods over 30 open-source systems (15 Kotlin/15 Java). The experiment compares different machine learning algorithms for the computation of human-readable code smell detection rules for Kotlin, where the JRip algorithm proved to be the best machine learning algorithm with 96% and 97% of overall precision and accuracy, validated at 10-fold cross-validation. Further, the severity of code smell at the class level is evaluated for prioritization of applications written in Kotlin and Java language. Moreover, the process of severity computation is semiautomated using the CART model, and thus, metric-based severity classification rules are achieved. The experimentation provides a complete understanding of prioritization of code smells in Kotlin and Java and helps to attain prioritized refactoring which will enhance the utilization of resources and minimize the overhead rework cost.