WebGuard

# WebGuard Annotation Dataset

WebGuard Dataset

This dataset contains web safety annotations for browser interactions. Each entry represents
an annotated action on a website with a risk level.

Fields:
- url: The URL where the action was performed
- description: Description of the action (may be null)
- tagHead: HTML tag type of the target element
- Screenshot: Google Drive link to screenshot view
- Annotation: Review classification (SAFE/UNSAFE/LOW/HIGH)
- website: Website name/category

## Dataset Summary

This dataset contains 5,999 web safety annotations for browser interactions.

## Data Fields

- `url`: The URL where the action was performed
- `description`: Description of the action (may be null)
- `tagHead`: HTML tag type of the target element
- `Screenshot`: Google Drive link to screenshot view
- `Annotation`: Review classification (SAFE/UNSAFE/LOW/HIGH)
- `website`: Website name/category

## Usage

```python
from datasets import load_dataset

# Load the dataset
dataset = load_dataset("osunlp/WebGuard")

# Access the data
for example in dataset["train"]:
print(f"URL: {example['url']}")
print(f"Description: {example['description']}")
print(f"Tag: {example['tagHead']}")
print(f"Screenshot: {example['Screenshot']}")
print(f"Annotation: {example['Annotation']}")
print(f"Website: {example['website']}")
print("---")
```

## Citation

```bibtex
@article{zheng2025webguard,
title={WebGuard: Building a Generalizable Guardrail for Web Agents},
author={Zheng, Boyuan and Liao, Zeyi and Salisbury, Scott and Liu, Zeyuan and Lin, Michael and Zheng, Qinyuan and Wang, Zifan and Deng, Xiang and Song, Dawn and Sun, Huan and others},
journal={arXiv preprint arXiv:2507.14293},
year={2025}
}

@inproceedings{zheng-etal-2024-webolympus,
title = "{W}eb{O}lympus: An Open Platform for Web Agents on Live Websites",
author = "Zheng, Boyuan and Gou, Boyu and Salisbury, Scott and Du, Zheng and Sun, Huan and Su, Yu",
editor = "Hernandez Farias, Delia Irazu and Hope, Tom and Li, Manling",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.emnlp-demo.20",
pages = "187--197",
}
```

## License

Creative Commons Attribution-NonCommercial 4.0 International

# WebGuard标注数据集
WebGuard数据集

本数据集涵盖针对浏览器交互的网页安全标注内容，每条数据对应网站上一项附带风险等级的已标注操作。

### 字段说明
- url：执行该操作的网页URL
- description：操作描述（可为空值）
- tagHead：目标元素的HTML标签类型
- Screenshot：指向截图的谷歌云端硬盘（Google Drive）链接
- Annotation：审核分类（SAFE/UNSAFE/LOW/HIGH）
- website：网站名称/类别

## 数据集概览
本数据集共包含5999条浏览器交互的网页安全标注数据。

## 数据字段
- `url`：执行该操作的网页URL
- `description`：操作描述（可为空值）
- `tagHead`：目标元素的HTML标签类型
- `Screenshot`：指向截图的谷歌云端硬盘（Google Drive）链接
- `Annotation`：审核分类（SAFE/UNSAFE/LOW/HIGH）
- `website`：网站名称/类别

## 使用方法
python
from datasets import load_dataset

# 加载数据集
dataset = load_dataset("osunlp/WebGuard")

# 访问数据
for example in dataset["train"]:
print(f"URL: {example['url']}")
print(f"Description: {example['description']}")
print(f"Tag: {example['tagHead']}")
print(f"Screenshot: {example['Screenshot']}")
print(f"Annotation: {example['Annotation']}")
print(f"Website: {example['website']}")
print("---")

## 引用
bibtex
@article{zheng2025webguard,
title={WebGuard: Building a Generalizable Guardrail for Web Agents},
author={Zheng, Boyuan and Liao, Zeyi and Salisbury, Scott and Liu, Zeyuan and Lin, Michael and Zheng, Qinyuan and Wang, Zifan and Deng, Xiang and Song, Dawn and Sun, Huan and others},
journal={arXiv preprint arXiv:2507.14293},
year={2025}
}

@inproceedings{zheng-etal-2024-webolympus,
title = "{W}eb{O}lympus: An Open Platform for Web Agents on Live Websites",
author = "Zheng, Boyuan and Gou, Boyu and Salisbury, Scott and Du, Zheng and Sun, Huan and Su, Yu",
editor = "Hernandez Farias, Delia Irazu and Hope, Tom and Li, Manling",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.emnlp-demo.20",
pages = "187--197",
}

## 许可协议
知识共享署名-非商业性使用4.0国际许可协议（Creative Commons Attribution-NonCommercial 4.0 International）