Re: [xliff] The Terminology data category

Hi Yves, 

thanks for bringing this up again.

Am 12.11.2014 um 13:12 schrieb Yves Savourel <ysavourel@enlaso.com>:

> Hi all,
> 
> There are still a few un-resolved aspects for the Terminology data category.
> 
> --- 1) The case of ref and value
> 
> The ITS local attribute termInfoRef is mapped to ref.
> 
> The ITS global rule can also have a termInfoPointer that points to an actual content equivalent to XLIFF Term annotation's value. A
> global rule can also have a termInfoRef which is equivalent to the local termInfoRef, and a termInfoRefPointer that points to a
> reference. The three attributes cannot exist at the same time.
> 
> So:
> 
> a) When you map from ITS local rules:
> 
> - You can only map to ref
> 
> b) When you map from ITS global rule:
> 
> - termInfoPointer can be dereferenced and put into value
> - termInfoRef can map to ref
> - termInfoRefPointer can be dereferenced and put into ref
> 
> A side effect of this is that--when mapping from existing ITS rules--you should never end up with ref and value at the same time on
> an ITS Term annotation.
> That is not the case for a 'normal' XLIFF term annotation where both at the same time is not forbidden.
> 
> Now if you create term annotations in XLIFF and want to merge them back into the original document:
> 
> - If the XLIFF term has just a ref: you can map it back to a local termInfoRef.
> 
> - If the XLIFF term has a value: You cannot map it back to a local rule (there is no way to store the value's content it in ITS).
> But you can map it back to a global rule, to termInfoPointer. Note that it's not easy at all to merge back: You have to create an
> actual element/attribute to store the value and point to that element/attribute in a global rule that you have to create also.
> 
> The question is:
> 
> Should we forbid or allow ref and value to co-exist for an ITS Term annotation in XLIFF?

How is the interpretation of term annotation in XLIFF that includes both ref and and value? Are both concatenated? If yes, one could say: when mapping back to the original file, combine the information and create in the original file a place for the information.  

> 
> - If we do: You cannot map both back to the original file.
> 
> - If we do not: How do we make a distinction between a 'normal' Term annotation (which can have both) and an ITS Term annotation?
> The only way I can think of is that it's considered an ITS annotation if at least one of the following occurs:
> 	- The type attribute is 'itsm:term-no'
> 	- There is also an itsm:termConfidence in that annotation.
> 	- That annotation is within the scope of an itsm:annotatorsRef with an entry for 'terminology'.
> 
> I don't like the last condition because that annotatorsRef could be define on any ancestor and include term annotations that are not
> necessarily ITS ones. So you could make--unwittingly--many entries invalid by just adding an annotatorsRef at the top of the
> document.
> 
> I'm wondering if we should simply provide a reminder note stating that if an annotation has both ref and value and if you want to
> merge the data back into the original format, you have to choose which of ref or value to create in your document as both cannot
> co-exists in an ITS-only world. As for the ITS rules file: maybe we can have the rule check if both exist and pick arbitrarily the
> ref to map back.
> 
> 
> --- 2) Current issue with value definition
> 
> Currently in the section "5.9.4.1 ITS Terminology Annotation" if the 2.1 draft, we say:
> 
> "The value attribute is OPTIONAL and contains a short definition of the term that was obtained by dereferencing the [ITS] defined
> termInfoPointer"
> 
> I think tis is wrong. One can have ITS terminology annotation that are not create from the original file, but in XLIFF directly. So
> the definitions (not just this one) should take that into account in the wording.
> 
> I would suggest something like:
> 
> "The value attribute is OPTIONAL and contains a short definition of the term.
> If the annotation is created from the source document the value can be obtained by dereferencing the termInfoPointer of a global
> rule.“
> 
+1.

Best,

Felix

> 
> Cheers,
> -yves
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe from this mail list, you must leave the OASIS TC that 
> generates this mail.  Follow this link to all your TCs in OASIS at:
> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php 
>

Received on Wednesday, 12 November 2014 13:31:47 UTC