Fake News Detection Research Using NLP (2017– 2025): A Bibliometric Analysis

Abstract

The rapid rise of fake news and misinformation, especially on social media, has led to a surge of interest in developing ways to detect it. This bibliometric study takes a close look at 1625 publications on fake news detection from 2017 to 2025, drawing data from Scopus, Web of Science, and Google Scholar. We utilized statistical analysis and visualization tools of Python and VOSviewer to explore trends in publications, identify leading authors and institutions, examine international collaborations, and uncover thematic patterns. Our results indicate a steady increase in research output, with India emerging as the most productive country, while the UK and US lead in citation impact. Keyword clustering points to three primary research directions: advanced models such as deep learning and transformers, socially relevant issues like COVID-19 and hate speech, and traditional machine learning approaches. This study brings fresh insights by updating earlier analyses with a focus on NLP and open 2025 dataset. However, despite these advancements, collaboration networks are still quite limited, and the field predominantly relies on English-language sources. This study outlines the landscape of the field and suggests future paths for more inclusive, multilingual, and interdisciplinary research in fake news detection.