AndroMetric: Bridging Multi-Dimensional Software Metrics and Mobile Application Security

AndroMetric is a large-scale dataset designed to support empirical research at the intersection of software metrics and Android application security. The dataset contains 39,243 Android applications, including 29,208 malware samples and 10,035 benign apps, collected from publicly available repositories and analyzed using a static analysis pipeline. For each APK, 31 static code metrics are extracted to capture multi-dimensional structural and behavioral characteristics. These metrics span six categories: (i) size and dimensional structure, (ii) object-oriented design properties, (iii) control-flow complexity, (iv) sensitive operation usage, (v) Android Intent patterns, and (vi) exception handling. In addition to metrics, the dataset includes relevant metadata to facilitate large-scale analysis. AndroMetric enables research on Android malware detection, comparative analysis of benign and malicious codebases, and security-aware software quality assessment. The dataset is intended for researchers and practitioners in mobile security, software engineering, and applied machine learning. If you use this dataset, please cite the associated paper describing the AndroMetric dataset. @inproceedings{AndroMetric, author={Sebastian Siedler and Karim Elish}, booktitle={23rd IEEE/ACM International Conference on Mining Software Repositories (MSR)}, title={AndroMetric: Bridging Multi-Dimensional Software Metrics and Mobile Application Security}, year={2026}} App and Malware Access Information: The apps and malware samples referenced in this dataset are real-world Android applications. To adhere to research ethics guidelines and comply with data sharing and redistribution policies, we do not directly distribute APK files. Instead, researchers are required to obtain the applications through the official AndroZoo repository and/or Drebin. AndroZoo: https://androzoo.uni.lu/ Drebin: https://drebin.mlsec.org/ Usage Instructions: Request access to the repository by following the instructions provided on the official website. Use the cryptographic hash values (e.g., SHA-256) included in our dataset to retrieve the corresponding application samples from the repository.AndroMetric是一款大规模数据集，旨在支撑软件度量与安卓应用安全交叉领域的实证研究。本数据集包含39243个安卓应用，其中恶意软件样本29208个、良性应用10035个，均采集自公开仓库，并通过静态分析流水线完成分析。 针对每个安卓应用包（APK），我们将提取31项静态代码度量指标，以捕获其多维度的结构与行为特征。这些指标涵盖六大类别：(i) 规模与维度结构，(ii) 面向对象设计属性，(iii) 控制流复杂度，(iv) 敏感操作使用情况，(v) 安卓意图（Android Intent）模式，(vi) 异常处理。除度量指标外，本数据集还包含相关元数据，以助力大规模分析工作。 AndroMetric可支撑安卓恶意软件检测、良性与恶意代码库的对比分析，以及兼顾安全性的软件质量评估。本数据集面向移动安全、软件工程与应用机器学习领域的研究人员与从业者。 若使用本数据集，请引用描述AndroMetric数据集的相关学术论文。 @inproceedings{AndroMetric, author={Sebastian Siedler 与 Karim Elish}, booktitle={第23届IEEE/ACM软件仓库挖掘国际会议（MSR）}, title={AndroMetric：搭建多维度软件度量与移动应用安全之间的桥梁}, year={2026} } 应用与恶意软件获取说明： 本数据集涉及的应用与恶意软件样本均为真实安卓应用。为遵循研究伦理准则，同时符合数据共享与再分发政策，我们不直接分发APK文件。研究人员需通过官方的AndroZoo仓库及/或Drebin获取应用样本。 AndroZoo: https://androzoo.uni.lu/ Drebin: https://drebin.mlsec.org/ 使用说明： 1. 请通过官方网站提供的流程申请仓库访问权限。 2. 使用本数据集包含的加密哈希值（如SHA-256）从对应仓库中检索所需的应用样本。