ChatBot-RAG

ChatBot-RAG is a Retrieval-Augmented Generation (RAG) based conversational agent that leverages LangChain, FAISS, and Streamlit to enable context-aware Q&A from custom data sources. This project demonstrates how to ingest documents, generate embeddings, store them in a vector database, and query them during chatbot interactions.

---

Tech Stack

• LangChain – Language model orchestration
• FAISS – Fast vector similarity search for document retrieval
• Streamlit – Interactive UI for the chatbot
• OpenAI / HuggingFace – LLM backbone (can be swapped)
• Python – Primary programming language

---

Directory Structure

ChatBot-RAG/
├── data/                         # Input documents (.txt files)
├── vectorstore/db_faiss/        # Persisted FAISS index
├── bot.py                       # Streamlit frontend for chatbot
├── create_memory_for_llm.py     # Embeds and indexes documents
├── connect_memory_with_llm.py   # Chains vector memory with LLM
├── test.py                      # Test scripts and utilities
└── requirements.txt             # Python dependencies

---

Getting Started

1. Clone the Repository

git clone https://github.com/rijul21/ChatBot-RAG.git
cd ChatBot-RAG

2. Install Dependencies

pip install -r requirements.txt

Ensure you have Python ≥ 3.7 and an OpenAI or HuggingFace API key set via environment variable.

3. Prepare Data

Place .txt documents inside the data/ directory.

---

Create Vector Memory

Run the script to generate vector embeddings and build the FAISS index.

python create_memory_for_llm.py

---

Connect Memory with LLM

Set up LangChain with the indexed data:

python connect_memory_with_llm.py

---

Launch Chatbot UI

Launch the chatbot with Streamlit frontend:

streamlit run bot.py

You’ll get an interactive UI where you can ask questions grounded in your uploaded documents.

---

Testing (Optional)

Run test.py to verify embedding and retrieval logic:

python test.py

---

Use Cases

• Internal knowledge base assistants
• Academic document Q&A bots
• Domain-specific helpdesks
• Custom data-aware chat interfaces

---