COMET Community Meeting Slides, 30 October 2025

COMET Community Meeting 30 October 2025 Improve the Quality of PID Metadata The Goal Collective Stewardship Trust through Transparency and Collective Governance COMET Model Foundational Principles doi.org/10.7269/C1TG6H Collective Benefit A rising tide lifts all boats COMET is an innovative facilitator Develops new capabilities whilst amplifying and connecting existing solutions, uniting community enrichments. COMET Organisers Public Knowledge Project CWTS, Leiden University California Digital Library DataCite COMET Organisers Pilot Enrichment Projects High quality curated metadata exists, but the benefits are isolated. De-siloing quality metadata How it works ●We normalize the institution's input data using the same method as our OpenAlex data ingestion. ●For each input work, we use fuzzy author and affiliation string matching (including name variants) to find overlaps and discrepancies. ●We capture both the raw and normalized affiliation strings from all matches. ●Use these extracted strings to query the database for all other instances of those affiliations, finding every work they appear on. Outputs ●Linkage Validation File ○DOI-by-DOI check showing if the institution's curated author-affiliation pair exists in OpenAlex. ○Flags discrepancies (e.g., institution says author is on a paper, but OpenAlex is missing the affiliation or ROR ID). ●Full Works List ○Uses the affiliation strings from the small input sample to capture all other works, returning the full list from OpenAlex with the same or similar affiliations, including those with incorrect or missing ROR IDs. Pilot Partners European Molecular Biology Laboratory (EMBL) Wageningen University & Research (WUR) •Matched ~90% of curated works with reconciliation strategy, found new works, with 5% missed because of author affiliation gaps • Remaining 5% potentially missed because of small input size (single year) or additional refinement needed on author affiliation matching • For all works from the test year (mached + missed), 83% have correct author affiliation ROR ID assignments, 12% appear to have an EMBL affiliation string but no ROR ID, remainder affected by same author affiliation parsing gap • Not just an OpenAlex issue! ROR needs more EMBL aliases to support matching in OpenAlex Results: EMBL matched incomplete or missing affiliation (bars match) needs investigation incomplete or missing affiliation (bars match) correct has affiliation Reconciliation Matched + missed •Complex curation workflow with source and resource type exclusion prevents straightforward reconciliation result analysis, but lots to learn from WUR’s example •Testing with example year and resource type, 92% of curated works have the correct affiliation and ROR ID assignment • 6% of works have affiliation gaps or errors that prevent reconciliation and ROR ID assignments • Gaps overlap with EMBL on source clustering and we found a sliver of new works! Results: WUR incomplete or missing affiliation (bars correct ROR ID assignment) missed ROR ID assignment correct ●Pilot demonstrates a viable, scalable strategy that creates value for both sides. ●For Institutions: ○Fast, automated validation of local curation. ○A way to find new, uncurated works. ○A clear, actionable list of data gaps to address. ●For Infrastructures: ○Receives targeted, reproducible feedback from a high-quality source. ○Provides an opportunity for institutions to use their existing work to improve the scholarly record. Takeaways ●We've flagged the parsing issues to OpenAlex for investigation, but they’re doing great! ●Future Work: ○Explore formalizing this reconciliation (e.g., harvesting assertions from OAI-PMH). ○Incorporate different inputs and fields. ●Special thanks to our partners at EMBL and WUR. ●The code for this pilot is open source (check it out!). Next Steps Community Perspective: Diego Restrepo from CoLav Open Discussion: Leveraging Curation from CRIS systems Join Us! •Contribute curated metadata and enrichment processes •Get involved in our pilot projects • Save the date for our next community meeting on 21 January at 4 PM (UTC). •Keep informed by signing up to our newsletter.