PhreshPhish
收藏资源简介:
PhreshPhish是一个大规模、高质量的钓鱼网站数据集,由OpenText研究机构收集,旨在解决现有钓鱼网站数据集质量不高、数据量不足的问题。该数据集通过自动化和人工标注的方式清理数据,确保其质量,并通过Hugging Face平台提供访问。数据集的创建过程涉及网页的收集、清洗和评估,旨在为钓鱼检测模型提供训练和评估。该数据集适用于钓鱼检测领域,旨在解决网络钓鱼攻击带来的经济和声誉损害问题。
PhreshPhish is a large-scale, high-quality phishing website dataset collected by OpenText Research, which aims to address the problems of poor quality and insufficient data volume in existing phishing website datasets. This dataset cleans and validates data through both automated and manual annotation methods to ensure its quality, and provides access via the Hugging Face platform. The dataset creation process involves web collection, cleaning and evaluation, and is intended to provide training and evaluation resources for phishing detection models. Tailored for the phishing detection domain, this dataset is designed to mitigate the economic and reputational damages caused by phishing attacks.
PhreshPhish 数据集概述
基本信息
- 许可证: Creative Commons Attribution 4.0 International (CC-BY-4.0)
- 规模: 100K < n < 1M
- 任务类别: 文本分类
- 数据集名称: PhreshPhish
- 版本: v1.0.0 (初始发布于2025-05-14)
数据集描述
PhreshPhish 是一个用于钓鱼网页检测的大规模、真实世界数据集和基准测试,包含钓鱼和良性HTML-URL对。
数据划分
- 训练集: 371,941 个样本
- 良性: 253,189
- 钓鱼: 118,752
- 测试集: 36,787 个样本
- 良性: 30,048
- 钓鱼: 6,739
基准测试
- 基准数量: 404 个
- 基准率范围: [5e-4, 1e-3, 5e-3, 1e-2, 5e-2]
使用方式
python from datasets import load_dataset
train = load_dataset(phreshphish/phreshphish, split=train) test = load_dataset(phreshphish/phreshphish, split=test)
许可与使用条款
- 许可: 仅限用于反钓鱼研究
- 许可链接: https://creativecommons.org/licenses/by/4.0/
引用
- 论文标题: PhreshPhish: A Real-World, High-Quality, Large-Scale Phishing Website Dataset and Benchmark
- 论文链接: https://arxiv.org/abs/2507.10854
- BibTex引用: bibtex @article{dalton2025phreshphish, title = {PhreshPhish: A Real-World, High-Quality, Large-Scale Phishing Website Dataset and Benchmark}, author = {Thomas Dalton and Hemanth Gowda and Girish Rao and Sachin Pargi and Alireza Hadj Khodabakhshi and Joseph Rombs and Stephan Jou and Manish Marwah}, year = 2025, journal = {arXiv preprint}, url = {https://arxiv.org/abs/2507.10854}, eprint = {2507.10854} }




