SentimentAI - NLP Analytics

A stunning web application for analyzing customer sentiment in mobile phone reviews using advanced NLP and Machine Learning techniques.

Lead Developer

2024

PythonFastAPIScikit-learnNLTKGensimJavaScript

View Live Source Code

Project Screenshots

SentimentAI - NLP Analytics screenshot 1

SentimentAI - NLP Analytics screenshot 2

SentimentAI - NLP Analytics screenshot 3

Overview

SentimentAI is a modern web application designed to analyze and visualize customer sentiment from product reviews. It focuses on comparing flagship smartphones (iPhone 15 vs Galaxy S24) using a suite of Natural Language Processing (NLP) models. The application features a glassmorphism-inspired UI and provides detailed confidence scores for each prediction.

Problem & Constraints

Understanding customer sentiment from text reviews can be challenging due to sarcasm, nuance, and volume. Standard tools often lack clear visualization or rely on single, opaque models. There was a need for a transparent, multi-model approach to see how different algorithms interpret the same text.

Solution

I developed a full-stack solution integrating multiple classical ML and NLP techniques: • **Multi-Model Inference** - Compares results from TF-IDF + Logistic Regression, Word2Vec + SVM, and FastText + SVM • **Interactive Visualization** - Displays confidence probability distributions for positive/negative sentiment • **Real-time Analysis** - Instant feedback on typed text • **Modern UI** - A responsive, glassmorphism design for an engaging user experience

Impact & Results

• **Transparent Analysis** - Users can see how different models agree or disagree • **Educational Value** - Demonstrates the strengths and weaknesses of different NLP embedding techniques • **Production Deployment** - Hosted on Vercel with a Python backend

What I'd Improve Next

• Add Transformer-based models (BERT/RoBERTa) • Support for custom dataset uploads • API endpoint for external developers • Multi-language support

Back to all projects Get in touch