On Using Disparate Scholarly Data to Identify Potential Members for Interdisciplinary Research Groups

Abstract

Supporting interdisciplinary research (IDR) requires detecting the expertise needed to solve complex problems and identifying researchers with that expertise. Universities have adopted various expertise systems, many of which use publications and keywords to identify experts. Research expertise is dynamic in nature as one's expertise may change over time. Relying solely on publications to infer research interests can be less effective in identifying potential collaborators as different types of scholarly activities demonstrate the change in research direction at different times. This paper uses disparate scholarly data to propose and evaluate different approaches for building research footprints and presents experimental results to show how these footprints perform in identifying potential members for IDR groups. Results indicate that grant data is a better predictor of IDR membership than publication data. The paper also describes two approaches for building IDR-specific classifier models, along with the accuracy of those models in identifying potential IDR group membership.