Integrating Behavioral Code Metrics and Sentiment Code Review Analysis for Efficient Defects Discovery

Integrating Behavioral Code Metrics and Sentiment Code Review Analysis for Efficient Defects Discovery

연구 분야: Software Development

학회: Asian Conference on Intelligent Information and Database Systems

Context: Defect discovery in software systems is a critical aspect of software engineering, but is often hindered by the complexity of codebases and team dynamics. Traditional methods like static analysis overlook the behavioral and human-centered aspects of development. Behavioral code metrics and sentiment analysis of code reviews offer complementary approaches that can address these gaps. Behavioral metrics identify high-risk hotspots in code, while sentiment analysis provides insights into the emotional tone of developer interactions, potentially revealing underlying issues in collaboration and process efficiency. Objective: The objective of this study was to integrate behavioral code metrics and sentiment analysis to enhance defect discovery processes in software development. This approach aims to link technical and human factors for a comprehensive understanding of defects-prone areas. Method Using real-world software projects, Author combined hotspot detection metrics, such as change frequency and code complexity, with sentiment trends extracted from code review comments. The analysis correlated hotspots and sentiment polarities to identify patterns associated with higher defect rates. Results: The findings reveal that hotspots with frequent negative sentiments in their associated code reviews exhibit disproportionately higher defect rates. This correlation underscores the importance of addressing both technical and emotional factors in software quality management. The integrated approach improved the prediction of defects and the prioritization of resources. Conclusions; Combining behavioral metrics and sentiment analysis provides a robust framework for defect discovery, balancing technical insights with team dynamics. This method enables more efficient resource allocation, improved code quality, and improved collaboration in complex software environments.