Re: [ISSUE-22] Provenance Revision Agent

Hi Dave, Yves, all,

I agree with Yves' comments wrt to global rules. If you need to express
complex information, it makes more sense to formulate that in a standoff
manner like with localization quality issue  or the "tool info" proposal,
see
http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/0137.html

I think also we had a discussion about pointer attributes before, see
"pointer" mentioned at
http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Jul/0276.html

- Felix

2012/9/21 Yves Savourel <ysavourel@enlaso.com>

> Hi Dave, all,
>
> Just a few notes on the updated proposal:
>
> -a) absolute path:
>
> In the first example you us its:transRevPersonPointer=”//dc.creator/”
> But that is not a path relative to selector="/html/body/par"
> The same is true in the EX-trans-rev-prov-rules.xml file for the HTML5
> example.
>
>
> -b) DC namespace:
>
> xmlns:dc="http://purl.org/dc/elements/1.1 in the first example is part of
> the content, it should be an attribute of <text>.
>
>
> -c) using rules for annotations:
>
> It also seems that some of those examples, and others in the specification
> (including some of mine probably), are a bit unrealistic from a real-life
> processing viewpoint.
>
> The idea that we would produce global rules (external or embedded) that
> use selectors to annotate a XML/HTML5 document seem not very practical. As
> soon as the document changes (for example a new <p> element is added above
> the existing one in this examples) the selector points to the wrong
> paragraph.
> In other words, using rules to annotate works if the document is
> 'read-only'.
>
> The main function of global rules, IMO, is to define stable information
> about a document, or even better, a type of document.
> We started to overload this with Localization Note, where we used rules
> for non-local annotations.
> And now in 2.0 where we have many data categories that are annotations, it
> seems we routinely assume that annotating a document using global rules and
> hard coded selectors is just fine. But I think it's often not a viable
> scenario in real life.
> IMO annotations should be local or standoff (a local attribute points to
> the annotation).
>
> My concern is that people from the HTML WG, and other reviewers, will look
> at this and wonder about the applicability of ITS.
>
> Cheers,
> -yves
>
>
>
> -----Original Message-----
> From: Dave Lewis [mailto:dave.lewis@cs.tcd.ie]
> Sent: Thursday, September 20, 2012 6:43 PM
> To: public-multilingualweb-lt@w3.org
> Subject: Re: [ISSUE-22] Provenance Revision Agent
>
> Apoligies, I forwarded the wrong version of this data category, please use
> the attached instead.
> Dave
>
> On 21/09/2012 01:29, Dave Lewis wrote:
> > Hi all,
> > Please find attached a revised proposal for the translation revision
> > agent provenance data category. Based on the feedback form the
> > previous call for concensus
> > (http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Jul
> > /0256.html) and discussions with David I've made the following
> > changes:
> >
> > - included the pointer and refPointer variants consistent with other
> > data categories
> > - to split the the global rule between transation agent proveance and
> > translation revision provence, I ended up actually splitting this into
> > two data categories. This post contains the translation_revision_
> > agent provenance. I'll send on the other, the translation agent
> > provenance shorlty, though this will follow the same pattern
> > - I allow only one value rather than multiple ones for each type of
> > agent and explicitly include attribute so the three types can be
> > differentiated - so the data types can be interpreted unambiguously
> > - slimmed down the description
> >
> > comments welcome. One open question is whether the naming of
> > 'translationRevision' (or 'transRev') could be better named 'postedit'
> > (or 'pe')?
> >
> > cheers,
> >
> > Dave
> >
>
>
>
>
>


-- 
Felix Sasaki
DFKI / W3C Fellow