Synthetic Legal Contract Clauses With Risk Annotations

Mr-Dintov's Legal Contract Clause Risk Corpus is a synthetic dataset engineered for clause-level contract risk classification and legal NLP research. It contains pairs of contract clauses representing dangerous and market-standard versions, built around US Delaware, UK English Law, and ICC International jurisdictions. The dataset was last updated on April 6, 2026.

Use Cases

• Fine-tune language models to classify clause risk levels using the annotated 'dangerous' and 'balanced' versions.
• Train models for phrase-level annotation tasks within contract clauses.
• Benchmark models on multi-jurisdictional legal text covering US, UK, and ICC law.

Strengths

• Dataset is engineered for specific legal NLP tasks like fine-tuning and classification.
• Provides structured comparisons between dangerous and market-standard clause versions.
• Covers multiple legal jurisdictions: US Delaware, UK English Law, and ICC International.

Limitations

• Dataset is synthetic, which may not fully represent the complexity and noise of real-world legal contracts.
• Specific volume of entries, row count, and annotation depth are unknown.
• Lacks metadata on license, organization, and original source methodology.

Provenance

Source

huggingface, author Mr-Dintov

Collection Method

Synthetically engineered dataset.

Time Range

null

Freshness

Last updated on April 6, 2026.

Geography

US Delaware, UK English Law, ICC International jurisdictions.