Code-Smell Detection as a Bilevel Problem

Abstract

Code smells represent design situations that can affect the maintenance and evolution of software. They make the system difficult to evolve. Code smells are detected, in general, using quality metrics that represent some symptoms. However, the selection of suitable quality metrics is challenging due to the absence of consensus in identifying some code smells based on a set of symptoms and also the high calibration effort in determining manually the threshold value for each metric. In this article, we propose treating the generation of code-smell detection rules as a bilevel optimization problem. Bilevel optimization problems represent a class of challenging optimization problems, which contain two levels of optimization tasks. In these problems, only the optimal solutions to the lower-level problem become possible feasible candidates to the upper-level problem. In this sense, the code-smell detection problem can be treated as a bilevel optimization problem, but due to lack of suitable solution techniques, it has been attempted to be solved as a single-level optimization problem in the past. In our adaptation here, the upper-level problem generates a set of detection rules, a combination of quality metrics, which maximizes the coverage of the base of code-smell examples and artificial code smells generated by the lower level. The lower level maximizes the number of generated artificial code smells that cannot be detected by the rules produced by the upper level. The main advantage of our bilevel formulation is that the generation of detection rules is not limited to some code-smell examples identified manually by developers that are difficult to collect, but it allows the prediction of new code-smell behavior that is different from those of the base of examples. The statistical analysis of our experiments over 31 runs on nine open-source systems and one industrial project shows that seven types of code smells were detected with an average of more than 86% in terms of precision and recall. The results confirm the outperformance of our bilevel proposal compared to state-of-art code-smell detection techniques. The evaluation performed by software engineers also confirms the relevance of detected code smells to improve the quality of software systems.

On the diffuseness and the impact on maintainability of code smells: a large scale empirical investigation

“…We chose not to use existing detection tools (Marinescu 2004;Khomh et al 2009b;Sahin et al 2014;Tsantalis and Chatzigeorgiou 2009;Moha et al 2010;Oliveto et al 2010;Palomba et al 2015a) because (i) none of them has ever been applied to detect all the studied code smells and (ii) their detection rules are generally more restrictive to ensure a good compromise between recall and precision and thus may miss some smell instances. To verify this claim, we evaluated the behavior of three existing tools, i.e., DECOR (Moha et al 2010), JDeodorant (Tsantalis and Chatzigeorgiou 2009), and HIST (Palomba et al 2015a) on one of the systems used in the empirical study, i.e., Apache Cassandra 1.1.…”

“…On the one side, researchers developed methods and tools to detect code smells. Such tools exploit different types of approaches, including metrics-based detection (Lanza and Marinescu 2010;Moha et al 2010;Marinescu 2004;Munro 2005), graph-based techniques (Tsantalis and Chatzigeorgiou 2009), mining of code changes (Palomba et al 2015a), textual analysis of source code (Palomba et al 2016b), or search-based optimization techniques (Kessentini et al 2010;Sahin et al 2014). On the other side, researchers investigated how relevant code smells are for developers (Yamashita and Moonen 2013;Palomba et al 2014), when and why they are introduced (Tufano et al 2015), how they evolve over time (Arcoverde et al 2011;Chatzigeorgiou and Manakos 2010;Lozano et al 2007;Ratiu et al 2004;Tufano et al 2017), and whether they impact on software quality properties, such as program comprehensibility (Abbes et al 2011), fault-and change-proneness (Khomh et al 2012;Khomh et al 2009a;D'Ambros et al 2010), and code maintainability Moonen 2012, 2013;Deligiannis et al 2004;Li and Shatnawi 2007;Sjoberg et al 2013).…”

The Scent of a Smell: An Extensive Comparison Between Textual and Structural Smells

“…Most of them are based on the analysis of the structural properties of source code (e.g., method calls) and on the combination of structural metrics [51], [56], [65], [67], [69], [72], [88], [95], [99], [103], while in recent years the use of alternative sources of information (i.e., historical and textual analysis) have been explored [74], [75], together with methodologies based on machine learning [3], [33] and search-based algorithms [17], [46], [47], [87].…”

Aspect Oriented Re-engineering of Legacy Software Using Cross-Cutting Concern Characterization and Significant Code Smells Detection

“…Although Extract Method could be possible solution, however, if the problem of scattering persists, it can accumulate many design anomalies [27]. Alternatively, the code could be moved into an aspect to get rid of fragility as aspects are known to modularize code cutting across the whole software architecture [35,36]. Similarly, we present our motivating example of God Class in Listing 2, where aspect-oriented re-factoring (AOR) seems to be a better solution.…”