See also: IRC log
<fsasaki> http://www.w3.org/2012/10/TPAC/
<fsasaki> http://www.w3.org/2012/10/TPAC/
<fsasaki> http://www.w3.org/2012/10/TPAC/#Registration
16/10 October registration deadline
Individuals need to register themselves
f2f will be on Thursday/Friday
Next event is March, Rome
<fsasaki> http://www.w3.org/2011/12/mlw-lt-charter.html
Proposed date for next meeting week of January 23/24 2013 in Frankfurt or 30/31
January 29-31 2013 Frankfurt
Next regular call day and time change to accommodate Dave on Monday's 15:00 UTC
Desire to 'fix' tool issue this afternoon
After next call take time to edit the specification
16: 00 UTC live editing session
Editing sessions on Monday's after usual call
MLW-LT PC meet on Wednesday 14:00 UTC
Start Monday calls this coming Monday 1 October
Continue with demonstrations
Dominic and Dave presenting
daveL: CMS Interoperability
Thousands of API's
CMIS standardized API
Standard published in 2010. Members inc. Adobe
Many implementors
Models repository as document, folder and relationship objects
2 Requirements for l10n: (a) ITS rules; (b) Readiness
Can apply same rule(s) to several documents?
Implemented data model in draft
Polling scheme to define readiness updates and notifications
Plans: extend ITS Rule types, integrate xliff, discuss extensions with CMIS-compliant vendors
Readiness essential to implementation
fsasaki: ITS Readiness and CMIS and objectives of Pedro and mhellwig coming together with common needs
<fsasaki> fsasaki: no need to have these now in the ITS 2.0 standard, but please move this forward as a joint effort and we can advertise the outcome of this also outside of ITS 2.0
Opportunity for Pedro/Cocomore/TCD to work together
Video link of CMS Lion in the Use Case section of draft
Dave's presentation finished
Presentation by tadej
tadej: What elements of a docuent
should be treated specially
... focus mainly on html
... TA Annotation maybe superceded by tool info
... html->object model->analysis->add
annotations->html
<fsasaki> demo is here http://aidemo.ijs.si/mlw/
tadej: two markups. First very
verbose, second uses single agent-ref div
... can agent-ref be combined with 'tool info'?
aidemo has rdfa output option
fsasaki: can people take advantage of tadej's process?
tadej finished
philr: can enterprise specific domain models be built?
tadej: yes, in principle
... not hard to construct multilingual databases
Part of terminology process
<fsasaki> discussion on how to integrate enrycher into tools, e.g. terminology systems, translation tools. Confidence of disambiguation annotation not related per se to term "yes" / "no"
Des: any other multilingual databases?
end of demos
fsasaki: Arle will send out the slides that he has compiled of the demos, all please review and comment
<fsasaki> scribe: fsasaki
dom: go through action points and
issues related to test suite
... will go through matrix - currently has only Yves' name as
implementor
... will go names to the categories, and will work with people
who put the names
phil: question
... people who are using a data category
... do I have to produce tests for my tool?
dave: will explain that now
(presentation about test suite)
(related issues and ACTION: issue-33, action-139, action-145)
dave: goal of conformance testing
is that it is transparent and reliable
... that is needed in the w3c process
... but it also helps that other implementors can come and
check wether their implementation does the real thing
... that is the test suite in the w3c lingo
... we do that per feature and data category
... essentially what is test is the (ITS) parsers
phil: when I built a dom dynamically, I don't produce any output - what should I do?
dave: takes yves example - he has
a model that splits the test output out
... there is also a representational aspect:
... even if there is one data category, but if there is one
data category tested by many company, that gives an image of
output
phil: what about generators and consumers?
dave: the conformance is really saying "I understand all the requirements" of ITS conformance
phil: so if we don't have two tests what will happen?
