Sequential Topic Modelling: A Case Study on One Health Conversation on Twitter

In the past few years, HIV, SARS, cryptococcal meningoencephalitis, and COVID-19 have been worsening. The world is exterminated by pandemic COVID-19, causing tremendous death tolls, economic chaos, and social disruptions. Since the COVID-19 pandemic, the wildlife trade has been seriously re-evaluated. Twitter, as a social media platform, can be a challenging place to collect data in the form of tweets that are currently attracting the attention of many people. Nevertheless, human beings find it relatively difficult to extract latent information from a set of texts to generate particular topics. The process of evaluating the topic model started with understanding its importance. As a next step, we reviewed existing methods for topic coherence, along with the available measures of topic coherence. In order to establish a baseline coherence score, we used Gensim to implement a default Latent Dirichlet Allocation (LDA) model and discuss ways to optimize the LDA hyperparameters.