Comments on "rdf:text A Datatype for Internationalized Text"

HP comments on rdf:text
http://www.w3.org/TR/2008/WD-rdf-text-20081202/


==== Summary

Issue 1: systems that do understand rdf:text may represent information in RDF in two different ways that other RDF systems not upgraded to rdf:text will interpret in two different ways.

Issue 2: The treatment of xs:string is at odds with the current RDF specifications.


==== Issue 1:

3.2 Abbreviations of rdf:text and xs:string Literals

Interoperability amongst existing RDF and SPARQL deployed systems should be given a higher priority.  There is no identified advantage to RDF to make this change to RDF.
 
Specifically, the only syntactic form for literals with language tags should be that defined in the current RDF specifications.  Adding an alternative form brings no advantage to RDF systems and full interoperability of OWL2 and RIF systems is achieved by converting rdf:text to literals with language tags ("text"@lang) on all output in RDF and converting to datatype rdf:text on input.

Two changes to reflect this, 
(sec 3)

"""
In addition to the RIF and OWL specifications, this datatype is expected
to supersede RDF's plain literals with language tags, cf. [5], which is
why this datatype has been added into the rdf: namespace.
"""
===>
"""
In addition to the RIF and OWL specifications, this datatype corresponds
to RDF's plain literals with language tags, which is why this datatype
has been added into the rdf: namespace. However, note that the RDF
syntax and semantics remain unchanged and that in an RDF/XML
serialization such literals MUST be serialized using the existing
RDF/XML syntax for plain literals with language tags [5].
"""

And in section 3.2: making the encouraged but optional "should" into "must"

"""
If an implementation supports abbreviation of literals, it SHOULD abbreviate the literals eagerly whenever possible.
"""
===>
"""
An implementation MUST support abbreviation of literals, and it MUST abbreviate the literals eagerly.
"""

== Impact on SPARQL

To show there is observable change in RDF, consider these SPARQL examples: APIs of many RDF toolkits would exhibit similar behaviour.

Does "xyz@en"^^rdf:text have a language tag?

In SPARQL currently:
lang("Padre de familia@es"^^rdf:text) ==> ""
datatype("Padre de familia"@es) ==> error

but with rdf:text interpretation presumably the results would be different.  "en" and rdf:text would be possible if one syntax had been converted to the other form.

http://www.w3.org/TR/rdf-sparql-query/#func-lang


If it does make this change, then this introduces an asymmetry in RDF and SPARQL processors because rdf:text must be handled specially.


==== Issue 2:

The treatment of xs:string should be made compatible with the RDF specifications.  In RDF, it is an entailment that xs:string and plain literals with no language tag are the same value ("RDF Semantics").  The working draft on rdf:text treats xs:string as abstract syntax; it is not.

An effect of the rdf:text proposal is that system may change RDF input in one form into the other form, causing differences in other unmodified RDF systems.
 
"RDF Semantics" gives Entailment Rules xsd 1a and xsd 1b for xs:string and plain literals without a language tag.  It is not a syntactic relationship.

:x :p "xyz" .

:x :p "xyz"^^xs:string .

are not the same graph as RDF abstract syntax.

"RDF Semantics" - http://www.w3.org/TR/rdf-mt/

"Resource Description Framework (RDF): Concepts and Abstract Syntax" -
http://www.w3.org/TR/rdf-concepts/


The suggested changes above would meet our concerns in this area as well.


==== Conclusion

While it would be useful to have a datatype approach for literals with language tags in RDF, this is not the point in time to introduce it as an RDF change or variation.  The goals of RIF/OWL2 can be met using existing RDF syntax and that would avoid any possible interoperability issues between RDF systems, depending on whether they do or do not implement rdf:text.

A working group note to capture this work as it relates to RDF so that it will be considered in future changes to RDF would be good.


--------------------------------------------
  Hewlett-Packard Limited
  Registered Office: Cain Road, Bracknell, Berks RG12 1HN
  Registered No: 690597 England

Received on Tuesday, 6 January 2009 14:35:14 UTC