A replication of ‘DeepBugs: a learning approach to name-based bug detection’

Winkler, Jordan; Agarwal, Abhimanyu; Tung, Caleb; Ugalde, Dario Rios; Jung, Young Jin; Davis, James C.

doi:10.1145/3468264.3477221

Cited by 2 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are noteworthy studies in the field of bug detection, such as [4], [5], and [6], but the focus of these works differs from what we address in this article. It is crucial to emphasize that bug localization and bug detection are distinct concepts and tackle separate issues.…”

Section: Introductionmentioning

confidence: 99%

Bug Localization Model in Source Code Using Ontologies

da Silva,

Garcia,

Botega

2023

IEEE Access

View full text Add to dashboard Cite

The bug location process aims to identify source code artifacts associated with reported bugs. Manual bug location is burdensome for programmers who must reproduce and analyze the bug to identify the defective artifact and perform necessary maintenance. Bug locating techniques classify and identify project-specific source code artifacts, narrowing the search space. These techniques often use machine learning methods, such as textual similarity, classification algorithms, and grouping of source code files based on bug report data. This paper proposes a bug location model that leverages semantic architectural knowledge through ontologies to infer new knowledge and retrieve information from bug reports. The model's performance is evaluated on six relevant open-source projects in C Sharp (AutoMapper, MsBuild, EfCore, AspNetCore, MQTTnet, and NLog). Experiments utilize the evaluation metrics Top N Rank of Files (TNRF), Mean Reciprocal Rank (MRR), and Mean Average Precision (MAP). The results demonstrate the significant efficacy of the proposed model. The model contributes to relieving the manual burden on programmers and enhances bug localization accuracy and efficiency by integrating architectural semantic knowledge represented through ontologies with machine learning. The evaluation results indicate the potential of the proposed model for improving the bug-fixing process in software development.

show abstract

Section: Introductionmentioning

confidence: 99%