A Data Mining Methodology for Detecting Conspiracy Theories from Scientific Articles: The Covid-19 Case

Abstract

The goal is to do a text mining analysis of all scientific publications and find out what journal and what aspects are studying about the conspiracy theories of Covid-19. For this purpose, all publications available in the National Center for Biotechnology Information (NCBI) database were consulted as they were peer-reviewed papers. Of all these papers, only the abstracts of each one were studied using artificial intelligence techniques to determine, for example, whether the subject is of importance depending on the journals where it has been published, and above all, what possible relationships could be extracted from the information published in them. In addition, the "Net Prevalence per Covid19" index was definedin those countries with a high value, greater campaigns should be sponsored to avoid the misinformation generated by Covid-19, although this comment should be verified in future publications. The main challenge was to unify the abstracts and for this purpose, a text summarizer was used under artificial intelligence schemes. The results obtained indicate the tendency of certain topics by the frequency of the words obtained where the focus on the conspiration are the Covid-19 vaccines, but further work is still needed to continue working on this methodology to unify the results.