ITS IG -- 11 Jun 2014

roll call

checking attendance

<Arle> \msg fsasaki Felix, I may not be able to pay attention for the entire call due to some urgent deadlines. If I do not respond to something I need to, please send me a Skype ping. Sorry about that.

<Tatiana> Good afternoon and greetings from Latvia! :)

<Tatiana> Cannot dial in and join you via IRC

hi Tatiana, still waiting for the call to be started by the organizer

http://lists.w3.org/Archives/Public/public-i18n-its-ig/2014Jun/0010.html

dave: Open Data Management for Public Automated Translation Services
... felix, I and others were at the LREC conference two weeks ago
... there was a workshop about infrastructure in language technology
... some things were academic, but otheres more practical
... andrjes Vasilijevs and Hans Uszkoreit talked about what the EU is planning to do in terms of concrete services
... one service will be automated machine translation
... the MLI project is gathering technical requirements for machine translation services
... I, felix and asun gomez perez talked to these guys after the workshop
... it seems that so far people have not looked at the data management side of things
... e.g. how to use open lexical resources
... and other items
... ITS does not have all solutions for this
... but some data categories can help, e.g. Terminology, Provenance, maybe even LQI
... I, Asun and Felix wrote a document
... https://www.w3.org/International/its/wiki/Open_Data_Management_for_Public_Automated_Translation_Services
... with high level requirements
... e.g. how to collect parallel texts, how to determin if the content has been created by MT etc.
... e.g. one use case being annotation of parallel text with ITS metdata
... we put this up on the ITS IG wiki to see if people have feedback
... and flesh out technical requirements, but also look into other groups like LD4LT and the LIDER project
... and then give that feedback to the MLI project
... any feedback so far?

serge: who is the driver of the MLI project?

<daveL> http://mli-project.eu/

dave: DCU is involved, coordinated by INMARK (based in SPAIN), various other companies and ELDA
... they started just to get going, since the EU had not yet announced CEF broadly

<daveL> https://www.w3.org/International/its/wiki/Open_Data_Management_for_Public_Automated_Translation_Services

[see partners list at http://mli-project.eu/?page_id=14 ]

dave: there is a lot of expertise on the language technology side of things and how to run e.g. cloud services
... there is also a call out to look into interop and standards
... it seems to assume that the services will be provided by the EU MT service
... from an interop point of view it sounds like it may be operated by the MT@EU project
... from an interop point of view it sounds like a good opportunity to promote ITS, linked data things, TBX / TMX as well

<Tatiana> Sorry, was away from keyboard for a few minutes: about MLi

<Tatiana> At the end of June there is a deliverable to be submitted by MLi

dave: andrejs Vasilijevs and Hans Uszkoreit said that the project is looking currently for input

serge: it is critical to include the data component in this

<Tatiana> MLi deliverable in June will recommend on the pilot services

serge: if the data component has been negleted we need to step in, make contributions

<Tatiana> I believe, there should be a tight collaboration between LIDER, MLi, and relevant initiatives within W3C

<daveL> +1

<Tatiana> Dave, yes, MLi now is gathering input from users

<daveL> https://www.w3.org/International/its/wiki/Open_Data_Management_for_Public_Automated_Translation_Services

felix: we are making the dicussion here in public also to get broad input from people not directly involved in MLi

<Tatiana> Thank you for this link, I will forward it to my colleague who is directly involved in the MLi work

<Tatiana> His task is to collect user requirements

<daveL> tatiana, what date would be good to aim for to submit this as a contribution to MLi after gathering further input, e.g. from GALA

serge: clear, we will contribute, and everybody should, see if things are missing etc.

<Tatiana> If it is OK with the ITS community, this doc might be shared with MLi (at Tilde, at least) and kept updating

<Tatiana> as soos as there is input from the ITS community

tatiana, sure, that is also why we do this in public - also, see dave's question on the timeline for feedback

christian: about the document you drafted

<Tatiana> In this way, both parties will be updated

christian: do you want to give that as a position statement of the ITS IG, or of the editors?
... that's a different aspect

serge: good point - should be an ITS IG position

christian: not necessarly - we need clarity - an IG position or compiled by members of the IG

serge: there is a lot of discussion now about European language cloud
... if we imagine in the future the European language cloud
... that should be the data model for ITS, MQM, industrial implementations
... that should be coordinated
... otherwise we will get a zoo of tools and mess with the data

dave: to answer christian:
... we put this up on the wiki, not proposing that this should be an ITS IG position statement
... happy to move that elsewhere if that is a conflict or out of scope
... the scope is broader than ITS, but I thought the ITS group is a good place to put it
... there is a lot of other people interested in this, also outside the ITS IG
... so if people want to contribute we can do that and e.g. add names of editors
... so would you feel happy to host the discussion here

christian: for me this is the ITS IG
... ITS1.0 and 2.0 fit well with automated translation services
... so discussions related to automated translation services should be mentioned here as well
... so bringing this into the group and discussing it here is great

discussion on where this should be discussed - ITS IG seems to be a good open forum for this

christian: if we want to have an official ITS IG statement - what would be the procedure ?

