U.S. Civil Code Question Answering Dataset

USLawQA contains legal texts from the United States Civil Code, structured for Information Retrieval and Question Answering research. The dataset was created by ArchitRastogi using custom web scrapers on uscode.house.gov and was last updated in December 2024.

Use Cases

• Train models to answer legal questions using passages from the U.S. Civil Code.
• Benchmark information retrieval systems on legal text passages and associated queries.
• Fine-tune large language models for legal domain adaptation using question-context pairs.
• Evaluate model performance on extracting answers from structured legal code sections.

Strengths

• Dataset size is categorized as between 100K and 1M entries.
• Built directly from the authoritative source uscode.house.gov.
• Specifically structured for legal QA and IR tasks, which is a defined niche.

Limitations

• Exact row count, column names, and sample size are unknown.
• Limited to the U.S. Civil Code, which may not generalize to other legal jurisdictions or code types.
• Potential temporal staleness if the scraped code sections are not the most recent.

Provenance

Source

United States Civil Code from uscode.house.gov.

Collection Method

Created using custom web scrapers.

Freshness

Last updated on 2024-12-02.

Geography

United States.