Add ModelIdentifier mapping to DataCite–PIDINST schema crosswalk

[strambc]

Request on behalf of the Imaging-PHD NSF funded project

The current DataCite-to-PIDINST mapping only accommodates the instrument model name, represented as a Description with descriptionType="TechnicalInfo". This approach loses structured, machine-actionable information and does not reflect the full capabilities of the PIDINST schema, which already includes a dedicated ModelIdentifier field.

This issue requests that the mapping be extended to include a representation of ModelIdentifier, enabling persistent identifiers for instrument models — such as RRIDs — to be carried through the DataCite mapping.

---

Background

The PIDINST schema supports the following structure for instrument models:

• Model/modelName — the human-readable name of the model
• Model/modelIdentifier — a persistent identifier for the model (e.g. an RRID)
• Model/modelIdentifier/@modelIdentifierType — the identifier scheme (e.g. RRID)

RRIDs (Research Resource Identifiers) are already used in the community to persistently identify instrument models, and their use is within scope of the PIDINST specification.

However, the current DataCite crosswalk maps only the model name:

PIDINST field	DataCite mapping
Model/modelName	descriptions/description with descriptionType=TechnicalInfo
Model/modelIdentifier	(not mapped)

This means that when instrument metadata is expressed in DataCite, any structured model identifier is silently dropped. A DataCite record cannot currently round-trip a modelIdentifier back into PIDINST.

---

Proposed change

Extend the DataCite mapping to include Model/modelIdentifier, for example using the relatedIdentifiers:

PIDINST field	Proposed DataCite mapping
Model/modelName	descriptions/description with descriptionType=TechnicalInfo (existing)
Model/modelIdentifier	relatedIdentifiers/relatedIdentifier with relationType="IsVariantFormOf" (or isDescribedBy) and relatedIdentifierType matching the identifier scheme

An alternative could be alternateIdentifiers, though relatedIdentifiers is more semantically expressive for capturing the model–instance relationship.

The exact relationType value is open for discussion — suggestions welcome.

---

Motivation

• Preserves machine-actionable, persistent identification of instrument models across schema representations
• Aligns the DataCite mapping with what the PIDINST schema already supports
• Enables round-tripping of modelIdentifier values (e.g. RRIDs) between PIDINST and DataCite records
• Supports interoperability with life science infrastructure and FAIR instrument metadata workflows

---

References

• PIDINST schema specification: Model/modelIdentifier
• RRID Initiative: https://www.rrids.org
• Existing DataCite–PIDINST crosswalk (link to relevant section)

[bandrow]

So according to the full PIDINST white paper schema at https://docs.pidinst.org/en/latest/white-paper/metadata-schema-recommendations.html

The way to handle the model is
ID | Property | Obligation | Occ. | Definition | Allowed values, constraints, remarks
6 | Model | R | 0-1 | Name of the model or type of device as attributed by the manufacturer | Element
6.1 | modelName | R | 1 | Full name of the model | Name field from RRIDE.g.‘Illumina HiSeq 3000/HiSeq 4000 System’
6.2 | modelIdentifier | O | 0-1 | Persistent identifier of the model | RRID identifierE.g.‘RRID:SCR_016386’
6.2.1 | modelIdentifierType | O | 1 | Type of the identifier | Free text; must be identifier typeE.g. ‘RRID’

[strambc]

@bandrow this is corret.
The issue is about the DataCite tp PIDINST mapping.

1. Mapping of PIDINST/Model
   In the current mapping (https://datacite-metadata-schema.readthedocs.io/en/4.6/mappings/pidinst/), PIDINST/Model is mapped to field 17.Description with 17.a descriptionType=TechnicalInfo.

However, this mapping does not appear to accommodate the modelIdentifier.