perm filename MAIL.AI[RDG,DBL]1 blob
sn#730117 filedate 1983-11-09 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00054 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00007 00002 Comments to Carl Hewitt, wrt his Omega system
C00027 00003 ∂TO BUCHANAN@SUMEX 13:35 29-Dec-81
C00043 00004 ∂TO %score csd.smith, csd.dietterich 13:37 3-Feb-82
C00053 00005 ∂Snail Mailed to Doug Hofstadter -- 28-June-82
C00058 00006 ∂17-Feb-83 0923 DIETTERICH@SUMEX-AIM CPM knowledge base
C00062 00007 ∂15:02 10-Mar-83: ml/subj ↑Paige
C00064 00008 ∂11-Feb-83 2134 SLEEMAN@SUMEX-AIM An alg for analogies
C00066 00009 ∂02-Sep-82 2335 Keith Wescourt <G.WESCOURT at SU-SCORE> end of week already
C00067 00010 17:53 31-Mar: ml/su Winston's Latest
C00068 00011 ∂17:36 9-Apr: grinberg@diablo/su Alto Account
C00069 00012 ∂07-Apr-83 2226 FRALEY.hp-hulk at UDel-TCP Arpanet/CSnet miracles
C00073 00013 ∂TO demmel@berkeley 15:44 7-Feb-83
C00083 00014 ∂21:11 2-Apr: csd.lenat@score/subj Upcoming talk
C00087 00015 ∂14:39 18-Apr: FRALEY.hp-hulk@UDel-TCP, genesereth@sumex/cc/su re:MRS
C00090 00016 ∂20 Apr 83 20:49:23-PST <BARR@SUMEX-AIM> friday
C00093 00017 ∂27 Apr 83 17:41:44-PDT <FEIGENBAUM@SUMEX-AIM> phone call
C00094 00018 ∂16:31 9-Mar-83: dekleer@parc, csd.lenat@score/subj Less Procedural Rules
C00099 00019 ∂28-Mar-83 1349 CSD.FRALEY@SU-SCORE April hikes
C00113 00020 ∂ 8 February 1983 11:11 est Johnson.RADC←MISu at RADC-MULTICS Test
C00116 00021 ∂15 Apr 83 09:40:42-PST <CSD.TAJNAI@SU-SCORE.ARPA> Sohio recruiting AI PHD students
C00129 00022 ∂11-Apr-83 1029 DEKLEER.PA@PARC-MAXC paper and talk.
C00134 00023 I have 2 strong anti-recommendations:
C00146 00024 ∂12:42 13-May: Lenat@score/su Statistics
C00161 00025 ∂16-May-83 1156 mccune@aids-unix Future
C00163 00026 ∂15:00 13-May: deKleer@Parc/su TEX "Answers"
C00176 00027 ∂13:49 17-May: rod/su Welcome Back
C00179 00028 ∂14:24 6-May: sleeman@sumex/su ML Workshop
C00184 00029 ∂13 Oct 1982 1633-EDT PAIGE at RUTGERS
C00199 00030 ∂04-May-83 1209 DIETTERICH@SUMEX-AIM Paper on EMACS and typewriting
C00207 00031 ∂13:46 31-May: dekleer@parc/su Confession
C00210 00032 ∂23-May-83 1633 FRALEY@SU-SCORE.ARPA next hike
C00214 00033 ∂12-Jun-83 2329 JF@SU-SCORE.ARPA partitions
C00221 00034 ∂14:57 15-Jun: dkanerva@sumex/su FYI - from ISI
C00225 00035 ∂12:50 12-May: HART@SRI-KL/su Get Together
C00229 00036 ∂17:19 14-Jun: jf, pressburger@kestrel/su Problem #2
C00231 00037 ∂Mon 13 Jun 83 15:29:05-PDT <BUCHANAN@SUMEX-AIM.ARPA> summer support
C00234 00038 ∂13:26 27-Jun: lenat@SUMEX-AIM/su Support - Financial and Emotional
C00237 00039 ∂17-Jun-83 1034 WESCOURT@USC-ISI Party
C00239 00040 ∂05-Jul-83 1709 CAB MacArthur
C00240 00041 ∂11:11 29-Jun: dietterich@sumex/su Meetings - past and future
C00243 00042 ∂20:12 5-Jul: dekleer@parc/su Sandwiched in
C00245 00043 ∂Sun 24 Jul 83 03:11:07-PDT Janet Coursey <JVC@SU-SCORE.ARPA> ai-list
C00246 00044 ∂10:59 28-Jul: MULLEN@SUMEX-AIM, genesereth@sumex/cc/su Re: Report
C00247 00045 ∂28-Jul-83 1336 JED jobs
C00248 00046 ∂13:16 26-May: bcm/su Prolog hereabouts
C00255 00047 ∂14:52 27-Jul: appelt@sri, de2smith@sumex, pressburger@kestrel/su Internal Frame Backpack
C00260 00048 ∂15:56 29-Aug: Frawle@mit-mc/su Re: IJCAI Msg
C00261 00049 ∂21:02 14-Sept: vanLehn@parc, brown@parc/cc/su In case you didn't see it...
