Skip to content

AshutoshRajGupta/SemanticCart-AI-Powered-Ecommerce-Semantic-Search-Assistant

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🛍️ AI Ecommerce Semantic Search Chatbot

An AI-powered ecommerce semantic search chatbot that retrieves products based on meaning and user intent instead of relying only on traditional keyword matching.


🚀 Project Goal

Traditional ecommerce search systems mainly depend on exact keyword matching.

This project demonstrates how Semantic AI Search, Vector Databases, and LLMs can improve ecommerce product discovery using:

  • Natural Language Queries
  • Semantic Retrieval
  • Embedding-Based Similarity Search
  • AI-Powered Query Understanding

❓ Problem Statement

Traditional ecommerce search systems struggle with:

  • Vague user queries
  • Conversational search
  • Intent understanding
  • Descriptive product searches

Example Queries:

  • comfortable shoes for long office walking
  • stylish back cover for gaming phone
  • lightweight shoes for daily jogging

Traditional search may fail because exact keywords may not exist.

This project solves that problem using semantic AI retrieval.


🎯 Selected Use Cases

1. Natural Language Product Discovery

Users search naturally instead of typing exact product names.

The AI chatbot understands user intent and retrieves semantically relevant products.


2. Ecommerce Search Enhancement

Traditional ecommerce search struggles with vague or descriptive queries.

The AI system improves retrieval using semantic embeddings and vector search.


3. Personalized Shopping Assistant

Users often do not know exact product names or categories.

The chatbot acts like an AI shopping assistant using conversational AI.


🧠 Core AI Concepts Used

  • Semantic Search
  • Vector Embeddings
  • Redis Vector Databases
  • Similarity Search
  • LLM Query Understanding
  • AI Retrieval Pipelines

🛠️ Tech Stack

Frontend

  • React
  • Vite
  • Axios

Backend

  • FastAPI
  • Python

AI / ML

  • Sentence Transformers
  • Groq LLM

Vector Database

  • Redis Vector Database

Data Processing

  • Pandas
  • NumPy

Deployment

  • Render
  • Vercel

📂 Dataset Information

The project uses an ecommerce product dataset containing:

  • Product Names
  • Brands
  • Product Types
  • Descriptions
  • Keywords
  • Colors
  • Model Names

The dataset was cleaned and preprocessed before generating embeddings.


🏗️ High-Level Architecture

React Frontend ↓ FastAPI Backend ↓ Groq LLM + Embedding Model ↓ Redis Vector Database ↓ Semantic Product Retrieval


🔄 Exact Application Flow

Step 1

User enters natural language ecommerce query.

Step 2

Frontend sends request to FastAPI backend.

Step 3

Groq LLM generates optimized ecommerce keywords.

Step 4

Sentence Transformer converts text into vector embeddings.

Step 5

Redis Vector Database performs vector similarity search.

Step 6

Most relevant products are retrieved.

Step 7

Backend returns structured product metadata.

Step 8

React frontend renders recommendation cards dynamically.


✨ Features

  • AI Semantic Search
  • Natural Language Product Discovery
  • Redis Vector Similarity Search
  • Structured Product Recommendations
  • FastAPI REST APIs
  • React Chat Interface
  • Full-Stack AI Architecture

📘 Key Learning Outcomes

Through this project I learned:

  • Semantic Search
  • Vector Embeddings
  • Redis Vector Databases
  • AI Retrieval Pipelines
  • FastAPI Backend Development
  • React Frontend Integration
  • Full-Stack AI Deployment
  • AI Product Design Thinking

🔮 Future Improvements

  • Product Images
  • Personalized Recommendations
  • Conversational Memory
  • Streaming Responses
  • Hybrid Search
  • Recommendation Ranking
  • Multi-Agent Ecommerce Assistant

About

An AI-powered ecommerce semantic search chatbot that retrieves products based on meaning and user intent instead of relying only on traditional keyword matching.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors