using owl:sameAs

[jpadfield]

cidoc-crm.examples/models/production/production-triples.csv

Lines 50 to 57 in 91a5c27

	_Blank Node crm:P2.has type crm:E52.Time-span
	_Blank Node crm:P2.has type aat:300379244
	_Blank Node crm:P82a.begin of the begin 1505-01-01 #xsd:dateTime
	_Blank Node crm:P82a.end of the end 1505-12-31 #xsd:dateTime
	_Blank Node rdfs:label 1505@en
	_Blank Node owl:sameAs wd:Q6153
	ngo:002-0432-0000 crm:P4.has time span _Blank Node
	aat:300379244 rdfs:label years@en

It has been suggested that people should avoid using owl:sameAs. It is assumed that this is because people do not always trust all of the properties assigned by others and thus by using owl:sameAs you are automatically accepting all of the potentially properties.

Do people have thoughts about this?

From comments originally made by @natuk

[jpadfield]

I think that this one might be a question left to the creation of software rather than one for modelling.

In this example we have a year defined within the local model and then defined in Wikidata. In this case it is actually the same year, so using owl:sameAs would seem correct.

However, when using the data or automatically looking for new information I would say that given namespaces such as WikiData should be allocated a level of trust. Hence if you fully trusted a namespace then all of the properties defined there could also be used. If you trusted the identification but not the properties then the namespace can still be used as a public reference PID, but the use of any additional properties could be limited.

I would argue that, for modelling, if entity A is actually the same as entity B then using owl:sameAs is correct and required. But owl:sameAs does not mean one has to trust all of the remote properties.