Using https://schema.org/IndividualProduct

[huberrob]

As proposed by @louatbodc an instrument instance can be described as a schema:IndividualProduct which can easily be linked to a given schema:Dataset using the isBasedOn property

{
    "@context": "http://schema.org",
    "@type": "Dataset",
    "@id": "doi.org/somedoi",
    "name": "Dataset links an instrument instance",
     "isBasedOn": {
     "@id": "pidinst.org/somePIDINST",
     "@type": "IndividualProduct",
     "name": "SBE 37 MCAT-IM-CT(P) 2490",
     "serialNumber": "2490"
     }
}

[louatbodc]

Hi @huberrob. ESIP science on schema suggest that datasets can be linked to physical samples (IGSN) through schema:hasPart. The same principle could be used for physical instruments.

[huberrob]

Dear @louatbodc ,

Using schema:isPartOf instead of schema:isBasedOn in the example snippet above produces a "IndividualProduct is not a known valid target type for the isPartOf property." Error at the schema.org validator

According to the documentation, the expected type for schema:isPart is schema:CreativeWork only. But for schema:isBasedOn we are allowed to use schema:CreativeWork, schema:Product and URL. Since schema:IndividualProduct is a subtype of schema:Product

I would therefore recommend to use schema:isBasedOn to link an instrument with a given schema:Dataset.

[louatbodc]

Hi @huberrob,

We actually recommend an instrument is described as a schema:CreativeWork and schema:IndividualProduct because it is something credited. As we are describing instruments as a CreativeWork, you could use hasPart.

Cheers
Lou

[huberrob]

Dear @louatbodc ,

yes this would correctly validate. However I doubt using_schema:isPartOf_ to indicate an instrument used to compile a dataset is semantically correct. Can an instrument really considered/described as being part of a dataset which was created e.g. using the output an this instrument?
Also, schema:isBasedOn does not perfectly match...

Actually with respect to instruments anyone would expect this would be linked via schema:measurementMethod but alas this only allows e.g. schema:DefinedTerm etc not CreativeWork etc..

This (snipped below) would validate but is probably overstreching schema.org ;) :

{
    "@context": "http://schema.org",
    "@type": "Dataset",
    "@id": "doi.org/somedoi",
    "name": "Dataset links an instrument instance",
     "measurementMethod": {
     "@id": "pidinst.org/somePIDINST",
     "@type": ["IndividualProduct","CreativeWork","DefinedTerm"],
     "name": "SBE 37 MCAT-IM-CT(P) 2490",
     "serialNumber": "2490"
     }
}

[louatbodc]

Hmmm.....I've seen the use of schema:Action gaining momentum in other areas which seems like a neat way of linking things and attributing that linkage. You might have to @graph the json-ld to make it work as the action would be a separate object where the schema:instrument will be the instrument PID.

You could then attribute the operator of the instrument for generating the dataset

[rduyme]

Hello here is additional information regarding link between Instrument and Dataset in schema.org context

• Science on Schema recommends to use prov:wasGeneratedBy : https://github.com/ESIPFed/science-on-schema.org/blob/main/guides/Dataset.md#indicating-a-software-workflow-or-processing-activity-provused-and-provwasgeneratedby

• CDIF recommends to use prov:wasGeneratedBy (CDIF recommendation)

• A proposition was done to add wasGeneratedBy in schema.org 2024/10 : Add wasGeneratedBy property schemaorg/schemaorg#3694

[huberrob]

Imo this would be an option only if schema.org included a wasGeneratedBy property which allows to use prov:Activity etc
However, since the current schema.org/Dataset does not provide such a property it would be quite unexpected for metadata parsers?

We should support @buttigieg ( I just did) proposal to add a schema:wasGeneratedBy property and probably also ask for a more specific schema:MeasurementAction which would make it more clear in what context the schema:instrument would be here.

This would enable @louatbodc s proposal above and perfectly mimic DCAT

I am not sure what the CDIF plans are with respect to specifications by now it seems to be a recommendation only. Probably parsers would need something like a schema.org branch or community profile in a similar way as RO-CRATE or bioschemas did to include additional properties.