AIList-REQUEST@SRI-AI.ARPA (AIList Moderator Kenneth Laws) (09/16/85)
AIList Digest Monday, 16 Sep 1985 Volume 3 : Issue 123
Today's Topics:
Linguistics - TaleSpin Story Generator Reference,
Humor - Psychotherapy,
AI Tools - LISP to C & More Xerox Announcements at IJCAI &
New Lists for TI Explorer Discussion &
Connectionist Network Simulator &
Discussion of AI Languages,
Information Retrieval - Interactive Encyclopedia &
Technical Foreign Language Material
----------------------------------------------------------------------
Date: Thu 12 Sep 85 12:00:25-PDT
From: Matt Heffron <BEC.SHAPIN@USC-ECL.ARPA>
Subject: Re:TaleSpin
Re: Children's story generator <7-Sep-85 Steve.Hoffman@CMU-CS-K>
The program you're interested in is TALE-SPIN, by James Meehan.
He is currently at:
James R. Meehan
Cognitive Systems Inc.
234 Church St.
New Haven, CT 06510
MEEHAN@YALE
- Matt Heffron
------------------------------
Date: Monday, 9-Sep-85 16:26:12-GMT
From: GORDON JOLY (on ERCC DEC-10) <GCJ%edxa@ucl-cs.arpa>
Subject: Analysis of ...
Re: Expert Systems In Psychiatry (Vol 3 # 116)
Psychotherapy is debugging for humans...
Gordon Joly.
[I'm not sure what the intent of this message is. I'll pass it along
under a "humor" label, but to avoid any "Polynomial" debacle would
like to point out that psychotherapy is not a joking matter. -- KIL]
------------------------------
Date: Thu 12 Sep 85 12:03:19-PDT
From: Matt Heffron <BEC.SHAPIN@USC-ECL.ARPA>
Subject: Re: LISP to C
Re: Lisp to C <20-Aug-85 duke@mitre.ARPA>
I know of (no endorsement, no experience with) a system called SILL
which is built on top of PSL on the Apollo. It can "... output standard
PASCAL (or other conventional languages). This capability enables the
user to port SILL developed code to other environments."
Contact:
SILMA Incorporated
1800 Embarcadero
Palo Alto, CA 94303
(415)-493-0145
-- Matt Heffron
From: Matt Heffron : Beckman Instruments Inc.
------------------------------
Date: 12 Sep 85 15:12 PDT
From: Fischer.pa@Xerox.ARPA
Subject: More Xerox announcements at IJCAI
The 1185 and 1186, though they can run IBM-PC software, do not use the
8086 to run Interlisp. The primary processor is microprogrammable,
constructed using bitslice chips. A second IO processor is, in fact, an
8086. IBM compatability is provided by a yet third processor, an 8086,
on a plug in board. Performance is about 15% greater overall than the
Dandelion; we're still benchmarking. These machines also have a new
larger 19" screen availible as an option, with more pixels at the same
density.
Also announced was a version of Quintus Prolog for D-machines, using
microcode support, which benchmarks at 51,000 LIPS.
Xerox Commonlisp was announced for 2nd quarter 1986. It is to be fully
integrated with the Interlisp environment.
(ron)
------------------------------
Date: Fri 13 Sep 85 15:16:25-PDT
From: Richard Acuff <Acuff@SUMEX-AIM.ARPA>
Subject: New Lists for TI Explorer Discussion
In order to facilitate information exchange among DARPA sponsored
projects using TI Explorers, two ArpaNet mailing lists are being
created. INFO-EXPLORER will be used for general information
distribution, such as operational questions, or announcing new
generally available packages or tools. BUG-EXPLORER will be used to
report problems with Explorer software, as well as fixes. Requests to
be added to or deleted from these lists should be sent to
INFO-EXPLORER-REQUEST or BUG-EXPLORER-REQUEST, respectively. All
addresses are at SUMEX-AIM.ARPA. These lists signify no commitment
from Texas Instruments or Stanford University. Indeed, there is no
guarantee that TI representatives will read the lists. The idea of
the lists is to provide communication among the users of Explorers.
-- Rich Acuff
Stanford KSL
------------------------------
Date: 13 Sep 1985 1557-EDT (Friday)
From: Hon Wai Chun <hon%brandeis.csnet@CSNET-RELAY.ARPA>
Subject: Connectionist Network Simulator
CONNECTIONIST NETWORK TEACHING / LEARNING TOOL
A connectionist network teaching / learning tool called AINET-1 is available
for distribution (educational and research purposes only) from the Computer
Science Department, Brandeis University.
AINET-1 is a graphic-oriented software package which can be used to
interactively create, manipulate, and experiment with connectionist
networks. Most commands are conveniently driven by a mouse. Nodes in
AINET-1 are shaded to reflect their activation levels. Once a network is
created, the user can run an animated simulation (network relaxation). In
the simulation, AINET-1 will change the various node shadings as the
activation levels change during each run cycle. After a simulation, the
user can plot the results or display tables of previous activation levels.
Networks can be stored into binary files and reloaded later for further
editing.
AINET-1 is intended to be used mainly as a learning tool to give the user a
flavor of how connectionist networks behave. A more sophisticated version,
called AINET-2 (under development), may be useful for development work.
AINET-1 is written in Symbolics Common Lisp and presently runs on Symbolics
Lisp machines (Release 6.0). The system is offered on a non-commercial,
non-disclosure, and as-is basis for a nominal fee.
The fee is $150 for universities, and $250.00 for laboratories. Interested
parties should send requests to (or call).
Hon Wai Chun hon@brandeis.csnet
Computer Science
Brandeis University
Ford 232A
Waltham, MA 02254
617-647-2650 or
617-647-2119 (main-office)
------------------------------
Date: 14 Sep 1985 07:21 EDT (Sat)
From: Wayne McGuire <Wayne%MIT-OZ@MIT-MC.ARPA>
Subject: Lisp vs. Prolog vs. ?
Predicting what language will be most used for AI is problematic since
there seems to be little agreement about what AI is. Let's assume,
however, that over the next decade or two ''AI'' will refer primarily
to expert systems.
In that case, one might speculate that perhaps the bulk of AI code
will be written neither in Lisp nor Prolog (not even an enhanced
Prolog which can elegantly manipulate and coordinate in the same
conceptual space multiple worlds, logics, and beliefs), but a
higher-level language, perhaps using Lisp and/or Prolog as a base.
One of the major tasks in the coming years will be tranferring the
expert knowledge from many domains--economics, medicine, sociology,
political science, literature, law, etc.--into expert systems.
Experts in these fields have dedicated their lives to mastering their
respective fields, not to learning the art of writing compact and
elegant Lisp code. These experts are not computer scientists or even
computer programmers.
This situation implies three methods for transferring the knowledge of
domain experts into computer programs:
(1) Natural language understanding systems which can translate
raw text directly into working knowledgebases. We probably won't see
such systems on a large scale, which can operate with any degree of
reliablity, for at least another twenty-five years, and perhaps much
longer.
(2) The laborious interviewing of domain experts by expert system
experts (if you will), and the transcription of those interviews into
programs.
(3) The direct encoding of knowledge into expert systems by
domain experts themselves.
If the third method does indeed become the preferred method for
writing expert programs, then whatever higher-level language and user
interface best supports that activity (and I doubt it will be Lisp or
Prolog, which are relatively low-level) will probably become the
language which is most widely used for AI.
------------------------------
Date: Mon, 26 Aug 85 08:22:25 edt
From: Ben Shneiderman <ben@MARYLAND>
Subject: The Interactive Encyclopedia System (TIES)
[Excerpted from IRList 1.8 by Laws@SRI-AI.]
The Interactive Encyclopedia System (TIES)
--- ----------- ------------ ------ ------
Ben Shneiderman, Department of Computer Science
Janis Morariu, College of Library and Information Services
University of Maryland, College Park, MD 20742
June 1985
The Interactive Encyclopedia System (TIES) has been under development at the
University of Maryland since Fall 1983. It allows novice users to explore
information resources in an easy and appealing manner. They merely touch (or
use arrow keys to move a light bar onto) topics that interest them and a brief
definition appears at the bottom of the screen. The users may continue read-
ing or ask for details about the selected topic. An article about a topic may
be one or more screens long. As users traverse articles, TIES keeps the path
and allows easy reversal, building confidence and a sense of control. Advanced
features include the ability to view an index of articles or print out arti-
cles of interest. [...]
