special interest working groups within SRKB

Tom Gruber <Gruber@Sumex-AIM.Stanford.edu>
Message-id: <2922600378-1912444@KSL-Mac-69>
Date: Wed, 12 Aug 92  02:26:18 PDT
From: Tom Gruber <Gruber@Sumex-AIM.Stanford.edu>
To: srkb@ISI.EDU
Subject: special interest working groups within SRKB

The SRKB working group has a broad range of possible topics and
activities, broad enough to consume a group discussion without
producing results.  That is why in Phase II we are going to focus on
ontologies as a sharing mechanism.

The working assumption is that it is possible to specify the
commitments and assumptions of a set of programs, (or more generally,
the representational primitives required for a set of knowledge-use
tasks) with a relatively small, carefully defined vocabulary.  For
each set of programs, or related tasks, there may be a family of
relevant ontologies, of which a developer might choose one or more to
adapt to her needs.  A second assumption is that we can specify these
commitments and assumptions at the knowledge level: in a declarative
form independent of symbol-level encodings within programs.

Our main job is to identify the areas where it would pay to
formalize these ontologies, with an understanding of why.  
Our secondary task is to provide a mechanism for identifying, 
analyzing, and distributing ontologies. 

Given the many possible areas for ontologies, a set of special interest
subgroups of the SRKB are being formed -- one per ontology area.  Each
subgroup is nominally headed by a two or three SRKB participants, but
all are open to anyone willing to contribute.

The subgroups that have been identified to date are as follows.

  * Problem-solving tasks and methods 
    (Mark Musen, Stanford;  Bob Neches, USC/ISI))

  * Engineering Modeling
    (Brian Falkenhainer, Xerox;  Tom Gruber, Stanford))

  * Top-level comprehensive ontologies (e.g., for NLP)
    (Ed Hovy, USC/ISI;  Doug Skuce, U Ottawa))

  * Semiformal representations of decision making and design rationale
    (Jeff Bradshaw, Boeing CS;  Jintae Lee, MIT;  and Charles Petrie, MCC))

  * Planning and Scheduling
    (Masahiro Hori, IBM Japan;  Don McKay, Unisys))

  * User interface reference model
    (Jim Foley, Georgia Tech;  Bob Neches, USC/ISI))

  * Manufacturing and enterprise models
    (Mark Fox, U Toronto;  Marty Tenenbaum, EIT))

The names in parentheses are those people who have been asked to lead the
discussion and activity in each subgroup; the membership of each subgroup is
to be established by communication with the leaders.  Some of the folks
within an area have been talking to each other for years; others will be
getting together for the first time.  Clearly, these assignments can change
to accommodate people's interests and commitments.  Also, anyone who is
willing to put in the effort can propose a new subgroup.  Currently there is
none for biomedical information, which is relatively advanced in the
knowledge sharing business; any takers?

Each subgroup is tasked with the following:

  0. Identify itself to the SRKB list with a short description of
    purpose and a call for participants.

  1. Characterize the tasks and uses for knowledge sharing/reuse in an
   area
   (by email at first; then a position paper if something technical
   emerges)

  2. Identify the existing ontologies in that area
  (first by name and authors; then collecting papers)

  3.  Make some ontologies, and/or example KBs, available online for
   public analysis 
   (first by collecting definitions in whatever form they currently
   exist; then by working with the authors to write formal definitions
   where possible)

  4.  Analyzing and comparing the ontologies
  (first by simple analysis of the names used, then by looking for
  opportunities for merging and consolidating).

The product of the subgroups will depend on the area in which they are
working.  In the engineering realm, for example, it might be possible
to get some fairly well-defined ontologies for behavior modeling,
since the domains of mathematics and physics have already received a
lot of formal treatment in the literature.  In the problem-solving
tasks and methods, there is still quite a bit of research to do in
characterizing these things, so the results will be more informal and
heterogeneous.

One thing that the chairs might want to do, after getting organized
within SRKB (step 0 above), is to write a call for ontology proposals,
to be distributed to researchers in specific domain areas. 
Each could overview the knowledge sharing problem for an area, describe
purpose in formalizing ontologies in that area, and perhaps describing
some example work.  An example of such a message follows.