Impactos da pandemia de COVID-19 em alunos de graduação: um estudo através da modelagem de tópicos neural em discussões do Reddit

Abstract

The COVID-19 pandemic has impacted the lives of many undergraduate students around the world, causing significant changes in their routines. The need for measures to combat the transmission of the virus, such as social distancing and migration to online classes, has caused a series of situations that affected students in the social, educational and emotional spheres. In order to help institutions plan effective policies, we sought to analyze these impacts through posts from undergraduate students, obtained by collecting discussions published on the social news site Reddit between January 1, 2020 and December 31, 2021. For this analysis, we used a neural topic model to apply contextual document embeddings – generated through a pre-trained language model – in the grouping of semantically similar posts, resulting in dense clusters, which had their thematic structures represented in the form of keywords. To help interpret the results, the most representative documents of each topic were observed and a diagram showing the hierarchy between clusters was generated, helping understand the structural similarity between documents across different topics. In addition, we explored the most popular posts about the observed themes, taking into account various periods. By analysing these factors and considering the most useful topics, the result was a study of the impacts of the COVID-19 pandemic on undergraduate students, where the main themes of each group of posts were observed and translated into understandable terms.