See also: IRC log
<Yves_> Sorry: I'm in a call. I'll be late joining the GTM session
presence: phil, arle, yves, christian, felix
comments 12 - change of "enabled flag"
felix: no need to discuss comments that mean change of markup (attributes, attribute names, values) - we can't do this for ITS2. You can defer such commments right away to ITS2.next
comment 15 - example of (in) accurate mapping - arle will provide that
comment 16 - the "must" needs to be a "must not"
comment 17 - plural for "informative mappings"
table for LQI information: was covered during last call. Arle will then re-write the "profile" row. The "enabled" information will be kept (see comment 12)
arle: comment one in lqi types
doc on order of the values is significant
... we want to leave them as is
... we had realized that some issues could be categorized as
multiple things
... so we had a principle that would guide selection
<chriLi> Sorry, my bandwith for GoToMeeting is too limited. It's even unstable for IRC.
[order of values] discussion above is related to this doc
arle: comment 2 - how to identify target and source for "ommission"
phil: tend it to be target only
felix: changing it to "target
only" would not influence implementations, so I think such a
change is ok
... but you as implementers need to be convinced of that :)
arle: propose to make this "target (t)" only
yves: the problem is here:
... when we do alignment - e.g. aligning things in which there
is only a source, no target
... or there is an addition to ... nothing?
arle: from the alignment perspective it is different - so you would prefer to leave the target *and* source?
yves: I see errors related to alignment that could need the "source"
phil: ok to me - so have this as target and source, and have examples, for both source and target
yves will write an example for the source usage, arle for the target
comment 3
arle: would mean a markup change, so leave this as is
comment 4
<Arle> • A text translated into German systematically uses û, ô, and â instead of the appropriate 'ü', 'ö', and 'ä'.
<Arle> • A text translated into German erroneously uses û, ô, and â instead of the appropriate 'ü', 'ö', and 'ä'.
<chriLi> The observation could also be made for the source language.
<Arle> • A German text erroneously uses û, ô, and â instead of the appropriate 'ü', 'ö', and 'ä'.
<chriLi> Thus, the "A text translated into ..." possibly could be changed to "Characters in the source or target language are not appropriate (e.g. due to an encoding issue)".
<Arle> Christian, for your comment number 5, we can't change the name of the markup at this stage. Are you OK with "formatting" instead of "rendering"?
<chriLi> Could/should we add a comment along the lines of "Strictly speaking, 'rendering' is not the appropriate term. It has been chosen to ..."?
<Arle> Proposal for issue #4: A German text erroneously uses û, ô, and â instead of the appropriate 'ü', 'ö', and 'ä'.
<chriLi> +1
<Arle> Christian, I thought you were asking for changing *to* "rendering"
<Arle> "Rendering" doesn't occur in the text. I'm confused.
<Arle> And "formatting" is what this is called in general language.
comment 5, see above. Needs to be looked at again
<chriLi> I was still at the "characters" value. Sorry for the confusion.
now looking at comment 6
<Arle> #6: Some tools may correct for differences in units of measurement to reduce false positives. -> Some tools may correct for differences in units of measurement to reduce false positives (e.g., a tool might adjust for differences in values between inches and centimeters to avoid flagging numbers that seem to be different but are equivalent).
<Arle> #6: Some tools may correct for differences in units of measurement to reduce false positives. -> Some tools may correct for differences in units of measurement to reduce false positives (e.g., a tool might adjust for differences in values between inches and centimeters to avoid flagging numbers that seem to be different but are in fact equivalent).
[at Arle & all: ommission in the source example - see comment 2 of lqi types doc - is now at http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Aug/0043.html ]
now comment 7
<Arle> We will need to ask for Christian for the source of confusion on number 7.
looking at no 7 again
discussion on what the problem with "pattern-problem" is
<Arle> Would something like this note work? “A simple example is the use of a regular expression to detect inappropriately repeated words.”
<chriLi> Could we a URL-related example?
<Arle> Christian, we can create such an example and add it. We would also add a note that the patterns are up to the implementing application and beyond the scope of this spec.
felix: for the wednesday call Arle will go through the word documents that Christian had sent out http://lists.w3.org/Archives/Public/public-multilingualweb-lt-comments/2013Aug/0000.html and make changes with change marks. we will go through these during the call
agreement on this approach
comment 8 - non-conformnance
<Arle> “A system may build a statistical profile of a reference corpus and a new text and use it to note dissimilarities between those profiles. In the event that the new text seems to be out of expected bounds, it may be flagged.”
<Arle> (As a note)
<chriLi> What would be a "statistical profile of a reference corpus"?
philr: not sure what part of the concept is hard to grasp for people
arle: my initial difficulty was:
I was not sure what (non)conformance against a corpus is
... because I lack the background
philr: I will try to re-word this to make it clearer
comment 9 - about "other" vs "uncategorized"
arle: both are needed
values
... this might require a note to the table as a whole to
explain why we have the two
christian: couple of dependencies
between the initial values
... e.g. saying "if you are unsure use the following value: .."
is always tricky
<Arle> Add a note to the table to explain when to use one and the other.
christian: so making very clear when to use uncategorized and when to use other would help
now looking through the table at the bottom of the lqi types comments doc
christian: this is the outcome of
a small survey about the understandibility of the
categories
... and the received importance of the categories
... can be read as: "terminology" is most easy to
understand
... "pattern-problem" was not easy to understand
... the people who gave input to the table had a look at the
whole text in the lqi types section
arle: we will go through this
table on the Wednesday call. For the call I will make the
changes before the call in the word doc, to discuss them
... thanks to all for taking the time, adjourned for now
<Yves_> bye, thanks.
This is scribe.perl Revision: 1.138 of Date: 2013-04-25 13:59:11 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/order of the values/comment one in lqi types doc on order of the values/ Succeeded: s/would/arle: would/ Succeeded: s/going through comments/going through comments in lqi main section/ Succeeded: s/needed/needed values/ Succeeded: s/doc/lqi types comments doc/ No ScribeNick specified. Guessing ScribeNick: fsasaki Inferring Scribes: fsasaki Present: phil arle yves christian felix Agenda: http://lists.w3.org/Archives/Public/public-multilingualweb-lt-comments/2013Aug/0000.html Got date from IRC log name: 19 Aug 2013 Guessing minutes URL: http://www.w3.org/2013/08/19-mlw-lt-minutes.html People with action items:[End of scribe.perl diagnostic output]