[EMOXG] Modality issues to be discussed

Dear all,

In fullfillment of ACTIONS 33 and 34 (http://www.w3.org/2005/Incubator/emotion/group/tracker/actions/33 ,http://www.w3.org/2005/Incubator/emotion/group/tracker/actions/34) I'm writing you this e-mail in order to discuss and get feedbacks on issues related to modality.
In the last conference call (http://www.w3.org/2005/Incubator/emotion/minutes/2008-10-02.html) there has been a general agreement on the annotation of modality. The basic proposal is based on the optional tag <modality> (a sub-tag of the main <emotion> element):

        <modality set="basic_modalities" mode="face"/>

The set of 'basic_modalities' could be composed of the following values: face, voice, body and text.
However, the main question is wether we want to add the distiction between medium and mode like in emma (http://www.w3.org/TR/emma/#s4.2.11). In this markup 'medium' indicates the broader medium (or channel) while 'mode' indicates the specific mode of communication. For the first attribute three values are allowed: acoustic, tactile, and visual. For each of these categories a variable list of modalities can be defined. For example the visual medium could comprise modalities like image, video, etc.

In the emoML we could have:

 <emotion>
        <category set="everyday" name="excited"/>
        <modality medium="acoustic" set="basic_modalities" mode="voice"/>
 </emotion>

The advantage of this annotation is that we have two different levels of annotation with different granularities that could be alternatively used. The drawback is a more complex formalism that could be in contrast with our intended principle of proposing simple annotations, unless more complexity is needed. So, taking into consideration your use cases, the question is wether this distiction could be useful or not.

The second issue regards the use of composite attributes for multimodality. We could use an explicit annotation like this:

<emotion>
        <category set="everyday" name="excited"/>
        <modality id="m1" set="basic_modalities" mode="face"/>
        <modality id="m2" set="basic_modalities" mode="voice"/>
        ...
</emotion>

Or, in alternative, a more compact annotation that makes use of composite values, i.e. lists of values separated by special chars (a blank in the following example):

<emotion>
        <category set="everyday" name="excited"/>
        <modality set="basic_modalities" mode="face voice"/>
        ...
</emotion>

In the latter case 'mode' is an attribute of type nmtokens (http://www.w3.org/TR/xmlschema-2/#NMTOKENS)

Regarding this particular annotation, it would be useful to have some feedbacks from the group in order to set a decision concerning this point. If you have any valid reasons to prefer one solution with respect to the other, please let the members know.

And now let's discuss....

Best Regards,
Enrico

Received on Friday, 10 October 2008 16:41:51 UTC