MLW-LT WG

welcome naoto

Naoto: from Japan, living in Irland
... work for CNGL at UL
... working on MT-related aspect of localization process
... quality of translation, etc. are of interest

Work schedule

Felix: will publish in 22nd oc OCT
... last edit on Oct-19
... lot of travel before, let's see what we can do.
... any new update for Provenance?

Dave: Phil sent me a new copy
... will look at it
... may need to adjust the example for tool reference
... Phil and I will try to discuss that tomorrow
... and post feedback to the list

Felix: so when ready?

Dave: maybe thursday

Felix: hopefuilly can do it before Seattle

Dave: indeed

<scribe> ACTION: DaveL to coordinate with Phil and send new draft to the list by Thursday [recorded in http://www.w3.org/2012/10/08-mlw-lt-minutes.html#action01]

<trackbot> Created ACTION-246 - Coordinate with Phil and send new draft to the list by Thursday [on David Lewis - due 2012-10-15].

<scribe> ACTION: Felix to put the provenance draft in spec before Seattle [recorded in http://www.w3.org/2012/10/08-mlw-lt-minutes.html#action02]

<trackbot> Created ACTION-247 - Put the provenance draft in spec before Seattle [on Felix Sasaki - due 2012-10-15].

Felix: Disambiguation?

Tadej: working draft going around
... Arle needs to look at the definitions
... will circulate new version after
... Dave wants to look at it too

Dave: yes

Tadej: I can ciculate current version

Felix: not needed
... let's wait for the final draft
... can we do that this week?

Tadej: yes, but may get updated
... and there is the global tool reference to deal with

Felix: important to have those definition in darft
... maybe we can deal with the tool reference afterward with editors
... so we just need the next stable version

Tadej: then yes
... For text analysis annotation things will depends on the tool reference
... examples will be different

Dave: was trying to apply this to use cases
... what about how to control Disambiguation?
... we could have a lot of annotations, but even where we don't want them
... but we can't control where

Tadej: Cocomore talked about this
... because it wasn't important for use case
... maybe a switch would be useful?

Dave: Ok, but maybe it's a best practice
... suggest to discuiss this on the list

Tadej: switch make sense to expose

Dave: other things may be to concidered

tadej: sure, need examples then. Let's discuss it on the list

Felix: we are running out of time for new feature.

Dave: Phil, quickly: have an action to work on Provance

Felix: separation of HTML and XML

jirka has action-220

scribe: HTML users have specific use cases to take into account
... we may need the draft to be freeze to work on that

Jirka: working on the schema today
... hope to have something by end of week
... examples could stay there
... idea is to specific the difference (notation, etc.)
... case-sensitive or not between HTML or XML
... such section will explain the differences
... and the table with all the names will help

Felix: next week is fine too

Jirka: will try for this week.

Felix: next is text about Localization Implementers

<scribe> ACTION: Felix to put the Localization Implementer case in draft by end of the week [recorded in http://www.w3.org/2012/10/08-mlw-lt-minutes.html#action03]

<trackbot> Created ACTION-248 - Put the Localization Implementer case in draft by end of the week [on Felix Sasaki - due 2012-10-15].

Felix: will do the NIF section this week too
... editorial notes: some reviews to follow up with
... would be good to have co-editors to go through that
... Shaun can you look at that?

Shaun: yes

Felix: will keep you up to date on the edits we do

Domain in HTML5

Felix: Empty string?
... seems the outcome is decided

Shaun: then how do you override?
... seems we have no default

Jirka: allowedCharacters=".*"

Yves: will add that to specs

Domain in HTML5

<fsasaki> "Domain in HTML5"

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Oct/0057.html

<fsasaki> scribe: fsasaki

yves: problem was that in some cases we could point in some cases we could point to keyword list of domain
... in some cases we point only to one domain value
... need to separate that
... idea is to provide the result of data category as a comma separated list
... that is built from the result of XPath expressions
... when you have different nodes of a single string, you do the same with a substring
... felix wrote an algorithm, I did that as a pseudo code and implemented it

<Yves_> scribe: Yves_

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Oct/0054.html

Felix: link is the pseudo code

<fsasaki> "Get the list of nodes resulting ...."

