RE: Theories v. Vocabularies (was: RE: [huml] May Minutes)

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

Title: Theories v. Vocabularies (was: RE: [huml] May Minutes)

The reason for using abstract types in the primary was to enable

substitution sets so one could create alternative theories or

vocabularies related by the types.   It isn't a strong set, but

adequate for experimentation.   The primary benefit is to list

the contexts that can affect the expression of the communication (sign)

and the interpretations.   XMLers realize of course, that the type

tree is extensible even in the secondaries.   We're on unknown

ground here, but I hope the examples we worked back in the

Phase 0 timeframe suggest some potential paths.  I think we

may have some concrete element types in there that need

to move to the secondaries because they are theory-level,

or in other terms, codelists.

 

I'm a bit numb right now, so we can come back to this topic

if it needs elaboration.  I think most of us who have been

working on HumanML for the last year or so understand

the research behind it and why the choices are made.  The

new faces may need some tutoring so they can avoid the

mistakes we made early on that led to the overhyping and

misunderstanding.   Consider it fortunate because we have

that as a shared example of how easily one can create

misunderstanding with the sincerest of intentions, and given

the amplifier that is the WWW, how quickly it comes back

for feeding.   Org must eat and it isn't always picky about that

being a healthy diet.  :-)

 

It is important though, that anyone who begins to evangelize

this work understand how thin it is at the moment.   We have

at the best, an approach and a shared understanding to

begin work with.  We have to share the roles of being

experimenters, implementors, and teachers.   I do hope that

the ranks of those who delve into HumanML begin to see it

not just as the artifact that is the PBS, but an experiment

in discovery of how to apply the principles of semiotics

to solving problems of communication.   There are many

theories but we need practical experience.

 

All of the hard work is ahead in doing things such

as creating libraries of 3D expressions, signs, that we can

then label and categorize to enable other applications such

as genre-centric communications to work with.   These

libraries, based on the references of the abstract types,

can be useful in the building of protocols.   One realizes

that a protocol and a genre are much the same sort of

thing.  As the works of Ekman et al prove, some of these

really are shared by all humans regardless of culture or

history, and that lends credence to their basis in evolution.

Darwin wins that one.   However, when used in aggregates,

these are cultural and Perce wins that one.

 

len

-----Original Message-----
From: Rex Brooks [mailto:rexb@starbourne.com]
Sent: Friday, May 23, 2003 2:58 PM
To: Bullard, Claude L (Len)
Cc: 'Rex Brooks'; huml@lists.oasis-open.org
Subject: Theories v. Vocabularies (was: RE: [huml] May Minutes)

You have my condolences. Been there. Done that.

Thanks also for giving me an opportuity to make a point about the work of our subcommittees, which is that it becomes more difficult as we get closer to making applications, or using the process of making applications as a way to show us what we need in our secondary base schemata. This is something I find difficult myself, and that is transposing the theory behind an application with the secondary vocabularies we will be developing. I think my recent work in creating a set of facial expressions for the chat application illustrates this.

It didn't quite sink in to me while I have been in discussions about the psychological theories that underly the manner in which emotional states function in humans, but there are, as Amir has pointed out and illustrates in the table of theories/theorists he posted and which I am attaching, numerous theories for this. The define, use and account for basic emotions differently and these theories can and will be used in very different applications based on the secondary vocabularies we build for this category of elements.

So the point I'm making is that we need to allow for those different usages, which means we must refrain from choosing a theory and adopting it. And, more importantly we need to be certain that we don't confuse the theory behind an application with the element definitions needed by that and other theories that we must, needs be, support.

Just so you all know, the theory that accounts for fairly well-acknowledged six basic, fundamental facial expressions is that of Ekman, Friesen and Ellsworth, the second in Amir's table. However, I did not select these expressions either because I believe this theory or because I really think that all other expressions can be formed by combinations of these six. I chose them solely based on my research as an artist, using"The Artist's Complete Guide to Facial Expression" by Gary Faigin from Watson-Guptill. I do hope that the full spectrum of facial expressions corresponding to the wide variety of human emotions within and across cultures, but that is too wide a variety to be certain about now. Lots of work.

Ciao,

Rex

At 12:24 PM -0500 5/23/03, Bullard, Claude L (Len) wrote:

It sounds like this was an exciting meeting.  I apologize for missing it.

