Chatbot_assignment

<<<<<<< HEAD

Chatbot_assignment

=======

🤖 Intelligent Customer Support Chatbot with RAG

A comprehensive AI-powered customer support chatbot built with LangChain, Flask, and FAISS. This system uses Retrieval-Augmented Generation (RAG) to provide accurate, context-aware responses to customer queries.

✨ Features

Part 1: Intent Classification

Multi-intent recognition: Billing, Technical, Account, Complaints, General
Entity extraction: Account numbers, dates, emails, product names
Confidence scoring: Provides confidence levels for each classification
Multi-intent handling: Detects and handles multiple intents in single message

Part 2: Knowledge Base Integration (RAG)

FAISS vector store: Fast similarity search for document retrieval
Semantic search: Understanding context beyond keyword matching
Document chunking: Efficient text splitting for better retrieval
Source attribution: Shows which documents were used for answers
Multiple knowledge domains: Billing, Technical, Account, General FAQs

Part 3: Conversation Management

Multi-turn conversations: Maintains context across messages
Session management: Separate conversation history per user
Memory integration: Uses LangChain's ConversationBufferMemory
Clarifying questions: Asks for clarification when needed
Graceful fallbacks: Offers human agent connection when uncertain

Part 4: Analytics & Improvement

Real-time metrics: Query count, response times, satisfaction scores
Intent distribution: Tracks which intents are most common
User feedback: Star rating system for responses
Performance monitoring: Response time analytics
Continuous learning: Data collection for model improvement

🏗️ Architecture

┌─────────────┐
│   User UI   │
│  (Browser)  │
└──────┬──────┘
       │
       ↓
┌──────────────────────────────────────┐
│         Flask Application            │
│  ┌────────────────────────────────┐  │
│  │  Intent Classification         │  │
│  │  - Pattern Matching            │  │
│  │  - Entity Extraction           │  │
│  │  - Confidence Scoring          │  │
│  └────────────────────────────────┘  │
│                                       │
│  ┌────────────────────────────────┐  │
│  │  RAG Pipeline (LangChain)      │  │
│  │  ┌──────────────────────────┐  │  │
│  │  │ Query Processing         │  │  │
│  │  └──────────────────────────┘  │  │
│  │  ┌──────────────────────────┐  │  │
│  │  │ Vector Search (FAISS)    │  │  │
│  │  │ - Semantic Retrieval     │  │  │
│  │  │ - Top-K Documents        │  │  │
│  │  └──────────────────────────┘  │  │
│  │  ┌──────────────────────────┐  │  │
│  │  │ LLM Generation           │  │  │
│  │  │ - OpenAI / Gemini        │  │  │
│  │  │         │  │  │
│  │  └──────────────────────────┘  │  │
│  └────────────────────────────────┘  │
│                                       │
│  ┌────────────────────────────────┐  │
│  │  Conversation Memory           │  │
│  │  - Session Management          │  │
│  │  - Context Tracking            │  │
│  └────────────────────────────────┘  │
│                                       │
│  ┌────────────────────────────────┐  │
│  │  Analytics Engine              │  │
│  │  - Metrics Collection          │  │
│  │  - Performance Tracking        │  │
│  └────────────────────────────────┘  │
└───────────────────────────────────────┘
         │              │
         ↓              ↓
┌────────────────┐  ┌──────────────┐
│  FAISS Index   │  │  Analytics   │
│  (Vector DB)   │  │  JSON Store  │
└────────────────┘  └──────────────┘

🛠️ Tech Stack

Backend: Flask (Python web framework)
LLM Framework: LangChain (Orchestration)
Vector Database: FAISS (Facebook AI Similarity Search)
Embeddings: HuggingFace Sentence Transformers
LLM Options:
- Google Gemini Pro (Free tier available)
Frontend: HTML5, CSS3, Vanilla JavaScript
Session Management: Flask Sessions

📦 Installation

Prerequisites

Python 3.8+
pip package manager
Git

Step 1: Clone the Repository

git clone <your-repository-url>
cd customer-support-chatbot

Step 2: Create Virtual Environment

# On Windows
python -m venv venv
venv\Scripts\activate

# On macOS/Linux
python3 -m venv venv
source venv/bin/activate

Step 3: Install Dependencies

pip install -r requirements.txt

Step 4: Create Required Directories

mkdir -p knowledge_base templates

Step 5: Add Knowledge Base Documents

Create .txt files in the knowledge_base/ directory:

billing.txt
technical.txt
account.txt
general.txt

(Sample documents are provided in the repository)

Step 6: Create FAISS Index

python ingest_documents.py

This will:

Load all documents from knowledge_base/
Split them into chunks
Create embeddings
Build and save FAISS vector store

Expected output:

Loading documents...
Loaded: billing.txt
Loaded: technical.txt
Loaded: account.txt
Loaded: general.txt
Loaded 4 documents
Splitting documents into chunks...
Created 152 chunks
Creating embeddings (this may take a few minutes)...
Creating FAISS vector store...
Saving vector store...
✓ Vector store created and saved successfully!

