Asaffa/sheriallm

Kenya Law Corpus

Dataset Summary

A comprehensive corpus of Kenyan legislation and case law scraped from Kenya Law — the official platform of the National Council for Law Reporting.

This dataset powers Iroh, an AI-powered Kenyan legal research assistant.

Dataset Statistics

• Total chunks: 62,000+
• Legislation: 546 Acts (every Act in the Laws of Kenya)
• Case law: Thousands of judgments from Kenyan courts
• Raw size: ~72MB (chunks), ~526MB (with embeddings)
• Languages: English (primary), Swahili (some judgments)

Data Structure

Each record contains:

{
  "chunk_id": "https://new.kenyalaw.org/akn/ke/act/...##chunk0",
  "text": "Full text of the chunk",
  "title": "Employment Act",
  "url": "https://new.kenyalaw.org/akn/ke/act/...",
  "type": "legislation | case_law",
  "section": "chunk_0"
}

Source

All data sourced from Kenya Law under the AKN (Akoma Ntoso) URL structure. Kenya Law publishes legislation and judgments under open access for public use.

Intended Use

• Legal AI research and RAG pipelines
• African legal NLP benchmarking
• Fine-tuning language models on Kenyan legal text
• Legal information retrieval systems

How It Was Created

1. Scraped legislation and case law from new.kenyalaw.org
2. Parsed HTML to clean text (JSON)
3. Chunked into 800-word segments with 100-word overlap
4. Embedded using sentence-transformers/all-MiniLM-L6-v2
5. Stored in ChromaDB for retrieval

Related Resources

• Iroh (Live App): https://huggingface.co/spaces/Daudipdg/iroh
• ChromaDB (with embeddings): Daudipdg/iroh-chroma-db — 526MB, ready to use with ChromaDB
• Raw chunks (this dataset): Daudipdg/kenya-law-corpus — 72MB, text only
• Built by: David Maina — Makini AI

License

CC BY-NC 4.0 — Free for research and non-commercial use. For commercial licensing contact: david@makini.tech

Citation

@dataset{maina2025kenyalaw,
  author = {Maina, David},
  title = {Kenya Law Corpus},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/Daudipdg/kenya-law-corpus}
}