Developing a Curated Topic Model for COVID-19 Medical Research Literature

Philip Resnik, Katherine E. Goodman, Mike Moran

Keywords: topic modeling, medical research literature, human-in-the-loop

TL;DR: We present work in progress on constructing a curated topic model for COVID-19 research literature

Abstract: Topic models can facilitate search, navigation, and knowledge discovery in large document collections. However, automatic generation of topic models can produce results that fail to meet the needs of users. We advocate for a set of user-focused desiderata in topic modeling for the COVID-19 literature, and describe an effort in progress to develop a curated topic model for COVID-19 articles informed by subject matter expertise and the way medical researchers engage with medical literature.