⚙️ Configuration

Environment Variables

Copy .env.example to .env:

cp .env.example .env

Edit .env with your configuration:

Option 1: Using OpenAI

MODEL_TYPE=openai
OPENAI_API_KEY=sk-your-api-key-here
OPENAI_MODEL=gpt-3.5-turbo
SECRET_KEY=your-random-secret-key

Option 2: Using Google Gemini (Free)

MODEL_TYPE=gemini
GOOGLE_API_KEY=your-google-api-key-here
GEMINI_MODEL=gemini-pro
SECRET_KEY=your-random-secret-key

Get free API key: https://makersuite.google.com/app/apikey

Option 3: Using Ollama (Local, Free)

MODEL_TYPE=ollama
OLLAMA_MODEL=llama2
OLLAMA_BASE_URL=http://localhost:11434
SECRET_KEY=your-random-secret-key

First install Ollama: https://ollama.ai/ Then run: ollama pull llama2

🚀 Usage

Start the Application

python app.py

The server will start at http://localhost:5000

Access the Chatbot

Open your browser and navigate to:

http://localhost:5000

📁 Project Structure

customer-support-chatbot/
├── app.py                      # Main Flask application
├── ingest_documents.py         # Knowledge base ingestion script
├── requirements.txt            # Python dependencies
├── .env.example               # Environment variables template
├── .env                       # Your environment variables (create this)
├── README.md                  # This file
├── analytics.json             # Analytics data storage
├── templates/
│   └── index.html            # Frontend UI
├── knowledge_base/           # Knowledge base documents
│   ├── billing.txt
│   ├── technical.txt
│   ├── account.txt
│   └── general.txt
└── faiss_index/              # FAISS vector store (generated)
    ├── index.faiss
    └── index.pkl

🧪 Testing Examples

Example 1: Billing Query

User Input:

I was charged twice for my subscription on account 123456789

Expected Response:

Intent: billing (high confidence)
Entities: Account number 123456789
Answer: Information about duplicate charges and refund process
Sources: Billing FAQ documents

Example 2: Technical Issue

User Input:

The app keeps crashing when I try to login and I'm getting an error message

Expected Response:

Intent: technical (high confidence)
Multiple topics detected: crashes, login, errors
Answer: Troubleshooting steps for login and crash issues
Sources: Technical support documents

Example 3: Account Management

User Input:

How do I reset my password and enable two-factor authentication?

Expected Response:

Intent: account (high confidence)
Multiple intents: password reset, 2FA setup
Answer: Step-by-step instructions for both tasks
Sources: Account management documents

Example 4: Multiple Intents

User Input:

I'm very frustrated because my payment failed and now the app won't open

Expected Response:

Primary intent: complaints
Secondary intents: billing, technical
Answer: Empathetic response addressing both issues
Sources: Multiple document categories

Example 5: General Query

User Input:

What integrations do you support?

Expected Response:

Intent: general
Answer: List of supported integrations
Sources: General information documents

📊 Analytics Dashboard

Access Analytics

Click the "📈 Analytics" button in the chat interface or navigate to:

http://localhost:5000/analytics

Metrics Tracked

Total Queries: Number of questions asked
Average Response Time: Mean time to generate responses
Satisfaction Score: Average user ratings (1-5 scale)
Intent Distribution: Breakdown of query categories

Analytics Storage

Analytics are stored in analytics.json:

{
  "total_queries": 45,
  "intent_accuracy": [],
  "response_times": [1.2, 0.8, 1.5, ...],
  "satisfaction_scores": [5, 4, 5, 3, ...],
  "intent_distribution": {
    "billing": 12,
    "technical": 18,
    "account": 10,
    "complaints": 3,
    "general": 2
  }
}

🔌 API Endpoints

POST /chat

Send a message to the chatbot

Request:

{
  "message": "How do I reset my password?"
}

Response:

{
  "response": "To reset your password: 1. Go to login page...",
  "intent": {
    "primary_intent": "account",
    "confidence": 0.95,
    "all_intents": ["account"]
  },
  "entities": {},
  "response_time": 1.23,
  "sources": [
    {"content": "Password reset process..."}
  ]
}

POST /feedback

Submit user feedback

Request:

{
  "rating": 5
}

Response:

{
  "status": "success"
}

GET /analytics

Get analytics summary

Response:

