Dutch CRIS repositories · publications, file status, licenses
This dashboard is a direct outcome of DURF (Dutch Repository Federation), Theme 3 – Full Text Capture. See the full roadmap on the DURF Gantt chart. It was built following discussions with the Theme leads for Theme 3 and Theme 2, Pascal and Rutger.
Proof of concept, not a live connection: this dashboard does not query the DuckLake catalog live. Every figure was fetched once via a direct DuckDB connection to the public SURF ORI DuckLake catalog and embedded in this page as a static snapshot. If new data lands in the catalog afterwards — institutions depositing files, harvests re-running, metadata edits — none of that will appear here. The numbers will drift from the live catalog over time; treat this as a demonstration of what's possible, not a monitoring tool, until it's wired up to refresh automatically or rebuilt from a fresh extract.
OA Switchboard exploration: the "Automation exploration routes → via OA Switchboard" filter and the manual-capture-cost KPI/chart below are a first pass at a cost–benefit case for whether OA Switchboard could build a service to help CRIS/repositories capture PDFs from its partner publishers. The "via OA Switchboard" toggle matches the first publisher name on a record against OA Switchboard's published participant list (live + preparing to go live, checked 1 Oct 2026) by case-insensitive substring — approximate, same caveat as the dashboard's other publisher normalization. The 7-minutes-per-record manual-capture estimate comes from one institution's own figures from their 2023 Taverne retro-opt-out exercise (~450 journal articles ≈ 52 hours); it hasn't been validated for other institutions or publication types — book chapters, in particular, are reported anecdotally to take noticeably longer per record than journal articles. Use the publication-type filter to scope the cost estimate to the types this rate actually reflects.
Source: SURF ORI DuckLake, schema cris.publications. Data aggregated by institution × publication type × year × open access status × file status × license × publisher × OA Switchboard match. Years 1995–2025 (2026 is still in progress).
Data currency: the cris.publications table in this catalog was last (re)loaded on 4 March 2026 (UTC, per the catalog's own snapshot log). Individual records also carry their own last-modified timestamp from the source repository (OAI-PMH header.datestamp), which ranges up to 15 April 2026 depending on the institution — see per-institution dates. Everything in this dashboard is a static snapshot as of those dates: if an institution deposits files, changes access rights, or otherwise edits records afterwards, that will not show up here until the catalog is reloaded and this dashboard is rebuilt from it.
Off by default. When on, restricts everything to publications from OA Switchboard partner publishers (live + preparing to go live, checked 1 Oct 2026) — the publishers OA Switchboard could plausibly negotiate automated PDF capture with.
Number of publications within the current filter selection
Open file, embargoed file, or no file (share %)
Percentage of publications with an openly accessible file, per year
Open / embargoed / no file per year (share %) — to track progress on the backlog
Hours to manually source and deposit a file for every currently-filtered "no file" record, by publication year, at 7 minutes/record (indicative € shown on hover, using the rate set in the KPI above). Use the filters — institution, publication type, year range, and the "via OA Switchboard" toggle — to scope the backlog to whatever a business case needs to compare.
Distribution of publication-level access rights
First license of the first file per publication (Taverne highlighted separately)
Number of publications per type (total and with file), sorted by total count
Open / embargoed / no file per publication type (share %)
Normalized publisher names (variants merged), sorted from low to high. Only publishers with at least 100 publications in the current selection. "No publisher" and "Other publisher" (long tail, not normalized) are excluded here. Red = low coverage, green = high coverage — candidates for further investigation appear at the top.