W3C home > Mailing lists > Public > public-xg-emotion@w3.org > October 2008

[EMOXG] Updated discussion document

From: Ian Wilson <ian@emotionai.com>
Date: Thu, 23 Oct 2008 10:43:10 +0100
Message-ID: <8376826c0810230243m7edd499cy3373d1e6e7ca8844@mail.gmail.com>
To: public-xg-emotion@w3.org
Hi All,
Core 1 - Agree with Christian, even our group is called "Emotion" and this
format is for everyone so the tag should be the most easily understood. Lets
also not forget that this has to be used around the world and most native
english speakers dont know what "affect" means.

Core 2 - We should have a repository, perhaps at W3? I like the Qnames idea,
simple and tidy.

Core 3 - I would suggest a "scale" attribute which could be used throughout
the schema :

            <arousal scale="unipolar" value="0.75" />  OR
            <dimensions scale="unipolar" set="emotionai">

Core 4 - Is there a less specific, more easily understood word than
"appraisals"? It seems very Scherer specific whereas "dimensions" is a very
general term.

Core 5 - Again, "action-tendencies" seems very specific to one model rather
than general? I would suggest just "actions" but maybe that is too general,
            Abstraction is good, so is a hierarchy, I use a hierarchy so a
top level is "Energy" (or use of energy) with food as a specific sub set.

Core 6 - I think everything should be part of a set (like dimensions) which
can contain one or more elements. I think most of what we are doing is
inherently complex and parallel and we should encourage that thinking

Core 7 - I am less convinced that there should be a separate intensity tag.
It does not seem intuitive that intensity is not tied to an element but is
on its own?

Core 8 - Agree with Catherine, I am currently using "attack, sustain, decay"
timing values in my work so having a start point and a duration for those
would be required (end point is then redundant). Standard format timing,
machine readable.

Meta 1 - Agree that confidence is a widely available attribute to go with

Meta 2 - Not sure about this one.

Links 1 - Links without semantics in an XML format seems odd, that seems
more like presentation unless the semantics are implicit from the structure
and tags of the XML.

Links 2 - Absolute timing seems like an edge case. How about having a
separate absolute time tag that specifies a absolute start point then all
other timings are relative (to that absolute). This would be like absolute
and relative coordinates.
             Prefer the SMIL format.
             End timing (or duration) is redundant but not a problem to have
             Timing may not require an anchor (it could be used to drive an
external system)

Links 3 - If there is a generally agreed upon set of semantic roles we can
fix to that set. The uri is ambiguous here and does sound like links 1, one
of these could be removed.

Global 0 - Go with the HTML format perhaps as it is so well known?

Scales - Abstract is good but difficult to use, I would tend to favor
positive/negative verbal scales


Ian Wilson
Emotion AI

e : ian@emotionai.com  |  w : www.emotionai.com  |  t : +44 (0) 704 040 3198

profile : www.linkedin.com/in/ianwilson

"Bringing the power of emotional communication to your products"
Received on Thursday, 23 October 2008 09:43:54 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:52:15 UTC