Sentiment Analysis Application

An advanced sentiment analysis web application built with Streamlit that analyzes text sentiment using NLTK's VADER and transformer-based models.

Features

• Real-time sentiment analysis of text input
• Detailed sentiment scoring on a scale from -1 (very negative) to 1 (very positive)
• Statistical metrics (average, median, standard deviation)
• Visual representation of sentiment distribution
• Sentiment gauge visualization
• Downloadable results in CSV format

Setup Instructions

Prerequisites

• Python 3.8 or higher
• pip (Python package manager)

Installation

1. Clone this repository
2. Install the required packages:

   pip install -r requirements.txt
   

3. Download required NLTK data:

   import nltk
   nltk.download('vader_lexicon')

Usage

Run the Streamlit app:

streamlit run sentiment_analysis_app.py

The application will open in your default web browser.

How to Use

1. Enter or paste text in the input area (left column)

   • Each line will be analyzed separately
   • For best results, enter complete sentences

2. View sentiment scores in the right column

   • Scores range from -1 (very negative) to 1 (very positive)
   • Color coding provides quick visual feedback

3. Below the main interface, you'll find:

   • Statistical metrics (average score and standard deviation)
   • A histogram showing the distribution of sentiment scores
   • A table with detailed results that can be sorted and filtered

4. The application updates in real-time as you type or modify the input text

How It Works

The application uses two sentiment analysis models:

1. NLTK's VADER (Valence Aware Dictionary and sEntiment Reasoner) - Optimized for social media text
2. DistilBERT transformer model - Used as a backup for more complex cases

For each input text, the app:

• Calculates sentiment scores
• Provides sentiment labels (Very Negative to Very Positive)
• Displays statistical metrics
• Creates visualizations of sentiment distribution
• Allows downloading results as CSV

Technical Implementation

The application uses a pre-trained DistilBERT model from Hugging Face that has been fine-tuned for sentiment analysis. The model outputs are converted to a continuous scale between -1 and 1 for intuitive interpretation.

License

This project is open source and available for educational and personal use.

Last Updated

April 13, 2025

Screenshots

[Add screenshots here]