C00263 00050 ∂20-Sep-83 0707 PETTY@RUTGERS.ARPA 1983 abstract mailing
C00270 00051 ∂20-Sep-83 1806 HOLTZMAN@SUMEX-AIM Copy of message sent to SUMEX.
C00272 00052 To: greiner@SAIL
C00275 00053 ∂15:46 29-Aug: jc50@cmua/su FollowUp-ette
C00277 00054 For Peter Hart's Benefit
C00280 ENDMK
C⊗;
Comments to Carl Hewitt, wrt his Omega system
∂TO CSD.HEWITT@SCORE 14:17 20-Nov-81
A mergable flavor of OF
In our discussion last Tuesday I suggested there may be situations
in which OFs could be merged together, as in
(AND (A C (of r1 vr1))
(A C (of s1 vs1)))
is
(A C (of r1 vr1) (of s1 vs1)).
We agreed that, for this to work,
it was necessary that r1 and s1 be independent of one another.
To avoid the "WITH" solution (that is, replacing one or both of the OFs with
WITHs), it is essential that there be r2 and s2 relations which are
interdependent on r1 and s1 respectively.
You challenged me to find some real examples of this.
How about
(A Number (of Addend1 3) (of Addend2 9)
(of Divisor1 2) (of Divisor2 6))
or
(A Person (of Employee# 7321) (of Company IBM)
(of StudentId 234) (of School Stanford))
or
(A Professor (of Level AssistantProf) (of Institution MIT)
(of Sabbatical-Loc Fairchild) (of Sabbatical-Year 1981-82))
[Of course we assume a person can work for various companies,
and attend more than one school; and that a professor can work at various
institutions, at different levels at each, and take more than one sabbatical
leave during his life.]
It seems clear to me we should allow
(AND (A Number (of Addend1 3))
(A Number (of Divisor2 6)))
is
(A Number (of Addend1 3) (of Divisor2 6))
All of the examples you gave had C as a relation of certain arguments,
as in Product or Sum.
In those cases there is only one "reasonable" class of OF statements possible
-- corresponding to those args.
However these examples above show classes are more general,
more than one set of non-binary relations are meaningful.
Two Possible Solutions:
1. There should be some mechanism for clustering these properties -- here
Addend1 and Addend2 should go together, as should Divisor1 and Divisor2.
OF can then have a merging rule similar to WITH's:
(AND (A C (of r1 v1))
(A C (of r2 v2)))
is
(A C (of r1 v1) (of r2 v2))
*** UNLESS r1 and r2 belong to the same cluster --
*** as would be the case if
r1 is (A Relation (with ClusterMate r2)).
[Alternately a converse relation ]
[ r1 is (A Relation (with IndependenceFrom r2)) ]
[could be necessary to make the connection should above.]
2. The "OF" clause could have an additional parameter, which does this clumping.
(AND (A C (of r1 v1 tag1))
(A C (of r2 v2 tag2)))
is
(A C (of r1 v1 tag1) (of r2 v2 tag2))
*** UNLESS (SAME tag1 tag2).
Comments?
Perhaps OF should retain its current definition, and this slightly
differnt attribution could be given another name, like Participates.
(This does spoil the monotonicity in length-of-characters found in
OF, With, WithEvery, WithUnique; but that's life.)
Russ
∂24-Nov-81 1108 Carl Hewitt <CSD.HEWITT at SU-SCORE> Re: A mergable flavor of OF
To: RDG at SU-AI
cc: csd.hewitt at SU-SCORE
Thanks for you comment. It think that it has a lot of merit and
deserves further thought. Perhaps it would be a good topic for discussion
next class.
Cheers,
carl
-------
∂TO chewitt@xx 22:00 19-Jan-82
(When this was first mailed, on 17 Dec 1981, to CSD.HEWITT@SCORE, it was
simply returned as undeliverable.)
∂TO CSD.HEWITT@SCORE 16:41 14-Jan-82
∂TO CHEWITT@SRI-KL 24-Dec-81
∂24-Dec-81 1749 The Mailer Daemon <Mailer at SU-SCORE> Message of 19-Dec-81 17:45:48
Date: 24 Dec 1981 1749-PST
From: The Mailer Daemon <Mailer at SU-SCORE>
To: RDG at SU-AI
Subject: Message of 19-Dec-81 17:45:48
Message failed for the following:
csd.hewitt at SU-SCORE: Undeliverable after 5 day(s).
