Vulnerability Datasets for Classification Performance Analysis

It contains 5 csv files extracted from NVD using CVE API. The files contain description and a severity rating. The 17-23.csv file contains 130430 vulnerability descriptions reported during 2017-2023. It can be used for model training. 2024.csv file contains 19876 descriptions reported in 2024. It may be used to test the model for temporally separated vulnerabilities. kev_april25.csv contains 1309 vulnerability descriptions present in the KEV (Known Exploit Vulnerability) catalog. It contains high-risk vulnerabilities and may be used for performance analysis. android.csv and windows.csv files contain 7379 and 6190 vulnerability descriptions respectively and may be used for performance analysis of classification models.本数据集包含5个通过CVE API（通用漏洞披露API，Common Vulnerabilities and Exposures API）从国家漏洞数据库（National Vulnerability Database，NVD）提取的CSV格式文件，所有文件均收录漏洞描述与严重程度评级。其中17-23.csv涵盖2017至2023年间上报的130430条漏洞描述，可用于模型训练。2024.csv包含2024年上报的19876条漏洞描述，可用于测试模型对时间隔离型漏洞的适配性。kev_april25.csv收录了已知利用漏洞（Known Exploit Vulnerability，KEV）目录中的1309条漏洞描述，此类漏洞均为高风险漏洞，可用于性能分析。android.csv与windows.csv分别包含7379条与6190条漏洞描述，可用于分类模型的性能分析。