Text classification technique for discovering country-based publications from international COVID-19 publications

Area of Science:

• Bibliometrics
• Text Mining
• Health Informatics

Background:

• The surge in COVID-19 research necessitates advanced analysis methods.
• Understanding global and national research trends is crucial for effective health strategies.

Purpose of the Study:

• To apply text classification techniques to identify country-based publications within international COVID-19 research.
• To analyze the consistency of research topics between Iranian and global COVID-19 publications.

Main Methods:

• Applied research utilizing text-mining techniques, including clustering (Latent Dirichlet Allocation - LDA) and text classification (Support Vector Machine - SVM).
• Analysis of COVID-19 publications from PubMed Central (PMC) between November 2019 and June 2021.
• Utilized Python and the scikit-learn library for text classification.

Main Results:

• Seven distinct topics were identified in both international and Iranian COVID-19 publications using LDA.
• The most prominent research area globally and nationally was "Social and Technology in COVID-19" (April 2021 and February 2021, respectively).
• Peak publication rates occurred in April 2021 (international) and February 2021 (national).

Conclusions:

• A common trend and consistency were observed between Iranian and international COVID-19 publications.
• Specifically, the topic of "Covid-19 Proteins: Vaccine and Antibody Response" showed parallel research and publishing patterns.