------------
Mail-from: ARPANET site SU-SCORE rcvd at 19-Dec-81 1745-PST
Mail-from: ARPANET site SU-AI rcvd at 17-Dec-81 1647-PST
Date: 17 Dec 1981 1428-PST
From: Russell Greiner <RDG at SU-AI>
Subject: Misc comments on Omega
To: csd.hewitt at SU-SCORE
I promised to send a list of comments/suggestions/etc on Omega.
These of course reflect my personal biases; but nonetheless point
to things which I feel point to potential problems for any representation
language.
As a zero-th level comment, I was quite impressed with almost all of Omega,
and especially by its strong theoritical foundations, indicated by the
rather powerful theorems which have been proven.
The flavor of my objections derive from my particular vantage point
(sorry to mix metaphors) based on work on RLL/MRS:
the fact that Omega cannot reason about itself, and (hence) cannot modify
its behaviour.
The list below should instantiate this idea:
1. It seems quite limiting not making the various "operators" of your system
(including IS, OF, A(n), WITH, ...) into full fledged citizens of Omega.
Even in a finished system it is nice to have a set of descriptions of
the constituent parts of a system; and in a being-built system -
especially one which claims to be "self-describing" (AAAI-1, p.157) -
having such information seems essential.
My experience with RLL has shown me the power of using a
representational scheme for helping in the design and construction of a
complex system -- even when that system is (the implementation for) the
underlying language itself.
2. A closely related point deals with the control mechanisms themselves,
including that part of the system responsible for making inferences.
In Omega these too are invisible and unalterably by the user.
There are many arguments for placing them under user control.
The primary reason is the help it will provide you in furthering
the design and implementation of Omega. Rather than tweak LISP or
Act1 code, you can use the power of the existent system in modifying
itself (ala LISP's bootstrapping capability, only at a higher level.)
Another reason of the (desired) versatility of Omega -- different users may
have slightly different tasks, each with different needs and constraints.
Many users may want something quite similar to what you are providing,
with but a small, but essential modification. (Perhaps he has a slightly
different type of operator he wanted to use (such as the OF' connective
I mentioned last letter), or he wants the matching performed in a different
order, for efficiency.)
If Omega provides but a single, static control mechanism, the users only
options are (1) to suffer with the current system, (2) get you to send out
a hacker to cons up this "new improved system" (adequate only until he finds
the next "limitation"), or (3) use some other representation language.
Writing Omega's control in Omega itself provides a fourth possibility:
let the user himself "personalize" his copy of the system, as often as he needs.
A final plus deals with modelling another "belief system" (see point 4).
If Omega is powerful enough to handle its own reasoning mechanism,
it should be relatively straightforward to encode a different scheme.
This would allow Omega, for next to nothing,
to "simulate" another inferencing system,
e.g. to figure what that individual would believe, given his premises.
These arguments are advocating a rather strong and potentially dangerous position:
of permitting the user to construct the particular inferencing capability he wants.
It may be difficult to permit this generality and still guarantee
that Omega's nice properties (including soundness, ...).
Perhaps the user could be permitted to provide some "guidance" to the inferencing
process, within some strict guidelines to insure these properties of the
modified system.
(Exactly how this can be done remains an interesting reseach issue.)
An yet easier goal would be to allow the user to simply "see" the theory behind
Omega's inferencing (to justify how it reached a given conclusion,)
perhaps augmented with some "implementation level" information,
describing which functions were used, and how.
At least would de-"black box" its behaviour, and render its processing more
understandable.
(By the way, members of RLL/MRS crowd are not alone in this view.
See FOL/MetaFol, Chi (Cordell Green's post-Psi project),
or recent work on necessary extensions to PROLOG,
to improve the behaviour of its unifier/resolver.)
3. Omega's current descriptions of "class entities" seems imprecise, if not
outright embarassing. (That is, what exactly is meant by the description
"Female" -- especially how it is tied to descriptions like Mary,
who IS (A Female).)
Note this is not just an esoteric whim - there are cases when the same thing,
X, must be used as both the individual X and as a class descriptor (a X).
Consider, for example, Flight37. It is a member of the class of flights
-- as in
Flight37 IS (a Flight (with Company PSA) (with Destination LA) ...)
At the same time we may want to discuss the plane which left the runway
on Sept 3. That is,
Flight37-on3Sep IS (a Flight37 (with DepartureDate 3-Sep)
(with Pilot Fred) ...)
(We could go on to discuss the many parts of this particular flight --
eg distinguishing the flight by which plane. Here we may be forced to
define Flight37-on3Sep-Step1 IS (a Flight37-on3Sep (with Plane p17)
(with from 2:47) (with to 3:03) ...)
Etc, etc, etc.)
The point is this is a natural way of talking about events, which Omega
seems unable to encode -- or at least has difficulty describing.
4. It is often useful to discuss various belief systems, each as an entity.
(eg Fred'sBelief IS (A Belief-System).)
It wasn't clear to me that Omega's "perspectives" could handle this --
could we describe similarities between Fred'sBelief and George'sBelief?
