Understand the Emotion
Behind Every Word
Analyze reviews, tweets, comments and messages using Machine Learning.
Real-time sentiment classification with confidence scoring.
Built with Production-Grade Technology
Every component is chosen for performance, accuracy, and scalability.
Machine Learning
Trained on 61K+ real-world Twitter samples across Positive, Neutral, and Negative classes.
FastAPI Backend
High-performance REST API with auto-generated Swagger UI documentation, CORS handling, and Pydantic validation.
TF-IDF Vectorization
Unigrams and bigrams with top 5000 vocabulary terms intelligently extracted to represent text numerically.
Logistic Regression
The best-performing model with 82.35% validation accuracy, outperforming Naive Bayes and Random Forest.
Responsive Design
Fully responsive glassmorphic UI that works beautifully on desktops, tablets, and mobile devices.
Real-time Prediction
Instant sentiment classification with probability-based confidence scores, returned in milliseconds.
How AuraSentiment Works
A clean, end-to-end NLP pipeline from raw text to sentiment prediction.
Twitter Dataset
61,000+ labeled tweets in Positive, Neutral, and Negative categories loaded from CSV files.
Text Preprocessing
URLs, punctuation, numbers, and special characters removed. WordNet Lemmatization applied.
TF-IDF Vectorization
Text transformed into numerical feature vectors using Term Frequency-Inverse Document Frequency.
Machine Learning
Logistic Regression classifier trained and evaluated against Naive Bayes and Random Forest.
Prediction + Confidence
FastAPI returns the predicted sentiment with a probability-based confidence score in real time.
Sentiment Analyzer
Type or paste any text below and discover its emotional tone instantly.
About AuraSentiment
A complete end-to-end AI application built for the DEVFORGE internship program.
Abdul Sami
AI/ML Intern · DEVFORGE
Built this end-to-end sentiment analysis application from dataset exploration through model training to a fully deployed FastAPI microservice with a premium frontend.