Bigbarry/WebWorldData_copy

WebWorldData 🌐

         

Overview

WebWorldData is a large-scale dataset of 1.06M web interaction trajectories collected from the open web, designed for training browser world models. It is the training data behind the WebWorld model series.

Each trajectory consists of sequences of (state, action, next_state) transitions, where states are represented as A11y Trees extracted from real websites using Playwright.

Dataset Statistics

Data Sources

The dataset is collected through a scalable hierarchical pipeline:

Data Format

Each sample is a multi-turn conversation in JSONL format:

{
  "messages": [
    {
      "role": "system",
      "content": "You are a web world model. I will provide you with an initial page state and a sequence of actions. For each action, predict the resulting page state.\nStrictly maintain the original format. Output only the full page state without explanations, code, or truncation."
    },
    {
      "role": "user",
      "content": "Initial Page State:\nRootWebArea 'Example Site'\n\t[1] banner ...\n\nFirst Action: 'click([32])'\n\nNext Page State:"
    },
    {
      "role": "assistant",
      "content": "RootWebArea 'Example Site - News'\n\t[1] banner ...\n\t[50] main ..."
    },
    {
      "role": "user",
      "content": "Continue the trajectory. Given the previous state, predict the next page state after this action.\n\nAction: 'fill([19], \"weather today\")'\n\nNext Page State:"
    },
    {
      "role": "assistant",
      "content": "RootWebArea 'Example Site - News'\n\t[1] banner ...\n\t[19] textbox ..., value='weather today' ..."
    }
  ]
}

Domain Distribution

The dataset covers diverse web domains:

Action Space

Trajectories use a unified action space as Python-style function calls:

Action distribution: Element interactions (83.4%), Browser & navigation (11.9%), Meta & control (2.4%), Keyboard (1.2%), Coordinate & mouse (1.1%).

Filtering & Safety

The dataset undergoes rigorous dual-stage filtering:

1. 1.Rule-based filtering: Website reachability checks, banned keyword filtering (pornography, gambling, violence), trajectory pruning for no-op transitions
2. 2.LLM-based URL filtering: Each URL scored across accessibility, content suitability, interactivity, and engineering quality
3. 3.Trajectory-level filtering: Max 30K tokens, max 30 turns, keyword safety checks

All data is collected from publicly accessible webpages in compliance with robots.txt protocols.

Usage

from datasets import load_dataset

dataset = load_dataset("Qwen/WebWorldData")

Intended Use

• —Training browser world models for web simulation
• —Generating synthetic trajectories for web agent fine-tuning
• —Research on world modeling, environment simulation, and agent learning

Limitations

• —Data is collected from publicly accessible webpages; residual PII may exist despite filtering
• —Web content is inherently non-deterministic (ads, A/B tests, dynamic widgets) — some trajectories may not be perfectly reproducible
• —Domain distribution reflects the composition of FineWeb and CCI 3.0 pre-training corpora

Associated Models

Citation

@misc{xiao2026webworldlargescaleworldmodel,
      title={WebWorld: A Large-Scale World Model for Web Agent Training}, 
      author={Zikai Xiao and Jianhong Tu and Chuhang Zou and Yuxin Zuo and Zhi Li and Peng Wang and Bowen Yu and Fei Huang and Junyang Lin and Zuozhu Liu},
      year={2026},
      eprint={2602.14721},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2602.14721}, 
}

License

This dataset is released under the Apache 2.0 License.