Or deduce that Fred didn't "believe" in Modus Ponens, based on the facts
found (and omitted) in this corpus of believed "truths"?
Point 2 above proposed formalizing Omega's own inference system, and encoding
it within Omega.
Any system capable of doing this could readily encode other schemes, and use
them to advantage. (EG resourse limited processing, or default reasoning, or ...)
5. How does Omega deal with existential variables? Suppose we know that
there is a tall blonde man with one black shoe.
How would this be stated in Omega?
Of course the expression
(a Person (with Gender Male) (with Size Tall) (with ...))
is lacking -- there may be no individuals which satisfy this description.
Perhaps adding the fact that this description "satisfies" the Individual
predicate would work, but I'll need to be convinced.
Permitting these existential entities opens a bundle of issues relevant to
intensional objects in general.
The statement
"The postman who ate dinner here, bit the dog."
should be stored differently from
"The postman who bit the dog, ate dinner here.".
This definitional/assertional dichotomy IS important;
and any system which simply ignores this point is avoiding a major issue.
It should be possible to state that a pair of descriptions both refer to the
same individual(s); and I do not feel the current SAME relation covers
all the intricacies of coreferentiality when dealing with intensional objects.
------
Mike provided a set of questions, designed to challenge the capacity of any
system which is trying to encode facts about the world. These include:
1. All purple mushrooms ar poisonous.
2. Everybody loves somebody. (both meanings)
3. Purple mushrooms have many attributes.
4. There are 23 purble mushrooms in the world.
5. Purple-Mushroom-31 has been accessed 12 times.
6. The fact that a purple mushroom is purple is definitional.
7. The integers mod 5 form a group under addition.
8. John gave Mary the book.
9. Apple-56 embodies the theory (or concept) of apples.
10. The color of all singleton sets is red.
11. The alleged thief hated the poor violinist.
12. Most elephants are grey.
13. John poured some of the liquid out of the bottle.
14. Tower of Hanoi (that is, the procedure for its solution,
together with other pertanent facts.)
15. The system's matching procedure is the lisp function Match#234.
You might enjoy the challenge of representing these in Omega.
-------
These points may be totally irrelevant to your task -- if so, feel free to
ignore this whole message. However, there are many language designers who
have found themselves haunted by their systems' inadequate treatment of
these issues, and wish they had been aware of these points when they were
building their systems. Anyway, (on that ominous note,) let me know if
you'd like to discuss any of these points.
Russ
-------
∂TO BUCHANAN@SUMEX 13:35 29-Dec-81
Things everyone should know...
Bruce -
This is part of the message I sent to Lindley:
-------
A friend asked me what sorts of things us AI types should know, outside
AI proper. (Yes, there is an inherent circularity in this, in that anything in
another field which is found truly relevant will eventually find itself
"borrowed/stolen" into some AI system...)
Anyway, my list would include (in no particular order)
1. Tarski - semantics
2. Whorf/Sapir - "you can't think what you can't say"
3. Wittgenstein - familiar resemblence
4. Philosophy of cognition - as in Behaviorism thru Abstract Functionalism
5. emic/etic distinction (from Pike (phonetics) on)
6. the idea of conceptual categories
a) basic idea, shown in works by Bower, Rosch
[& Shepard/Halperin in music, Simon for Chess, ...]
b) Miller's 7+/-2 paper - for short term memory (of these categories)
c) Kay's basic color terms
7. Rough idea of neurophysiology
a) visual perception -- [Heubel and Weisel, Gregory, ...]
b) centers of Speech (as shown by aphasaics)
c) split brain (Sperry et al)
8. Notion of a scientific paradigm (as expounded by Kuhn, etc)
[I'm skipping many other important things which most CS people encounter anyway.
Included are
Chomsky - notion of grammar,
van Neumann - theory of games,
Shannon - information theory
superficial psychology - especially behavior "production rules"]
----
All of these seem to say something about intelligent behaviour, or analysis
of such.
Any additions/deletions/modifications?
------
Russ
∂29-Dec-81 1439 Buchanan at SUMEX-AIM Re: Things everyone should know...
I think you should circulate your list around the HPP and see what
additions you get. My file <buchanan>HPP.DIS at Sumex is a current
distribution list.
I'd add a number of things:
* propositional logic
* predicate logic
* set theory
* axiomatization of theories
* the Copernican revolution
* Mendelian genetics
* Watson & Crick
* chess
* checkers
* backgammon
* organic chemistry
* Newtonian mechanics
* relativity
* statistics
Polya
... <how does one ever stop?>
I think you were looking for "one-liners" like 7+/-2. Maybe these
expand into a few of those.
bgb
-------
∂TO BUCHANAN@SUMEX 15:29 6-Jan-82
A little knowledge ...
That's quite a list!
Your generator is clearly well tuned -- but its apparent
I never defined the tester.
My goal was to enumerate things which a competent AIer need never know,
but which might, nonetheless, prove useful in designing and implementing
an arbitrary AI program.
