zongqing0068/EvolvingWorld-Books-Gutenberg

Dataset Card for EvolvingWorld-Books-Gutenberg

Dataset Summary

This dataset provides the 57 source book texts used for EvolvingWorld data construction. The books were selected from Goodreads' Best Books Ever list and obtained from Project Gutenberg.

Dataset Structure

Each row is a JSON object with the following fields:

• —title: the title of the book
• —author: the author of the book
• —content: the full text content used for data construction

Example:

{
  "title": "Pride and Prejudice",
  "author": "Jane Austen",
  "content": "..."
}

Source and Selection

The selected books are chronologically narrated literary works. This supports EvolvingWorld's scene-by-scene state-tracking pipeline, where later scenes can serve as look-ahead references for grounding character and world-state changes.

Intended Use

This dataset is intended for academic research and reproducibility, especially for:

• —literary world simulation
• —role-playing agents
• —character and world-state modeling
• —data construction for long-horizon narrative simulation

License and Copyright Notice

The original works are believed to be public domain in the United States and were obtained from Project Gutenberg. Copyright status may vary by jurisdiction. Users outside the United States should verify the copyright status of each work under their local laws before redistributing or using the texts.

This dataset is provided for non-commercial academic research and reproducibility. Users are responsible for ensuring that their use complies with applicable laws and the terms of the original source providers.

Citation

If you use this dataset, please cite the paper: EvolvingWorld: An Open-Schema Framework for Co-Evolving Role-Play Agents and World Model in Interactive Literary World.

Languages

The dataset is primarily in English (en).