Correct non-english labels in OWL-Time RDF representation

[dr-shorthair]

Originally mentioned by @chris-little in https://github.com/w3c/sdw/issues/1173#issuecomment-652525127

Nobu and Kyoung-sook made suggestions about the Japanese and Korean language tags which imply that the original tags may have been misconstrued. To make sensible corrections, I need to understand the use and intent of the multiple language tags - that is, the skos:prefLabels. My ontological ignorance is now showing.

DayofWeek is tied to the Gregorian Calendar and the labels like below are not problematic (though Korean is missing):

:Friday rdf:type :DayOfWeek ;
rdfs:label "Friday"@en ;
skos:prefLabel "Freitag"@de ;
skos:prefLabel "Friday"@en ;
...
skos:prefLabel "星期五"@zh ;
skos:prefLabel "金曜日"@ja ;

The durations unitSecond, unitMinute, unitHour, unitDay, unitWeek, unitMonth and unitYear have been translated very literally into Chinese, Japanese and Korean. Hence the one or yi meaning a 'unit'. All the other skos:prefLabels in the other languages are just the name, like a Unit of Measure. Is this a good way to think of these labels? Would they be used with numeric values?

If so, then unitYear skos:prefLabel "一年"@ja ; used to express three years, would end up as the rather strange "三 一年" (3 unitYear) rather than "三 年" (3 year) .

:unitYear
rdf:type :TemporalUnit ;
rdfs:label "Year (unit of temporal duration)"@en ;
skos:prefLabel "1 년"@kr ;
skos:prefLabel "一年"@ja ;
skos:prefLabel "Jahr"@de ;
skos:prefLabel "rok"@pl

I hope my concerns make sense.

[dr-shorthair]

I believe the issue concerns the labels for temporal 'units of measure' listed at https://github.com/w3c/sdw/blob/gh-pages/time/rdf/time.ttl#L1150 et seq. The Japanese and Korean translation has taken the localName 'unitYear' etc and turned it into a label 'one year'. While this is arguable correct for the atomic unit, when combined with a count, you would get a construction like '3 one year' rather than '3 year' .

However, I'm not sure how far we can push this - is it expected that the prefLabel from the RDF gets dropped directly into a natural language serialization? But underlying is that the label should not be generated from the local-name - in many ontologies the local name is opaque e.g. a UUID. So if the Korean and Japanese speakers think that the 'unit' or '1' element shoudl be dropped, then definitely it should.

Either way, while of course we should do our best internationalising the documentation, I don't think this is a deal breaker as we are only looking at labels, which are annotations and not actionable semantics.

[chris-little]

Touch pad trouble

[tguild]

@dr-shorthair and @chris-little , what are your thoughts? Should we go ahead and publish the two IG Time Notes tomorrow with this outstanding and fix the rdf representation later or hold off?

[dr-shorthair]

This erratum is unrelated to time-aggregates or time-entity-relations.

Things got confused because it was picked up during the review of the two notes, but the issues are in artefacts related to the original OWL-Time, not the new pieces of work.

[tguild]

Thanks for clarifying, will try to get those published tomorrow