Filimon-H/ethiopian-tax-law-assistant
Domaine:
natural language processingType de record:
software
Créateur:
Fil
Hôte:
AI-powered legal assistant that answers questions about Ethiopian tax and commercial laws using a Retrieval-Augmented Generation (RAG) pipeline. Built with FastAPI, LangChain, and official legal documents including proclamations, regulations, and textbooks
# 📟 Ethiopian Tax Law Assistant
A RAG-based legal chatbot that helps users query the Ethiopian tax code—including VAT, Turnover Tax, and Income Tax—by retrieving relevant chunks from legal PDFs and generating accurate, cited answers using Groq’s LLaMA3-8B model. Built using LangChain, FastAPI, Streamlit, and ChromaDB.
---
## 📌 Project Overview
**Goal:** To make Ethiopian tax regulations more accessible and understandable to professionals, citizens, and businesses by providing an AI-powered legal assistant.
**Problem:** Navigating Ethiopia's complex tax codes is time-consuming and requires legal expertise.
**Solution:** A Retrieval-Augmented Generation (RAG) chatbot that accepts natural language tax-related questions and responds with answers grounded in official legal documents.
---
## 🧠 Architecture & Tools
| Layer | Technology | Purpose |
| ---------------- | ----------------------- | --------------------------- |
| LLM | Groq (LLaMA3-8B) | Answer generation |
| Chunking | LangChain | Semantic text splitting |
| Embedding | Sentence Transformers | Dense vector representation |
| Vector Store | ChromaDB | Document similarity search |
| Backend API | FastAPI | Handles `/ask` endpoint |
| Frontend UI | Streamlit | Simple chat interface |
| Containerization | Docker & Docker Compose | Reproducible deployment |
---
## 📂 Project Structure
```bash
.
├── docker-compose.yml
├── .env.example
├── requirements.txt
├── src/
│ ├── api/main.py # FastAPI app
│ ├── config/load_config.py # Loads .env & API keys
│ ├── ingestion/document_loader.py # PDF loader & chunker
│ ├── embedding/embed_store.py # Embedding and vector store
│ └── pipeline/retrieval_qa.py # RAG logic
├── streamlit_app/app.py # Streamlit UI
```
---
## 🥪 How to Run Locally
## …