RE: Issue-55: XLIFF mapping - Terminology and termInfoPointer

Hi David,

The current mapping is nice. but when you try to actually implement it you run into many issues.
So I'm just trying a lot of different things right now. Not saying we must use one or the other.

For example: using the comment attribute for termInfoRef or the result of termInfoPointer seems fine, but how do you know the difference when going from XLIFF to ITS? How actually can you map from a plain text comment to ITS locally? (do we actually need to have a local its:locInfo in ITS after all)?

> UL and TCD are working with
> <mrk mtype="term"> (and comment for inforef) as per the mapping agreed here:
> http://www.w3.org/International/multilingualweb/lt/wiki/XLIFF_Mapping
> This also says that mrk should be used as well at the structural level, which 
> makes the solution uniform. [Also terminology at structural level does not 
> seem to be frequent enough to warrant a different solution].

For round-trips moving the information to an inline element is a bit complicated:

For HTML you would go from:

<p its-term='yes''>text</p>

to 

<p><span its-term='yes'>text</span></p>

It's doable, but changes the content and comes with a lot of side effects: new element has impact on TM matches later, difference when comparing original and translations, etc.

For XML is even more complicated:

<myNS:para its:term='yes'>text</myNS:para>

to... what?

<myNS:para><its:span term='yes'>text<its:span></myNS:para> How do you know you can have an its:span there?

<myNS:para><myNS:span its:term='yes'>text<myNS:span></myNS:para> How do you know what element to use for the inline markup?

Actually, this is a general issue for XLIFF to XML+ITS: if the original ITS markup is using global rules to map elements to ITS data categories, it'll be very hard to create back an XML document that is valid. one would have to provide the mapping backsomehow.

cheers,
-yves


> I see that you are using both okp:itsTermInfo and note to 
> convey the term definition. While it seems that duplicating the info 
> in a core element makes the information more likely to survive. 
> I do not think that we should recommend using the duplicity.

The note is the read-only way we represented terms so far. I'm just keeping there for now. It may survive as an option for backward compatibility (how many CAT tool understand <mrk mtype='term'> today?). 


-----Original Message-----
From: Dr. David Filip [mailto:David.Filip@ul.ie] 
Sent: Wednesday, February 20, 2013 4:55 AM
To: Yves Savourel
Cc: public-multilingualweb-lt@w3.org
Subject: Re: Issue-55: XLIFF mapping - Terminology and termInfoPointer

Yves,

UL and TCD are working with
<mrk mtype="term"> (and comment for inforef) as per the mapping agreed here:
http://www.w3.org/International/multilingualweb/lt/wiki/XLIFF_Mapping
This also says that mrk should be used as well at the structural level, which makes the solution uniform. [Also terminology at structural level does not seem to be frequent enough to warrant a different solution].

In the mappings you provide in the samples you use mrk mtype="its-x"
combined with its:term="yes" [combination of a user defined type with an custom namespace attribute] which seems less interoperable with general XLIFF editors and more complicated.

I see that you are using both okp:itsTermInfo and note to convey the term definition. While it seems that duplicating the info in a core element makes the information more likely to survive. I do not think that we should recommend using the duplicity.
I guess that we should decide EITHER for note OR mrk comment [my preference] from the core repertoire and not introduce a custom attribute..
Eventually we could say that mrk comment is to be used as canonical mapping, while the duplicity in note is optional.

Rgds
dF



Dr. David Filip
=======================
LRC | CNGL | LT-Web | CSIS
University of Limerick, Ireland
telephone: +353-6120-2781
cellphone: +353-86-0222-158
facsimile: +353-6120-2734
mailto: david.filip@ul.ie


On Wed, Feb 20, 2013 at 12:31 AM, Yves Savourel <ysavourel@enlaso.com> wrote:
> Hi Felix,
>
>> just to understand: is this a question that might lead to a change in 
>> ITS2 or only in the mapping approach?
>
> No, just the mapping I think.
> Except if there is a need to have some local attribute its:termInfo in ITS, but apparently there was none so far since ITS 1.0.
>
> As for the mapping: it makes things a bit more complicated, but doable (I think).
>
> Mārcis, Dave, David: I've updated the XLIFF output of the test files for Terminology with a tentative mapping:
> https://github.com/finnle/ITS-2.0-Testsuite/tree/master/its2.0/xliffsa
> mples/inputdata/terminology
> Any thoughts?
>
> Thanks,
> -yves
>
>

Received on Wednesday, 20 February 2013 12:26:33 UTC