By "need never know" I meant things which an AI qual examiner would not ask.
This description alone is way too encompassing --
given the premise that any knowledge is intrinsically beneficial,
that list would need to include essentially everything, (outside of AI proper).
I want to exclude facts which pertain only to specific domains
(eg about organic chemistry, or chess).
While they are essential for coding up the rules for this particular domain,
their utility is limited to that application.
I was looking for more general types of facts.
Hence, while I would not include chess and checkers,
I did include the theory of games itself.
This guideline prunes the first four suggestions
(propositional logic, predicate logic, set theory and axiomatization of theories)
from your list, as these are legal tender for AI qual questions.
I "objected" to the games chess, checkers and backgammon on the basis of
specificity; the same criterion rules out organic chemistry, Newtonian mechanics
and relativity.
I wasn't sure why you included Watson and Crick -- their work, while quite
interesting, seems irrelevant. Only Molgen-ers would have any use for it;
and, except to glean an historical perspective, even their interests would
lie more with the more recent developments in DNA research.
Similarly Medelian genetics presents a rather small picture. His results
are, by today's standards, superficial; and his method is now known to be bogus.
The Copernican revolution, per se, seems inconsequential. The concept of
scientific revolutions, however, is clearly important, and was listed.
Statistics is an important area to science in general.
Its intersection with AI,
including things like CF in MYCIN, Zadeh's fuzzy set theory, and
Bayesian "a postiori reasoning",
are already required knowledge for any AIer.
What other areas do you consider pertanent?
The same sort of comment applies to Polya's work.
Much of his contributions are already in the required part of AI --
especially his usage of things like analogy and heuristics.
Anyway, I did mention these last two things in the "things already known"
section.
Further suggestions? Perhaps you could indicate some relevant works on the
science/philosophy of causation, or temporality, or ...
Thanks,
Russ
∂TO CSD.PCOHEN, CSD.DIETTERICH, STT 12:17 9-Jan-82
Other interesting things
I'm sorta compiling a list of things which a competent AIer need never know,
but which might, nonetheless, prove useful in designing and implementing
an arbitrary AI program.
By "need never know" I meant things which an AI qual examiner would not ask.
This description, by itself, is way too encompassing --
given the premise that any knowledge is intrinsically beneficial,
that list would need to include essentially everything, (outside of AI proper).
So please exclude those facts which pertain only to specific domains
(eg about organic chemistry, or chess).
While they are essential for coding up the rules for this particular domain,
their utility is limited to that application.
I'm trying to generate general types of facts.
Hence, while I would not include chess and checkers,
I did include the theory of games itself.
The list so far is
1. Tarski - semantics
2. Whorf/Sapir - "you can't think what you can't say"
3. Wittgenstein - familiar resemblence
4. Philosophy of cognition - as in Behaviorism thru Abstract Functionalism
5. emic/etic distinction (from Pike (phonetics) on)
6. the idea of conceptual categories
a) basic idea, shown in works by Bower, Rosch
[& Shepard/Halperin in music, Simon for Chess, ...]
b) Miller's 7+/-2 paper - for short term memory (of these categories)
c) Kay's basic color terms
7. Rough idea of neurophysiology
a) visual perception -- [Heubel and Weisel, Gregory, ...]
b) centers of Speech (as shown by aphasaics)
c) split brain (Sperry et al)
8. Notion of a scientific paradigm (as expounded by Kuhn, etc)
Other considered things:
Something on causality (by who ever did it? - like Plato)
Or Hermaneutics (Heidinger(sp) ...)
[I'm skipping many other important things which most CS people encounter anyway.
Included are
Chomsky - notion of grammar,
van Neumann - theory of games,
Shannon - information theory
Polya - heuristics, math education
superficial psychology - especially behavior "production rules"
statistics - probability, distributions]
----
All of these seem to say something about intelligent behaviour, or analysis
of such.
Any additions/deletions/modifications?
------
Thansk,
Russ
∂09-Jan-82 1359 Tom Dietterich <CSD.DIETTERICH at SU-SCORE> Re: Other interesting things
To: RDG at SU-AI
cc: CSD.DIETTERICH at SU-SCORE
I'm not sure what to make of your list. Why exactly are you creating
such a list? Are you trying to build a reading list for yourself?
Someone else?
Your list includes many different items at different levels of detail.
For instance, you mention an entire area, like "Philosophy of
cognition", and then talk about specific parts of that area, like
conceptual categories. Some things are technical (e.g., tarksi,
chomsky) while others are philosophical. Here is an alternate
organization:
A. Logical and mathematical models of cognition:
- Tarskian semantics
- Chomskian grammars and theory of language
- Von Neumann's theory of games
- Shannon's mathematical theory of information
- Polya's speculations on heuristics and math education.
- Probability and statistics.
