ChatBot With Memory

RAG chatbot with persistent conversation history and PDF document querying.

---

What it does

Upload a PDF and ask questions about it. The chatbot maintains conversation context across turns, retrieves relevant document chunks using vector embeddings, and generates answers via Google Gemini.

---

Stack

Python LangChain Google Gemini FAISS Streamlit HuggingFace Embeddings

---

Quick start

git clone https://github.com/umran666/ChatBot_With_Memory.git
cd ChatBot_With_Memory
pip install -r requirements.txt

Create .env with your Google API key:

GOOGLE_API_KEY=your_key_here

Launch:

streamlit run app.py

Upload a PDF in the sidebar and start asking questions.

---

Architecture

PDF Upload → Text Extraction → Chunking → Embeddings → FAISS Index
                                                           ↓
User Query → Retrieve relevant chunks → LLM → Answer + Sources

---

Key features

• Conversation memory — retains history for context-aware follow-ups
• PDF Q&A — extract and search document content via RAG
• Semantic search — HuggingFace embeddings + FAISS for fast retrieval
• Source transparency — shows which document sections informed each answer

---

Configuration

Parameter	Default	Description
k (retrieval)	4	Number of document chunks to retrieve
temperature	0.3	LLM response creativity
chunk_size	1000	PDF text split size
chunk_overlap	200	Overlap between chunks

---

Use cases

• AI document assistant
• Research paper Q&A
• Company knowledge base chatbot
• Resume / portfolio analyzer
• Legal document search

---

Requirements

• Python 3.8+
• Google Gemini API key (free tier available)
• Internet connection

---

License

MIT