W3C home > Mailing lists > Public > public-multilingualweb-lt@w3.org > November 2012

[ACTION-311]: Check with group about harmonization of interval values - mtconfidence, disambiguation, locquality issue, loc precis

From: Dr. David Filip <David.Filip@ul.ie>
Date: Fri, 23 Nov 2012 10:23:17 +0000
Message-ID: <CANw5LK=UtBX0JkzNhbE=oJ4foVT+e+GpYPRA+gAxNCWVZKGm3A@mail.gmail.com>
To: public-multilingualweb-lt@w3.org
Hi all, especially Owners of mtconfidence, disambiguation, locquality
issue, loc precis,
on the editing meeting of Nov 20, we noticed that throughout the spec we
used several ways how to describe rational number value ranges between 0
and 1 inclusive. We previously settled on giving a floating point number,
but this was really implemented only on mtConfidence.

In the editting meeting we settled on a more precise and bullet proof text
now contained in the draft here:
http://www.w3.org/International/multilingualweb/lt/drafts/its20/its20.html#mtconfidence
[A required mtConfidence attribute with] a value that represents [the
translation confidence score] as a rational number in the interval 0 to 1
inclusive.
[The bracketed parts are obviously category specific]
This has been later made absolutely unambiguous with the XML schema
reference:
 The value follows the XML Schema decimal data
type<http://www.w3.org/TR/2004/REC-xmlschema-2-20041028/#decimal> with
the constraining facets
minInclusive<http://www.w3.org/TR/2004/REC-xmlschema-2-20041028/#rf-minInclusive>
set
to 0 and maxInclusive<http://www.w3.org/TR/2004/REC-xmlschema-2-20041028/#rf-maxInclusive>
set
to 1.

@Owners of mtconfidence, disambiguation, locquality issue, loc precis,
please react and state weather the above approach is OK for specifying
<0;1> rational intervals as values for your data category.
If you answer a simple "yes", or if you do not take any action by the next
editing call time, the editors will harmonize as described above. If you
need something else, you need to make the point why your value should be
understood differently and exactly specify - so that it fits into the
current draft - how the value range should be given.
*Note:*
0-1 inclusive intervals will be mostly displayed as percentages by
implementing tools (and you are free to say so), the rationale for defining
the value as a rational interval is that all these scores are at the
baseline ratios comparing the present state with an ideal result
represented as 1. 0 represents absolute non-conformance and all ratios that
can be obtained in between are valid values, this value range would be
unduly restricted if we settled for instance on percentage with up to 2
decimal digits. Tools are of course free to make their rounding and display
as that. There might be discussion if zero is a valid value but we did not
bother to exclude it and there is no harm in having it, On the other hand
it is important to have 1 that is why the interval is defined as including
the boundary points.

Thanks for your attention
dF

Dr. David Filip
=======================
LRC | CNGL | LT-Web | CSIS
University of Limerick, Ireland
telephone: +353-6120-2781
*cellphone: +353-86-0222-158*
facsimile: +353-6120-2734
mailto: david.filip@ul.ie
Received on Friday, 23 November 2012 10:24:26 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 9 June 2013 00:25:03 UTC