Summify

Summify is a summarization project with two separate applications:

• Abstractive Summarizer: A Streamlit app that summarizes input text and uploaded documents using a transformer model.
• Extractive Summarizer: A Flask app that summarizes text and article URLs using spaCy and LexRank-style sentence scoring.

Repository Structure

• Abstractive-Summify-main/: Streamlit-based abstractive summarizer
• Extractive-Summify/: Flask-based extractive summarizer

Features

• Summarize pasted text
• Summarize uploaded PDF and DOCX files in the abstractive app
• Summarize article content from a URL in the extractive app
• Display readability metrics in the extractive app

Technologies Used

• Abstractive app: Streamlit, Transformers, Torch, PyPDF2, python-docx
• Extractive app: Flask, spaCy, Sumy, BeautifulSoup, textstat
• Language model: sshleifer/distilbart-cnn-12-6

Clone the Repository

git clone https://github.com/AnshShah-18/Summify.git
cd Summify

Setup and Run

1. Abstractive Summarizer

cd Abstractive-Summify-main
python3.10 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
streamlit run app.py

Notes:

• The app runs locally at http://localhost:8501 by default.
• The first summarization request may take longer while the model downloads and caches.

2. Extractive Summarizer

cd Extractive-Summify
python3.10 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python -m spacy download en_core_web_sm
flask --app app.py run

Notes:

• The Flask app runs locally at http://127.0.0.1:5000 by default.
• If you want to use a different port, run flask --app app.py run --port 5001.

Usage

Abstractive App

1. Open the Streamlit app.
2. Choose either Summarize Text or Summarize Document.
3. Paste text or upload a .pdf / .docx file.
4. Click the summarize button to generate a summary.

Extractive App

1. Open the Flask app.
2. Paste text into the text summarizer, or provide a URL.
3. Submit the form to generate an extractive summary.
4. Review the readability metrics shown in the UI.

Important Notes

• This repository does not use a single root requirements.txt; each app has its own dependency file.
• The current abstractive app uses a BART-based summarization model rather than T5.
• Summary length is currently determined by the app logic and is not user-configurable in the UI.

Future Improvements

• Add one-click setup scripts for both apps
• Unify the project structure and documentation
• Add tests for startup and summarization flows
• Improve model and UI consistency across both apps

Contributions

Contributions are welcome. Feel free to open an issue or submit a pull request.