COVID-19 Article Classification Using Word-Embedding and Different Variants of Deep-Learning Approach

ABSTRACT

The COVID-19 pandemic has changed the way we go about our everyday lives, and we will continue to see its impact for a long time. These changes especially apply to the business world, where the market is very volatile as a result. Requirements of the people are changing rapidly, as are the restrictions on transport and trade of goods. Due to the intense competition and struggles brought about due to the pandemic, acting first on profit opportunities is crucial to businesses doing well in the current climate. Thus, getting the relevant news in time, out of the huge number of COVID-19 related articles published daily is of utmost importance. The same applies to other industries, like the medical industry, where innovations and solutions to managing COVID-19 can save lives, and money in other parts of the world. Manually combing through the massive number of articles posted every day is both impractical and laborious. This task has the potential to be automated using Natural Language Processing (NLP) with Deep Learning based approaches. In this paper, we conduct exhaustive experiments to find the best combination of word-embedding, feature selection, and classification techniques;and find the best structure for the Deep Learning model for article classification in the COVID-19 context. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.