I do understand that HumanML is intended for end user consumption.
However, what humans think and what apps "think" will "hopefully" coincide
at some point "not to far off". This is the point I'm trying
to make. While VR and AI are just two of the application areas that
would use HumanML, they are likely to be very demanding ones...
Again, I will attempt to view it from a non-apps perspective...
Rob
Re: HumanML is designed to represent
what the human thinks...not what the app thinks.
Ranjeeth Kumar Thunga wrote:
This
discussion seems to make very clear the need of the new VR_AI focus that
Rob has brought to the table. HumanML,
in and of itself, is not meant to be for apps designer consumption, but
rather for end-user consumption. I'm being a bit carefree with my
labels here, but basically I mean that HumanML is designed to represent
what the human thinks...not what the app thinks. Of
course, the app designer could use CSS or XSL or code to style/transform
a pure HumanML document in a conducive fashion, but the point is is that
how HumanML is rendered or transformed is not the concern of HumanML...on
its own. Then whose
concern is it? It certainly can't be ignored or left on its own...no
one would use it. This
brings us to explicitly defining a bridge, as Rob describes in the outline,
and this API of sorts is a fundamental aspect of the project to clarify....maybe
we can call it 'HumanAPI'? I know you used the definition in a slightly
different context Rob, but I think it falls under the exact problem area
we have been discussing thus far. -----------Ranjeeth
Kumar Thunga
----- Original Message -----
Sent: Monday, October 08, 2001 7:09
PM
Subject: Re: HM.VR_AI
I don't see how we can "completely" divorce ourselves from how apps
designers will think about HumanML (being a hard core apps designer myself),
I may require a bit of "training" to pull out of that mode.
The way I look at things, is that if a language does not allow for those
things I need to (or want to) do, I'm not going to bother with it regardless
of how noble the cause. I hope that we remain "cognizant" of how
complex HumanML may have to be in order to be of real value. "We"
are, after all about the most complex system going...
Yes, this should be discussed in a requirements thread.
Thanks Rex for the refocus,
Rob
Hmmn,
A rewind may be in order here. I think we need to discuss requirements
in a requirements thread, and I should have made that distinction way back
when Ranjeeth spoke of a litmus test of what is within the purview of HumanML
in relation to VR_AI: Goals and Overview. Unfortunately, I was so pleased
at the prospect of a genuine argument over values and datatypes that I
neglected to point that out. When I responded to Len's request for clarification,
I said that HumanML needs to look to such areas as VR and AI, (as well
as all others that we identify as important), for THEIR requirements of
HumanML, and see if we can accommodate them or at least not create vocabularies
that contradict them. Somehow that got to a discussion what is allowable
as HumanML. I think we ought to be careful of where we jump, as into conclusions,
and mea culpa, me too. IMHO, we're trying to do something we should not
be doing when we get into thinking about specifying primary, let alone
secondary, audiences for HumanML. Isn't that the realm of apps builders?
Ciao,Rex At 6:03 PM -0400 10/8/01, Ranjeeth Kumar Thunga wrote:
The cultural/historical/political/personal_belief
IDENTITY is what differentiates definitions from another...there is no
such thing as a precise word definition except it has attached a particular
identity with it.
Below
is a trashy example (I admit) but makes the point pretty clearly about
a markup language for human consumption, and as Rob mentioned it could
be numbers or letters.
Hypothetical
example
--------------------------------
For "geek"
sub-culture, happiness may be represented from 0% to 100%
For "literary"
sub-culture, happiness may be enumerated as "cathartic" "jubilent" or "ecstatic"
For "pop_psychology"
sub-culture, happiness may be enumerated as digits from a scale of 1 to
10
The job
of mapping these to rendering languages is not the job of HumanML at a
base level, and not necessarily are represented in higher level constraints
either. I think we may need to scope out a "mapping/transformation"
deliverable as part of HumanMarkup (as if we don't have enough ;-))...nonetheless,
as Len mentioned, we are building for a future...a future which may take
10 years to see the final implementation.
I propose
the following...
HM.Requirements:
HumanML
must represent, in an explicit and clear fashion, the values as they make
sense to the conveyers themselves themselves, as best as possible within
the limits of XML design...no one else needs to be consulted, or should
be consulted, except the designated authority creating these self-definitions.
Ranjeeth
Kumar Thunga
----- Original Message -----
From: Bullard, Claude
L (Len)
To: Rob Nixon
Cc: Rex Brooks
; Ranjeeth Kumar Thunga ;
OASIS Comment
Sent: Monday, October 08, 2001 5:27 PM
Subject: RE: HM.VR_AI: Goals and Overview : HumanML_VR_AI
Facilitator
Of course
the dilemma there is secondary receivers (or multiple receivers past the
intended receiver). For example, the word "crusade" has a
loose
meaning in the west. To a person from the Middle East, it has a pejorative
or emotional meaning. So code lists are developed that
have
cultural attributes. This means the sender tries to ascertain
the use of the code in each view (where the view is an aggregation of
culture,
history, etc.) and the receiver attempts to determine the scoping of the
original message (intent: was it local meaning of sender,
or was
it intended to inflame the secondary receiver). For example,
one can envision an interface in which a gesture or word is expressed
in a
view consisting of personal time, historical time, culture, etc. joined
to a set of all possible receivers ranked by the intention of the
sender
for a given receiver to get this message in primary or secondary roles
(or any set of roles you can envision). This would return
a graph
where that gesture is the topic and all of the receiver interpretations
are linked nodes with some visualization technique
(say
color coding) that ranks the interpretations according to some other dimension
(criticality, danger, affinity, whatever). This
would
make it possible to explore different interpretations and pick one that
meets the local politic. Remember, the system doesn't
find
a "true" meaning; it enables one to choose a meaning.
In a
more formal communication, say a process constrained set of messages, one
creates a protocol. This means that the potential
interpretations
and the potential receivers are much more limited enabling a much more
predictable behavior as long as everyone
sticks
to the a priori rules for the protocol. Such contract-constrained
communications usually include a phase similar to
what
is described above in which a set of message types are proposed, contracted,
and limited in the interpretation such that
the
response behavior can be observed and validated as belonging to an acceptable
range. Otherwise, if outside the range,
the
system punts to a negotiation node to enable it to determine the next move.
len
-----Original Message-----
From: Rob Nixon [mailto:robnixon@execpc.com]
Sent: Monday, October 08, 2001 4:03 PM
To: Bullard, Claude L (Len)
Cc: Rex Brooks; Ranjeeth Kumar Thunga; OASIS Comment
Subject: Re: HM.VR_AI: Goals and Overview : HumanML_VR_AI
Facilitator
I should clarify that we will need support for both a
numeric (not Borg only) value, AND human readable value. But
the values need to "mean" something tangible across all human cultures
as Len mentions.
--
Rex
Brooks
GeoAddress:
1361-A Addison, Berkeley, CA, 94702 USA, Earth
W3Address:
http://www.starbourne.com
Email: rexb@starbourne.comTel:
510-849-2309
Fax: By Request
|