Classification of covid related articles using machine learning

Abstract

Covid 19 pandemic has placed the entire world in a precarious condition. Earlier it was a serious issue in china whereas now it is being witnessed by citizens all over the world. Scientists are working hard to find treatment and vaccines for the coronavirus, also termed as covid. With the growing literature, it has become a major challenge for the medical community to find answers to questions related to covid-19. We have proposed a machine learning-based system that uses text classification applications of NLP to extract information from the scientific literature. Classification of large textual data makes the searching process easier thus useful for scientists. The main aim of our system is to classify the abstracts related to covid with their respective journals so that a researcher can refer to articles of his interest from the required journals instead of searching all the articles. In this paper, we describe our methodology needed to build such a system. Our system experiments on the COVID-19 open research dataset and the performance is evaluated using classifiers like KNN, MLP, etc. An explainer was also built using XGBoost to show the model predictions.

Keywords: COVID-19; Explainability; KNN; MLP; Text classification; XGBoost.