Software Quality through the Eyes of the End-User and Static Analysis Tools: A Study on Android OSS Applications

Abstract

Source code analysis tools have been the vehicle for measuring and assessing the quality of a software product for decades. However, recently many studies have shown that post-deployment end-user reviews provide a wealth of insight into the quality of a software product and how it should evolve and be maintained. For example, end-user reviews help to identify missing features or inform developers about incorrect or unexpected software behavior. We believe that analyzing end-user reviews and utilizing analysis tools are a crucial step towards understanding the complete picture of the quality of a software product, as well as towards reasoning about the evolution history of it. In this paper, we investigate whether both methods correlate with one another. In other words, we explore if there exists a relationship between user satisfaction and the application's internal quality characteristics. To conduct our research, we analyze a total of 46 actual releases of three Android open source software (OSS) applications on the Google Play Store. For each release, we employ multiple static analysis tools to assess several aspects of the application's software quality. We retrieve and manually analyze the complete reviews after each release of each application from its store page, totaling 1004 reviews. Our initial results suggest that having high or low code quality does not necessary ensure user overall satisfaction.