dom: come to that in a minute
dave: validation is also helpful,
and use case demos
... these are two activities distinct to conformance
testing
... e.g. a nice use case doesn't mean that you contributed to
the conformance process
tadeJ: testing makes sure that everybody makes the data model right
dave: data categories are independent
http://www.w3.org/International/multilingualweb/lt/drafts/its20/its20.html#EX-idvalue-attribute-1
felix: example above shows what do to as an implementor for idValue: there is a MUST statement about xml:id, and the test checks that. so implementor need to decide: global and local, and then go into the data category (sub) section and check for the MUST statements, then implement the related tests
leroy explaining the test suite
leroy: output is a tab-delimited output, easier to compare than previous XML format
felix: test suite will also be helpful to explain ITS behaviour, e.g. with regards to inheritance
dom: we now need to get
commitments of implementors
... who is interested to contribute to the test suite for
going through the test suite commitments
<scribe> ACTION: felix to check what to do with directionality and ruby [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action01]
<trackbot> Created ACTION-228 - Check what to do with directionality and ruby [on Felix Sasaki - due 2012-10-03].
<scribe> ACTION: dom to circulate table for testing and everybody to fill in by friday next week , that is 4. October [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action02]
<trackbot> Sorry, couldn't find user - dom
<scribe> ACTION: Dominic to circulate table for testing and everybody to fill in by friday next week , that is 4. October [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action03]
<trackbot> Created ACTION-229 - Circulate table for testing and everybody to fill in by friday next week , that is 4. October [on Dominic Jones - due 2012-10-03].
CLOSING NOW:
<scribe> ACTION: leroy to put test suite on githup [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action04]
<trackbot> Created ACTION-230 - Put test suite on githup [on Leroy Finn - due 2012-10-03].
close issue-33
<trackbot> ISSUE-33 Test suite design closed
close action-139
<trackbot> ACTION-139 Check options for test suite design closed
close action-145
<trackbot> ACTION-145 Think about a round tripping test suite data package closed
close action-207
<trackbot> ACTION-207 Get some feedback on test suite input and output from HTML WG closed
<scribe> ACTION: leroy to create tests for its:param [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action05]
<trackbot> Created ACTION-231 - Create tests for its:param [on Leroy Finn - due 2012-10-03].
<Yves_> Scribe: Yves
<Yves_> First proposal : Logrus (Serge)
<Yves_> Serge: Idea is to be practical
<Yves_> .. we have many clients, different formats,
<Yves_> .. and have to deal with a zoo of tools
<Yves_> .. a lot of disconnect between tools
<Yves_> .. data is the key for those standards
<Yves_> .. we need to remember the human user is the key as well
<Yves_> .. tagged files are not WYSIWYG
<Yves_> .. users need context, and this issue is getting more and more important
<Yves_> .. fragmentation is leading to a need for more skilled users
<Yves_> .. HTML5 is a simple universal standard
<Yves_> .. idea is: Work In Context System (WICS)
<Yves_> .. human need context, and therefore need an help page, with text only
<Yves_> .. and without tags
<Yves_> .. can be prepared from the source (XML, XLIFF, etc.)
<Yves_> .. the output can be represented there
<Yves_> .. can also involve reference material, term base etc.
<omstefanov> What is speaker's name?
<Yves_> .. HTML5+ ITS + Javascript for human view
olaf, it is serge gladkoff
<Yves_> .. view can be viewed from any browser
<Yves_> .. 2 stage process: proprietary tagged file (with ITS), then HTML5 + ITS
<Yves_> .. helper page can be used at any time
<Yves_> .. for example: client reviewers
<Yves_> .. not being in context may cause issue for reviewers
<Yves_> .. helper page would be in a single format
<Yves_> .. this has several advantages like no WYSIWYG, more efficient, etc.
<Yves_> .. ITS data categories and that format: Terminology, Translate, Loc Note, Text Analysis, Disambiguation, External resource
<Yves_> serge then describes part of the roadmap details.
