Copy Link
Add to Bookmark
Report
AIList Digest Volume 4 Issue 151
AIList Digest Wednesday, 18 Jun 1986 Volume 4 : Issue 151
Today's Topics:
Queries - References on Natural Language & Aristotle &
Prolog Optimization & P-Shell & Knowledge Acquisition &
Expert Systems for Clinical Neuropsychological Assessment &
Expert System Validation and Verification &
Expert-Ease & Recursive Fixed-Point Solvers &
Cheeseman's Automatic Expert System Induction,
Psychology & Physics - Inside Out & Dr. Who
----------------------------------------------------------------------
Date: 2 Jun 86 16:56:00 PST
From: seismo!nwc-143b.ARPA!sefai
Subject: References on Natural Language???
[Forwarded from IRList Digest V2#26 by Laws@SRI-AI.]
I am investigating literature that will hopefully help me on my
master's thesis. Without being too specific, the topic centers around
schemes for representing natural language in a computer system. So far,
my list of references includes:
1. Handbook of Artificial Intelligence, Barr and Feigenbaum
2. NETL: A System for Representing and Using Real-World
Knowledge, Fahlman
3. Human Information Processing, Lindsay and Norman
4. A Theory of Syntactic Recognition for Natural Language,
Marcus
5. Principles of Artificial Intelligence, Nilsson
6. Basic English (series), Ogden
7. The Cognitive Computer on Language, Schank with Childers
8. Computer Models of Thought and Language, Schank and Colby
9. Artificial Intelligence, Winston
10. A Handbook of English Grammar, Zandvoort
I'd appreciate any good references others have come across and
I'd be more than happy to send out the list afterwards.
Gene Guglielmo
sefai@nwc-143b
[Note: Thank you for the offer of collecting references. You have
quite an unusual assortment of works! I encourage you to look at
"Introduction to Modern Information Retrieval" by Salton and McGill
and "Information Retrieval, 2nd ed." by C.J. VanRijsbergen for a
rather different perspective. Let us know more details of your plans
when you become more focused. - Ed]
------------------------------
Date: Sat 7 Jun 86 23:38:21-PDT
From: Ali Ozer <ALI@SU-SCORE.ARPA>
Reply-to: ali@score,taran@sushi
Subject: Curious about Aristotle, "Knowledge Processor"...
[Forwarded from the Stanford bboard by Laws@SRI-AI.]
In p.19 of June 4 Campus Report, there is a short 2-column article
titled "Knowledge processor named Aristotle pays a visit." The article
says... "Modeling computer architecture after the human nervous system,
a Stanford graduate has developed Aristotle, a unique knowledge
processor. ... Modeled on synapses, the junctions between nerve cells,
Aristotle encodes information in fundamental units ranging from a
single character to a word, then a sentence, and finally a paragraph. ...
``You teach Aristotle like a child,'' he [John Voevodsky, the inventor]
said. ``Characters first, then words and sentences.'' ... Aristotle can
perform several tasks. It was first trained to turn a light on and
off, then to ring a bell, and finally to blow a whistle. ... "
Anyway, if you're curious from the above, you should get your hands on
a Campus Report and read the whole article. This machine just
sounds fascinating, but there isn't any technical information about it
in the paper. Does anyone out there know more about this? The
article makes it sound like this processor provides an approach to
intelligence that could easily replace most of the current AI techniques!
But, I don't know much about AI, and I certainly know very little about
this Aristotle, so I just don't know... If anyone has more info or
knows where there is more written about this "knowledge processor,"
I would like to hear about it.
Very curious about things I should not be curious about during
finals week, but am,
-Ali
------------------------------
Date: 13 Jun 86 04:25:15 GMT
From: sdcsvax!sdcrdcf!burdvax!psuvax1!gondor!hou@ucbvax.berkeley.edu (Po Hou)
Subject: A.I.(expert systems)
I am studying application of prolog on expert systems.
Is the following fact correct ?
(1) when a predicate is used recently then it will be used in the future
with higher possibility than those predicates that are not used recently.
(i.e. it is similar to working set concept of virtual memory.)
For example,
predicate call p(a,Y,Z) gets a set {(X,Y,Z)| X=a } , then what is the
possibility that p(a,Y,Z) is called again ?
(2) what is the user behavior to use a expert system ?
(3) frequently used knowledge will be used with a higher possibility ?
------------------------------
Date: 13 Jun 86 22:49:29 GMT
From: sdcsvax!noscvax!kanemoto@ucbvax.berkeley.edu (Nelson T. Kanemoto)
Subject: P-Shell Query
I'm looking for information on P-Shell, described in the article:
"Programming in P-Shell", by Newton S. Lee, IEEE Expert,
pg. 50-63, in the recent Summer 1986 issue.
If anyone knows the cost, availability, or any other information concerning
P-Shell, please send me a message:
kanemoto@nosc.arpa
Thanks in advance,
Nelson T. Kanemoto
Computer Sciences Corporation
NOSC Hawaii
------------------------------
Date: Fri, 13 Jun 86 14:52:26-1000
From: Jimmy Y. Cheng <cheng%humu@nosc.ARPA>
Subject: Knowledge Acquisition
I'm interested in the knowledge acquisition of the domain
knowledge from an expert to an engineer. Can anyone help me in
locating an article or reference to people working in this area? Any
help would be greatly appreciated. Since this the bottleneck in
building an expert system, any progress would be a boon to AI.
------------------------------
Date: 15 Jun 86 17:31:48 GMT
From: ucbcad!nike!topaz!harvard!ut-sally!ut-ngp!gknight@ucbvax.berkeley.edu
(Gary Knight)
Subject: Expert systems for clinical neuropsychological assessment.
A few weeks ago I posted an inquiry concerning my interest in the current
state of research and development on expert systems for clinical
neuropsychological assessment. I received several replies, some of which led
to some very useful material.
I would now like to re-post that inquiry, seeking still further input
from anyone who has such information and did *not* respond before. So . . .
Does anyone have information they can share with me
on research or development work with respect to
expert systems for application to clinical neuro-
psychological assessment? If so, please reply by
mail and I'll post a summary, including all previous
replies.
Thanks very much.
--
Gary Knight, 3604 Pinnacle Road, Austin, TX 78746 (512/328-2480).
Biopsychology Program, Univ. of Texas at Austin. "There is nothing better
in life than to have a goal and be working toward it." -- Goethe.
------------------------------
Date: Mon, 16 Jun 86 15:58 ???
From: PENN%NGSTL1%ti-eg.csnet@CSNET-RELAY.ARPA
Subject: Expert System Validation and Verification
I am doing a literature search on the validation and
verification of expert systems. I have found a few
articles manually, however, my database searches
(INSPEC, COMPENDEX, etc.) haven't been helpful.
I am getting more on the use of expert systems to
test other computer software than procedures/
methods for validating expert systems!
If you have any pertinent information, or some
good sources with carry-over potential to
expert systems I would appreciate being
contacted. In return I will be happy to
furnish you with the final literature search
information. Thank you!
Mary Penn
Knowledge Engineer
TI-Artificial Intelligence Laboratory
(214) 343-7667
P.O. Box 660246 M/S 3645
Dallas, TX 75266
PENN%NGSTL1@TI-EG.CSNET
[One validation effort was carried out by John Reiter for the HYDRO
expert system (an extension of Prospector that he, Rene Reboh, and
John Gashnig developed). Reiter used scattergrams and rank correlations
to compare various actual parameters with those predicted by the
system. The final SRI report was "Development of a Knowledge-Based
Interface to a Hydrological Simulation Program," May 1982, but I
believe most of the validation effort was documented in John's
dissertation. -- KIL]
------------------------------
Date: 16 Jun 86 18:58:19 GMT
From: ihnp4!houxm!mtuxo!mtgzy!jis@ucbvax.berkeley.edu (j.mukerji)
Subject: Info wanted on Expert-Ease
I just read a glossy on Expert-Ease, which is based on an inference engine
developed by Donald Michie at Edinburgh University. I would appreciate any
comments about it (good or bad) from anyone who has used it. I am
considering buying it, and of course would like to know whether it is all
that it is touted to be. If there is sufficient interest I will summerize
responses to this message in this newsgroup.
Thank you.
Jishnu Mukerji
AT&T Information Systems
Middeltown NJ
ihnp4!mtgzz!jis1
------------------------------
Date: Mon, 16 Jun 86 11:31 EDT
From: DSTEVEN%clemson.csnet@CSNET-RELAY.ARPA
Subject: Recursive fixed point solvers.
We are looking for a program to solve for fixed points of
recursive equations. Actually, any help will be greatly
appreciated.
Thanks in advance
Steve
(803) 656-5880
------------------------------
Date: Mon 16 Jun 86 14:00:05-PDT
From: Tom Garvey <Garvey@SRI-AI.ARPA>
Subject: Re: Seminar - Automatic Expert System Induction (NASA Ames)
Does this mean that Cheeseman has at long last implemented something,
or is this going to be more of the same old theoretical maximum
entropy stuff over high-order probability distributions that would
not only eliminate the need for an expert but also make it impossible
for the expert to provide the necessary information. Presumably, an
expert system with no experts is misnamed, and systems for statistical
analysis have been around for a long time.
Cheers,
Tom
------------------------------
Date: Sat, 7 Jun 86 19:00:43 bst
From: gcj%qmc-ori.uucp@Cs.Ucl.AC.UK
Subject: Re: Inside Out
>From: majka@ubc.CSNET.UUCP
>> ...Einstein's theory of general relativity, which models the cosmos
>> as a 4 dimensional pseudo-Riemannian spacetime. ...
