Hybrid Text Summarizer with Flask and Streamlit

This project is a hybrid text summarization application combining extractive (TF-IDF) and abstractive (Hugging Face Transformers) techniques. It features a Flask backend for API support and a Streamlit frontend for an interactive user experience. Users can upload text files and receive a concise summary using advanced NLP methods.

---

Table of Contents

• Overview
• Features
• Technologies Used
• How It Works
• Setup and Installation
• Usage
• Output
• Future Enhancements
• Acknowledgments

---

Overview

This project provides a hybrid approach to summarization by combining:

Extractive Summarization: Selects key sentences using TF-IDF scores. Abstractive Summarization: Generates summaries using Hugging Face’s facebook/bart-large-cnn model. Hybrid Summarization: Refines extractive summaries through abstractive techniques for improved results.

---

Features

-File Upload Support: Accepts text files for summarization. -Hybrid Summarization: Combines extractive and abstractive methods. -Interactive UI: Streamlit frontend for user-friendly interactions. -API Integration: Flask backend for scalable summarization services.

---

Technologies Used

• Backend: Flask
• Frontend: Streamlit
• NLP Model: -SpaCy for sentence tokenization. -Hugging Face Transformers (facebook/bart-large-cnn) for abstractive summarization. -Scikit-learn's TfidfVectorizer for extractive summarization.
• Libraries:
  • Requests for API communication. -Streamlit for frontend development.

---

How It Works

1. Input:

   • User uploads a text file via the Streamlit interface.

2. Summarization:

   • The text is processed and tokenized using SpaCy. -Extractive summarization ranks sentences by relevance using TF-IDF scores. -Abstractive summarization generates refined summaries using Hugging Face's model. -The hybrid approach combines both methods for optimal results.

3. Output: -The summary is displayed in the Streamlit app.

---

Setup and Installation

Prerequisites

• Python 3.8 or higher
• Pipenv or pip for Python dependency management

Installation Steps

1. Clone the Repository:

   git clone <https://github.com/siddhinarayan09/Text-Summarizer-NLP>
   cd Text-Summarizer-NLP

2. Set Up the Backend: Create a virtual environment and install dependencies:

    pip install -r requirements.txt
   

3. Start the Flask App:

     python app.py
   

4. Run the Streamlit Frontend:

     streamlit run streamlit_app.py
   

Usage

Launch the Streamlit app. Upload a text file for summarization. Click "Summarize" to generate a hybrid summary. View the results dynamically in the app.

Output

The application produces:

Future Enhancements

Support for Multiple File Formats: Add support for PDFs and Word documents. Advanced Visualizations: Use Plotly for interactive visual outputs. Authentication: Allow user accounts to save and access summaries. Language Support: Extend the model to handle multilingual text.

Acknowledgments

Hugging Face: For providing the facebook/bart-large-cnn model. SpaCy: For efficient text preprocessing. Streamlit: For building an interactive frontend. Flask: For lightweight backend API development.

Inspiration from NLP projects and article summarizers.

Thank you for exploring the project:)