代码安全漏洞cve增强版数据集

security-cve 数据集描述 security-cve 数据集是对原始 ReposVul 数据集进行清洗和优化后的版本，是一个包含 6,134 条 CVE 条目的高质量集合，涵盖 C、C++、Java 和 Python 中的 1,491 个项目。它提供了多粒度漏洞信息，从仓库级到行级。清洗过程提高了数据质量，使其适合用于训练和评估用于漏洞检测的机器学习模型。参考https://arxiv.org/pdf/2401.13169将该论文的数据集做增强，标记出漏洞代码特征。

Dataset Description for the security-cve Dataset
The security-cve dataset is a cleaned and optimized version of the original ReposVul dataset. It is a high-quality collection of 6,134 CVE entries, covering 1,491 projects developed in C, C++, Java, and Python. It provides multi-granularity vulnerability information, ranging from repository-level to line-level. The data cleaning process enhances overall data quality, rendering it suitable for training and evaluating machine learning models for vulnerability detection. This dataset has been enhanced and annotated with vulnerability code features by referencing the dataset construction approach detailed in the paper available at https://arxiv.org/pdf/2401.13169.

搜集汇总

背景与挑战

背景概述

该数据集是一个高质量的安全漏洞集合，包含多种编程语言的CVE条目，提供详细的漏洞信息和修复建议，适用于机器学习模型的训练和评估。