Dear all,
Please find below this week's meeting minutes.
Best,
Lucía
...........
Attendance: Rodolfo, Yoshito, Lucia, we have quorum.
Administration
R: I move to approve May 20, meeting minutes – https://groups.oasis-open.org/discussion/meeting-minutes-18
Y: I second.
R: Meeting minutes approved.
Technical
XLIFF 2.2. Committee Specification.
https://docs.oasis-open.org/XLIFF/XLIFF-core/v2.2/cs01/XLIFF-core-v2.2-cs01-part1.html
Promotion activities of the new version.
Webpage updates.
Rodolfo: I have received the review of the current content of the main page of the TC webpage made by Lucia, there are many changes that required almost changing the whole page. I will do that in the incoming days.
L: Thank you, Rodolfo.
Webinar.
Y: I have included information about some of the changes.
L: Thank you, Yoshito.
New translation memory standard.
R: The first thing we need to define is what a translation memory should do, and how we can use XLIFF to convey it. We can use fragments of XLIFF for that. It would be great if we provide an implementation and I can do that. Preparing a library for that. It would be opensource.
L: It sounds like a great idea.
Y: It sounds like a good idea.
R: it could have two parts: a specification, and a practical implementation.
Y: Inline tags are an issue. From the use of translation memory, for our case it is not important that these tags are kept. In our implementation. We need to consider these issues.
R: For translation recovery it helps a lot. Translation memory should be useful to retrieve useful matches, and the other use case I have in mind is concordance search, specially for multilingual documents. It helps to me to see, for example, to see how something was translated in Portuguese when translating into Spanish. For me it is important that tags are in XLIFF and not in TMX to avoid conversion.
Y: A third use is that traditional translation memory is segment by segment. But XLIFF can have units. And this could be useful for IA training, they require corpus that carries context information.
R: Yes, I agree. What do you think about starting working in github?
Y: As we will start analyzing it, we might realize that some tags are not needed. In many cases, inline elements are annoying. That kind of things. Of course, implementation should handle this type of stuff.
R: I will start defining what translation memory should be and how XLIFF could be used for that.
Y: This is an important decision. XLIFF is bilingual, but TMX can be multilingual.
R: Yes, we could use this format to use XLIFF to make a translation memory multilingual.
L: I think you said that you were going to have a look at TMX 2 that you were involved in the past.
R: The first change was to use Schema instead of DTD. TMX 2 would have to be rewritten to adopt the current XLIFF elements, so it would not be a feasible thing.
Y: there are tags that might be just noise as Yoshito said, and we might drop those.
L: We might need to update the charter to include this work.
R: sure, we can also update other aspects of the charte.
L: Roll call: do you all agree to include this information in the new proposal of the charter?
Yes: Rodolfo, Yoshito, Lucia.
L: Thank you, I will prepare a document with the changes in the charter and start the renewal process of the charter.
L: what about provenance metadata?
R: we can store that, it can be indeed very useful also for IA training.
Y: we are not changing the semantics, we would just be limiting for translation memory uses.
R: the XLIFF document will still be XLIFF, this will be something else for another purpose. We need one element that could glue information coming from different languages. It will not be an XLIFF document.
L: What about naming it, because translation memory is so closed to TMX?
R: we do not need a specific name.
L: But if we have a different root element, we are basically creating a new language.
R: it could be just memory or collection.
Y: it would not be a good idea to have a different entity that differentiates from XLIFF.
R: another idea would be to have several XLIFF files in a zip files.
Y: I think the important thing is that we do not create another TMX format.
R: Absolutely. We are just providing a way of using XLIFF for translation memory. I was thinking about creating a document explaining XLIFF as a translation memory. We should explain how different XLIFF could be regroup . Let me start writing and we can discuss it.
Y: I can also add my ideas to it.
R: that would be great.
L: no new business, meeting adjourned.
------------------------------
Lucía Morado Vázquez
Researcher and lecturer
University of Geneva
------------------------------