Dear all,
Please find below today's meeting minutes.
Best,
Lucía
....
Attendance: Mihai, Yoshito, Rodolfo, Lucia. We have quorum.
Administration
R: I move to approve september 17, meeting minutes – https://groups.oasis-open.org/discussion/meeting-minutes-7
Y/L: I second.
R: Meeting minutes approved.
Technical
XLIFF 2.2 public review. https://groups.oasis-open.org/discussion/invitation-to-comment-on-xliff-v22-csd01-1
See comments that we received: : https://groups.oasis-open.org/communities/community-home/digestviewer?CommunityKey=f7b70a54-5dd7-4ea9-9d6f-018dce262ff9
R: We have received four comments so far. 3 comments were already answered, one is left. Mihai, can you handle as it is related to your module?
M: I am not sure if understanding correctly.
R: he is specifically asking for the plural module.
M: It is in the spec. We have other places with similar information. We do not have a machine readable format.
R: This topic has already been raised. I have already answered to the first question about machine readable related question.
M: for plural at least, there are constraints. The keywords come from an Unicode standard and we cannot promise that new terms are added in the future.
R: Yes, I see. I think you should answer it. (Rodolfo shows the specific question)
M: It is not well-known is invented. It is written for this document. I see, I will answer it after the call.
R: This one is critical. Because the public review ends tomorrow.
R: I understood your text. If we provided the answers and the public review ends tomorrow. We can contact Kelly to move on with the next phase. She might have to create a ballot, it will take a month.
R: We did not get any severe criticism.
New translation memory standard.
R: I had a conversation with A. Lommel, we were talking about AI and MT, and the big problem that MT has is that the lack of context. And I mentioned that we were discussing a TM format based on XLIFF. And it seems that it could be well accepted.
Y: we used translation models in the past. But with LLM, they might need more than the segment (context as you said). This is a kind of a chance for XLIFF, and my motivation to propose this kind of document level memory.
R: Your idea was welcomed by Arle. We could implement something like that.
M: Placeholders can also be an issue for MT. There is nothing like that in the training data.
R: That is right, the conversion from TMX always resolves into losing something when we talk about inline elements.
Y: we should start deciding the needs, what is missing and what we want to achieve with this new format.
R: That is very important. If you use TMX, it is very difficult to have incontext matches. With XLIFF that is not difficult. It will be interesting to do something about it.
M: One of the ideas was to start with something based on XLIFF.
R: Yes, that is right. We are talking something about memory. That is something that we could draft as we wait for the standard.
R: we can publish it as a note.
M: It would be interesting to see the differences between TMX and XLIFF. For example, in TMX you can have more than two languages.
R: An XLIFF file is bilingual. But if we talk about a memory container that uses XLIFF elements, we could have several languages.
Y: I do not think that multilingual is the key issue.
R: I use multilingual TMX all the time. Concordance searches in multilingual languages can be useful.
Y: For consistency for sibling languages and dialects. You might not retranslate completely. Handling multilingual languages, I can understand the user case.
M: Realistically, if the French translators needs to make the fixing, it is not going to make it in the TMX. It is a pain to have multiple people working on the same file.
R: You use the TMX file for exchanging the data not for working on the file itself.
M: I think we should start with the use cases. What you can do with XLIFF and TMX and what you cannot do.
R: With XLIFF you can do almost everything.
M: If we cannot make a compelling list of things that cannot be done with XLIFF, we can conclude that XLIFF can be used.
R: we can already do everything with XLIFF.
Promotion and Liaison
M: (Message format) It is still a technical draft. The standard might be ready by Spring.
L: No new business. Meeting adjourned.
------------------------------
Lucía Morado Vázquez
Researcher and lecturer
University of Geneva
------------------------------