See also: IRC log
last meetings minutes approved
social media setup - what is the state?
david: social setup is managed between felix, arle and richard - no need to track the item
arle: like to see an ongoing action item for doing the social media outreach
david: arle proposed to regularly
review this - I'll change it to "ongoing social media
... new due date will be 18 May - title says "ongoing"
now open action items
felix: new due date for https://www.w3.org/International/multilingualweb/lt/track/actions/77
felix: ACTION-77 - due next Monday
dave: new deadline should be 20 May
dave: two parts - one what pedro
did, one suggestion I made what we meant about trigger
... had some feedback from Moritz about separating that from progress state
... two things remain: I'll update that section on that proposal
... so that we have separate data categories or values
... we need to decide: what are the values of processes?
david: you will do the update by Monday?
tadej: there was my theory that
the text analytics annotation would be a general
... I think there should be a superclass, but it should not be text analytics annotation, since that implies that this is automatic
... the general class would be to refer to an entity
... there is still an open question on the mtDisambiguation, the rest is quite clear
dave: it's OK to have the issues open, as long as they are in the notes
pedro: we can use very basic
... apart from the question "automatic generation or not"
... the difference for me is: are you tagging a part of the real content (terminology, named entity) or whether you are adding information, a kind of classificator
... you can add that additional information with a certain scope
... this is a very basic distinction
tadej: we have also domain and
genre, in the end they say the same thing
... they basically say "what is the topic of the document?"
... e.g. "advertising text"
... that might be the same piece of information for the MT disambiguation problem
pedro: two distinctions according
to the semantics - would be great to have feedback from piek
... you can have two types of information: subject or semantic keyword
... it is true that the domain metadata covers that
... we have to see if we are going to add semantic primitives appart from domain
dave: intention of domain data
category could cover the whole document than just a term
... we might want to keep domain separately
... but that wouldn't stop you for terminology to bind a domain declaration to some terminology
... otherwise we need a terminology domain
... e.g. a news report mentioning terminology for the medical domain
david: how does disambiguation data to genre?
dave: mt disambiguatino relates to terms I think, on the term level
tadej: Mtdisambigation is
supposed to mark up fragments of text with additional
information that would help me
... is there something about mtdisambiguation that is not covered by domain?
need to involve thomas from lucy and declan from DCU in the discussion
david: you mentioned people who
understand that from the MT point of view
... we can also solicit feedback from asia online
... this will be local text fragments for MT in realtime
... asia online could give good feedback on this
arle: I'll give another try on this
david: some people in the working group, tadej, can you reach out to people like thomas from lucy and declan from dcu, to get their feedback by e.g. next week?
tadej: yes, will do
arle: will contact asia online guy again
david: new deadline is next friday
dave: might be defered until we
have consolidated the data categories
... next week
dave: sent a mail out last night
dave: I listed data categories,
trying to keep the list up to date
... some are ITS 1.0 requirements, we have to do that
... next column: I recorded IDs of people who responded to the questionnaire
... terminlogoy, context and translate came up a lot
... in terms of difficulties, sometimes hard to see whether we understand the requirement and the use case
... or there is a consensus problem
... some technical knowledge sometimes hasn't emerged yet, e.g. mt disambiguation
david: so a lot of interest in
... process triger
... id value is difficult
... context is important, but very difficult
... apart from the initial post we haven't discussed
... your action item is completed, now everyone should have a look to review it
felix: propose an implementation commitment column
<Milan> BW: Disable (0 Mbit) 0 b/s (0 bytes) down, 0 b/s (0 bytes) up
<scribe> ACTION: daveL to add an implementation committment column to the table and ask people to populate it [recorded in http://www.w3.org/2012/05/04-mlw-lt-minutes.html#action01]
<trackbot> Created ACTION-82 - Add an implementation committment column to the table and ask people to populate it [on David Lewis - due 2012-05-11].
pedro: reference to provenance group - the related action belongs to dave
dave: yes, still working on
... there will be some decisions about what we will have in the document, what external etc.
... I'll try and address that before early next week
pedro: I have my name also in the
"autolanguage processing rule"
... need to clarify the purpose of this
<scribe> .. pending is also the ACTION-64
<trackbot> ACTION-64 -- David Lewis to discuss with pedro about providing feedback to best-practices group -- due 2012-05-08 -- OPEN
pedro: we only need to provide our input about best practices
dave: I'll follow up with you, I
talked to Richard about that, to get it published it needs a
kind of committment
... to get it published you need to be involved
pedro: won't have time, just FYI
felix: please keep me in CC, I am also in the i18n core group and can maybe give additional information
dave: we have a few weeks before
we send a first draft out
... I am prioritzing the consolidation
... there is other issues like populating the use cases, to make it more understandable
... some requirements are stubs, that haven't been elaborated on
... should we make more effort to make the document more readable
<Arle> Felix: There is no real rule about that. Think about what your most important audience is. Think that most of your readers will know what directionality is and it is enough to have a pointer to ITS 1.0 or HTML dir attribute, that is sufficient.
<Arle> .. If there are different audiences then it is harder and you have to explain everything. But normally the requirement documents explain their audience and target only them.
dave: we should be writing this to target people in the localization industry
<Arle> Felix: I think one value of the group is that we should make outreach to people where the localization industry is normally not involved. E.g., Dave will meet people from the Microsoft browser side of things. If you want HTML5 integration, it would help the localization industry.
<Arle> .. But you need to communicate it intelligibly to the browser community. It is not something for the requirements document, but for a general strategy we should involve people from the HTML working group.
<Arle> .. We should involve Richard, e.g., with his blog post on the translate attribute.
<Arle> .. It is OK to have the loc focus, but don't treat the fact that this is happening in the W3C as just a coincidence.
david: yes, it's an ultimate benefit for the downstream of the categories are taken up
dave: best way to do that is to
focus on explaining which sections focus on which product
... it is better to focus on getting somthing like arle's table so that they see which part is important for which audience
felix: agree, having a table with pointers to different audiences is a good approach
david: process diagrams will help too
arle: maybe on monday we can discuss that as well
felix: will continue the liaison discussion on Tuesday offline - sorry for my strong mail to the public list, will make "peace again" with David