Mapping the "long tail" of research funding: A topic analysis of NSF grant proposals in the Division of Astronomical Sciences

TL;DR

This study analyzes NSF astronomy grant proposals to understand the distribution of small, heterogeneous 'long tail' data, revealing funding and publication patterns across different research topics with implications for research policy.

Contribution

It introduces the concept of 'Topic Investment' and applies Latent Dirichlet Analysis to explore funding distribution and research output in astronomy, highlighting the 'long tail' dynamics.

Findings

Different topics show varied funding levels and publication patterns.

Funding distribution aligns with the 'long tail' theoretical framework.

Dynamics of funding and research output reflect heterogeneity in research topics.

Abstract

"Long tail" data are considered to be smaller, heterogeneous, researcher-held data, which present unique data management and scholarly communication challenges. These data are presumably concentrated within relatively lower-funded projects due to insufficient resources for curation. To better understand the nature and distribution of long tail data, we examine National Science Foundation (NSF) funding patterns using Latent Dirichlet Analysis (LDA) and bibliographic data. We also introduce the concept of "Topic Investment" to capture differences in topics across funding levels and to illuminate the distribution of funding across topics. This study uses the discipline of astronomy as a case study, overall exploring possible associations between topic, funding level and research output, with implications for research policy and practice. We find that while different topics demonstrate different funding levels and publication patterns, dynamics predicted by the "long tail" theoretical framework presented here can be observed within NSF-funded topics in astronomy.