<<<<huh - in what sense is Tarskian semantics, or Probability & Statistics,
related to models of cognition?>>>>
B. Philosophical discussions relevant to cognition
[Basically all of philosophy is relevant, in my opinion, with the
possible exception of religious philosophy and ethics.]
- Analytic philosophy of meaning and reference: Whorf/Sapir,
Wittgenstein, and other work (causal theories of
reference--Boyd).
- Philosophy of cognition: Behaviorism thru Abstract Functionalism.
Conceptual categories (Bower, Rosch, Miller, Kay)
- Epistemology, especially philosophy of science: Kuhn, Bohm,
Shapere, Suppe.
- Hermaneutics, Maturana
- Philosophy of perception (including philosophy of
art--representation). emic/etic distinction (?).
<<<<what of politics? Why not exclude this, if you exclude religion?
Do you not know the emic/etic distinction? It has to do with cultural biases,
and it's effects of perception...>>>>
C. Psychology of cognition
[You left out a lot here, don't you think?]
- Miller, Bower, Simon (I don't know how to split these guys
between philosophy and psychology).
- "superficial psychology, esp. production rules?"
D. Biology of cognition
- visual perception -- [Heubel and Weisel, Gregory, ...]
- centers of Speech (as shown by aphasaics)
- split brain (Sperry et al)
I think the important thing for an AI person to know is to have a fair
understanding of what the issues are in each of these areas, so that
any AI theories that you develop don't sound outright stupid to people
in these other fields. A case can be made for knowing some of the
results (as well as the issues raised) in psychology. I regard such
data as merely suggestive. Most of it is gathered under highly
restricted assumptions, assumptions I am not willing to grant.
--Tom
-------
Oh, so that's what I was trying to do!
∂TO %score csd.smith, csd.dietterich 13:37 3-Feb-82
Constraining Constraints
I don't quite buy DE2's suggestion that any proposition is a constraint.
First, I think the term "constraint" implies more than just any fact.
Tell me in what way the proposition (= 4 4) constrains a curcuit design;
or (IF (AND A B) B), or, for that matter, (Exists X. ¬X=X)?
It seems the constraint must (ie is constrained) to be relevant to the problem
at hand -- at least in the same language.
(Dave did mention that a constraint is always wrt some problem or context --
I'm just refining that idea of context a bit. Or rather begging it one further
degree: We now "just" need to know what is relevant...)
Secondly, I think there are some constrains which are difficult (if not absolutely
impossible) to express in predicate calculus.
I would claim that the frame assumptions form a constraint on (the solution to)
the M&C problem. Considering the difficulties JMC has had in formulating his
circumscription stuff, there is clearly no (know) natural ways of expressing
this restriction, formally.
In this same category are "inferencing constraints".
An extreme case is the resource limited processor TW talks about.
Clearly their behaviour CONSTRAINS the problems they can solve; and the
nature of the solution obtained as well.
(The Whorf/Sapir hypothesis can likewise be viewed as a constraint.)
Yes, I agree these types of constraints could be represented;
but not in a simple, domain-level, first-order predicate calculus system.
-----
It seems premature to split hairs now on a precise definition for this
loosely (mis)defined term "constraint".
Naively speaking, a constraint is something which helps refine/restrict/verify
a solution to a problem. As TGD mentioned, it might be incorporated in the
generator (causing it to produce only a subset of the candidates it would
otherwise suggest),
as well as the more obvious place, in the tester (as a predicate which any
candidate must pass).
[Dendral's 1/2-order theory, which guides the planner, is a constraint as well.]
Taking a behaviouristic view, let's examine what sorts of things one can do
with a constraint (or two).
----
One can *Post* a constraint (on the solution to a problem).
[Generate?]
-- this renders many potential solution candidate unacceptable...
-- either they won't be generated, or they'll be immediately pruned.
One can *Merge* two or more constraints.
-- Both constraints must now be satisfied.
One can *Propogate* the effects of a constraint.
One can *Apply* a constraint to something.
One can *Satisfy* a constraint.
One can *Resolve* a constraint.
One can *Simplify* a constraint.
One can make *Explicit* a constraint.
One can *Violate* a constraint.
[Refute]
One can *Justify* a constraint.
[Rationalize]
----
Now what? Anyone want to consider, formally, what each of the operations do
to the implicit search space?
Comments?
Russ
Tom, Dave: Once you've added your respective comments, could you forward to
the rest of the Sussman group? (I don't have the distribution list.) Thanks.
∂03-Feb-82 2311 David E. Smith <CSD.SMITH at SU-SCORE> Re: Constraining Constraints
To: RDG at SU-AI
cc: csd.smith at SU-SCORE, csd.dietterich at SU-SCORE
In-Reply-To: Your message of 3-Feb-82 1337-PST
You misinterpretted my statement: Not every proposition is a constraint for
every problem. The relevance of a proposition to the solution of a problem
is however impossible to determine a priori. Take for example (from the
Smith & Genesereth draft paper on the subject) the proposition that you own
a dog named Herby, and the problem of deciding whether to vote for Carter or
Reagan. It is impossible to determine whether these facts will enter into your
solution ("constrain it") without going through the full motion of solving the
problem. In making the decision one might wish to consider the ultimate
ramifications of the politicians policies on dogs and their owners.
