Contextual Insights

Contextual Insights aims to shed light on how public datasets are used in scientific research, i.e., who is using them, how and for what purposes, and where in the world. Building on these insights, it aims to democratize data by supporting its discovery and determination of its fit for purpose. Target stakeholders include researchers, policymakers, and the broader public, helping them understand the real-world impact of open data.

By connecting datasets to their broader ecosystem, i.e., publications, authors, institutions, software and tools, etc., we aim to surface patterns that would otherwise be impossible to see. We also aim to facilitate new levels of data reuse, accountability and attribution, and evidence-informed data trust.

Technical Approach

Contextual Insights builds on the Democratizing Data project. The underlying technical approach draws on machine learning and natural language processing, leveraging our partnership with Digital Science to conduct full-text analysis of scientific publications from theDimensions database and to identify mentions of specific public datasets within publication text. These mentions are then linked into a rich metadata knowledge graph that includes authors, institutions, topics, citation counts, tools, and used to derive publication counts, author networks, and usage statistics.

The resulting data-usage descriptors, i.e., structured records that capture the dataset-use ecosystem across the research literature, power search, filtering, and visualization features throughout the platform, and enable users to discover related datasets and determine their fit-for-purpose based on shared usage patterns. As the number of matched publications can get large, optimization, such as focusing on the top-cited publications, is used to manage the time/resource requirements. The results shown on the Contextual Insights Dashboard, such as publication counts, active authors, and institutional coverage, reflect this linked analysis and are updated as new publications are processed. Note that results may include false positives and omissions, and the dashboard supports user feedback on errors in the insights presented. The overall research methodology underlying Contextual Insights is described in [1].

Acknowledgements

This research is supported in part by the National Science Foundation through an EAGER grant [2].

References

[1] Chenarides, L., Ladislau, R., Parashar, M., Porter, S., & Lane, J. Data-usage descriptors as search metadata: the case of food security data and the National Data Platform (2015–2025). Research Square.

https://doi.org/10.21203/rs.3.rs-8569040/v1

[2] Parashar, M., “Collaborative Research: EAGER: NAIRR Pilot Demonstration: Integrating Democratizing Data Services into the National Data Platform”, NSF Award No. 2440195, 2025.

https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2440195