Workshop on Semantic Distance

Apologies in advance to those who receive multiple copies of this
announcement.  If interested, please note *Contact Information* 
below.

*****************************************************************


Preliminary Announcement

International Workshop on

SEMANTIC DISTANCE

Nov 10-12, Washington DC Area

Abstract: NIST plans a focused, scenario-driven interdisciplinary 
workshop of international experts to determine best foundations for 
applied research and standards.

The Manufacturing Systems Integration Division of the U. S. National 
Institute of Standards and Technology (NIST), with the anticipated 
support of the National Science Foundation (NSF), will sponsor a 
workshop on the difficult problem of "measuring" semantics. The 
practical application domain is information exchange in the 
manufacturing enterprise - primarily of models and information about 
models. The applicability of resulting research and related standards 
is expected to be broad, certainly extending to "knowledge 
management."

The sponsors aspire to identify existing formal methods, their 
relative strengths and weaknesses. They intend to develop a roadmap 
toward standards and technologies that benefit industrial users. 
Preliminary efforts in this area have revealed significant open 
issues and resulting controversies; therefore, something more than a 
state of the art review is sought. Participants will be encouraged to 
synthesize approaches based on a scenario and associated issues.

The location will be at the NIST campus, Gaithersburg MD, north of 
Washington DC. An honorarium for participants is anticipated, but not 
yet guaranteed.

*Scenario Outline: Characterization of Semantic Conveyance.*

One party has modeled information, perhaps about a state-rich 
collaborative process involving many partners. That party transmits 
information to another whose native methods, (and perhaps ontology) 
differ. Probably the recipient is a collaborator in the process 
modeled.

In the general case the semantics will be "translated" imperfectly. 
Both parties will wish to know how perfect was the semantic 
conveyance; if imperfect, each party wishes to know the extent or 
"distance" of the imperfection. Do the imperfections matter or were 
the losses "unimportant?" What might this mean for downstream 
communication?

These questions may be a matter of extent (simple tolerances) or 
context (whether important details were lost). NIST's special concern 
is the development of metrics to characterize the conveyance, 
translation and situating of such semantics.

Issues may include:

- Whether to characterize an absolute "semantic space," and/or to 
characterize "end-points" in order to determine distance. 
Alternatively whether a simpler difference can be determined.

- What theory is the best foundation and what expressions of that 
theory have a workable balance with usability; should there be 
competing theories;  is a new interdisciplinary approach required?

- Are there different types of semantic distance with unique costs 
and consequences?

- Can the candidate solutions be extended to the problems of indexing 
and search to support model component libraries.

- Can the solution be scaled to more general problems from the domain 
of manufacturing?

- What does it mean to maintain and certify relevant standards? How 
do the notions map to existing standards and practices in the 
manufacturing enterprise? What should be the relationship with 
developing notions for a "semantic web."

*Contact Information*

The workshop is by invitation only, limited to 25 participants. Ted 
Goranson is handling logistics and is the point of contact for 
invitations and questions, tedg@sirius-beta.com, 757/426-6704. The 
NIST point of contact is Al Jones, jonesa@cme.nist.gov, 301/975-3554.

*Schedule*

The workshop is planned for two and a half days. The morning of the 
first two days is blocked for introduction to the problem and 
presentations by participants. These should not be stock 
presentations; they are "off the record" and should speak to the 
specific nature of the problem. (Pre-workshop discussion with the 
organizers would be desirable.)

The afternoons will be occupied by two parallel breakout sessions: 
one will be concerned with comparative approaches, their tradeoffs 
and benefits; possible synthetic approaches; mappings to existing 
practice; and the nature of the resulting metric. A second group will 
tackle the problem from the metrics perspective, working backward. 
They will be concerned with issues of practice (with a heavy emphasis 
on test cases), validation, use of the metrics and maintenance of the 
standard(s).

On the second day, the groups will swap topics and moderators. 
(Moderators will be announced in a later flier, together with an 
outline of possible working issues.) The morning of the third day 
will synthesize the results and produce near and longer term 
recommendations. Steps toward a high payoff program are expected.

-- 

*****************************************************************

Evan K. Wallace
Manufacturing Systems Integration Division
NIST
ewallace@nist.gov

Received on Thursday, 31 July 2003 10:10:39 UTC