See also: IRC log
<fsasaki> agenda at http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2012Nov/0081.html
<trackbot> ACTION-263 -- David Filip to summarize the options and the recommendations related to HTML parsing workflow in the CMS, see discussion at http://www.w3.org/2012/11/01-mlw-lt-irc#T10-13-41 -- due 2012-11-08 -- OPEN
Felix: This was discussed in face-to-face in Lyons. David, do you have an update or will you get to it?
David: Yes, I finished this week.
<trackbot> ACTION-281 -- David Lewis to ask for use cases of data category-specific confidence scores -- due 2012-11-12 -- OPEN
Felix: Dave, do you have an update on ACTION 281?
Dave: After discussion with ???,
I don't think the text analytic annotation works by itself. If
you apply the same annotation to a span for different purposes,
we run into overwrite problems. Flags don't help unless you got
... The other suggestion, made last week, is that MT confidence is fundamentally different. MT should not be merged with the others.
... Should we consider having a confidence score in those data categories? E.g., for terminology, disambiguation, domain. Although domain is very complex.
... Disambiguation and terminology might have their own optional confidence attribute. If you include it you need an ITS rules defined in the document for that data category.
Felix: Do you need more time?
Dave: We don't have a conclusion, so yes. Can we add an optional elements to terminology from a backward compatibility standpoint?
Dave: Then I'll propose it on the list.
Felix: First week of December for
... The dates are on the wiki.
Felix: After the last call, we
have six weeks to get comments from outside and then we will
discuss at our face to face.
... Christian, can you introduce yourself?
Christian: I work for SAP, been
working on ITS for 15 years. Worked on multilingual content
production. Background on NLP. My background that was noted was
with terminology, work on interchange.
... Also worked on content interchange similar to XLIFF.
... Worked on XLIFF.
... Contributed to ITS 1.0.
David: I missed the last XLIFF TC
call, but I believe the last call covered extensibility in
<mrk> and <note>. Yves, can you update us?
... I'm worried that we may not be able to extend <mrk>.
Yves: It would be better to have extensibility, but people would be happy with a module.
David: The problem is that will take too long, so our markup would be illegal until that is defined.
Yves: I think you will find a lot
of opposition to extending <mrk> extensibility.
... If we can't add extensibility from our side that is an XLIFF problem. We should assume we can do it.
Dave: You wanted to keep a
pointer in for standoff markup.
... Scroll down in that link to where provenance is covered. There is something called phase group.
<fsasaki> search for "Or (standoff markup)"
Dave: You could write a rule that uses pointers in provenance to map to these. These would be in addition to stand-off. The question is if we worry about doing this explicitly.
Des: A question about this. Looking at the document, it looks like the mappings are fairly stable for XLIFF 1.2.
Yves: Probably yes. There is no show-stopper. I think most are mapped correctly.
David: I have an action to color-code things that are dependent on unstable things. But we needed to map this to see if there are show-stoppers.
Felix: Skip this topic.
<fsasaki> ACTION: felix to update the wiki with 2013 planning [recorded in http://www.w3.org/2012/11/12-mlw-lt-minutes.html#action01]
<trackbot> Created ACTION-285 - Update the wiki with 2013 planning [on Felix Sasaki - due 2012-11-19].
Felix: Think about any other venues we should go to. Also consider location for the final workshop in 2013.
Felix: Proposed that we replace this table in 6.1 with what Shaun and Olaf-Michael started. Thoughts?
Shaun: The table there now has information not present in my table. Combining them would make a table that is too wide.
Dave: Both have useful
information, but combining them makes the table too big.
... Jirka provided an algorithm to do the mapping from XML attribute names to HTML names.
Felix: So you think that Shaun and Olaf-Michael’s table as an appendix?
<Des> Arle, for minuting purpose, my question was specific to the XLIFF 1.2 specification. I understand the 2.0 spec is still in a state of flux. Apologies for my poor audio on the question.
<fsasaki> ACTION: shaun to put data category table into non-normative appendix in the spec [recorded in http://www.w3.org/2012/11/12-mlw-lt-minutes.html#action02]
<trackbot> Created ACTION-286 - Put data category table into non-normative appendix in the spec [on Shaun McCance - due 2012-11-19].
Shaun: For some of the categories there are many different properties. Do any of the categories have complex defaults, or can we do it just on a data category case-by-case basis?
Felix: I think data categories are fine as the basis.
Felix: Who can summarize the discussions?
Leroy: If you look at LocNote, the discussion is whether we have LocNote and LocNotePointer or just locNote.
<fsasaki> various versions are summarized here http://lists.w3.org/Archives/Public/public-multilingualweb-lt-tests/2012Nov/0019.html
<fsasaki> (versions = with or without "pointer")
Shaun: If you want to know
whether the test output identifies a pointer, the suite doesn't
test whether you are able to find the relative expression from
the pointer. That's another test of capability.
... But I'm not sure what the output from the test would be, a path or the value.