<Yves_> the result would be a working prototype of WICS generator
<Yves_> Tadej: so if I select a node i would get the annotation for that node?
<Yves_> Serge: that's the idea
<Yves_> .. the idea is tho improve productivity, so we show what helps the user to work fast
<Yves_> .. so the info needs to be determined, tailored
<Yves_> .. the important is to provide the info in visual form
<Yves_> Jan: stage1 seem to be where the issues would be
<Yves_> .. looks a lot like LocStudio a bit. Not easy to implement, but powerful
<Yves_> Serge: yes, that's what we split stage1 from stage2
<Yves_> Serge: this would be for a given client
<Yves_> Arle: what would be the output of stage1?
<Yves_> Serge: it would not be open source
<Yves_> .. output can be anything
<Yves_> Arle: XLIFF?
<Yves_> Serge: yes.. possibly
<Yves_> Felix: so you would go from XLIFF to HTML5, not trivial
<Yves_> Serge: not sure the middle would be XLIFF
<Yves_> dF: XLIFF would be good because there are already tools
<Yves_> Des: what about context? how do you get this? (style guide, etc.)
<Yves_> Serge: our understanding is that it can be set with ITS
<Yves_> Des: context is a big problem, so how it's propagated
<Yves_> .. if it's just text there is no context
<Yves_> Serge: not in this case, it's glossary, terms, etc.
<Yves_> dF: in XLIFF 2.0 there is a way to output HTML directly
<Yves_> .. providing full WYSIWYG is impossible, but basic stuff is.
<Yves_> .. so that preview feature in 2.0 is doing that.
<Yves_> Pedro: we tend to do automatic tasks.
<Yves_> .. human task is needed, but where and how
<Yves_> .. but not all metadata can be added automatically (e.g. loc note, etc.)
<Yves_> .. so intervention of the human is important to keep in mind
<Yves_> .. also translator are used to CAT tools
<Yves_> Dave: it's call a viewer, but can the user add things?
<Yves_> Serge: rules can be added to stage1
<Yves_> Dave: but then viewer is not doing this?
<Yves_> Serge: good question. But that's a too big of a scope it seems.
<Yves_> .. it's an HTML editor, and feedback loop is hard
<Yves_> Felix: would the tagged file +ITS would be valid ITS?
<Yves_> .. no I meant the final file HTML5.
<Yves_> Felix: stage1 is proprietary
<Yves_> .. any way to make it open-source?
<Yves_> Serge: part maybe, but not all
<Yves_> .. we'll pick specific client/format
<Yves_> .. the converter will do the conversion will be done by in-house tools, not linked to ITS/HTML5
<Yves_> .. if intermediate file is describe then it's ok, anyone can generate the same file and use the viewer
<Yves_> Felix: many are creating HTML5+ITS, can they use this viewer?
<Yves_> Serge: yes, they should
<Yves_> .. source can be HTML5
<Yves_> Pedro: thought stage1 is what we are doing
<Yves_> dF: important differences. here output is bilingual
<Yves_> Felix: here the output is very specific
<Yves_> des: here the HTML5 is not the original format
<Yves_> .. just a container
<Yves_> Pedro: the problem, is the middle file. we produce that
<Yves_> Serge: that's good then
<Yves_> Felix: the issue is people seems to do stage1 already
<Yves_> .. so we need only stage2
<Yves_> serge: the funded project is just stage2
<Yves_> felix: so cocomore, etc. would be input for the viewer
<Yves_> Serge: we would have to agree on the middle format
<Yves_> Pedro: if scope is stage2, if editing the ITS tags could be done it would be better.
<Yves_> dF: that would be a huge task: feeding back is difficult
<Yves_> serge: yes.
<Yves_> .. main goal is to show productivity gain through viewer
<Yves_> .. a big part is to see what is effective
<Yves_> Pedro: see this as very useful
<Yves_> .. users have to be trained
<Yves_> .. it take long to do this
<Yves_> Serge: helper page has many applications
<Yves_> .. we want a general example. not just a single case like Drupal.