Felix: there are some example too
... take 2mn to look at it
... any thoughts?

dF: should be working in XML
... looking now the HTML case
... not sure the issue of multiple keywords is resolved

Felix: idea here is to provide the list

<fsasaki> yves: goal is to get list of values

<fsasaki> .. not a single value

<fsasaki> .. after that you add to it

<fsasaki> david: domain should be one domain

<fsasaki> yves: but in HTML you can have several of them

<fsasaki> david: so are we happy with several keywords?

<fsasaki> yves: you can map them to one values

<fsasaki> felix: via domainMapping attribute

<fsasaki> david: so there is a method to parse multiple keywords and link a mapping

<fsasaki> .. so I can produce m:1 and m:n mapping

dF: ok, then I should be ok for n->1 or n->m mappings

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Oct/0054.html

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Oct/0057.html

Felix: this email above has some examples

df: we can only reduce?

Felix: yes

df: no sure about this
... seems that to be able to have several domain is a problem?
... should we be able to set priority

Dave: we talked about this, Declan noted it wasn't really possible to specify that
... difficult to have a set way to do this
... there are not really good practices for this

<Ankit> Yeah, in MT, usually a document will belong to one domain..

dF: there is an asymmetric aspect in the data category
... we may need to do that for XML too

Felix: you can already

<fsasaki> yves: you can point to the same type of keyowrds in xml too

<fsasaki> .. if you want to have a list of separated values

<fsasaki> felix: would have an xpath expression that would point to several nodes

dF: I see.
... should we agree n->1 is the best practice?
... m->n mapping complicate things

Felix: it depends on what you testing

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Oct/0057.html

Felix: see in example there where domainMapping is used

dF: see you can do conformance test
... but business case may be more tricky
... sevral domains may not be useable

Felix: so how to move forward?

dF: asking simply if the n->1 mapping is required

Dave: seems that it's more useful to know all domains than to redice them to one
... reducing is fine, but using all is fine as well
... mapping/reduction is complicated to do
... not sure we can solve this on the ITS side

Naoto: from MT viewpoint having one domain is simpler
... multiple doamins would be already in multiple documents
... different domains = different models

dF: think you can specify multiple domain in the same document

Felix: should continue this discussion in list

dF: we are using domain
... supporting multiple would be a big change
... not sure we can be conformant than

Shaun: sounds like issue is where the data comes from
... like HTML keywords
... maybe that could be a best practice: primary set first

and then MT can just take the first one

Felix: sound like a good solution
... best practice allows to use things like keywords then
... but need more discussion probably

<scribe> ACTION: Yves_ to add the Domain algorithm in draft asap [recorded in http://www.w3.org/2012/10/08-mlw-lt-minutes.html#action04]

<trackbot> Sorry, couldn't find Yves_. You can review and register nicknames at <http://www.w3.org/International/multilingualweb/lt/track/users>.

<scribe> ACTION: Yves to add the Domain algorithm in draft asap [recorded in http://www.w3.org/2012/10/08-mlw-lt-minutes.html#action05]

<trackbot> Created ACTION-249 - Add the Domain algorithm in draft asap [on Yves Savourel - due 2012-10-15].

AOB

Felix: any othe rtopic?

Phil: sent some question about toolRefs
... could we talk about it Dave? on Wednesday call

Dave: yes

Felix: I can help if you want, let me know.

Phil: late morning tomorrow might work for me.

Pedro: two questions:
... for unstable data categories: should we wait for implementation?
... and other question: are ready for readiness

Felix: please focus on data categories that are important, even if not stable
... for readiness, please go ahead

Dave: agree with felix

Pedro: ok

Felix: main call is done

<fsasaki> Jirka, we lost you on gotomeeting, it seems

<fsasaki> ACTION: felix to make sure that the its20 draft says the right thing in sec. 1.1.1 about direcitonality and ruby - if we make them non-backward compatibly [recorded in http://www.w3.org/2012/10/08-mlw-lt-minutes.html#action06]

<trackbot> Created ACTION-250 - Make sure that the its20 draft says the right thing in sec. 1.1.1 about direcitonality and ruby - if we make them non-backward compatibly [on Felix Sasaki - due 2012-10-15].

<fsasaki> http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Oct/0004.html