VELVET: a noVel Ensemble Learning approach to automatically locate VulnErable sTatements

Ding, Yangruibo; Suneja, Sahil; Zheng, Yunhui; Laredo, Jim; Morari, Alessandro; Kaiser, Gail E.; Ray, Baishakhi

doi:10.1109/saner53432.2022.00114

Cited by 26 publications

(32 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To evaluate the effectiveness of COSTA, we adopt the most popular and largest vulnerability dataset, Big-Vul [16], which contains 15,478 vulnerable statements and 83,844 non-vulnerable statements. Our vulnerable/nonvulnerable statement classification results show that our context-based approach outperforms the state-of-the-art approaches [17,19,20] up to 96% in F1-Score and 167% in Accuracy. In addition, by ranking the statements according to their suspiciousness scores, COSTA improves these approaches from 30% to 100% in Top-1 Accuracy.…”

Section: Introductionmentioning

confidence: 84%

“…How accurate is COSTA in localizing vulnerable statements? How is it compared with the state-of-the-art approaches [17,19,20]?…”

Section: Experimental Methodologymentioning

confidence: 99%

“…For dependence context, COSTA conducts both backward and forward slicing in the Program Dependence Graph (PDG) to obtain statements impacting/being impacted by s via control and data dependencies. In this work, both the AST and the PDG are parsed by Joern [23] which is a popular source analysis tool widely used in software engineering research [10,11,13,17,20]. The surrounding context is approximated by k statements preceding and k statements succeeding s in the function.…”

Section: Context Extractionmentioning

confidence: 99%

“…In order to narrow down the search space, several recent studies [17][18][19][20] have focused on fine-grained, i.e., statement-level vulnerability detection. However, there are two main limitations in these approaches.…”

Section: Introductionmentioning

confidence: 99%

“…In addition, by ranking the statements according to their suspiciousness scores, COSTA improves these approaches from 30% to 100% in Top-1 Accuracy. In particular, the Top-1 Accuracy of COSTA is 42.15% while these figures of VELVET [20], LineVul [19],…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Context-based Statement-level Vulnerability Localization

Nguyen¹,

Vo²

2022

Preprint

View full text Add to dashboard Cite

The number of attacks exploring software vulnerabilities has dramatically increased, which has caused various severe damages. Thus, early and accurately detecting vulnerabilities becomes essential to guarantee software quality and prevent the systems from malicious attacks. Multiple automated vulnerability detection approaches have been proposed and obtained promising results. However, most studies detect vulnerabilities at a coarse-grained, i.e., file or method level. Thus, developers still have to spend significant investigation efforts on localizing vulnerable statements. In this paper, we introduce COSTA, a novel context-based approach to localize vulnerable statements. In particular, given a vulnerable function, COSTA identifies vulnerable statements based on their suspiciousness scores. Specifically, the suspiciousness of each statement is measured according to its semantics captured by four contexts, including operation context, dependence context, surrounding context, and vulnerability type. Our experimental results on a large vulnerability dataset show that COSTA outperforms the state-of-the-art approaches up to 96% in F1-score and 167% in Accuracy. COSTA also surpasses these approaches up to two times in Top-1 Accuracy. Especially, COSTA obtains about 80% at Top-3 Recall. In other words, developers can find about 80% of the vulnerable statements by investigating only three first-ranked statements in each function.

show abstract

Section: Introductionmentioning

confidence: 84%

“…How accurate is COSTA in localizing vulnerable statements? How is it compared with the state-of-the-art approaches [17,19,20]?…”

Section: Experimental Methodologymentioning

confidence: 99%