Automated topic naming to support cross-project analysis of software maintenance activities

Abstract

Researchers have employed a variety of techniques to extract underlying topics that relate to software development artifacts. Typically, these techniques use semi-unsupervised machine-learning algorithms to suggest candidate word-lists. However, word-lists are difficult to interpret in the absence of meaningful summary labels. Current topic modeling techniques assume manual labelling and do not use domainspecific knowledge to improve, contextualize, or describe results for the developers. We propose a solution: automated labelled topic extraction. Topics are extracted using Latent Dirichlet Allocation (LDA) from commit-log comments recovered from source control systems such as CVS and BitKeeper. These topics are given labels from a generalizable cross-project taxonomy, consisting of non-functional requirements. Our approach was evaluated with experiments and case studies on two large-scale RDBMS projects: MySQL and MaxDB. The case studies show that labelled topic extraction can produce appropriate, context-sensitive labels relevant to these projects, which provides fresh insight into their evolving software development activities.

Exploring the Prevalence and Evolution of Android Concerns: A Community Viewpoint

“…Hindle et al [32], for instance, explored the way NFRs changed over time and examined the difference in emphasis for such issues in MySQL and MaxDB systems. They found that NFRs issues decreased as the systems matured, and interests in such issues for these two systems were driven largely by external technology forces (e.g., the release of new software that needed integration into these products).…”

“…Alipour et al [42] also included NFRs words to aid with the detection of duplicate bugs. Hindle et al's [32] study, while being somewhat similar to the work that is undertaken here, only examined one version of each of two systems, and was aimed mostly at demonstrating how automated topic extraction techniques are relevant to understanding pass events in software projects. In our context we aim to understand how the Android stakeholders' quality-related concerns shifted over time:…”

“…Further, Boehm et al's model [54] has also been widely considered during the assessment of NFRs and software quality issues. In order to evaluate how NFRs issues evolved in the Android OS and to understand whether and how the community's priorities and focus on stakeholders' concerns shifted over time, we considered a combination of these models as used by Hindle et al [32] to classify Android's NFRs related n-grams. In the taxonomy we used each of the six ISO-9126 terms (efficiency, functionality, maintainability, portability, reliability and usability), resulting in an extended list of words under each category.…”

A Survey on Mining Software Repositories

“…Hindle et al proposed a technique to automatically extract labeled topic to help software maintenance activity in the form of supervised and semi-unsupervised approach [100]. They used only commit comments and not functional requirements but NFR (nonfunctional requirements) are targeted.…”

Using heuristics to estimate an appropriate number of latent topics in source code analysis

“…Additionally, some topics recur over the life of a project. The authors later demonstrated [17] how many of these topics can be named using an automated approach.…”