My father passed away and there were family duties and trying to keep

focused at work.

 

HumanML as it stands, is the set of categories of information that

can be collected by observation or other instrumentation about the

contexts of a communication.  These can then be used in post-analysis

frameworks, or with more risk, in real time feedback systems to

augment the use of selectors for alternative responses.   This can

be applied to real human communications or to simulations such as

the application to X3D H-anim figures.    Because it is a set of

categories or types of information, application languages have to be

developed that focus on the precise aspects of the communication to 

be observed or instrumented.   These application languages will have

stronger semantics or interpretations than the weak ontology that

is the HumanML primary base schema.   This is in accordance with

the works of language experts such as John Sowa with regard to

latticed theories and their emergence from use of language. Because

each theory is distinct but related through the lattice, the level of

detail of information to be collected can be different in each theory

or, application language.   The primary base relates these by type

but only weakly constrains the interpretation.   So, HumanML is a

good basis for experimentation, but not a final or ultimate solution

to the challenges of improving human communication in the face

of noise or deliberate obsfuscation.   It has been shown, however,

(See Paul Ekman et al), that such techniques can reliably reveal such noise

and deliberate obsfuscation, so the challenge is not in proving the

categorization, but in realizing the instrumentation in concordance

with libraries expressed in the derived application languages, in

short, instrumenting a theory.

 

len

-----Original Message-----
From: Rex Brooks [mailto:rexb@starbourne.com]
Sent: Friday, May 23, 2003 11:51 AM
To: huml@lists.oasis-open.org
Subject: [huml] May Minutes

Here are the minutes for this month's meeting. Please review and post any corrections and/oradditions.

Here are the minutes for this month's meeting:

May 21, 2003

Teleconference meeting of the OASIS HumanMarkup Technical Committee.

USA Toll Free Number: 888-576-9014
USA Toll Number: +1-773-756-0201

Roll Call:
Voting Members:

Rex Brooks
Rob Nixon
Ranjeeth Kumar Thunga
Sylvia Candelaria deRam

Invited Visitor:

Russell Ruggiero

Minutes taken by TC Secretary Rex Brooks

Meeting convened12:05 p.m. Eastern Time

Previous meeting minutes accepted.

This meeting was held on the third Wednesday of the Month.

We did not have a quorum.

As usual, some of these items were actually discussed in a different sequence from this  summary.

Old Business:

Because we did not have a quorum, and also had no official business which needed a voting quorum, the meeting was largely comprised of reports on implementations, subcommittee work, liaison opportunities, outreach endeavors and branstorming.

Rex updated the TC on the gradual development of an expressed need across TCs to request OASIS to allow subcommittees to recruit special experts who need not be TC members to address the more focused and specialized needs of some subcommittees for adequately performing the tasks of their charters. This is somewhat different from the current policy that requires special dispensation from the OASIS CEO for "Invited Experts." This may or may not be relevant to our subcommittees, but the growing numbers of subcommittees was also pointed out as a similar development in the TCs in which Rex participates, and in which he has established informal liaisons.

Our invited visitor, IT industry writer Russell Ruggiero, was introduced and he informed us about the excellent efforts he has pursued in writing and publishing his first article about the HumanML initiative in WSReview (Web Services Review), a publication of The Thomson Corporation. Russell said that he plans to expand and cntinue this work handing off information about HumanML to high-end media, vendors and and analysts so that we can begin to get some traction in the marketplace and bring attention to the value of our efforts. Unanimous approval was evidenced for this work, and any momentum this can bring to the work of the TC.

Ranjeeth mentioned that Owen Ambur of the federal government's xml.gov working group may get involved with the Conflict Resolution Subcommittee, through a connection between Owen's interest in freedom of information through improved records-keeping and HumanML's ability to expand contextual human information and improve accuracy of that information.

Rob said that our newest recruit, graduate student Amir Youssefi will be working with him on the VR-AI Subcommittee's work following the generally high level of communications between them. Rob also mentioned that he had been thinking of changing the name of the subcommittee to avoid the negative baggage carried by the term "Artificial Intelligence." Sylvia pointed out that there also needed to be some reference to the field of cognitive science, and aggreed that AI carries too much unfavorable associations and connotations. They agreed to pursue this offline.

