JuDGE Chinese Legal Judgment Benchmark Dataset

*This dataset supports online use.Click here to jump.

JuDGE (Judgment Document Generation for Chinese Law System) is a benchmark dataset for legal document generation designed for the Chinese legal system. This dataset formalizes judgment document generation as a conditional text generation problem. Given a case fact description (Fact), the goal is to generate a structurally coherent and legal judgment document (Judgment Document).

The dataset aims to improve the performance of the legal document generation model through high-quality annotated data, especially its capabilities in legal reasoning and document writing. It is suitable for various application scenarios such as legal intelligent systems, automatic generation of legal documents, and legal question-and-answer systems.

Data Structure

• Case ID:  Unique identifier for the case.
• Fact:  A summary of the key facts of the case, limited to 1,000 Chinese characters.
• Full Document:  Full judgment document.
• Reasoning:  A detailed description of the legal reasoning process.
• Judgment:  Final judgment and penalty (within 1000 to 3000 Chinese characters).
• Sentence:  Length of sentence.
• Fine:  Amount of fine.
• Crime Type:  The type of crime involved.
• Law Articles:  An index to the legal provisions cited in the judgment.