<Yves_> dF: see the values through a condensed material
<Yves_> .. everybody is using the same references
<Yves_> Pedro: some data categories are editable is needed
<Yves_> Felix: those are two different use cases
<Yves_> .. stage2 built on top of what some of use do.
<Yves_> Serge: some client use only MT-based web site.
<Yves_> Felix: is stage2 valuable?
<Yves_> Jan: valuable for demonstration
<Yves_> .. not here today
<Yves_> Serge: if user can annotate, he may ament the glossary for example, and regenerate the helper page
I think Jan said: its valuable, as a helper page. The more enhanced usage (editing translations, making corrections etc.) is a different taks, more complex and not in stage 2
<Yves_> Serge: idea is to push ITS as possible
<Yves_> .. instead of using javascript
<Yves_> .. usability is important. e.g. no pop-ups
<Yves_> .. user needs context, color coding, no pop-up.
<Yves_> dF: pop-up cold be interactive
<Yves_> Tadej: using this one could see for example different meaning to do correction.
<Yves_> Felix: need to get a general opinion
<Yves_> .. anyone in favor or not.
<Yves_> Dave: strong feeling that seeing such HTML5 page would be useful
<Yves_> Jan: visualization of the ITS data.
<Yves_> .. it's useful
<Yves_> Arle: also an example of something tools can implement directly
<Yves_> Tatiana: like the idea.
<Yves_> .. context was one of the top requirement we saw in our reasearch
<Yves_> Felix: the tool information is important, we can talk about that.
<Yves_> Second Proposal: Init (Sebastian)
<Yves_> Sebastian: main focus of ]Init[ is government
<Yves_> .. many clients are multilingual (Europe languages)
<Yves_> .. we discovered ITS in May. it's a process.
<Yves_> .. our proposal is ITS in open/libre-office
<Yves_> .. we create documents in DE, then they are translated
<Yves_> .. but we have a break between the autor and the translator
<Yves_> .. data categories we'd like to use:
<Yves_> .. Translate
<Yves_> Locale Filter
<Yves_> .. Terminology (for candidate)
<Yves_> .. Localization Note
<Yves_> .. idea is to add button in O/L-Office
<Yves_> .. so ITS markup could be added in document
<Yves_> .. this would be a plugin, an extension of O/L-Office
<Yves_> .. author would add ITS tags in content
<Yves_> .. then we could save this into OD format.
<Yves_> .. we could also possibly extend Okpai to support ITS in ODF filter
<Yves_> .. for planning: development until Dec.
<Yves_> .. reporting experience after.
<Yves_> .. will need UI to do insertion of the ITS marker
<Yves_> .. save to XLIFF, and merge back
<Yves_> .. the UI would be in English (and German)
<Yves_> .. will not consider MT and Asian languages
<Yves_> .. it would be an open-source project
<Yves_> .. will do either open or Libre-office
<Yves_> .. possibly re-using okapi functionality
<Yves_> Jirka: so when I annotate file the data go outside.
<Yves_> .. seems that the project should be able to save the ITS data in the OD file
<Yves_> .. extensibility too
<Yves_> .. is available
<Yves_> Shawn: interested too. if tags would be in ODF then i can use it too.
<Yves_> Pedro: using two files would be difficult
<Yves_> .. agree that data should be in ODF
<Yves_> dF: yes, ITS markup would need to go in ODF
<Yves_> dF: don't see a reason to have extraction from O/L-Office
<Yves_> Sebastian: ok, I'm taking the feedback.
<Yves_> Jirka: you can use custom attribute in ODF with some restrictions
<Yves_> .. those may be lost if ODF application does not support those extension
<Yves_> Sebatian: you think extension mechanism will allow for this?
<Yves_> Jirka: I do not know.
