ahmedBargady/MIAF_DomainDetection_Infrastructure_Datasets

MIAF: Domain Detection Infrastructure Datasets

This dataset collection serves as the standardized evaluation benchmark for MIAF. It consists of nine classification datasets derived from four major public malicious-domain benchmarks.

Overview

We evaluate MIAF across nine classification datasets derived from four public malicious-domain benchmarks:

• DomainRadar (Hranicky et al., 2025)
• ThreatFox (Abuse.ch, 2025)
• DeepURLBench (Schvartzman et al., 2024)
• MACFormer (2025)

These datasets cover malware detection, phishing detection, and long-tailed malware-family attribution. The benchmarks span diverse characteristics in terms of scale, class balance, label granularity, and metadata availability, enabling an evaluation of the efficacy of infrastructure metadata under varied operating conditions.

Data Splits and Reorganization

To establish a consistent evaluation baseline, we have reorganized each benchmark into task-specific subsets and partitioned them into disjoint training, development, and testing sets using a 60/20/20 ratio. This stratification was applied to every individual class within the datasets to ensure a fair and consistent baseline across all evaluated methods.

All datasets that lacked native infrastructure metadata (ThreatFox, DeepURLBench, and MACFormer) have been enriched using the uniform EDC pipeline.

Dataset Characteristics

Dataset	Task	Samples	Classes	Mal. % (Train/Val/Test)	Metadata Source
DomainRadar	Malware (binary)	341,935	2 (legit / malicious)	19.1%	Native
	Phishing (binary)	441,207	2 (legit / malicious)	37.3%	Native
	Malware family (MC)	311,906	61 (legit + 60 families)	11.3%	Native
ThreatFox	Malware (binary)	176,065	2 (legit / malicious)	43.2%	EDC-enriched
	Malware family (MC)	176,065	65 (legit + 64 families)	43.2%	EDC-enriched
DeepURLBench	Malware (binary)	375,917	2 (benign / mal)	24.1%	EDC-enriched
	Phishing (binary)	362,347	2 (benign / phishing)	21.3%	EDC-enriched
MACFormer	Malware (binary)	350,532	2 (legit / malicious)	32.9%	EDC-enriched
	Phishing (binary)	266,500	2 (legit / malicious)	11.8%	EDC-enriched

Structure

Inside each dataset folder, you will find:

• train.csv: Training split (60%)
• dev.csv: Development/Validation split (20%)
• test.csv: Testing split (20%)
• *_ip.csv: Corresponding infrastructure metadata (IP records) for each split.

Reproducibility

To facilitate reproducibility, we release this consolidated corpus of standardized metadata as a public resource.

Link: https://huggingface.co/datasets/ahmedBargady/MIAF_DomainDetection_Infrastructure_Datasets