Recognize rdf:dirLangString

[afs]

This closes #59.
This closes #65.
Replacement for PR #60.
See also w3c/rdf-star-wg#139

This PR includes updates for the "Semantic conditions for literals table" and "RDFS semantic conditions".

---

Preview | Diff

[gkellogg]

On tiny grammar change. Otherwise, looks good.

[pfps]

I believe that there needs to be a WG decision to add this. I created w3c/rdf-star-wg#139 for this purpose.

(The request changes flag is only to prevent yet another rogue merge.)

[afs]

I believe that there needs to be a WG decision to add this.

There has been this WG resolution

support initial text direction in RDF Language-Tagged Literals

and this WG resolution

Add base direction as a forth element of Literals

and this WG resolution

Accept RDF-Concepts issue "text direction"

---

RDF 1.1 Semantics has it's own description of datatypes - it needs updating.

Why is rdf:langString required?

RDF 1.1 Semantics calls out rdf:langString as a normative "exception" because it has a special value mapping from RDF term to value. It gives a fixed (normative) mechanism for rdf:langString that includes handling the language tag in a case insensitive manner. Semantic extensions must agree on this exception.

RDF 1.2 Semantics must do so for rdf:dirLangString with a compatible mechanism, adding dealing with base direction. This PR uses the same text wording style to emphasis this.

The same is true for rdf:dirLangString in order to enact the WG resolutions, making it available for semantic extensions with the same interpretation for all semantic extensions.

Otherwise, the text of RDF Semantics need significant revision because it talks in general terms about the datatype mapping lexical space to value space.

[pfps]

Part of this PR is

RDF processors are not required to recognize any datatype IRIs other than
xsd:string,
rdf:langString, and
rdf:dirLangString
I don't believe that the requirement to recognize directional language strings has been decided.

[afs]

Why is rdf:langString required?

[gkellogg]

I believe the need to support rdf:dirLangString semantics is implicit in the other approvals. Indeed, if it were not a recognized datatype with conditions based on rdf:langString it would be inconsistent. The Working Group's time is quite limited, and easily gets taken up on perm-threads. No need to add further burden to go over something that is so clearly necessary. If there are reasonable objections, they can be made here, or in #60.

[afs]

Why is rdf:langString required?

No answer has been forthcoming.

I believe that because rdf:langString is required then rdf:dirLangString is required because it has everything special about the value space that rdf:langString has.

Therefore this can go into the doc - if there are issues (1) state what they are (2) add an editors note.

[pfps]

Why is rdf:langString required?

No answer has been forthcoming.

I believe that because rdf:langString is required then rdf:dirLangString is required because it has everything special about the value space that rdf:langString has.

Therefore this can go into the doc - if there are issues (1) state what they are (2) add an editors note.

rdf:langString is historical.

[TallTed]

I want to make sure that discussion continues on my emdash suggestions which are marked "resolved" above, but which were not applied and which I cannot now mark "unresolved":

1. Recognize rdf:dirLangString #64 (comment)
2. Recognize rdf:dirLangString #64 (comment)

[afs]

rdf:langString is historical.

What evidence do you have for that?

rdf:langString was introduced in RDF 1.1 and it continues in 1.2. It has been part of the discussions around literals with base direction which the WG has resolved to accept.

Do not confuse it with rdf:PlainLiteral
BiDirection text is not in-scope.

[pfps]

From RDF 1.1 Semantics

The datatype IRIs rdf:langString and xsd:string MUST be recognized by all RDF interpretations.

An RDF interpretation recognizing D is a D-interpretation I where D includes rdf:langString and xsd:string, and which satisfies:

So the requirement to recognize rdf:langString is historical. No WG decision has been made to add rdf:dirlangString here.

[afs]

There has been this WG resolution

"support initial text direction in RDF Language-Tagged Literals."

RDF Semantics does not currently support initial text direction (now "base direction").

[pfps]

I do not view that resolution as requiring that the datatype be part of RDF semantics. It could just as easily be an optional datatype, like rdf:JSON.

[afs]

rdf:dirLangString is part of the RDF data model.

"""
A literal in an RDF graph consists of two, three, or four elements, as follow:"
...
if and only if the datatype IRI is http://www.w3.org/1999/02/22-rdf-syntax-ns#dirLangString,... a base direction that MUST be either ltr or rtl.
"""

However for rdf:JSON::
"""
This section defines additional datatypes that RDF implementations MAY support.
"""

See also the change note in semantics section 5.

[pfps]

The first part is some weak indication that directional strings should be recognized in RDF entailment, but I don't view it as conclusive. The second part is new in 1.2 and I don't believe that it has any bearing here.

I think all that is required is a quick WG vote as to whether directional strings should be part of RDF entailment.

[pfps]

The first part is some weak indication that directional strings should be recognized in RDF entailment, but I don't view it as conclusive. The second part is new in 1.2 and I don't believe that it has any bearing here.

I think all that is required is a quick WG vote as to whether directional strings should be part of RDF entailment.