Rasa E-commerce Assistant

Overview

An AI-powered chatbot for e-commerce platforms built using the Rasa framework. Handles customer queries related to order tracking, product information, and general support through natural language conversations.

Features

• Order tracking with ID validation
• Product search and inventory checking
• Return and refund policy assistance
• Payment and shipping information
• Context-aware and intent-based conversation flows

Tech Stack

• AI/NLP: Rasa Open Source 3.x, spaCy
• Frontend: React.js (18+)
• Backend: Node.js + Express
• Language: Python 3.8+

Setup Instructions

# Clone the repository
git clone <repo-url>
cd rasa-ecommerce-assistant

# Python environment
python -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`

# Install Python dependencies
pip install -r requirements.txt

# Install frontend dependencies
cd src
npm install
cd ..

# Train and run the assistant
rasa train
rasa run --enable-api --port 5005 &
rasa run actions --port 5055 &
cd src
npm start

Supported Intents

• track_order: Track orders by order ID
• return_policy: Provide return/refund details
• product_search: Discover available products
• shipping_info: Offer shipping and delivery details
• payment_methods: List accepted payment options

Entity Recognition

• Order IDs
• Product names
• Shipping types
• Payment methods

Project Structure

rasa-ecommerce-assistant/
├── actions/            # Custom action code
├── data/               # NLU training data, stories, and rules
├── models/             # Trained Rasa models
├── src/                # React frontend
├── config.yml          # Rasa pipeline and policies
├── domain.yml          # Intents, entities, slots, responses
├── endpoints.yml       # Action server and tracker store config
└── requirements.txt    # Python dependencies

Performance Metrics

• Response latency: < 200ms
• Intent classification accuracy: 95%+
• Handles up to 50 concurrent users (test environment)

License

This project is licensed under the MIT License.