The TIES authoring software guides the author in writing a title, brief defin-
ition (5-25 words), text (50-1000 words, typically), and synonyms for each
article title. The author marks references in the text by surround them with
a pair of tildes. TIES collects all references, prompts the user for synonym
relationships, maintains a list of articles, and allows editing, addition, and
deletion of articles. A simple word processor is embedded in the authoring
software, but users can create articles on their own word processor, if they
wish. There are no commands to memorize, every operation is done by selection
from options on the screen. [...]
In the study comparing the arrow keys (maybe better termed "jump" keys because
the cursor would jump to the closest target in the direction pressed) to the
mouse, the arrow keys proved to be and average of 15% faster and preferred by
almost 90% of the subjects. We conjecture that when there are a small number
of targets on the screen and when jump keys can be implemented, they provide a
rapid, predictable, and appealing mechanism for selection. [...]
We did find touchscreen was easier to teach people than arrow keys,
and that arrow keys were a solid winner against the mouse when there
are a small number of large targets on the screen.
-- Ben Shneiderman
[A mouse could be programmed for similar "jump selection", of course.
It would be interesting to know whether this would be more distracting
than helpful. -- KIL]
------------------------------
Date: 13 Sep 1985 1209-PDT (Friday)
From: eugene@AMES-NAS.ARPA (Eugene Miya)
Subject: Technical foreign language material
Lately, there have been significant technical advances from non-English
speaking countries: Japan and the Continent. How many know the Japanese
equivalent to the CACM? What is the German equivalent of the IEEE?
It is too easy to say that such organizations and publications are
not significant. We have been accused of parochialism.
Our problems in the computer industry are rather unique as colleagues in other
fields such as nuclear fusion report that most of their colleagues are,
for all practical purposes, forced to come to the U.S. This is not the case
with computing
Just as we have file servers and process servers, we have a
distributed system. Our greatest resource are not the machines, but
the people with special skills. To this end I propose the following:
Propose:
1) to identify individuals who are capable of providing simple
translation. It would help if the Universities could do this.
Perhaps, Universities could get assistance from foreign language
departments.
2) Identify various foreign language publications of technical interest.
Quickly identify articles of wide interest. This information could
be posted to general interest Usenet newsgroups such as net.research
and net.mag as well as the special interest groups such as the AI List,
net.lang, and so forth. We should not create news groups, but work
on top of existing groups.
3) Help fund subscription and translations. Perhaps, individuals
without technical translation expertise can get together to pay for
technical translations [commercial], and/or help fund the subscription
of those with technical translation expertise.
Dymond@nbs-vms.ARPA has started an info-japan and a nihongo discussion
group on the ARPAnet, but it would be difficult to get Usenet
participation. I specifically do not want to create new newsgroups.
This structure can be placed atop the existing news group structure.
The Usenet has several advantages for the circulation of this type
of material: 1) it has the links into Japan, Korea, Australia,
Germany, France, and the rest of Europe not on the ARPAnet. 2)
since there is no global authority, industrial companies can participate
more easily. 3) There are a diversity of news groups which make news
dissemination easier: net.mag for instance is used for posting
the TOCs of various publications, ideal for this type of dissemination.
Other significant groups include:
net.ai, net.cse, net.announce, net.physics, net.arch, net.math, net.mag,
net.research, net.bio, net.graphics, net.wanted, net.nlang
It appears our most critical needs are in the Eastern Asian languages
such as Japanese, Korean, and Chinese. Other useful work would include
French, German, and the other European languages. We have to look to
the Universities for much of our assistance, but private organizations
and government can also help. We can certainly make inquires.
The Usenet extends into Japan, France, and other non-English native
countries. We must take benefit of these contributors.
Similarly, we can contribute to these countries by tagging significant
English language documents.
I am willing to act as a clearing house for determining finding
individuals and groups, and specific journals. For this purpose,
I am giving my address an ARPA/uucp gateway. Send the mail inquiries
there. More in a couple of weeks.
From the Rock of Ages Home for Retired Hackers:
--eugene miya
NASA Ames Research Center
{hplabs,ihnp4,dual,hao,decwrl,allegra}!ames!amelia!eugene
eugene@ames-nas
------------------------------
End of AIList Digest
********************