Λεπτομέρειες βιβλιογραφικής εγγραφής
| Τίτλος: |
基于多粒度图卷积网络与自适应特征融合的 软件缺陷预测方法. (Chinese) |
| Alternate Title: |
Software defect prediction with multi-granularity graph convolution and adaptive feature fusion. (English) |
| Συγγραφείς: |
金海波, 刘雪婷, 肖成龙 |
| Πηγή: |
Application Research of Computers / Jisuanji Yingyong Yanjiu; Sep2026, Vol. 43 Issue 9, p2749-2757, 9p |
| Θεματικοί όροι: |
Graph neural networks, Contrastive learning, Defect tracking (Computer software development), Machine learning, Programming language semantics |
| Abstract (English): |
To address the shortcomings of existing methods in deep representation of code semantics and structure, this paper proposed a software defect prediction method based on multi-granularity graph convolutional network and adaptive feature fusion. The method combined GraphCodeBERT, a multi-granularity graph convolutional network, and contrastive learning to extract fine-grained semantic and structural information of code. GraphCodeBERT generated semantic-structural joint features through code sequences and variable node sequences. The multi-granularity graph convolutional network further extracted hierarchical structural semantics from local to global based on these features. Contrastive learning enhanced the discriminability of the obtained features. Finally, an adaptive gating mechanism fused handcrafted features for classification. Experimental results on the PROMISE dataset show that the proposed method achieves 3. 3% and 2. 0% improvements in F1 -score over the best baselines in within-project and cross-project defect prediction scenarios, respectively, validating the effectiveness of multisource feature fusion. [ABSTRACT FROM AUTHOR] |
| Abstract (Chinese): |
针对现有方法在代码语义与结构深层表征上的不足, 提出一种多粒度图卷积网络与自适应特征融合的 软件缺陷预测方法。 该方法结合 GraphCodeBERT、多粒度图卷积网络和对比学习, 提取代码的细粒度语义与结 构信息。 GraphCodeBERT 通过代码序列与变量节点序列, 生成语义-结构联合特征; 多粒度图卷积网络在此基础 上分层提取局部至全局的结构语义; 对比学习进一步增强上述特征的判别性; 最后, 通过自适应门控机制融合手 工特征进行分类。 在 PROMISE 数据集上的实验表明, 所提方法在项目内与跨项目缺陷预测场景下的 F1 -score 分别较最优基线提升 3. 3% 和 2. 0%, 验证了多源特征融合的有效性。 [ABSTRACT FROM AUTHOR] |
|
Copyright of Application Research of Computers / Jisuanji Yingyong Yanjiu is the property of Application Research of Computers Edition and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) |
| Βάση Δεδομένων: |
Complementary Index |