Code Visualization through Clothing Metaphor: An Empirical Software Comprehension Study

In a world where software systems grow continuously, adding a considerable amount of features after periods of time, it is important to know that the teams can also vary, with different people leaving or joining the process. When new people join a software product, resources are spent understanding the program, a process that could use improvements. This paper proposes a software visualization plugin for object-oriented systems, CodeVestimenta, created as a new metaphor in the form of clothing. Each class of the system is drawn as a specific piece of clothing, with each piece of clothing depicted with various features depending on seven software quality metrics (Lines of Code, Number of Class Members, Number of Class Methods, Number of Child Classes, Depth of Inheritance Treen, Response for Class, and Coupling Between Objects) widely used in software applications. CodeVestimenta was created with the objective of enhancing the program comprehension and understandability of the code by the developer by finding parts of the code that are the application’s core or are more complex than others. An empirical study was conducted to validate CodeVestimenta. Participants (41 in total) had diverse backgrounds in terms of industry experience and software metrics, and were divided into two groups (Text Group and Visual Group) having different available resources for code comprehension (code, metrics values, and visual representation). The program comprehension tasks were built around four major scopes that were embedded into ten questions. The results show that CodeVestimenta helps developers to understand parts of the system that are not visible at first glance, with improvements in the speed of this process and the level of understanding.