RE: HumanMarkup Reservations

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

You are right that for some datasets, the distinction isn't relevant

insofar as describing say, the color of a hat.

 

It can make a difference when it comes to identity.   The differences

have a lot to do with the "rights" of a resource.  That is a hairy concept

for some, but for anyone who deals in public records, it is an everyday

problem.  An avatar doesn't have the same rights as Len or Ranjeeth.

However, if Len or Ranjeeth have an avatar, they have rights (copyright,

for example).   A juvenile's arrest record and the profiles made by a

prison psychologists on the other hand, have extensive handling

requirements.   In a sense, that isn't HumanML's problem but a

business rule problem.

 

Len
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

-----Original Message-----
From: Ranjeeth Kumar Thunga [mailto:rkthunga@humanmarkup.org]

As Len pointed out in YahooGroups a while back as well, I think that
approaching from different perspectives will ensure the robustness of the
markup descriptors we create.  We are describing the same information, but
coming from different backgrounds and applications in mind, we ensure that
the data is indeed independent from the applications which processes it.

For example, there will not be a black and white distinction between 'real
human' and 'virtual human' characteristics.  Online agents may be designed
to represent some default characteristics, in addition to characteristics of
its real-life counterpart.

That's the approach I think we have now.  As time and viewpoints are
incorporated, this certainly may change completely.  However, with the tools
that we now have available (and you would probably be able to speak with
more authority), we can design very specific, explicit mappings between
different human frameworks as represented by different HumanML modules.