About this demo
A minimum viable dataset repository for UW–Madison. Datasets are stored in UW–Madison Research Object S3, described with Croissant metadata, served over the Pelican Platform, and indexed in an internal PostgreSQL database.
Discover
For end users. Search and filter published datasets, then copy a snippet
that loads one over Pelican with Hugging Face datasets.
Showing 12 datasets per page.
Publish
For library data stewards. Describe a CSV, upload it, generate and validate Croissant metadata, and record the dataset.
Planned
- Semantic search with BEAR
- DOI minting via DataCite
- Dataset versioning and backups
- NetID-scoped dataset keys, so
load_uw_data("netid/dataset")works - User-submitted datasets with an approval workflow