Yves: For 1.0, we put the string value.
Felix: I was first for having pointers, but am fine with removing them.
Leroy: Second question. Should we remove its-* from local html output and have it aligned with the global output for html?
Yves: That's probably better since you don't know where information is coming from.
Leroy: Do I keep its- or get rid
... So I'll standardize to the XML output.
<DomJones1> arle: do you want me to scribe?
<pnietoca> I believe that the agreement al Lyon was we use neither its- nor its: am I correct?
Felix: Dave sent a draft. See the link.
Felix: Related to the tool information topic. This looks stable.
<fsasaki> ACTION: arle to edit mtconfidence [recorded in http://www.w3.org/2012/11/12-mlw-lt-minutes.html#action03]
<trackbot> Created ACTION-287 - Edit mtconfidence [on Arle Lommel - due 2012-11-19].
<fsasaki> ACTION: felix to check xpath epxressions in mtconfidence draft, after arle's edits [recorded in http://www.w3.org/2012/11/12-mlw-lt-minutes.html#action04]
<trackbot> Created ACTION-288 - Check xpath epxressions in mtconfidence draft, after arle's edits [on Felix Sasaki - due 2012-11-19].
<fsasaki> close issue-41
<trackbot> ISSUE-41 mtConfidence closed
Felix: Discussed a while ago, but
propose to close this issue for now. May come back on a
... Hope we could have actions for data category implementers to look at specific categories to demonstrate their use.
... I would like to separate these issues. But over Christmas we may get more issues.
Dave: For MT Confidence score,
that is the only one that I think you need a global rule so you
can give confidence to attribute text. People might want to
look at that one.
... If you want to include pointers to support things, XLIFF examples would be very good/useful.
... I'll post these suggestions on the mailing list.
Felix: Let's move forward to having more data category-specific issues. Can we close the general issue?
Dave: Data categories with
pointers need to be written up so that we don't mix up pointers
and normal global rules.
... Is that still an objective?
Felix: It is still in the draft as a comment. So we will get to it.
<fsasaki> close issue-51
<trackbot> ISSUE-51 There are too many pointer attributes and global rules in general closed
<fsasaki> ACTION: felix to make action items for data category owners about global rules [recorded in http://www.w3.org/2012/11/12-mlw-lt-minutes.html#action05]
<trackbot> Created ACTION-289 - Make action items for data category owners about global rules [on Felix Sasaki - due 2012-11-19].
<fsasaki> "Example 85: Annotating an issue in XML with local standoff markup and a global rule"
<fsasaki> Example 86: Annotating an issue in HTML with local standoff markup
<fsasaki> <script type=application/xml id=its-standoff-1>
Felix: We want to use the XML
content in the script instead of its- attributes.
... The standup markup would no longer be available as its- attributes.
<fsasaki> arle: seems reasonable
Phil: I've checked and I think this was OK.
<dF> Reviewed the mtConfidence wording as we speak, looks good to me.. also the toolRef wording
<fsasaki> "id=its-standoff-1" - change to "lq1"
Arle: Does that mean that each markup will be its own script?
Felix: Provenance is widely relevance. Can we switch this to agentProvenanceProvenance?
<leroy_> my second question for the record was to remove its-* from local html output and have it aligned with the global output for html. It was decided to remove its-* and align the output.
<fsasaki> "translationProvenanceRecordsRef" > "ProvenanceRecordsRef"
<fsasaki> "transPerson"> "Person"
<fsasaki> ACTION: felix to update spec with regards to proveance, using daveL's input [recorded in http://www.w3.org/2012/11/12-mlw-lt-minutes.html#action06]
<trackbot> Created ACTION-290 - Update spec with regards to proveance, using daveL's input [on Felix Sasaki - due 2012-11-19].
Felix: That's the latest thread/draft
Dave: This is ongoing. The draft
is to gain clarity. In Lyon we thought to use an IRI.
... Felix, you asked what we gain from it. It is difficult to agree on what the content of the toolsRef would be. We could encode things in the IRI, but it would be difficult. If we don't say what the IRI points to.
... I don't think we have consensus yet, but the content I made is a basis for discussion.
Felix: Since we've not had time to consider this, let's leave it for the next call. Then next week we can add to the spec.
Dave: Should it be able to refer to non-ITS data categories?
Yves: Let's leave undefined things out of the spec.
Felix: XLIFF doesn't stop us from
... The editorial notes need to be addressed.
... If we move to last call in December, we have to be fine with the content with no open issues. There are still many explanatory sections that are not up to date. Esp. 1 and 2.
... We need to explain that non-normative sections will be worked on during the last call.
... For those who are new to the group, please make comments about normative features. Because there is a lot of mail traffic, look at the draft and see what needs more work in the normative sections.
... Please remember the upcoming calls.
Editing call is November 14, 20, 27, 28.
Felix: I can't make an editing call today.
Arle: I can't make it today either.
Felix: Thanks to you all.