Sentiment Analysis Application

Project Overview

This project is a Sentiment Analysis Application. The application allows users to input text, analyzes the sentiment of the input using a pre-trained model, and visualizes the results through various charts.

Features

• User-friendly interface for text input
• Sentiment analysis using the Hugging Face API
• Multiple visualizations of sentiment analysis results:
  • Pie chart
  • Bar chart
  • Horizontal bar chart
  • Donut chart
• Integration with GPT-2 for text generation (not implemented in the current version)

Technologies Used

• Python 3.8+
• Streamlit
• Plotly
• Hugging Face API
• Requests library

Project Structure

|-- your_project_folder/
    |-- app.py              # Main Streamlit app to interact with the user and visualize results
    |-- gpt_api.py          # Handles the GPT-2 text generation API calls
    |-- visualization.py    # Contains functions to visualize sentiment analysis results
    |-- README.md           # This documentation file
    |-- requirements.txt    # Dependencies for the project

Setup and Installation

Creating a Virtual Environment

It's recommended to use a virtual environment to manage dependencies for your project. Here's how you can set it up:

1. Make sure you have Python 3.8 or newer installed on your system.

2. Install virtualenv if you haven't already:

   pip install virtualenv
   

3. Navigate to your project directory:

   git clone https://github.com/maheshmm7/Sentiment-Analysis-ChatBot.git
   

4. Create a new virtual environment:

   python -m venv venv
   

5. Activate the virtual environment:

   • On Windows:

     venv\Scripts\activate
     

   • On macOS and Linux:

     source venv/bin/activate
     

Installing Dependencies

Once your virtual environment is activated, you can install the project dependencies:

1. Install the required packages:

   pip install -r requirements.txt
   

Setting Up the Project

1. Clone the repository (if you haven't already):

   git clone https://github.com/maheshmm7/Sentiment-Analysis-ChatBot.git
   

2. Set up your Hugging Face API key:

   • Sign up for a Hugging Face account and obtain an API key
   • Replace the headers variable in app.py with your API key:

     headers = {"Authorization": "Bearer YOUR_API_KEY_HERE"}

3. Run the Streamlit app:

   streamlit run app.py
   

API Endpoints

The application uses the following Hugging Face API endpoints:

1. Sentiment Analysis API:

   • URL: https://api-inference.huggingface.co/models/cardiffnlp/twitter-roberta-base-sentiment
   • Method: POST
   • Headers: {"Authorization": "Bearer YOUR_API_KEY_HERE"}
   • Body: {"inputs": "Your text here"}

2. GPT-2 Text Generation API (not currently implemented in the main app):

   • URL: https://api-inference.huggingface.co/models/openai-community/gpt2
   • Method: POST
   • Headers: {"Authorization": "Bearer YOUR_API_KEY_HERE"}
   • Body: {"inputs": "Your prompt here"}

Application Flow

1. User Input:

   • The user enters text into the Streamlit interface.

2. Sentiment Analysis:

   • The input text is sent to the Hugging Face Sentiment Analysis API.
   • The API returns sentiment scores for positive, neutral, and negative sentiments.

3. Data Processing:

   • The application processes the API response, extracting sentiment labels and scores.

4. Visualization:

   • The processed data is used to create various charts (pie, bar, horizontal bar, and donut) using Plotly.

5. Display Results:

   • The Streamlit interface displays the sentiment analysis results and visualizations to the user.

Example Usage

1. Ensure your virtual environment is activated.

2. Launch the application:

   streamlit run app.py
   

3. In the text area, enter a sentence or paragraph to analyze. For example:

   I absolutely love this new restaurant! The food is delicious and the service is excellent.
   

4. Click the "Analyze Sentiment" button.

5. View the results:

   • You'll see the sentiment scores displayed as text.
   • Below the text, you'll find four different charts visualizing the sentiment distribution.

6. Try different inputs to see how the sentiment analysis changes:

   The weather today is quite gloomy and it's making me feel a bit down.
   

   I'm not sure how I feel about this movie. It had some good parts, but also some boring scenes.
   

Future Improvements

• Implement GPT-2 text generation functionality
• Add more advanced NLP features
• Improve error handling and user feedback
• Optimize performance for larger text inputs

Author

• RANGALA MAHESH
• maheshrangala7@gmail.com
• +919032017670

Acknowledgments

• Hugging Face for providing the sentiment analysis model
• Streamlit and Plotly for the interactive web interface and visualizations