{
  "total_queries": 45,
  "avg_response_time": 1.15,
  "avg_satisfaction": 4.2,
  "intent_distribution": {
    "billing": 12,
    "technical": 18,
    "account": 10,
    "complaints": 3,
    "general": 2
  }
}

POST /clear

Clear conversation history

Response:

{
  "status": "success"
}

🔧 Troubleshooting

Issue: "FAISS index not found"

Solution: Run python ingest_documents.py to create the index

Issue: "API key not found"

Solution: Check your .env file has the correct API key for your chosen model

Issue: "Module not found" errors

Solution: Ensure virtual environment is activated and run pip install -r requirements.txt

Issue: Slow response times

Solutions:

Use faster model (e.g., gpt-3.5-turbo instead of gpt-4)
Reduce chunk size in retrieval
Use local Ollama model
Check internet connection

Issue: Port already in use

Solution: Change port in app.py:

app.run(debug=True, port=5001)  # Use different port

Issue: Ollama not responding

Solutions:

Ensure Ollama is running: ollama serve
Check base URL in .env
Verify model is downloaded: ollama list

🎨 Customization

Adding New Knowledge Base Documents

Create a new .txt file in knowledge_base/
Add your content
Run python ingest_documents.py again
Restart the Flask app

Modifying Intent Patterns

Edit the INTENT_PATTERNS dictionary in app.py:

INTENT_PATTERNS = {
    'billing': [
        r'\b(bill|invoice|payment)\b',
        # Add more patterns
    ],
    'custom_intent': [
        r'\b(your|patterns|here)\b',
    ]
}

Changing UI Theme

Edit the CSS in templates/index.html:

/* Change color scheme */
background: linear-gradient(135deg, #YOUR_COLOR1 0%, #YOUR_COLOR2 100%);

Adjusting Retrieval Settings

In app.py, modify:

vectorstore.as_retriever(search_kwargs={"k": 5})  # Return top 5 documents

🚀 Deployment

Deploy to Heroku

# Install Heroku CLI
heroku login
heroku create your-app-name
git push heroku main
heroku config:set OPENAI_API_KEY=your-key

Deploy to Railway

Connect GitHub repository
Add environment variables in dashboard
Deploy automatically

Deploy to AWS/GCP

Use Docker or traditional deployment methods. Ensure environment variables are set securely.

📝 License

This project is open source and available under the MIT License.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📧 Support

For issues and questions:

Create an issue on GitHub
Email: your-email@example.com

🙏 Acknowledgments

LangChain for the amazing framework
OpenAI, Google, and Ollama for LLM access
HuggingFace for embeddings
Facebook AI for FAISS

Built with ❤️ using LangChain, Flask, and FAISS

61f035c (Initial commit)

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
faiss_index		faiss_index
knowledge_base		knowledge_base
templates		templates
.env.example		.env.example
README.md		README.md
analytics.json		analytics.json
app.py		app.py
check_available_models_google_api.py		check_available_models_google_api.py
ingest_documents.py		ingest_documents.py
requirements.txt		requirements.txt

Prateekgithubgeek/Chatbot_assignment

Folders and files

Latest commit

History

Repository files navigation

Chatbot_assignment

🤖 Intelligent Customer Support Chatbot with RAG

📋 Table of Contents

✨ Features

Part 1: Intent Classification

Part 2: Knowledge Base Integration (RAG)

Part 3: Conversation Management

Part 4: Analytics & Improvement

🏗️ Architecture

🛠️ Tech Stack

📦 Installation

Prerequisites

Step 1: Clone the Repository

Step 2: Create Virtual Environment

Step 3: Install Dependencies

Step 4: Create Required Directories

Step 5: Add Knowledge Base Documents

Step 6: Create FAISS Index

⚙️ Configuration

Environment Variables

Option 1: Using OpenAI

Option 2: Using Google Gemini (Free)

Option 3: Using Ollama (Local, Free)

🚀 Usage

Start the Application

Access the Chatbot

📁 Project Structure

🧪 Testing Examples

Example 1: Billing Query

Example 2: Technical Issue

Example 3: Account Management

Example 4: Multiple Intents

Example 5: General Query

📊 Analytics Dashboard

Access Analytics

Metrics Tracked

Analytics Storage

🔌 API Endpoints

POST /chat

POST /feedback

GET /analytics

POST /clear

🔧 Troubleshooting

Issue: "FAISS index not found"

Issue: "API key not found"

Issue: "Module not found" errors

Issue: Slow response times

Issue: Port already in use

Issue: Ollama not responding

🎨 Customization

Adding New Knowledge Base Documents

Modifying Intent Patterns

Changing UI Theme

Adjusting Retrieval Settings

🚀 Deployment

Deploy to Heroku

Deploy to Railway

Deploy to AWS/GCP

📝 License

🤝 Contributing

📧 Support

🙏 Acknowledgments

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages