wjb@tantalum.eds.com (Bill Biesty) (04/30/91)
In article <1991Apr18.130550.8014@vax1.tcd.ie> rwallace@vax1.tcd.ie writes:
->I'm looking for information on a program called EURISKO written by Douglas
->Lenat around 1982, it's a program that uses heuristics to learn information
->specific to the domain it's applied to. I'd be grateful for any information
->anyone could send me.
Try: Artificial Intelligence 21, North-Holland, 1983.
There's two articles by Lenat. One about AM, precursor to Eurisko
and one about Eurisko. These are actually chapters in the book.
Also and article called: "Heuretics: Theoretical and Experimental Study of
Heuristic Rules." but my copy doesn't have the compilation volume. My
guess though if my memory serves me well is taht it was in a Proceeding of
one of the conferences of AI. (They're all named similarly and near each
other in the library.) Probably in the early '80s as the latest of the
references at the end is 1982.
And then in the recently published _Readings in Machine Learning_, edited by
Shavlik and Dietterich, Morgan Kauffman, 1990, there's "The Ubiquity of
Discovery".
Note Lenat has a large number of articles that are very small that just rehash
what's said in the above with rarely some new examples.
Hope this helps.
Billminsky@media-lab.media.mit.edu.MEDIA.MIT.EDU (Marvin Minsky) (04/30/91)
In article <1991Apr29.210353.24849@edsr.eds.com> wjb@tantalum.eds.com (Bill Biesty) writes: >In article <1991Apr18.130550.8014@vax1.tcd.ie> rwallace@vax1.tcd.ie writes: >->I'm looking for information on a program called EURISKO written by Douglas >->Lenat around 1982, it's a program that uses heuristics to learn information >->specific to the domain it's applied to. I'd be grateful for any information >->anyone could send me. There's also a follow-up experiment by Kenneth Haase, who replicated a large part of EURISKO with various variations and alternate reformulations. He is kwh@media-lab.media.mit.edu
jro@dcs.exeter.ac.uk (Jonathan Rowe) (04/30/91)
In article <1991Apr18.130550.8014@vax1.tcd.ie> rwallace@vax1.tcd.ie writes: >I'm looking for information on a program called EURISKO written by Douglas >Lenat around 1982, Try 3 articles by Lenat "The Nature of Heuristics I, II, III" in Artificial Intelligence, vols 19, 21. (1982) Jon Rowe.
wognum@utis70.cs.utwente.nl (Nel Wognum) (05/07/91)
CALL FOR PARTICIPATION
Explanation generation for knowledge-based systems:
what do users need and want in practice?
(a workshop at IJCAI-91)
There is a large discrepancy between research in explanation
generation and the explanations actually provided by current
knowledge-based systems. In research the importance of explanation,
especially of rich and flexible explanations, has frequently been
emphasized and a variety of system architectures has been proposed.
Also, techniques from natural language research and dialogue have been
studied to improve the explanation capabilities of knowledge-based
systems.
In practice, however, most current systems still have very restricted
explanation capabilities or do not even provide any explanation at
all. Yet, we know that some of these systems are rather successful.
Are their users really satisfied with the system-user interaction? If
so, is it because most knowledge-based systems today deal with a
specific type of tasks, domains, or a user group which hardly ever
requires explanation? Will users still be satisfied with the
restricted explanation capabilities of current knowledge-based systems
when these systems deal with different types of tasks and more complex
domains?
Answers to the following questions would help us to decide which
techniques to employ to support the appropriate level of system-user
interaction.
- What system characteristics (e.g. task, domain, organizational
impact) influence the type of explanation and system-user
interaction?
- What user characteristics influence the type of explanation and
system-user interaction?
- What types of explanation (e.g. superficial, summary, deep)
are needed for what type of systems and users?
- What type of interaction (e.g. one-shot answers vs.
dialogue-based) is needed for what type of systems and users?
We would like to address these topics in the workshop. To encourage
interaction and a broad exchange of ideas, the workshop will be kept
to a moderate size, limited to 30 participants. Provocative
statements will be presented by the organizing committee. All
participants will then be invited to contribute to the discussion in a
plenary session. The workshop will be held on Sunday, August 25, and
will last a half day.
We would like to encourage researchers who have addressed the topics
mentioned above or have made empirical studies on these topics to
submit three copies of an extended abstract, not exceeding 1000 words,
no later than May 15, 1991 to the chairperson (address below).
Submitted abstracts should include 1) the title of the paper 2) the
author's name, address, phone number, and where possible E-mail and
fax number 3) the issue(s) addressed in the abstract. All submitted
abstracts will be reviewed by the organizing committee.
Acceptance notices will be mailed to the authors by June 10, 1991.
Camera-ready papers should be received at the address below no later
than June 28, 1991. Workshop proceedings will be distributed to the
participants at IJCAI.
The workshop fee will be AUD$85.
ORGANIZING COMMITTEE:
- Dr. Nel Wognum, University of Twente, The Netherlands,
E-mail: wognum@cs.utwente.nl
- Dr. Cecile Paris, USC/ISI, U.S.A.,
E-mail: paris@vaxa.isi.edu
- Dr. Dianne Berry, University of Oxford, United Kingdom,
E-mail: dberry@vax.oxford.ac.uk
CHAIRPERSON: Dr. Nel Wognum
University of Twente
Department of Computer Science
P.O.Box 217
7500 AE Enschede
The Netherlands
Fax: +31 53 339605
SCHEDULE: Abstracts received: May 15
Notice of acception: June 10
Camera-ready papers received: June 28