Abstractive Text Summarization for Covid-19 News with Data Augmentation

Abstract

Since the spread of Corona Virus disease or Covid-19 at the end of 2019, there has been an extensive amount of news about Covid-19 and it takes a long time for humans to read the news, process it and retrieve important information from it. Therefore, automatic text summarization is necessary in this matter as it can help us process information faster and use it to make better decisions. Currently, there are two main approaches to automatic text summarization: extractive and abstractive. Extractive text summarization is conducted by identifying important parts of the text and extract a subset of sentences from the original text. Abstractive text summarization is closer to human's method as it is the reproduction or rephrasing based on interpretation and understanding of the text using natural language processing techniques. In this paper, we present text summarization of Covid-19 news using abstractive method to be close to human's method of summary. We also apply data augmentation in the pre-processing part to be an example case of working with data that are not perfect or diverse enough.