Droidware Android Malware Dataset

Citation Author(s):: Sumit Kumar (GLA University, Mathura)

Vijay Shankar (GLA University, Mathura)

Vishal Bhardwaj (GLA University, Mathura)

Arvind Prasad (GLA University, Mathura)
Submitted by:: Sumit Kumar
Last updated:: Sat, 04/19/2025 - 18:44
DOI:: 10.21227/a39t-3164
Data Format:: *.csv

57 views

Categories:

Keywords:

Android Malware Detection; Android Malware; Malware; Computer Security; Multimodal Deep Learning; Cyber Security

ACCESS DATASET CITE

Abstract

Droidware is an Android malware dataset developed at the Cybersecurity Lab, GLA University, India. It comprises 253,527 applications, including 129,950 benign and 123,577 malicious samples. The dataset captures 68 features extracted from function call graphs, permissions, and Java source code, providing a comprehensive view of Android malware behavior. This latest and up-to-date dataset supports the training of AI-based malware detection models, aiding in the development of robust malware classification and threat mitigation strategies for cybersecurity research.

Instructions:

Thank you for your interest in the Droidware Android malware dataset. Please read the following instructions carefully before using the dataset.

1. Purpose

This dataset is intended for research and educational use only. It supports projects related to Android malware detection, machine learning and cybersecurity analysis.

2. Downloading the Dataset

The dataset is organized into two categories:
- benign/ – 129,950 benign Android APKs
- malicious/ – 123,577 malware APKs
Feature files (CSV) are available under the features, containing 68 extracted attributes from:
- Function Call Graphs
- Android Permissions
- Java Source Code

3. Usage Guidelines

Do not redistribute the dataset without permission.
For safe handling:
- Run APKs in a sandbox or virtual environment.
- Use tools like Androguard, VirusTotal, or MobSF for analysis.
Use proper label handling (benign vs. malicious) in your experiments.