|
|
--- |
|
|
license: mit |
|
|
title: Saarthi |
|
|
sdk: gradio |
|
|
emoji: ⚡ |
|
|
colorFrom: indigo |
|
|
colorTo: gray |
|
|
app_file: app.py |
|
|
pinned: false |
|
|
short_description: AI bot for Legal Help & ID Guidance – Anytime, Anywhere |
|
|
sdk_version: 5.34.2 |
|
|
python_version: 3.11 |
|
|
--- |
|
|
|
|
|
|
|
|
# 🧠 Saarthi – Connecting Every Citizen to Justice and Identity Rights with AI |
|
|
|
|
|
Saarthi is an open-source AI-powered web assistant designed to simplify **legal awareness** and **identity document assistance** for the common man. Built using Gradio, Sentence Transformers, and local knowledge bases, Saarthi bridges the gap between complex legal processes and everyday citizens. |
|
|
|
|
|
> ⚖️ Whether you're confused about your rights or unsure where to begin your ID application, Saarthi is here to guide you — anytime, anywhere. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🧭 Why Saarthi Matters |
|
|
|
|
|
In India, many citizens don’t know where to begin when it comes to legal rights or ID corrections. Saarthi makes complex legal and bureaucratic processes understandable and accessible — especially for underrepresented communities. |
|
|
|
|
|
It’s not just an AI project. It’s a public good. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🔍 Features |
|
|
|
|
|
### 🟠 NyaySetu – Legal Q&A Chatbot |
|
|
> _"Accurate, easy-to-understand legal help — anytime, anywhere."_ |
|
|
|
|
|
- Ask natural language questions about Indian laws (e.g., _"What is Section 498A?"_) |
|
|
- Powered by **semantic retrieval** using MiniLM embeddings |
|
|
- Trained on curated legal Q&A datasets |
|
|
- Instant responses from offline-embedded knowledge (no API needed) |
|
|
|
|
|
### 🪪 PehchaanSetu – Identity Document Guide |
|
|
> _"Empower yourself. Know the process. Own your identity."_ |
|
|
|
|
|
- Get step-by-step guidance for applying, correcting, or reprinting: |
|
|
- Aadhaar |
|
|
- PAN Card |
|
|
- Voter ID |
|
|
- Caste Certificate |
|
|
- Driving License |
|
|
- Passport |
|
|
- Ration Card |
|
|
- All guidance is loaded dynamically from local folders |
|
|
- Dropdown-based interface with **zero typing needed** |
|
|
|
|
|
--- |
|
|
|
|
|
## 🧠 How it Works |
|
|
|
|
|
### 🔹 Legal Q&A Flow |
|
|
1. Load precomputed MiniLM embeddings from `qa_embeddings.pkl` |
|
|
2. Compute embedding for user query |
|
|
3. Use cosine similarity to find the best match |
|
|
4. Return the legally accurate, beginner-friendly response |
|
|
|
|
|
### 🔹 ID Assistance Flow |
|
|
1. User selects ID Type and Purpose |
|
|
2. `id_assist_interface.py` dynamically constructs path |
|
|
3. Loads `.txt` or `.md` from corresponding folder |
|
|
4. Renders guidance with basic formatting |
|
|
|
|
|
--- |
|
|
|
|
|
## 🚀 Try it Out (Hugging Face Space) |
|
|
|
|
|
> 👉 [Live Demo on Hugging Face Spaces](https://huggingface.co/spaces/raviix46/Saarthi) |
|
|
|
|
|
--- |
|
|
|
|
|
## ⚙️ Installation (For Local Development) |
|
|
|
|
|
```bash |
|
|
git clone https://github.com/raviix46/Saarthi.git |
|
|
cd Saarthi |
|
|
pip install -r requirements.txt |
|
|
python app.py |
