pdf pdf |
|---|
Dataset For Indian Legal Knowledge Base
A curated, bilingual (English + Hindi) legal corpus for AI-powered Indian contract analysis — built for RAG pipelines, LLM fine-tuning, and legal NLP research.
About This Dataset
This dataset is the knowledge backbone of LegalEagle — an AI-powered contract review platform for Indian startups and freelancers. It contains Indian statutes, contract templates, landmark case references, and clause examples, curated specifically for retrieval-augmented generation (RAG) in the Indian legal domain.
All government statutes included are in the public domain (Government of India publications).
Dataset Structure
dataset/
├── acts/ # Indian Statutes (English + Hindi)
│ ├── indian_contract_act_1872_eng.pdf
│ ├── indian_contract_act_1872_hindi.pdf
│ ├── arbitration_conciliation_act_1996.pdf
│ ├── arbitration_conciliation_act_1996_hindi.pdf
│ ├── specific_relief_act_1963_eng.pdf
│ ├── specific_relief_act_1963_hindi.pdf
│ ├── sale_of_goods_act_1930_eng.pdf
│ ├── sale_of_goods_act_1930_hindi.pdf
│ ├── indian_stamp_act_1899_eng.pdf
│ ├── indian_stamp_act_1899_hindi.pdf
│ └── information_technology_act_2000.pdf
│
├── templates/ # Real Contract Templates
│ ├── nda_template.pdf
│ ├── employment_agreement.pdf
│ ├── vendor_agreement.pdf
│ └── sample_employment_agreement_template.pdf
│
├── cases/ # Landmark Case References
│ ├── niranjan_shankar_golikari_case.pdf
│ ├── wrongful_termination_employment.pdf
│ ├── wrongful_termination_employment_case_2.pdf
│ ├── arbitration_cases.pdf
│ └── indemnity_contract_section_124.pdf
│
├── clauses/ # Clause-Level References
│ ├── clause_type_reference.pdf
│ ├── risky_patterns_clause.pdf
│ ├── non_compete_clause.pdf
│ ├── non_compete_clause_2.pdf
│ └── arbitration_clause.pdf
│
└── knowledge_base/ # Structured Text Files
└── clause_examples.txt # Good vs Bad clause examples
Contents Summary
Indian Statutes Covered
| Act | Year | Language |
|---|---|---|
| Indian Contract Act | 1872 | English + Hindi |
| Arbitration and Conciliation Act | 1996 | English + Hindi |
| Specific Relief Act | 1963 | English + Hindi |
| Sale of Goods Act | 1930 | English + Hindi |
| Indian Stamp Act | 1899 | English + Hindi |
| Information Technology Act | 2000 | English |
Contract Templates Covered
| Template | Type |
|---|---|
| Non-Disclosure Agreement (NDA) | Bilateral, India-specific |
| Employment Agreement | With e-stamp, Indian jurisdiction |
| Vendor Agreement | B2B, Indian startups |
| Sample Employment Template | General purpose |
Clause Types Covered
- Non-Compete Clauses
- Termination Clauses
- Indemnity Clauses (Section 124 ICA)
- Confidentiality / NDA Clauses
- IP Assignment Clauses
- Payment / Invoice Clauses
- Dispute Resolution / Arbitration Clauses
- Force Majeure Clauses
Intended Use
Primary Use Case
RAG (Retrieval-Augmented Generation) pipelines for Indian contract review:
from huggingface_hub import hf_hub_download
from sentence_transformers import SentenceTransformer
import fitz
# Download a file
path = hf_hub_download(
repo_id="d-riti/Dataset-For-Indian-Legal-Knowledge-Base",
filename="acts/indian_contract_act_1872_eng.pdf",
repo_type="dataset"
)
# Extract text
doc = fitz.open(path)
text = "\n".join(page.get_text() for page in doc)
Suitable For
- RAG pipelines for legal Q&A
- LLM fine-tuning on Indian legal text
- Legal clause classification
- Contract risk analysis
- Legal NLP research
- Indian law chatbots
Not Suitable For
- Actual legal advice (not a substitute for a lawyer)
- Jurisdictions outside India
- ❌Real-time case law (static dataset)
Recommended Extraction Pipeline
import fitz # PyMuPDF — for text PDFs
import pdfplumber # for tables in contracts
def extract_legal_pdf(pdf_path: str) -> str:
full_text = ""
# Extract text (handles English + Hindi)
doc = fitz.open(pdf_path)
for page in doc:
page_text = page.get_text()
if len(page_text.strip()) > 50: # skip image/stamp pages
full_text += page_text + "\n"
# Extract tables (for agreement templates)
with pdfplumber.open(pdf_path) as pdf:
for page in pdf.pages:
for table in page.extract_tables():
for row in table:
row_text = " | ".join(
str(cell) if cell else "" for cell in row
)
if row_text.strip():
full_text += row_text + "\n"
return full_text.strip()
Languages
| Language | Content | Script |
|---|---|---|
| English | Statutes, templates, cases, clauses | Latin |
| Hindi | Statutes (parallel corpus) | Devanagari |
All Hindi PDFs contain selectable text (not scanned) — no OCR required.
Legal Disclaimer
- All Indian government statutes are sourced from official Government of India publications and are in the public domain.
- Contract templates are included for educational and research purposes only.
- This dataset does not constitute legal advice.
- Users are responsible for verifying the accuracy and currency of all legal content.
Citation
If you use this dataset in your research or project, please cite:
@dataset{legaleagle_indian_legal_kb_2026,
author = {Riti Dube},
title = {Dataset For Indian Legal Knowledge Base},
year = {2026},
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/d-riti/Dataset-For-Indian-Legal-Knowledge-Base},
note = {Bilingual Indian legal corpus for RAG and legal NLP}
}
Related Project
This dataset powers LegalEagle — an AI contract review platform for Indian startups.
- Stack: FastAPI + React + LangChain + RAG + Groq (Llama 3)
- Features: Clause-by-clause risk analysis, Indian law citations, PDF report export
- GitHub: github.com/Ritidube/legaleagle
Author
Riti Dube
- GitHub: @Ritidube
- LinkedIn: linkedin.com/in/riti-dube
Built with ❤️ for the Indian legal-tech ecosystem
- Downloads last month
- 198