<Yves_> Sebastian: need to know between Libre or Open Office
<Yves_> Shawn: LibreOffice is a fork of OpenOffice. Then OpenOffice becaome part of Apache Software
<Yves_> .. LibreOffice is an out-of-the-box application, OpenOffice now is a "platform" developers can add to.
<Yves_> dF: what is the overlap?
<Yves_> Shawn: it was a full fork, both sides continued to merge from each others
<Yves_> .. on occasion.
<Yves_> jan: what is the one you use
<Yves_> Sebastian: we are using openOffice 2.1
<Yves_> .. but use LibreOffice for other clients
<Yves_> Jan: seems LibreOffice would be the choice then.
<Yves_> .. other platform could be extended like this too. here the idea is to demostrate ITS usage.
<Yves_> Jirka: maybe the LibreOffice could save as OOXML
<Yves_> .. could speedup adoption
<Yves_> Sebastian: for XLIFF, 1.2 or 2.0?
<Yves_> Felix: if you relay on Okapi, then use what is there.
<Yves_> Third Presentation: Tilde (Tatiana)
<Yves_> Tatiana: main business LSP, terminology services, Lingual technologies, and open Linguistic Infrastructure (Tilde META-SHARE node, META-NORD)
<Yves_> .. refocusing as Terminology as a service (TaaS)
<Yves_> .. now developing web-based services, MT, indexing, annotation, etc.
<Yves_> .. our first proposal was very extensive
<Yves_> .. outr goal now is to filter and understand the requirement of the LT-Web project
<Yves_> .. see that LT-Web need to be validated
<Yves_> .. our idea is to help in usability, interoperability, productivity
<Yves_> .. several use cases:
<Yves_> First: Simple Terminology Annotation
<Yves_> Data categories: Language Information, Locale Filter, Terminology
<Yves_> .. possibly a confidence attribute
<Yves_> .. unsupervised identification of term candidates
<Yves_> .. in HTML5, then presented to users 9highlighted)
<Yves_> .. third aspect is a web service with a web application
<Pedro> scribe: Pedro
Proposal 2: Enhanced terminology annotation and simple machine translation
Includes: term recognition, and
existing terminology resources.
... annotate terminology and interactive interface
... support consumption by machine translation
... for existing language resources there is a problem to find
the way to reuse and share them
... ITS DC Considered: Language information, locale filter,
termionology and translate
... benefits: production and consumption of ITS
... one of the existing resource can be used in the 2nd
proposal is eurotermbank
... largest terminology bank in Europe
... more benefits: reusing existing resources like this
bank
,. Proposal 3: Enhanced Termonology Annotation and Enhaced MAchine TRanslation
scribe: Includes the two previous
proposals, extend the MT enhancement
... .. ITS DC added Element whithin Text
... Additional benefits: enhance translation quality
Des: existing termionlogical resources are used for disambiguation tasks?
Tatiana: yes, it is one of the purposes
Tadej: it also to apply to terms?
Tatiana: yes, we use terminololgy datacategory and we explore disambiguation and some other
Tadej: terminology datacategory is not modelled for candidates, we should revise.
Des: candidate can be domain specific.
Tadej: to use confidence for this is interpretation.
Felix: this is about term
annotation and not entity annotation, and there is some
overlap
... but there is a diiference annotation workflow
Tadej: yes
Segey: There is any kind of API
to access existing resources?
... In the case os EU bank?
<Arle> scribe: Pedro
Tatiana: yes, it is a resource senstive scenario, does not take any glossary.
Sergey: why? there are many other valid glossaries.
Tatiana: maybe with teh same architecture we can in short time allow to compile other new resources
FElix: in Enrycher tadej has this approach.
Tadej: yes
Sergey: It is an open source application?
Tatiana: now it is under discusion, but it will be a service.
Felix: independently is open source it can be a service, like Drupal does
Tatiana: in the busienss model it iwll be a service level in the future, of course
Felix: Perhaps the best user strategy can be discussed.
