November 2022

​

In 2022, we engaged in a research project with CWTS and several OA Switchboard publishers. Results were presented at the 27 October webinar (recording here): Collaborating to Unlock the Power of PIDs - A how-to webinar from OA Switchboard partners and participants on opportunities to capture and maximise usage of identifiers
 

The research question was: “What are the opportunities to use ROR id’s available at source to enrich openly available author affiliation data?”

​

Results and Conclusions:

1. There is potential:

   • Some publishers have what they regard to be ‘validated ROR id’s’ in other systems than the ones currently feeding Crossref

   • ROR id’s can be derived from at least two independent algorithms (i.c. ‘smart matching’ and OpenAlex)
      

2. It depends how you look at things, how good/bad something is...
    

3. There are (quality) issues with:

   • Data

   • Algorithms

   • Systems
      

4. There are fundamental questions/issues:

   • Upstream or downstream?

   • Machine or human?

   • Who is responsible?

   • Interoperability of systems

​
 

Recommendations and Next Steps

• For institutions:

  • If your institution isn’t covered (correctly) in the ROR registry, let them know:
    http://curation-request.ror.org

  • If you receive OA Switchboard messages not meant for you, please share that feedback with the sender/publisher, as well as OA Switchboard (for learning, metadata and algorithm improvements, etc).
     

• For publishers:

  • Improve affiliation text quality with as much structure as possible (e.g. separated department, institution name, address, country) and any PID (“garbage in – garbage out”).

  • Review/decide business rules and quality ambition and thresholds in ‘smart matching’ (part of custom connector). Be critical on what the algorithms give you back: it is a ‘suggestion’ (85% is generally regarded to be ‘good'). Consider adding a human quality-check component.

  • Explore low-hanging fruit / options to enrich your own Crossref records with ROR id’s (existing option/procedure to update own records).
     

• For ROR:

  • Increase coverage of institutions (e.g. geographical coverage (e.g. China)), with correct labels.

  • Improve affiliation matching search service (algorithm), i.c. scores/thresholds, splitting.
    Note: the November 2022 release gives significantly better results.
     

• For OA Switchboard open source ‘smart matching’ module:
  (available in OA Switchboard for API-publishers, and in publishers’ custom connectors)

  • Improve algorithm based on updated ROR affiliation matching search service (algorithm), e.g. exact matches.

  • Add ROR id’s to the ‘stop-list’.

  • Use more/multiple algorithms.
     

• Vendor systems: interoperability