The problem of conceptualization in god class detection: agreement, strategies and decision drivers

Abstract

Background:The concept of code smells is widespread in Software Engineering. Despite the empirical studies addressing the topic, the set of context-dependent issues that impacts the human perception of what is a code smell has not been studied in depth. We call this the code smell conceptualization problem. To discuss the problem, empirical studies are necessary. In this work, we focused on conceptualization of god class. God class is a code smell characterized by classes that tend to centralize the intelligence of the system. It is one of the most studied smells in software engineering literature. Method: A controlled experiment that extends and builds upon a previous empirical study about how humans detect god classes, their decision drivers, and agreement rate. Our study delves into research questions of the previous study, adding visualization to the smell detection process, and analyzing strategies of detection. Result: Our findings show that agreement among participants is low, which corroborates previous studies. We show that this is mainly related to agreeing on what a god class is and which thresholds should be adopted, and not related to comprehension of the programs. The use of visualization did not improve the agreement among the participants. However, it did affect the choice of detection drivers. Conclusion: This study contributes to expand empirical evidences on the impact of human perception on detecting code smells. It shows that studies about the human role in smell detection are relevant and they should consider the conceptualization problem of code smells.

Exploring decision drivers on god class detection in three controlled experiments

“…The drivers' choice is more related to personal conceptualization of the smell than to technical aspects. This helps to characterize what we called the code smell conceptualization problem [19,21].…”

“…FinG aimed to address a set of context-aspects on god class detection, such as agreement, effort, strategies, and others. We had already partial results in [19,20,21]. In this paper, we focused only in the investigation of drivers adopted by the participants.…”

“…Other data analysis researchers also addressed the problems with Kappa values [8,17,24]. We discuss this issue in [19,21].…”

Building empirical knowledge on the relationship between code smells and design patterns: An exploratory study

“…On the other hand, manual inspection of code smell, beside being error prone, is impacted by subjectivity. Santos et al 64 call this issue "code smell conceptualization problem." Even other variables related to the development process impacting design quality, such as organizational culture or time-to-market, are hard to control and they also impact the evaluation.…”

Towards Customizing Smell Detection and Refactorings

“…Similarly to Class Data Should Be Private, the lowest one was obtained by NB. Note that the coverage of metrics also could implicate the effectiveness obtained by the techniques as seen in Complex Class, however some metrics, as Lines of Code, have higher influence when detecting God Class (74,79). Besides that, God Classes are usually associated with high values of the metrics.…”