Unfortunately whether a proposition bears upon the solution of a problem
depends on what other facts are present in the database. One can imagine
two ways out of this difficulty: 1) Explicit meta-knowledge about what
things are relevant to what problems, i.e. knowledge about the independence
of various "subjects", and 2) Meta level proofs (arguments) that it is not
syntactically possible for given statements to participate in the satisfaction
of a given constraint. These arguments are relatively easy for such statements
as "a=a".
Second, noone ever said anything about "first order" or "base level". In fact
when we read McDermott I think you'll see that his "policies" (which certainly
agree with my intuative notions of constraint) are meta-level propositions
about "orderrings" and resource limitations.
I found your statement about the adequecies/inadequecies of predicate calculus
to be totally off the wall. If you can imagine some language which is not
fundamentally based upon the notion of a "relationship" then I would love to
hear about it. Noone has yet managed to construct a language in which things
could be clearly said which cannot be said using meta-level or omega-order
predicate calculus. (In case you're inclined to jump on my back, remember
that I'm talking about language and not implementation data structures. My
statements do not rule out the use of analogical representations such as bit
vectors and arrays.)
Your list of things that can be done with constraints is basically what I set
out to do 6 months ago when I got ragged at the backhanded way various
reserchers were using the word constraint, and the term "constraint propagation"
I will try to get a draft of my attempts your way soon. Basically, in order
to make any sense out of those operations one must develop a formal "theory"
of problem solving (McDermott also has such a theory which I do not understand)
and figure out what each of these notions means in that context. In my view
many of these notions collapse or become vacuous, but that is too long a story
for a single message.
-- de2
-------
∂Snail Mailed to Doug Hofstadter -- 28-June-82
29 June 1982
Doug -
Thanks for the book -- I will indeed look through it.
You might be interested to hear about the "CPU" organization, to which
several of my friends belong. (I enclosed some of their information.)
Apparently Terry Winograd is involved, along with many other people at PARC.
(The community at large is quite interested as well -- "The Final Epidemic"
is showing locally almost every evening; and a recent (27/V/82)
Sigma Xi Lecture was on this topic -- by Drexel, of SLAC.)
One other note: I attended a talk given by Dr Dement, of the Sleep Research
Center here at Stanford. He began his talk by advertising this "TFE" film;
and later explained why: He has had two prominent dreams over the years
-- dreams whose realism were sufficient to cause him to change some aspect of
his (waking) life.
The first was some 20 years ago, when he dreamt he had terminal lung cancer.
He felt, in his dream, that he had cheated his wife and children;
and if only, only he had quit...
When he awake he resolved to quit smoking, a vow he has maintained until the
present. His other relevent dream experience was of a nuclear holocaust.
After this he joined Physicians for Social Responsiblity.
Anyway, thanks for your brief comments on my analogy paper.
I do offer a sincere apology for not citing your column.
In earlier drafts I used your "first lady of England" example in more detail
at a second place in my report, and cited your work there. That later
reference was later deleted. Sorry.
I will re-read your Chapter 19 in GEB -- it was several years ago when
I read your book; over which time I've obviously forgotten much of its
contents.
As to your "AI is ignoring me" anguish: That does seem unfortunate;
as you are indeed an eloquent spokesman. I don't know how representative
I am of that community, but I can volunteer my reaction when I read your
letter: I didn't realize you were particularly seeking an AI audience.
I, for one, have yet to read any (non-trivial) discription of your relevant
work. I'm referring here to SeekWhence in particular, which I know about
only from personal chats, and some quick asides in your column.
Neither of these of real AI-forums. Perhaps the AI community would
view you differently if they knew the projects on which you are working,
and of your approaches. Is there some reason your work hasn't appeared in an
article in, say, the AAAI proceedings? Another suggestion would
be to overview the current work at Indiana in the "Artificial Intelligence"
journal. (A la the description of Rutgers in the current issue.)
etc...
These are just simplistic ideas; but may provide another's
perspective on your situation...
I agree we should get together and discuss analogy-ish ideas
-- either in person or by mail. How should we arrange this "virtual meeting"?
Take care,
Russ
∂17-Feb-83 0923 DIETTERICH@SUMEX-AIM CPM knowledge base
To: rdg@SU-AI.ARPA
cc: DIETTERICH@SUMEX-AIM.ARPA, csd.genesereth@SU-SCORE.ARPA, buchanan@SUMEX-AIM.ARPA
Russ,
Presently, my CPM knowledge base is made up of two text files:
GRAMMAR.CPM and LOGIC.CPM. Take a look at the logic file first. It
defines the objects (files, etc.) of the domain and the primitive
actions that can be performed on them. Then look at the grammar file.
