Data Science DC Apr 2022 Meetup: Network-based Trajectory Analysis of Topic Interactions COVID Lit скачать в хорошем качестве

Data Science DC Apr 2022 Meetup: Network-based Trajectory Analysis of Topic Interactions COVID Lit

To discover an insightful meaning from the abundant COVID-19 literature, we focused on summarizing topics that were mostly researched and also sought to trace the relationships among these topics. Therefore, we utilized a Topic Model to extract topics and estimate the association among topics by modeling the Topic Model’s result with a Network Model. Moreover, we visualized the interactions among topics to help readers understand it more intuitively. Specifically, we used the Biterm Topic Model (BTM) to extract the topics from the COVID-19 literature. We then proceeded to the revised version of the Latent Space Item Response Model (LSIRM) in order to estimate the latent positions of topics based on the key words from each topic. As a result, we were able to visualize the interactions among topics based on their latent positions which helped to understand the associated studies in COVID-19 without depending on the expertise of each and every researcher. Speaker bio: Yeseul Jeon is a Ph.D. candidate in applied statistics and data science from the Yonsei University. Her statistical methods research focuses on modeling dependency structured data using spatial or network models to estimate their correlated structure. Her goal is to create statistical approaches for evaluating dependent structure data in order to expose hidden information in a more comprehensible manner while minimizing information loss.