Mapping Research Topics at Multiple Levels of Detail

This is a Preprint and has not been peer reviewed. The published version of this Preprint is available: This is version 3 of this Preprint.


Download Preprint

Supplementary Files

Sara K Lafia , Werner Kuhn, Kelly Caylor


Institutional reviews typically rely on scientometrics, like the h-index and impact factors of their participants, to assess research productivity. Productivity is not the only review criterion however, and scientometrics can be difficult to generate and compare in multidisciplinary settings. “Distant reading” methods from the Digital Humanities can complement the current quantitative evaluation paradigm; these methods support qualitative narratives, comprehension, and discovery of knowledge by arranging vast bodies of text into graphs, maps, and trees. To test this idea, we apply distant reading methods to a multidisciplinary body of research authored by 240 researchers from the Earth Research Institute (ERI) at UC Santa Barbara over the past decade. We model cross-disciplinary topics of research publications and projects emerging at multiple levels of detail. From these, we design maps that reveal the latent thematic structure of multidisciplinary research. ERI’s researchers use and evaluate these maps of research topics in the context of an institutional review to “read” ERI’s body of research at a distance, i.e. at multiple levels of detail. We find that our approach strengthens the institutional review process by exposing thematic expertise, relationships between researchers, topical distributions and clusters of work, and the evolution of these aspects over time.



Geography, Physical Sciences and Mathematics, Social and Behavioral Sciences, Spatial Science


data discovery, decision support, institutional review, knowledge representation, natural language processing, serendipity, spatialization, topic modeling


Published: 2020-07-12 12:30

Last Updated: 2020-08-31 12:15

Older Versions

CC BY Attribution 4.0 International

Add a Comment

You must log in to post a comment.


There are no comments or no comments have been made public for this article.