Scholarlink

<p>Software Fault Prediction (SFP) is a key step in enhancing the quality and accuracy of source code. This study examines various combinations of software code metrics and CodeBERT approaches, aiming to extract valuable information from both code and comments for comprehensive code embedding. This study employs advanced code-embedding methodologies, incorporating code and comments and utilizing the state-of-the-art CodeBERT model. The study uses the stacking classifier ensemble approach, with logistic regression and XGBoost as the ultimate estimators, alongside 22 classifiers across 41 diverse datasets. The results indicate that the proposed approach successfully predicts faulty source codes based on accuracy and Area under Curve (AUC). Through the experiment, the proposed approach achieved an accuracy of 80.39% and an AUC of 0.7458.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A comprehensive approach to enhance fault prediction through code-comment analysis with CodeBERT

  • 期刊论文

  • 发表日期：

    2025年10月8日

查看全文
  • Monika Yadav,
  • Lov Kumar,
  • Vishal Passricha

摘要

Software Fault Prediction (SFP) is a key step in enhancing the quality and accuracy of source code. This study examines various combinations of software code metrics and CodeBERT approaches, aiming to extract valuable information from both code and comments for comprehensive code embedding. This study employs advanced code-embedding methodologies, incorporating code and comments and utilizing the state-of-the-art CodeBERT model. The study uses the stacking classifier ensemble approach, with logistic regression and XGBoost as the ultimate estimators, alongside 22 classifiers across 41 diverse datasets. The results indicate that the proposed approach successfully predicts faulty source codes based on accuracy and Area under Curve (AUC). Through the experiment, the proposed approach achieved an accuracy of 80.39% and an AUC of 0.7458.