A Seq2Seq model with Attention that reads a food review and generates a short summary.
The model is pretty bad, it was only trained on a 100 summaries so it can't make a really good summary.
Sample test:
Test Loss: nan
Sample predictions:
Example 1
Pred: first okay party especially rather oils lemonlime pot pot fat work approved transfer deliberately company remember loved wholesome home method
True: flavor
Example 2
Pred: first okay party especially starving doesnt sits whenever taffy year expensive hour five tea agree regularly method visits abdominal sensitive
True: stomach
Before you begin, ensure you have the following installed:
- pandas
- numpy
- python-dotenv
- nltk
- torch
- scikit-learn
- beautifulsoup4
Configuration options can be set via environment variables or a config file:
REVIEWS_DATA={Path_to_review_data}
git clone https://github.com/Troppy2/text-summarization.git
cd [project_name]
pip install python-dotenv numpy pandas nltk torch scikit-learn beautifulsoup4