>
>*pseudo*-Riemannian? I think you mean Semi-Reimannian, and that applies
>to the metric, not the spacetime.
>
>---
>Marc Majka
>
OK, take your pick, but it must be a pseudo/semi-Riemannian spacetime, so
that you can have null distances; ie the metric on the manifold must have
differing signs eg (+1,-1,-1,-1), ie in Minkowski space. (Note that in GR,
all spacetimes are locally Minkowski). The manifold must be Hausdorff and
differentiable to arbitrary order, ie C-infinity.
Apart from differential geometry, the keyword is *model*. Spacetime is not
any type of Riemannian manifold. Newton did not need differential geometry
to form a model of gravity. Did spacetime suddenly curve when Einstein
discovered general relativity? What is interesting is the leap from the
intuitive idea of the apple falling because it is `pulled' by the earth,
to the non-intuitive idea of the apple falling because nothing holds it up.
It falls along a timelike geodesic, the shortest (4-dim) distance between
two points.
And there is nothing intuitive about quantum chromo-dynamics, at least not
to me.
Gordon Joly,
ARPA: gcj%maths.qmc.ac.uk%cs.qmc.ac.uk@cs.ucl.ac.uk
UUCP: ...!seismo!ukc!qmc-ori!gcj
------------------------------
Date: Fri, 6 Jun 86 14:10 EST
From: STANKULI%cs.umass.edu@CSNET-RELAY.ARPA
Subject: more inside and out
Another response to phayes AILIST vol 4 # 125 and subsequent replies on the
intuitions of tardis inside and out... particularly Ken Laws reply. i will
include references to relevant episodes which test the limits of tardis
functions.
the tardis is not a portal to another dimension. Gallifreyan temporal
mechanics are particularly limited to our 4-dimensional universe (they call it
N-space) in its operation. theoretically the time lords can go to any time and
place in N-space but they accept informal constraints they call "time laws"
which they try to enforce to lower the occurrence of paradox phenomema. but
even the high council of time lords will violate these regulations once in a
while at great expense of energy ('the five doctors' peter davidson). the
fundamental piece of Gallifreyan technology is called a dimensional stabalizer
which was discovered by Omeger and perfected by an engineer called Rasilon.
the metauniverse of dr. who is at least five dimensional. there have been
times when the doctor's tardis has been transported through accident into other
parallel 4D universes where it functions with different precision than in
N-space. the doctor (jon pertwee) had this happen once when repairing the
tardis console and later the tardis was thrown into E-space by a stellar
accident for a number of episodes (tom baker). E-space was a much smaller 4D
universe which was collapsing instead of expanding.
punching holes in the side of a tardis has happened. in 'terminus' (peter
davidson) the tardis was breaking apart in transit and attached to the side of
a space vehicle. the doctor and companions came and went from the tardis
through an unstable hole in the wall of nyssa's room. the hole acted just like
a door, but they could not control its opening and closing.
there is no fundamental reason why the inside of a tardis is always larger
than the outside. the relative dimensions of inside and out are uncoupled.
the 'outer plasmic shell' is controlled by a chameleon circuit and can be any
size. the outside could be larger than the inside. tom baker once designed an
exterior the size of the pyramid of cheops but since his chameleon circuit was
broken, it reverted to the police box. the master once had his tardis
materialize around a Concorde SST ('time flight' peter davidson). there is no
reason why a tardis outside could not be the size of a shoe box or postage
stamp, except that a humanoid could not exit the craft in such case. an error
in 'logopolis' (tom baker) caused it to become three feet high, trapping the
doctor inside. a tardis can also jettison portions of its interior space in
emergency ('castrovalva' peter davidson).
some other interesting properties have arisen in the 20+ year series. if a
tardis is turned over on its side, there is a control which can rotate the
interior so the floor orients with gravity ('time flight'). when a tardis
materializes, it incorporates the space it appears in. the master's tardis
contained the original SST inside his own. a dimensional anomaly arises when
one tardis materializes around another tardis ('logopolis'). the dimensional
stabilizer works by folding one dimension into another-- apparently a
point-for-point mapping mechanism. they call this 'block transfer
computation'. if one tardis incorporates another one, they are both in danger
of losing external reference. since they both contain the same folded space,
they both contain each other. it is possible to walk from the outer one through
the inner one to the outer one... like infinite regression in a hall of
mirrors.
for one of the longest running dramatic series in history, the BBC staff of
writers is to be admired for their conceptual detail in metauniversal design.
their spacetime mechanics have interesting and plausible ramifications on a
different order of magnitude than purely child fantasy like alice through the
looking glass. the limitations of temporal technology, genetic regeneration,
metalinguistic translation, and even the sonic screwdriver make the series
intriguing beyond the fun of watching.
stan
------------------------------
End of AIList Digest
********************
