An Application of Latent Dirichlet Allocation (LDA) on Covid19-related Tweets by Ejay Cho ’21, Statistics Colloquium, Monday, April 26, 1:30 – 2:15 pm, meet at https://meet.google.com/abm-rhgs-fmb.
Abstract: In this talk, I will discuss the Latent Dirichlet Allocation (LDA), a topic model that identifies the latent topics in a set of documents based on word frequency data. It also helps find topics that best describe each document. Of the various approaches to inference in LDA, I will examine Gibbs sampling, a Markov chain Monte Carlo algorithm. Then I will share an application by Xue et al. (2020) which analyzed millions of Covid19-related tweets to identify patterns and themes. This analysis is useful in exploring public discourse and people’s reactions and concerns regarding the pandemic.