Tatiana: shchedule Proposal 1: 6 months, Proposal 2 and 3: to be done in 1 year
Sergey: which languages?
Tatiana: English as the source
language.
... with rich morphology for inflection, etc.
... and other 2 languages with all the morphology. In the
future it will be able to add other languages.
Felix: additional confidence
attributes were discussed, and need to be stable
... for this prototyping this will help.
... but do you plan so be able to see before 6 months?
Tatiana: we have to see, but yes.
Felix: opinions about these three proposals?
<fsasaki> scribe: fsasaki
pedro: three different proposals
(Logrus, ]init[, Tilde)
... all final users implied
... this takes care about the "human side" of the usage of the
metadata
... so the Tilde scenario will help users to annotate metadata
in the content
declan: also for MT it might be useful - monolingual or multilingual terminology information
pedro: if we have time, the groups that we are talking to (translators, terminologists, testing)
tatiana: we can also leverage our localization and terminology partners
<Pedro> Pedro scribe/
<scribe> scribe: pedro
Tatiana: we will not perform all the 3 proposal, but to get your vision
,, and to select the one applies better
Felix: Does it is possible to modify Proposal 1
Sergey: do not understimate Proposal 1
dF: it is different in the different languages
Tataina: maybe the terminology
use case will be richer if we combine part of proposal
... but we do not want only to produce metadata but to consume
it
... terminology per se is not enough
Felix (coordinator): Maybe to collaborate with other exinting partners
scribe: we have to see if it is
Matrix or Lucy, or another one.
... we have to think about it.
... they can make experiences
... We need to have the feeling if it is better proposal 1, 2
or 3
Tatiana: we have to decide and going back to you
Jim: In terms of preferences I will stack in this orden: 1, 2,3
Sorry, wasnt Jim, but Jan
Declan: in proposal 3 you cover more
Felix: Of course, but the budget
<fsasaki> scribe: pedro
<fsasaki> scribe: felix
<fsasaki> pedro: we plan to finish most of the stuff middle next year
<fsasaki> .. so that we have time for showbusiness
<fsasaki> tatiana: the 3rd one would be one year, the others shorter
<scribe> scribe: felix
<scribe> scribe: pedro
Tatiana: a possibility is a mixture
FElix: we can discuss
later.
... there are now two other things before Coffee breack
... the current proposals are on the table and no time for
looking other
... decissions will be there in a short time:around 2 weeks
dF: we need the economical proposal details
Felix; yes, from Init we have, but we must have also a budget breakdown from Logrus and Tilde
scribe: Having it by next Monday is fine
Coffe break
<scribe> scribe: pedro
<fsasaki> https://www.w3.org/International/multilingualweb/lt/track/issues/open
<fsasaki> https://www.w3.org/International/multilingualweb/lt/track/issues/open
<fsasaki> scribe: dave
<fsasaki> scribe: daveL
yves: for many data categories,
know the tool is useful and for some it is really
importnat
... but the problem is some data cats operate best at a local
level, where the tool ref is a big overhead
... felix proposed a standoff format detailing the tool
... but this made it difficult to reference the tool.
... another approach is using xml:id, but this could clash with
the many other usages of xml:id.
Yves so an alternative proposal is a tool reference that bind the data category id
scribe: that could be used to
overrid the tool binding, but not the referneced data category
at a local level
... but this still needs to point to the standoff tool
specification file
<fsasaki> external tool information is here: http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Sep/att-0189/tool-its-info-example.xml
scribe: but this could be done by replacing the textual tool ID bound to the data category to tool with a URI
df what if you have two MT in the document
yves, you just define that locally to differentiate
shaun - could we allow tool refs for data categories outside of the ITS one
yves - yes , perhaps with a name space like prefix to data category extension
David and Tadej indicate this addresses thier tool identfication needs
pedro: asks whether this can help with evaluation of MT output
declan: those aren't confidence scores
Des; asks about updating the tool id
felix: asks also how to do this
in HTML5
... possibly using a similar mechanism to quality issue using
its:span
jirka: could also put it in a script element
<fsasaki> ACTION: shaun to create a list of canonical data category names [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action06]
<trackbot> Created ACTION-232 - Create a list of canonical data category names [on Shaun McCance - due 2012-10-03].
shaun: this could be a RDF link, but this requires an RDF tool chain rather than an XML one
jirka: says that using script is the only way currently of including XML in HTML
felix: shows how to do equivalent in the quality issue example
yves: seems like a hack
jirka: possible, but it whats possible in html currently.
