akhiyusuf/legal-ai-nigeria
Domain:
digital infrastructureRecord type:
software
Creator:
akh
Host:
# Legal QA System: Advanced RAG for {{COUNTRY}}
A production-ready Legal Question Answering system utilizing Advanced Retrieval-Augmented Generation (RAG).
## Features
- **Structure-Aware Chunking**: Preserves titles, chapters, and sections.
- **Knowledge Graph**: Entity extraction (spaCy) and NetworkX graph for multi-hop reasoning.
- **Hybrid Retrieval**: Combines Vector Search (Chroma) and Graph Traversal.
- **Reranking**: Uses Cross-Encoders for precision.
- **Source Proof**: Screenshot generation for PDF sources using `pdf2image`.
- **Auto-Discovery**: Automatic discovery of official legal portals for {{COUNTRY}}.
## Prerequisites
- Python 3.10+
- Poppler (system dependency for `pdf2image`)
- Linux: `sudo apt-get install poppler-utils`
- Mac: `brew install poppler`
## Installation
1. Clone the repository and navigate to the directory.
2. Install dependencies:
```bash
pip install -r requirements.txt
python -m spacy download en_core_web_sm
```
3. Configure your environment in `.env`:
```env
GROQ_API_KEY=your_key
COUNTRY={{COUNTRY}}
VECTOR_DB={{VECTOR_DB}}
```
## Getting Started
1. **Initial Ingestion**:
Run the ingestion script to discover and index laws for {{COUNTRY}}.
```bash
# Trigger via API
curl -X POST localhost
```
2. **Run the Application**:
```bash
python -m backend.main
```
Open `localhost` in your browser.
## Architecture
1. **Discovery**: `SourceDiscovery` identifies official .gov domains.
2. **Indexing**: Chunks are stored in ChromaDB; Entities are stored in a NetworkX graph.
3. **Retrieval**:
- Semantic search finds similar text.
- Graph traversal finds related provisions (multi-hop).
- Results are fused and reranked by a MiniLM Cross-Encoder.
4. **Generation**: Groq (Mixtral) generates answers grounded in the retrieved context with citations.
## Disclaimer
This tool is for informational purposes only. It is NOT a substitute for professional legal advice.