New KIF Manual, Interlingua Working Group Meeting, Etc.
Richard Fikes <fikes@sumex-aim.stanford.edu>
Date: Mon, 8 Jun 1992 10:04:25 PDT
From: Richard Fikes <fikes@sumex-aim.stanford.edu>
Subject: New KIF Manual, Interlingua Working Group Meeting, Etc.
To: Interlingua-working-group@ISI.EDU
Message-id: <MacMS.57641.16838.fikes@sumex-aim.stanford.edu>
Interlingua Working Group member,
Last Friday, we sent to you via U.S. Mail a draft of the (long awaited)
Knowledge Interchange Format (KIF) Version 3.0 Reference Manual for your
review and comment. Both the language and its formal description in the
manual have been significantly upgraded for the 3.0 version. (The currently
available reference manual is dated march 1991 and describes version 2.2 of
the language.) Thus, your consideration of this 3.0 manual is of particular
importance to the ongoing interlingua development process.
This message reprises the information in the cover letter sent out with the
manual.
>>>We need your immediate response regarding your availability for a
meeting
>>>of the working group at the upcoming AAAI National Conference.
In particular, we would like to have a meeting of the working group focused on
consideration of objectives and plans for future working group activities,
rather than technical discussions of details of the language specification.
The meeting could begin with brief status reports of working group activities
such as development of KIF translators and use of KIF in testbed experiments
in order to set the context for the discussions regarding future directions.
Given the demands on everyone's time at the conference, I suspect we will need
to limit the meeting to a few hours (perhaps an evening including dinner).
Let us know your availability for such a meeting, including whatever detailed
constraints on your time that week you know about at this point.
Highlights of what's new in the 3.0 language and its specification include the
following:
% Functions and relations are treated as objects in the universe of
discourse and defined as sets of lists of objects. Thus, KIF sentences can be
used to describe properties of functions and relations (e.g., transitivity).
A vocabulary for that purpose has been added to the language (see chapter 8 of
the manual), including a"lambda" function-forming term and a "kappa"
relation-forming term.
% The expressivity of the definitions language has been expanded to support
both "complete" and "partial" definitions of object, function, and relation
constants. Partial definitions can specify an arbitrary "defining axiom" that
restricts the possible denotations of the constant being defined. In
addition, partial definitions can optionally be restricted to being
"conservative" so that when the defining axioms they specify are added to any
given collection of sentences not containing the constant being defined, they
do not logically entail any additional sentences not containing the constant
being defined.
% An axiomatic semantics is being developed for the entire language and is
used throughout the manual to specify the language. In most cases, the KIF
definitions language is used to define function, relation, and object
constants.
The 3.0 manual is the result of significant work and review by several members
of the working group. However, this distribution is our first attempt to
obtain feedback and input from the entire group. Our plan is to provide the
working group with an opportunity to review and discuss this draft, do another
editing pass on the document based on that discussion, and then make the
resulting document available to members of the general interlingua
distribution list. Since there is widespread interest in KIF and many ongoing
requests for the reference manual, we are interested in making a 3.0 manual
available as soon as possible. So, I encourage you to give this draft your
attention right away, at least enough to assure yourself that you feel
comfortable having your name listed on the title page as a collaborator. Keep
in mind that this manual is a "living document" that we all will continue to
have the opportunity to evolve and refine.
In order to enable working groups to discuss proposals they are developing,
the Knowledge Sharing Initiative's Advisory Board has approved the formation
of "members only" distribution lists for that purpose. We have established
such a list for the interlingua working group called
Interlingua-working-group@isi.edu. Note that this message was sent to that
group. The list consists of the authors and collaborators listed on the title
page of the enclosed manual. In addition, as a matter of policy, all Advisory
Board members are eligible for membership on such lists. The Advisory Board
consists of Bob Neches, Ramesh Patil, Richard Fikes, Tim Finin, Mike
Genesereth, Tom Gruber, Peter Patel-Schneider, Marty Tenenbaum, Bill Swartout,
and Jay Weber. We encourage you to use Interlingua-working-group@isi.edu to
discuss this draft and any other issues regarding the working group that are
appropriate for a forum that is smaller and less public than the general
interlingua list.
Regarding this draft, in addition to your personal feedback, we solicit your
recommendations of who from outside the working group we might ask for
additional reviews. Our goal is to assure that the technical content of each
chapter has been carefully reviewed for accuracy and appropriateness by an
expert who is not one of the authors of the chapter.
Regards,
Richard and Mike
-------