Through the Topic Glass: Reimagining the Research Landscape through Topic Modeling

Date and Location

Apr 28, 2015 - 4:00pm
Network Science Lab, Bldg 434, Rm 122


Rafael Melendez-Rios
IGERT Trainee
Computer Science Department, UC Santa Barbara


Corpora of scientific papers have long been the focus of research with the intent to uncover social phenomena such as influence, innovation, and the spread of ideas. However, previous attempts have generally relied on citation networks for modeling and data mining purposes, visualizing citation or authorship as the principle bonds between any pair of documents. While important insights have been gained through these methods, there remain exciting opportunities to apply approaches that involve modeling of thematic contents of these texts. Specifically, academic publications can be modeled as a dynamic network of objects whose relationships are based on the degree to which they share the same statistically-modeled themes, or ‘topics’. In our spring quarter module, we will explore how to define and construct meaningful topic networks that will allow future researchers to identify research innovation, emerging subdisciplines, or ‘hot’ topics based on academic publications, expanding the frontier in scientometrics research and providing useful metrics to inform science policy.