Chinese Legal Domain Corpus for Industry Model Training

IndustryCorpus Law is a Chinese-language text corpus focused on the legal domain, constructed by BAAI to address data quality issues in industry model training. The dataset was processed using 22 specialized data operators and was last updated in July 2024. Specific volume and size details are not provided in the available metadata.

Use Cases

• Fine-tune language models for legal document classification using domain-specific text features.
• Train models for legal text summarization on corpus paragraphs and sections.
• Improve named entity recognition for legal entities and citations within the corpus text.
• Benchmark model performance on domain-specific tasks like legal query answering.
• Augment pretraining data for models targeting Chinese legal advisory applications.

Strengths

• Constructed with 22 specialized industry data processing operators.
• Explicitly targets high-quality, domain-expert data for legal applications.
• Dataset page was updated in July 2024, indicating recent maintenance.

Limitations

• Data volume and sample size are unspecified, preventing assessment of scale.
• Potential geographic and linguistic bias as the corpus is focused on Chinese law.
• Lack of provided column names or sample data hinders immediate structural understanding.

Provenance

Source

BAAI (Beijing Academy of Artificial Intelligence).

Collection Method

Constructed and processed using 22 industry data operators.

Time Range

null

Freshness

Last updated 2024-07, indicating recent activity.

Geography

Presumably China, based on domain focus and author.