Russell mentioned in connection with the brainstorming that accompanied this discussion that he was open to writing on new topics that any of us wish to point out to him.

Ranjeeth reiterated his request for State-Department-Related Use-Cases for the purpose of attracting attention and possible funding from that source through his recent contacts.

One aspect of this outreach effort, Russell pointed out, was the current emphasis in the corporate world on Return On Investment (ROI) accounting and decision-making, and the concurrent emphasis in government on the issues surrounding terrorism. These are the current interests, according to his observations, that Russell suggests it would be wise to address.

In this context, outreach efforts such as State Department Use-Cases, Rob noted that there are tiers of understanding related to State Department interests, and that we might be well advised to take that into account. He added that it is important to make sure we do not set up a misperception that HumanML will solve entire problems or is even capable of solving large parts of many-layered, historically intertwined problems such as the ongoing Israeli-Palestinian conflict.

Russell noted that, in this context, in direct support of Rob's concern, that emphasizing the role of HumanML as a global equalizer is best expressed by focusing on HumanML as an improvement in the exchange of information. By focusing on that, we defuse emotional context within the expressions of one person or group to another.

Ranjeeth agreed with that and added that it is probably a good idea to emphasize the idea that HumanML can help "explore" issues to find areas where HumanML can improve accuracy and understanding.

In a wide ranging discussion such as this, it is sometimes difficult to place comments from the Secretary's notes into their correct sequence, and the point of this paragraph is a case in point. I will italicize that point when I get to it. What happened is that Rob actually mentioned earlier on that one aspect of State-Department-related issues is that often individuals and groups have vested interests in preventing the kind of improved accuracy that HumanML seeks to bring to communications, particularly in Conflict Resolution terms. We did not pursue this aspect precisely because it came earlier and the nature of the discussion swept on from the point prior to this paragraph.

It seemed more important to put it above, where the context of discussing how to best portray HumanML makes it particularly important.

The discussion moved on to discuss ways to portray Use Cases.

Rob noted that in negotiations between sets of negotiators for two groups seated opposite each other at a conference table, where lead negotiators address each other directly while representative members of their respective groups are present, the situation is rife for feedback loops which amplify the emotional responses of each group and can lead to catastrophe.

Ranjeeth mentioned that he thought it was important to emphasize in any description of such a Use Case that these were "congitive" feedback loops fed by each groups intentions, understandings, etc.

Sylvia started to offer an analysis from her perspective to illustrate this feedback loop, but it was not immediately clear what she meant.

It was however, agreed that Sylvia would post this simplified cognitive semiosis analysis so that we could apply it as a framework for diagramming interactions in the use cases we provide since all use cases will likely involve two, or more, groups led in any interactions by individuals speaking for those groups, and so we can have a common method of explaining and analyzing examples.

Our discussion had taken us beyond our normal time limit, which Ranjeeth pointed out to us when he requested that Rex summarize the specific issues he had wanted to raise in relation to the Human Physcial Characteristics Description Markup Language Subcommittee.

Rex said that the issues were two. The first issue, coordinating with the development of a need for specific kinds of experts in other subcommittees, had been addressed, and the second issue, that his research had led him to organize the work of the HPCDML SC to produce an RDF Schema first reflecting the paper on convergent standards he posted the previous Saturday, and possibly using the registries of the ebXML TC and/or UBL TC as well, to organize the standards he has discovered. The upshot being that original work for the HPCDML SC if needed should be done after the existing standards are coordinated.

Before adjouning, it was agreed that the Conflict Resolution Subcommittee and the to-be-renamed VR-AI Subcommittee will present their charters at the next TC meeting.

The meeting adjourned at 1:49 p.m. Eastern Time.   

--

Rex Brooks

GeoAddress: 1361-A Addison, Berkeley, CA, 94702 USA, Earth
W3Address: http://www.starbourne.com
Email: rexb@starbourne.com

Tel: 510-849-2309

Fax: By Request

-- 

Rex Brooks
GeoAddress: 1361-A Addison, Berkeley, CA, 94702 USA, Earth
W3Address: http://www.starbourne.com
Email: rexb@starbourne.com

Tel: 510-849-2309

Fax: By Request