Classification of covid related articles using machine learning.

Area of Science:

• Computational biology
• Medical informatics
• Natural Language Processing

Background:

• The COVID-19 pandemic created a surge in scientific literature, making information retrieval challenging for researchers.
• Existing search methods struggle to efficiently filter relevant studies amidst the vast volume of publications.

Purpose of the Study:

• To develop a machine learning system for classifying COVID-19 research abstracts.
• To enable researchers to easily identify and access relevant articles from specific journals.

Main Methods:

• Utilized Natural Language Processing (NLP) text classification techniques.
• Developed a system to categorize abstracts based on their content and associated journals.
• Experimented with classifiers including K-Nearest Neighbors (KNN) and Multilayer Perceptron (MLP).
• Incorporated an XGBoost explainer for model prediction interpretability.

Main Results:

• The proposed system effectively classifies COVID-19 abstracts by journal.
• Performance was evaluated using standard machine learning classifiers on the COVID-19 open research dataset.
• The system demonstrated utility in organizing and searching scientific literature.

Conclusions:

• Machine learning-based text classification offers an efficient solution for navigating COVID-19 research.
• The developed system aids researchers in targeted literature review, saving time and effort.
• This approach facilitates better access to scientific knowledge during global health crises.