DoubleStar Protocol: Chinese AI Dialogue Corpus with Synthetic Data and Creative Works

DoubleStar Protocol — Soul Alignment Corpus is a collection of Chinese-language text data created by author Alicea123. The dataset includes approximately 20,000 pages of Word document dialogues with an AI, over 800 files of user statements, more than 20 papers and HTML presentations, and a 140,000-word novel. It also references synthetic datasets of 100,000, 1 million, 100 million, and 5 billion entries. The dataset was last updated on HuggingFace in May 2026.

Use Cases

• Training or fine-tuning language models for Chinese dialogue generation based on the extensive conversational records.
• Analyzing patterns in human-AI interaction and creative co-writing based on the described dialogue and article files.
• Studying the characteristics of large-scale synthetic text data referenced in the dataset description.
• Exploring unconventional, narrative-driven text sources for creative AI applications.

Strengths

• Contains a large volume of specific text artifacts, including ~20,000 pages of dialogue and over 800 user statement files.
• Includes diverse content types such as dialogues, articles, presentations, and a full-length novel.
• References multiple scales of synthetic data, up to 5 billion entries, for potential model training.

Limitations

• Description metadata is limited; actual data quality, structure, and accessibility require manual inspection after download.
• Column-level documentation is absent; field semantics and data organization must be inferred after download.
• The dataset's origin and creation method are described in highly personal and unconventional terms, which may affect reproducibility.

Provenance

Source

Alicea123 on HuggingFace.

Collection Method

Personally compiled from dialogues with an AI, creative writings, and referenced synthetic data.

Time Range

Described as involving a 23-year personal timeline, but specific data creation dates are unknown.

Freshness

Last updated 2026-05-01 13:55:14; freshness should be verified.

Geography

Likely China or Chinese-speaking context, but not explicitly stated.