Position paper for the W3C Workshop on Emotion Markup Language

Takeshi Natsuno <1>, Masao Isshiki <2> and Kazuyuki Ashimura <2>
(<1> Keio University/Dwango, <2> W3C/Keio)

We are interested in the W3C Emotion Markup Language (EmotionML) as a
standardized representation of emotions and emotion-related states for
Web services.

There is a very popular video sharing service in Japan named "Nico
Nico Douga" (=Smile Smile Video) [1] provided by Niwango, a subsidiary
of Dwango company.  Using the Nico Nico Douga service, users can
upload, view and share video clips.  The unique point of Nico Nico
Douga different from the other video sharing services I believe is
that users can add comments and annotations not only to video clips
which they themselves have uploaded but also those somebody else has
uploaded.  After comments are added they will be superimposed on the
video clips [2], and shared with all the other viewers.  So we are
very interested in the following workshop topics, especially on manual
annotation of material involving emotionality.

For Nico Nico Douga, some simple set of tags named "Nico script" [3]
is currently used to control user annotations for Nico Nico Douga
videos.  For example, the following tag will show a text "Great :) !"
at the center of the display for 10 seconds 30 seconds after the video
starts:
[[
30:@10 ue hidari big red:＠テキスト Great :) !
]]

We have reviewed the current EmotionML specification [4] but are still
not 100% sure about the following points:

1. Does EmotionML have sufficient expressive power for actual emotion
   annotation service like Nico Nico Douga?

2. Is part of or all the EmotionML easy enough to use for that kind of
   service?

3. Is the current list of recommended vocabularies enough for the
   service?

So we would like to attend the workshop and see the possible
application of EmotionML to the Nico Nico Douga service.  Basically
the service requires synchronization with video timestamps, and so we
are especially interested in EmotionML in the context of "References
and time (section 2.4)" and "Use with SMIL (section 5.2.3)".
Regarding vocabulary for emotion status, we are not sure whether
defining typical categories of emotion information would be really the
best solution or not, since people use various text to express their
emotional status and the vocabulary could be infinite though sometimes
people do use typical emoticons.

[1] Nico Nico Douga: http://en.wikipedia.org/wiki/Nico_Nico_Douga
[2] Example video (need Nico Nico account): http://www.nicovideo.jp/watch/sm12094678
[3] http://nicowiki.com/%E3%83%8B%E3%82%B3%E3%82%B9%E3%82%AF%E3%83%AA%E3%83%97%E3%83%88.html
[4] EmotionML: http://www.w3.org/TR/emotionml/
[5] 2.4 References and time: http://www.w3.org/TR/emotionml/#s2.4
[6] 5.2.3 Use with SMIL: http://www.w3.org/TR/emotionml/#s5.2.3