Re: [Relevant for all] options for the RDF representation of ITS 2.0

Hi Felix,
Thanks for laying out the possible options in this clear way, apologies 
for the long response:

On balance, my preference is for option 1, i.e. a non normative 
reference to NIF.

To explain my reasoning, firstly I'd like to make clear that I am 
personally heavily investing in advancing the role of RDF and linked 
data in the language services industry. It is core (together with ITS 
and XLIFF)  to the interoperability R&D strategy in CNGL. This autumn we 
will start a STREP (FALCON) with three SME partners looking specifically 
at leveraging RDF within localization tool chains and I will also be a 
partner in a new CSA (LIDER) around linguistic linked data, together 
with several of those involved in developing NIF.

The inclusion of NIF in the ITS spec is therefore an important outcome 
for us.

However, the relative benefits of it being in there as normative, as 
compared to an informative part of the spec are small overall. This is 
especially so when compare to any risk in delaying the progression of 
the spec altogether. ITS2.0 contains several other features that can be 
important for advancing the benefits of RDF (specifically the ref 
attributes in terminology, text analysis, provenance and possible LQI), 
but only when it reaches recommendation stage.

In general it is relatively early days in the uptake of RDF. So the 
normative vs. informative status of NIF in the spec will, I feel, have 
fairly little influence on those considering implementation, at least 
compared to XML and HTML.

If we pursue option 1, however it would be useful to lay out the 
necessary changes to the spec.

 From a quick check, we'd need to: change the NIF wording in the 1.0-2.0 
difference overview in 1.4; change the wording in 2.7.1 to make the 
non-normative status clear; change the wording of the NIF related 
clauses in 4.3 and 4.3; change the status of section 5.7, perhaps moving 
it to its own informative section and perhaps consider combining section 
5.7 with annex F (NIF-to-ITS mapping?).

The one danger is that with these changes someone could argue why do we 
need the NIF mapping in the spec at all, citing the XLIFF mapping to 
argue that without a normative grounding this should be in best practice 
rather than in the spec, like the XLIFF mapping.

I would strongly argue against this based on the interest of ourselves 
and others in the WG, as demonstrated via the ITS-NIF mapping 
implementations and test suite features already in place (which is still 
ongoing for XLIFF). It may also be that the charter status of RDF as an 
internal liaison makes it arguably more appropriate for inclusion in the 
spec compared to XLIFF, which is an external dependency (felix?).

I hope that's helpful, I'll be on the call tomorrow to talk about it 
further.

Regards,
Dave

On 12/08/2013 07:42, Felix Sasaki wrote:
> Hi all,
>
> this mail is relevant for the general progress of ITS 2.0. Please have 
> a look even if you are not interested in the RDF representation of ITS 
> 2.0.
>
> At
> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Aug/0009.html 
>
>
> I had explained changes to be done to move ITS 2.0. forward. The 
> change "make NIF a *non* normative reference" is actually just one 
> option to reply to this requirement from our charter
> http://www.w3.org/2012/09/mlw-lt-charter.html
>
> "The MultilingualWeb-LT WG will assure that the metadata approach 
> being developed is allowing a conversion to RDF, to foster integration 
> of MultilingualWeb-LT metadata into the Semantic Web."
>
> This requirement does not say that we define a normative approach to 
> allow for that conversion. My mail at
> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Aug/0009.html 
>
> was suggesting to use NIF as the non normative approach.
>
> With this mail I want to bring all options clearly in front of the 
> working group and see what you think. Please have a look at let's 
> decide on Wednesday how to move forward. Until then, the edit 
> announced in the 0009 mail is on hold.
>
> So the options are
>
> 1) Have a non-normative reference to NIF, as suggested in the 0009 mail
>
>
>
> 2) Intent to have a standardized, that is normative RDF representation 
> of ITS2. This could then not be NIF. It could be
>
>
> 2a) something based on NIF, e.g. moving the six URIs that we rely on 
> (+ the ontology file?)
>
>          1. 
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#Context
>          2. 
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#RFC5147String
>          3. 
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#beginIndex
>          4. 
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#endIndex
>          5. 
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#referenceContext
>          6. 
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#isString
>          7. The ontology file that defines these URIs (= RDF classes +
>             properties)
> http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core/version-1.0/nif-core.ttl 
>
>
> into the W3C namespace and define the URIs + the ontology as normative 
> part of ITS2. But it could also be
>
>
> 2b) something completely different, yet to be defined. Issue 
> https://www.w3.org/International/multilingualweb/lt/track/issues/18 
> made clear that it cannot be RDFa.
>
>
> Above options are hard to evaluate since we have the EU funding based 
> timeline. But to move forward we need a working group opinion. Please 
> state your thoughts in this thread.
>
> Best,
>
> Felix
>

Received on Tuesday, 13 August 2013 11:24:06 UTC