search

dret / I-D

Internet Drafts I've authored or contributed to.
17 stars 13 forks source link

Double check the @id for hreflang in the example JSON-LD context #140

Closed dret closed 4 years ago

dret commented 4 years ago

( couldn't assign this issue to @stain, so assigned it to myself instead... ) the JSON-LD context in https://tools.ietf.org/html/draft-wilde-linkset-06#appendix-B currently uses https://www.w3.org/ns/activitystreams#hreflang as the @id for hreflang, but uses DC terms for most other concepts. is there a reason to refer to the activity streams vocabulary for just this one concept, instead of using DC like for all other concepts?

stain commented 4 years ago

I think @hvdsomp found that - I guess because hreflang is a direct match for the Linkset meaning. Dublin Core Terms would have dct:language which is also appropriate - I am fine with changing it if you think it makes it simpler with a single vocabulary.

An alternative mapping to schema.org https://schema.org/LinkRole would use https://schema.org/inLanguage

dret commented 4 years ago

On 2020-06-26 11:04, Stian Soiland-Reyes wrote:

I think @hvdsomp https://github.com/hvdsomp found that - I guess because |hreflang| is a direct match for the Linkset meaning. Dublin Core Terms would have dct:language https://www.dublincore.org/specifications/dublin-core/dcmi-terms/#http://purl.org/dc/terms/language which is also appropriate - I am fine with changing it if you think it makes it simpler with a single vocabulary.

it was actually @philarcher pointing it out when we recently talked about GS1's plans to start using linksets.

An alternative mapping to schema.org https://schema.org/LinkRole would use https://schema.org/inLanguage

i guess ideally you rely on as few vocabularies as possible, no? if so, maybe the DC term would be the best choice?

hvdsomp commented 4 years ago

Hey @stain, I don't think it was me but I am going to assume it was @csarven who suggested https://www.w3.org/ns/activitystreams#hreflang as the @id for hreflang. I am sure it's a correct approach, but - as @dret indicates - it's probably better to use as few (widely known) vocabularies as possible, especially since this is for an example context provided FYI in an Appendix. So, I would definitely be in favor of using dct:language.

dret commented 4 years ago

On 2020-06-29 08:35, Herbert Van de Sompel wrote:

Hey @stain https://github.com/stain, I don't think it was me but I am going to assume it was @csarven https://github.com/csarven who suggested |https://www.w3.org/ns/activitystreams#hreflang| as the |@id| for |hreflang|. I am sure it's a correct approach, but - as @dret https://github.com/dret indicates - it's probably better to use as few (widely known) vocabularies as possible, especially since this is for an example context provided FYI in an Appendix. So, I would definitely be in favor of using dct:language.

out of curiosity i checked https://www.w3.org/ns/activitystreams#hreflang on sameas.org but nothing shows up. but then again, the whole exercise of deciding "when are two concepts really the same" is a bit of an ontological snake pit to begin with....

larsgsvensson commented 4 years ago

out of curiosity i checked https://www.w3.org/ns/activitystreams#hreflang on sameas.org but nothing shows up. but then again, the whole exercise of deciding "when are two concepts really the same" is a bit of an ontological snake pit to begin with....

Slightly OT: Since as:hreflang and dct:language are different individuals, they should not be owl:sameAs-ed anyway...

dret commented 4 years ago

On 2020-06-29 09:44, Lars G. Svensson wrote:

Slightly OT: Since as:hreflang https://www.w3.org/ns/activitystreams#hreflang and dct:language http://purl.org/dc/terms/language are different individuals, they should not be |owl:sameAs|-ed anyway...

i guess i looked in the wrong place. same as is for "real identity" and not for "just equality", i assume. whatever differentiates these (hence the ontological snake pit remark)...

csarven commented 4 years ago

Background: https://github.com/dret/I-D/issues/89#issuecomment-461192217 :)

Yes, using dcterms:language instead works.

philarcher commented 4 years ago

Sorry - should have jumped in last week. Yes, we (GS1) thought dcterms:language made sense for all the reasons discussed here. I'm tempted by schema:inLanguage (my rule of thumb - if it's in schema.org and the relatively loose semantics don't kill you, use it) but, well, we're looking at DC for most of this. Hmm... the tyranny of choice.

hvdsomp commented 4 years ago

Used the DC Terms URI for language in JSON-LD context.