Text Summarization is one of the classic problems that can be solved using Natural Language Processing. This project is an attempt towards preprocessing the text data for text summarization and demonstrating the two approaches towards Text Summarization using Natural Language Processing, namely
- Extractive Summarization
- Abstractive Summarization
The Extractive Summarization aims to select the most important highlights or sentences on the basis of a score that is determined by the count of words and other optional parameters such as the length of a sentence.
Abstractive Summarization is an attempt towards training the model used for summarization to generate a summary similar to the way a human would do it. This type of a Generative Model can be trained using a Deep Learning based technique known as the Attention mechanism. For this project, n LSTM-based Encoder-Decoder based Model combined with the Bahdanau Attention Mechanism has been used to perform this task.
Drive Link for the Summarizer Module:
https://drive.google.com/drive/folders/1zjeOXsXktqCW2z1JbiCNhVR_8fD0G6_V?usp=sharing
Download this module as a zip file from Google Drive and paste it in the WebApp folder.
Data Files:
https://drive.google.com/drive/folders/1bCVurULnzKDiEnk-MKQ9E999uiv3zquS?usp=sharing