Second element form here: ncbo/bioportal_web_ui#439 (comment)
From @ckindermann
Part 1 was #114
From "The example at hand" to "This alone makes mod.tll non-conformant to the OWL spec." @ckindermann shows that this not ok to use owl:equivalentProperty to represent the mappings btw MOD properties.
I totally agree with this and the demosntration is ok to me.
In fact, this is the main reason why @biswanathdutta and me orignally used a super loose property (dct:relation) to capture the "mappings" btw MOD suggested properties and the "other possible ones" identified to represent such an information about an artefact. This was the case untill MOD1.4.
Originally, MOD intended to review all the possibles vocabularies that could be used to describe ontologies (now SemanticArtefacts) and was trying to : (i) avoid minting any new metadata property if a similar would be available in an existing classic/standard semantic web vocabulary ; (ii) capure the "equivalent relation" (in https://hal-lirmm.ccsd.cnrs.fr/lirmm-01852080 we explain that we don't call them mappings voluntary) so tools can rely on MOD to unified metadata model for ontologies (typically what we did in AgroPortal). This was not MOD job to "map/align" all the semantic web vocabularies out there (impossible task!). So the use of dct:relation was loos enough to "show there are other similar properties in other semantic web vocabularies" but not entail any semantics.
A reminder of this time is till in the property template:
|
# owl:equivalentProperty RELATED PROPERTY IN ANOTHER METADATA (not owl:equivalentProperty); |
"RELATED PROPERTY IN ANOTHER METADATA (not owl:equivalentProperty)"
When, as a larger group, we moved to MOD2 (then later 3) eventually owl:equivalentProperty was introduced and I did not took the time to explain in details why it was not ok.
Note that the "mappings" where also move to another repo: https://github.com/FAIR-IMPACT/MOD-mappings with the perspective to work on them with a SSSOM tool.
My proposition would be now to roll back to dct:relation for representing loose equivalences in the properties.
Opinions?
Second element form here: ncbo/bioportal_web_ui#439 (comment)
From @ckindermann
Part 1 was #114
From "The example at hand" to "This alone makes mod.tll non-conformant to the OWL spec." @ckindermann shows that this not ok to use
owl:equivalentPropertyto represent the mappings btw MOD properties.I totally agree with this and the demosntration is ok to me.
In fact, this is the main reason why @biswanathdutta and me orignally used a super loose property (
dct:relation) to capture the "mappings" btw MOD suggested properties and the "other possible ones" identified to represent such an information about an artefact. This was the case untill MOD1.4.Originally, MOD intended to review all the possibles vocabularies that could be used to describe ontologies (now SemanticArtefacts) and was trying to : (i) avoid minting any new metadata property if a similar would be available in an existing classic/standard semantic web vocabulary ; (ii) capure the "equivalent relation" (in https://hal-lirmm.ccsd.cnrs.fr/lirmm-01852080 we explain that we don't call them mappings voluntary) so tools can rely on MOD to unified metadata model for ontologies (typically what we did in AgroPortal). This was not MOD job to "map/align" all the semantic web vocabularies out there (impossible task!). So the use of dct:relation was loos enough to "show there are other similar properties in other semantic web vocabularies" but not entail any semantics.
A reminder of this time is till in the property template:
MOD/mod.ttl
Line 688 in 303ea04
"RELATED PROPERTY IN ANOTHER METADATA (not owl:equivalentProperty)"
When, as a larger group, we moved to MOD2 (then later 3) eventually
owl:equivalentPropertywas introduced and I did not took the time to explain in details why it was not ok.Note that the "mappings" where also move to another repo: https://github.com/FAIR-IMPACT/MOD-mappings with the perspective to work on them with a SSSOM tool.
My proposition would be now to roll back to dct:relation for representing loose equivalences in the properties.
Opinions?