Legal Pretrain: Vietnamese Legal Text Corpus for Language Models

Vietnamese legal documents extracted and preprocessed from official sources like laws, codes, and government regulations. The dataset is curated by VLSP2025-LegalSML and was last updated on July 3, 2025. All documents are written in Vietnamese and reflect the real-world language of Vietnam's legal system.

Use Cases

• Fine-tune legal-domain language models based on official Vietnamese legal texts.
• Train named entity recognition models for legal concepts based on government regulations.
• Develop text classification systems for legal document types based on the corpus of laws and codes.
• Benchmark machine translation models for specialized legal terminology between Vietnamese and other languages.
• Analyze linguistic patterns and structure in formal Vietnamese legal writing.

Strengths

• Documents are sourced from official, publicly available legal texts, suggesting authenticity.
• Texts are preprocessed, which likely indicates a degree of cleaning or standardization.
• The dataset is specifically curated for training legal-domain language models, indicating a focused purpose.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Publicly available sources such as laws, codes, and government regulations from Vietnam.

Collection Method

Extracted and preprocessed.

Freshness

Last updated 2025-07-03 06:24:03; freshness should be verified.

Geography

Vietnam