felix: for interest groups there are no formal voting procedures

david: if the IG wants to make collective statements the note has to go through the process

http://www.w3.org/TR/xml-i18n-bp/

www.w3.org/TR/

felix explaining the concept of a W3C note

serge: this group is very active and elaborate, working on ITS etc.
... if we formulate general terms - how do we see as a group the proper implementation of a language cloud, and publish this as a note

felix: a bit afraid that the time is not sufficient to discuss and publish a note

christian: would be helpful to have a more focused link
... I see link to MLi project page - but where to find background info that is in the doc
... is there such a targeted link, or are there workshop notes?

dave: we can get Andrjes MLi notes, which were fairly general
... there is EU information - I can dig out a link about the current call from the EU for feedback

<scribe> ACTION: daveL to find out what the timeline is for giving feedback (and felix too) to check if ITS IG would have time to produce an official W3C note [recorded in http://www.w3.org/2014/06/11-i18nits-minutes.html#action01]

<trackbot> Created ACTION-47 - Find out what the timeline is for giving feedback (and felix too) to check if its ig would have time to produce an official w3c note [on David Lewis - due 2014-06-18].

christian: CEF is a funding scheme from the EU
... they will ask to create several infrastructues, one is related to public automated translation services
... so you suggest to give some input of baseline requirements

dave: broadly, though I am not sure it will be a public procurement process
... the preparatory call says: the expection says that the MT@EU will be the basis
... it won't necessarily a fully public tender

felix: will schedule another call next week so that we can continue this and other topics

CSS attribute selectors

http://lists.w3.org/Archives/Public/public-i18n-its-ig/2014Jun/0001.html

http://www.w3.org/TR/selectors-nonelement-1/

jirka: is there an implementation already using CSS selectors?

felix: I am aware of bluegriffon IIRC, but that development is not moving forward it seems

jirka: the motivation of the draft is to use CSS selectors with ITS
... so it needs some implementations
... don't remember who originally proposed this, e.g. some implementation or a general comment

<scribe> ACTION: felix to check with Cocomore if they would be interested in implementing the CSS attribute selector [recorded in http://www.w3.org/2014/06/11-i18nits-minutes.html#action02]

<trackbot> Created ACTION-48 - Check with cocomore if they would be interested in implementing the css attribute selector [on Felix Sasaki - due 2014-06-18].

<Arle> Sorry, I need to go. Speak to you all later.

felix: may also be interesting for Logrus
... no timeline for the draft - without implementations it just will not move forward?

jirka: right
... they won't stop us from working on this due to some deadlines

Review of FEISGILTT outcomes

http://www.localizationworld.com/lwdub2014/feisgiltt/Program.pdf

david: we had a good working event
... there will be proceedings published via localization focus issue
... deadline is end of July for camera ready papers
... there was interesting discussions
... there was XLIFF symposium and linked data track
... most interesting thing was requirement gathering
... e.g. doing object model + json serialization for XLIFF
... that was nicely aligned with requirements on future of ITS that was discussed in Marid, MLW WS
... we also discussed ITS mapping into XLIFF
... there was some discussion of XLIFF module dependencies
... ITS support can be in several modules, since the support will be in several categories

<Tatiana> Have to leave already, sorry! Talk to you later and looking forward to reading the minutes. Bye! :)

david: once the have ITS on the XLIFF TC agenda we can discuss details

felix: we can discuss this next week again if that makes sense

david: about the linked data topic:
... several people said that licensing of linked data is an important topic is important

dave: also, several people said not so much publishing of linked data but modeling with linked data is important
... as we discussed via mail with Arle about MQM mappings
... RDF modeling works well to build broad consensus and have specializations in sub classes etc.
... need to move forward with people who don't have much experience with writing ontologies
... other important point is API discussion
... also looking into JSON-LD for the payload
... that relates to XLIFF / API discussion as well
... people said: a data model is great - but to use it you will need an API as well
... for different use cases, like in HTML5

david: people from XLIFF community got really interested in linked data agenda
... which was good
... was successful in bringing together the two communities and detecting themes and topics

dave: discussion and feedback went really well
... me and felix are writing up some notes on this as well

aob

will have another call next week, same time, to discuss at least three topics: ITS/MQM, Open Data Management for Public Automated Translation Services, and ITS/XLIFF

- DRAFT -

ITS IG

11 Jun 2014

Attendees

Contents

roll call

CSS attribute selectors

Review of FEISGILTT outcomes

aob

Summary of Action Items

Scribe.perl diagnostic output