QueryBot 🧠💬

QueryBot is a smart, memory-aware chatbot powered by LangChain, Google Gemini, and Django. It understands natural language queries, maintains context across conversations, generates and executes SQL queries, and can optionally perform image-based similarity searches using FAISS and ResNet50.

---

🚀 Features

• 💬 Natural language to SQL conversion via LLM
• 🧠 Entity Memory for multi-turn contextual conversations
• 🗃️ PostgreSQL database integration with live query results
• 📊 Structured JSON outputs with tabular display flags
• 🖼️ (Optional) Visual similarity search using FAISS + ResNet50
• 🔐 Environment-based API key and path configuration

---

📂 Project Structure

querybot/
├── chatapp/
│   ├── chat\_logic.py          # LLM + LangChain chat logic
│   ├── response\_tools.py      # SQL execution + formatting
│   ├── feature\_extractor.py   # TensorFlow ResNet50 embedding extractor (optional)
│   ├── find\_similars.py       # FAISS-based visual similarity search (optional)
│   ├── views.py               # Django API endpoints
│   └── urls.py                # Routes
├── querybot/                  # Django project settings
│   ├── settings.py
│   ├── urls.py
│   └── ...
├── static/
│   ├── assets/
│   │   └── main\_logo.png
│   ├── css/
│   │   └── chat.css
│   ├── js/
│   │   └── chat.js
│   └── json/
│       ├── data.json
│       └── output\_data.json
├── templates/
│   └── chat.html              # Chat frontend template
├── .env                      # Environment variables (API keys etc.)
├── db.sqlite3
├── manage.py
├── requirements.txt
└── README.md

---

📦 Requirements

django
langchain
langchain-google-genai
google-generativeai
numpy
tensorflow
pickle5
psycopg2-binary

⚠️ Optional for image similarity: faiss-cpu Install with: pip install faiss-cpu

---

🧠 Chatbot with Entity Memory

• Uses LangChain’s ConversationEntityMemory for persistent, context-aware conversations
• Generates SQL queries from natural language and executes on connected DB
• Returns JSON responses including the SQL query and display flags, e.g.:

{
  "reply": "Showing sales data for the last month.",
  "sql_query": "SELECT * FROM sales WHERE sale_date >= CURRENT_DATE - INTERVAL '30 days';",
  "is_query_generated": 1,
  "table_display": 1
}

---

🖼️ Optional: Visual Similarity Search

• Extracts image embeddings using TensorFlow ResNet50 (feature_extractor.py)
• Searches for visually similar items with FAISS index (find_similars.py)
• Returns top matches from metadata JSON (data.json or similar)

Example usage:

from feature_extractor import extract_features
from find_similars import find_similar_items

features = extract_features("sample.jpg")
results = find_similar_items(features, top_k=10)

---

⚙️ Setup Instructions

1. Clone the repo

git clone https://github.com/your-username/querybot.git
cd querybot

2. Create & activate virtual environment

python -m venv venv
source venv/bin/activate   # Windows: venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Configure .env

Add your API keys, e.g.:

MODEL_API_KEY=your_gemini_api_key

5. Run the Django server

python manage.py migrate
python manage.py runserver

Open http://localhost:8000/chat to test the chatbot UI.

---

🔧 Optional: Image Search Setup

• Place your FAISS index file (your_index_file.index) and metadata pickle (your_metadata.pkl) in the project root or configured path
• Keep your image metadata JSON (data.json) updated
• Image search feature activates only if those files are present

---

🤝 Integration & Extensibility

• Connect with any frontend (React, Vue, mobile apps) via Django REST API
• Extend with voice-to-text or other AI modules easily
• Adapt database backend with minimal changes

---

🧑‍💻 Maintainer

Roshan Kumar B
AI & Machine Learning Engineer | Django Backend Developer
Specializing in building scalable backend systems integrated with machine learning, natural language processing, and computer vision.

---

📄 License

MIT License — free for personal and commercial use.

---

⚙️ Developer Ready. LLM-Powered. Visual-Aware. QueryBot is your smart assistant for advanced contextual queries.

---