MASCOT-Android
收藏资源简介:
MASCOT-Android是由马里兰大学巴尔的摩分校研究团队精心构建的Android恶意软件源代码数据集,旨在为安全研究提供高质量、可追溯的原始代码资源。该数据集包含1093个从GitHub平台收集的恶意软件源代码样本,时间跨度从2011年1月到2025年1月,覆盖后门、僵尸网络、钓鱼邮件等多种恶意软件类型,每个样本均经过人工审核并附带完整的元数据信息。数据集的创建过程通过关键词搜索、作者关联和代码分叉的雪球采样方法实现,并采用基于README文档的TF-IDF特征和LinearSVC分类器构建自动化收集管道,准确率达到96.28%。该数据集主要应用于恶意软件行为分析、代码重用模式研究、LLM辅助恶意软件开发检测以及符号信息对恶意软件检测性能的影响评估等领域,为解决Android恶意软件源代码稀缺性和分析深度不足的问题提供了重要基础。
MASCOT-Android is a high-quality, traceable Android malware source code dataset meticulously constructed by the research team at the University of Maryland, Baltimore, designed to provide premium, traceable raw code resources for cybersecurity research. This dataset contains 1093 malware source code samples collected from GitHub, spanning from January 2011 to January 2025, covering a wide range of malware types including backdoors, botnets, and phishing malware. Each sample has undergone manual review and is accompanied by complete metadata. The dataset was developed through keyword searching, author association, and snowball sampling leveraging code forking. An automated collection pipeline was built using TF-IDF features extracted from README documents and a LinearSVC classifier, achieving an accuracy of 96.28%. This dataset is primarily utilized in research domains such as malware behavior analysis, code reuse pattern investigation, LLM-assisted malware development detection, and the evaluation of the impact of symbolic information on malware detection performance. It serves as a critical foundation for addressing the challenges of scarce Android malware source code and insufficient analytical depth.

- 1MASCOT-Android: A Curated Dataset and Automated Collection Pipeline for Android Malware Source Code Specimens马里兰大学巴尔的摩分校 · 2026年



