Characterizing Software Maintainability in Issue Summaries using a Fuzzy Classifier

Chen, Celia; Shoga, Michael; Boehm, Barry

doi:10.1109/qrs.2019.00029

Cited by 3 publications

(2 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These patterns illustrate the recurrent linguistic rules that users are likely to use when reporting bugs or requesting new features. Motivated by these heuristic linguistic patterns, we proposed a fuzzy classifier [31,38] that aims to identify the maintainability subgroup SQ concerns expressed in issue summaries. Based on the definitions of these patterns, a set of 24 initial fuzzy rules was generated by heuristically identifying them from subgroup SQ definitions and practice guidelines.…”

Section: Background Studiesmentioning

confidence: 99%

See 1 more Smart Citation

A Large Empirical Study on Automatically Classifying Software Maintainability Concerns from Issue Summaries

Chen¹,

Shoga²

2021

Adv. sci. technol. eng. syst. j.

Self Cite

View full text Add to dashboard Cite

Software maintenance contributes the majority of software system life cycle costs. However, existing approaches with automated code analysis are limited by accuracy and scope. Using human-assessed methods or implementing quality standards are more comprehensive alternatives, but they are much more costly for smaller organizations, especially in opensource software projects. Instead, bugs are generally used to assess software quality, such as using bug fixing time as an estimate of maintenance effort. Although associated bug reports contain useful information that describe software faults, the content of these bug reports are rarely used. In this paper, we incorporate quality standards with natural language processing techniques to provide insight into software maintainability using the content of bug reports and feature requests. These issues are classified with an automated approach into various maintainability concerns whose generalizability has been validated against over 6000 issue summaries extracted from nine open source projects in previous works. Using this approach, we perform a large empirical study of 229,329 issue summaries from 61 different projects. We evaluate the differences in expressed maintainability concerns between domains, ecosystems, and types of issues. We have found differences in relative proportions across ecosystem, domain and issue severity. Further, we evaluate the evolution of maintainability across several versions in a case study of Apache Tomcat, identifying some trends within different versions and over time. In summary, our contributions include a refinement of definitions from the original empirical study on maintainability related issues, an automated approach and associated rules for identifying maintainability related quality concerns, identification of trends in the characteristics of maintainability related issue summaries through a large-scale empirical study across two major open source ecosystems, and a case study on changes in maintainability over versions in Apache Tomcat.

show abstract

Section: Background Studiesmentioning

confidence: 99%

“…This study depends on the model developed in [38]. The accuracy of classification is subject to the limitations and threats to validity detailed in the prior work.…”

Section: Threats To Validitymentioning

confidence: 99%