There is a fundamental choice regarding representing researcher identifiers in LDAP entries and SAML attribute assertions:
- Craft per-issuer attributes, appropriately named, for example:
- eduPersonOrcid (http://orcid.org/ )
- eduPersonScopusId (http://www.elsevier.com/online-tools/scopus )
- eduPersonProfilesRnsId (http://profiles.catalyst.harvard.edu/ )
- add to this list
- Prefix "eduPerson" to the attribute name or not (e.g., eduPersonOrcid or just orcid)
- Create a generic attribute for researcher identifiers
- For example, create "eduPersonResearcherId"
- Then decide how to indicate the issuer for each of the values
- the traditional eduPerson way would be to create a string with the raw identifier value, append "@" followed by a standardized string per issuer
- e.g., http://orcid.org/0000-0002-3174-899X@orcid; 77863@profilesRns; 15737449500@scopus
- other conventions
See, for example, Roland Hedberg email "Re: [refeds] ORCID attribute?"
- urn:schac:personalUniqueId:int:orcidID:xyz
- the traditional eduPerson way would be to create a string with the raw identifier value, append "@" followed by a standardized string per issuer
Issues and tradeoffs of proposed approaches
- ORCID CTO: "ORCID prefers the use of a URI to describe the ORCID iD - http://orcid.org/xxxx-xxxx-xxxx-xxxx"
- If we opt to indicate issue via ...@issuer, we buy into maintaining a registry of issuer identifiers; That is one cost of the approach
- Come up with a lightweight process that allows us to create "experimental" new attributes. One key would be a lightweight method of requesting and issuing OIDs in the eduPerson attribute OID arc for proposed and experimental attributes