兰州大学机构库 >信息科学与工程学院
Code classification with graph neural networks: Have you ever struggled to make it work?
Yu, Qingchen1; Liu, X(刘忻)2; Zhou, QG(周庆国)2; Zhuge, Jianwei3; Wu, Chunming1
2023-12-15
Online publication date2023-07
Source PublicationEXPERT SYSTEMS WITH APPLICATIONS   Impact Factor & Quartile
ISSN0957-4174
Volume233
page numbers15
AbstractCode classification is a meaningful task with plenty of practical applications. Combined with recently popular graph neural networks (GNNs), a body of research attempts to address the code classification problem with the help of fruitful achievements from the deep learning (DL) area. We systematically investigate the practices of existing works on the GNN-based code classification task and find out three important but often overlooked questions. (1) Existing works usually illustrate the effectiveness of GNNs in code classification tasks by contrasting them with non-graph baselines. But the contribution of the message-passing mechanism, which is at the core of GNNs, has not been shown on its own. So how does the message-passing mechanism help code classification? (2) Various problem formulations, model architectures, and learning objectives have been suggested in the literature to apply GNNs to code classification. In practice, how should we make choices among them? (3) One of the most prominent applications of code classification is automated vulnerability detection. However, learning-based vulnerability detection has not been widely accepted as a practical approach. How does GNN-based code classification perform in this task, especially in real-world scenarios? A comprehensive experimental study is conducted around these questions to evaluate the performance and feasibility of the GNN on different datasets and scenarios. Results suggest it is not easy to make the GNN perform well on the code classification task in the face of the intricate nature of programs. Interesting findings about the effectiveness of GNNs with different training objectives are reported for the first time in this paper, providing insights that can inform future research on code classification with GNNs. © 2023 Elsevier Ltd
KeywordCodes (symbols) Deep learning Message passing Network coding Network security Classification tasks Code classification Deep learning Graph neural networks Message-passing Model learning Network-based Problem formulation Software vulnerabilities Vulnerability detection
PublisherElsevier Ltd
DOI10.1016/j.eswa.2023.120978
Indexed ByEI ; SCIE
Language英语
WOS Research AreaComputer Science ; Engineering ; Operations Research & Management Science
WOS SubjectComputer Science, Artificial Intelligence ; Engineering, Electrical & Electronic ; Operations Research & Management Science
WOS IDWOS:001049594300001
EI Accession Number20233014442926
EI KeywordsGraph neural networks
EI Classification Number461.4 Ergonomics and Human Factors Engineering ; 716.1 Information Theory and Signal Processing ; 723 Computer Software, Data Handling and Applications ; 723.1 Computer Programming ; 723.2 Data Processing and Image Processing ; 723.4 Artificial Intelligence
Original Document TypeJournal article (JA)
Citation statistics
Document Type期刊论文
Identifierhttps://ir.lzu.edu.cn/handle/262010/532097
Collection信息科学与工程学院
Corresponding AuthorLiu, Xin; Wu, Chunming
Affiliation
1.College of Computer Science and Technology, Zhejiang University, China;
2.School of Information Science and Engineering, Lanzhou University, China;
3.Institute for Network Science and Cyberspace, Tsinghua University, China
Corresponding Author AffilicationSchool of Information Science and Engineering
Recommended Citation
GB/T 7714
Yu, Qingchen,Liu, Xin,Zhou, Qingguo,et al. Code classification with graph neural networks: Have you ever struggled to make it work?[J]. EXPERT SYSTEMS WITH APPLICATIONS,2023,233.
APA Yu, Qingchen,Liu, Xin,Zhou, Qingguo,Zhuge, Jianwei,&Wu, Chunming.(2023).Code classification with graph neural networks: Have you ever struggled to make it work?.EXPERT SYSTEMS WITH APPLICATIONS,233.
MLA Yu, Qingchen,et al."Code classification with graph neural networks: Have you ever struggled to make it work?".EXPERT SYSTEMS WITH APPLICATIONS 233(2023).
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Altmetrics Score
Google Scholar
Similar articles in Google Scholar
[Yu, Qingchen]'s Articles
[Liu, Xin]'s Articles
[Zhou, Qingguo]'s Articles
Baidu academic
Similar articles in Baidu academic
[Yu, Qingchen]'s Articles
[Liu, Xin]'s Articles
[Zhou, Qingguo]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Yu, Qingchen]'s Articles
[Liu, Xin]'s Articles
[Zhou, Qingguo]'s Articles
Terms of Use
No data!
Social Bookmark/Share
No comment.
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.