Leveraging data citations to respond to libraries’ data evaluation needs

DOI: 10.60804/yxna-f837

Libraries play an important role in driving adoption of open data, and they have an increasing focus on understanding dataset usage and impact. This information is key in order for institutions to gain information about the diverse research outputs contributed by their researchers, and to inform institutional strategy around research data management. Librarians also rely on insights on data usage to enhance their data advocacy efforts, showcase and support the expertise of the data workforce, and advance recognition and reward of data contributions.

In spite of the importance of these insights, reporting on the reach and impact of data often represents a challenge for academic libraries, due to the resources required to identify connections between datasets and other outputs, and the fact that this information is often obscured, incomplete, or siloed across different platforms and systems. One response to this need is to leverage data citations as indicators of dataset utilization: having access to data citations at scale would greatly facilitate libraries’ ability to report on data contributions from their institution.

To showcase the value that data-citation analysis can provide to institutions, we collaborated with Jamie Wittenberg (University Libraries, University of Colorado Boulder) and Kristi Holmes (Galter Health Sciences Library and Learning Center, Northwestern University Feinberg School of Medicine) to analyse the use of datasets from Northwestern University and the University of Colorado Boulder, utilizing data citation information from open aggregates like the Data Citation Corpus. The analysis involves data citations for both institutions identified through the current store of citations in the Data Citation Corpus, as well as data citations on Europe PMC for articles that had at least one affiliation corresponding to one of the two institutions.

Jamie and Kristi presented this analysis at the recent ACRL 2025 conference, and we share the key insights below.

Data usage trends over time

Data citations for both institutions show an increase particularly in the 2020-2023 period, with a peak of 4000 data citations in 2021. The top three most frequently used data repositories for both universities were dbSNP Reference SNP, the Protein Data Bank, and the European Nucleotide Archive. This reflects the high level of data use in fields such as structural biology and genomics, but also, the fact that Europe PMC focuses on biomedical fields, which are also areas currently strongly represented in the Data Citation Corpus. 

Data citations per repository over time 2013-2024.

Data-intensive research areas

The analysis of the journals where the datasets are cited and the research disciplines for the repositories where the datasets are hosted highlights several data-intensive research areas for each institution. Highly-used datasets from Northwestern University relate to medical research (e.g., cancer, immunology, infectious diseases), biochemistry and molecular biology and neuroscience. This is also reflected by the fact that datasets are primarily associated with life sciences and biomedicine journals, as well as interdisciplinary journals. When looking at datasets from the University of Colorado Boulder, data-intensive research areas include environmental sciences (e.g., microbial ecology, polar research), molecular biology and genetics, and plant sciences. This is also reflected in the fact that data used from this institution is associated with field-specific journals, particularly in environmental and earth sciences.

Citing disciplinary areas for data citations associated with each institution.

Advancing data evaluation to inform institutional data strategy

The above highlight some of the insights that data citations can already provide to libraries about the uses of research data. There are many more insights that libraries seek to gain in relation to data outputs, and we continue our work on the Data Citation Corpus so that it can respond to those institutional evaluation needs. As immediate steps, we are working on incorporating data citations from additional sources, and exploring steps to enrich the metadata for data citations, for example in relation to affiliation information.

The tools to access and analyse data citations should go hand-in-hand with institutional processes that evaluate and recognize data contributions. To this end, Make Data Count is also collaborating with HELIOS Open on the ‘Implementing data evaluation at institutions’ Working Group to produce resources that support institutional data-evaluation processes, and showcase examples of data evaluation and impact in academia.

We invite librarians to explore the data file for the Data Citation Corpus for insights about data usage at their institutions, and welcome feedback and suggestions to make the Corpus as useful a resource as possible for institutional data evaluation.