<Jirka> https://developer.mozilla.org/en-US/docs/DOM/DOMParser
felix: current recommendation is to put the tool info xml into script in html
<Yves_> var doc = parser.parseFromString(stringContainingXMLSource, "application/xml");
<Jirka> var parser = new DOMParser();
<Jirka> var doc = parser.parseFromString(stringContainingXMLSource, "application/xml");
<Jirka> // returns a Document, but not a SVGDocument nor a HTMLDocument
felix: should we do the same in quality issue example
yves: agrees
<scribe> ACTION: felix to update quality issue to use the same solution [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action07]
<trackbot> Created ACTION-233 - Update quality issue to use the same solution [on Felix Sasaki - due 2012-10-03].
felix: so how do we manage conformance?
yves: would need to specific for appropriate data category that they MUST support tool info
dF: agrees its a MUST for MTconfidence, but that does need to be universal?
felix: explains the current
conformance rules and suggest that is an application supports
mt confidence MUST support tool info and other MAY support
it
... this doesn't need a conformance test, but we should do this
to ensure ease of conformance for this option
df: but is it is optional, then it wont get implemented
tadej: its option for disambiguation, since this could be manually generated, so tool is not necessarily importNT HERE
dave: also an option for for proveance agent data categories
df: so should it be left option or made specific to mt confidence score
felix: but its still useful for other, just not mandatory, but would loose that if it was specific to mtconfidence score
tadej: the tool info is important
for text analytics
... so you can give the tool info even without the confidence
scoe
jan: asks how confident gorup is with confidence score being only valid inter-segment per engine rather than between engines
declan: had been discussed but that is currently what we can to and we wanted to keep it simple
<Jirka> if <script> is too much for us, we can also introduce new element like its-tool-info and define it in an applicable specification. As long as element will not have any text content it will not create any rendering and processing problems to browsers.
shaun: but could we just make the confidence a tuple of mt identifier concatenated with confeince scoe
yves: but this will still be long, so we loose a node by don't save characters, engine ids will still be verbose
dave: could be used for specifying person repsonsbile for generating data category also
<scribe> ACTION: yves to document the tool info [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action08]
<trackbot> Created ACTION-234 - Document the tool info [on Yves Savourel - due 2012-10-03].
action daveL to talk to phil about impact on quality and provenance
<trackbot> Created ACTION-235 - Talk to phil about impact on quality and provenance [on David Lewis - due 2012-10-03].
<scribe> ACTION: tadej to consider this for text analysis [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action09]
<trackbot> Created ACTION-236 - Consider this for text analysis [on Tadej Štajner - due 2012-10-03].
felix: to summarise this needs extra mechansim, addition to conformance sequence, and addition to test suite
all: it was all felix's fault
<scribe> ACTION: felix to check if we can have two and half days to have an implementors meeting include since we missed it today [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action10]
<trackbot> Created ACTION-237 - Check if we can have two and half days to have an implementors meeting include since we missed it today [on Felix Sasaki - due 2012-10-03].
<fsasaki> ACTION: felix to make sure that the beginning of the lyon meeting is reserved for developers [recorded in http://www.w3.org/2012/09/26-mlw-lt-minutes.html#action11]
<trackbot> Created ACTION-238 - Make sure that the beginning of the lyon meeting is reserved for developers [on Felix Sasaki - due 2012-10-03].