The grammar will be a little tricky for you to figure out, because I'm
still actively working on it and some things are just plain wrong.
The basic intuition is the following. Imagine that I give you a
PHOTO.LOG of the interaction between a user and CPM. Now, you must
make sense of it. This grammar lets you parse that interaction string
and infer what happened. (Of course the grammar can be used for
generation, etc., but it is easiest to think of it in terms of LR
parsing).
The hypothetical trace that Mike was showing you is stored in the file
TRACE.FID. This should be largely self-explanatory. The paragraph of
analogy statements that I was quoting today appears in my proposal
just before the hypothetical dialog there.
My understanding of the meeting today was that you are going to look
at this as a possible example of your problem of understanding an
analogy. I'm assuming that you and I will proceed on these problems
independently. Basically, I'm going to go ahead and choose a
reasonable starting state for my UNIX knowledge base (thus
side-stepping the whole question of understanding this analogy). I'll
still be dealing with analogy as an aid to interpreting output from
UNIX and as a guide to the plausibility of alternative theories of
UNIX.
--Tom
-------
∂17-Feb-83 1914 DIETTERICH@SUMEX-AIM CPM knowledge base files (addendum)
To: rdg@SU-AI.ARPA
cc: DIETTERICH@SUMEX-AIM.ARPA
Russ,
I just finished a major editing session on the CPM knowledge base
files. You will probably want to grab the new versions of them from
SUMEX as they are much improved.
--Tom
-------
∂15:02 10-Mar-83: ml/subj ↑Paige
Ben (Moszkowski, BCM@SAIL) sent these two pointers to Articles by Paige
on program optimizations by incremental transformations.
(He may know of other articles as well.)
-----
@Article(PaigeKoenig82,
Author = "R. Paige and S. Koenig",
FullAuthor = "Robert Paige and Shaye Koenig",
Title = "Finite differencing of computable programs",
Journal = "ACM Transactions on Programming Languages and Systems",
Number = 3,
Volume = 4, Month = JUL, Year = 1982, Pages = "402--454")
@InProceedings(PAIGE83,
Author = "R. Paige",
FullAuthor = "Robert Paige",
Title = "Transformational programming -- Applications to algorithms and systems",
Booktitle = "Tenth Annual ACM Symposium on Principles of Programming Languages",
Address = "Austin, Texas",
Month = JAN, Year = 1983, Pages = "73--87")
-----
Bob can be reached as Paige@Rutgers.
Enjoy,
Russ
∂11-Feb-83 2134 SLEEMAN@SUMEX-AIM An alg for analogies
To: greiner@SUMEX-AIM.ARPA
cc: sleeman@SUMEX-AIM.ARPA
Russ, I'd like your comments on a simple-minded alg I am
contemplating proposing its in <SLEEMAN>TMPPP1.MSS
I'd like to get this off soon so am early response
would be appreciated- I'd hope to catch you today at SIG-LUNCH
D
-------
∂TO sleeman@sumex 13:25 14-Feb-83
AN ALoGy
Derek --
I did read through the proto-proposal, and some a few minor
comments. Will you be in MJH today? Or did you want to call me
(x7-1863)? ... or whatever.
Russ
∂14-Feb-83 1702 SLEEMAN@SUMEX-AIM Re: AN ALoGy
To: RDG@SU-AI.ARPA
In-Reply-To: Your message of Mon 14 Feb 83 13:25:00-PST
Russ, I tried calling to collect your pearls of wisdom. Have decided to
pedal home in the light and continue there. Will try to call you from
there, or please call me at 856 4149.
I'm keen to get this off shortly
D
-------
∂02-Sep-82 2335 Keith Wescourt <G.WESCOURT at SU-SCORE> end of week already
To: rdg at SU-AI
Reply-to: Wescourt@ISI
sorry, i never got back to you about our tentative lunch
but I've been beset by work and forgetfulness. i
definitely want to get together soon-- maybe next week
i'll be more on the ball.
have a good weekend.
keith
-------
17:53 31-Mar: ml/su Winston's Latest
Mike -
Do you have Pat's best (= latest?) report on analogy?
Doug wants to hand out copies of that report to the Learning Seminar;
and my only remaining hard copy is circa 1981.
Thanks,
Russ
∂31-Mar-83 2207 ML
I have a report on using critics, which I believe is circa 1982 (I'll check
later). Try mailing BORIS%MIT-OZ a message, Pat doesn't read his mail file,
but BORIS would pass a message along. Better yet, he would mail a copy of
Pat's latest (I don't think the work on function and structure is what
DOug is interested in).
∂17:36 9-Apr: grinberg@diablo/su Alto Account
Milt -
How am I supposed to use the Altos, sans account? Is there
now an HPP account? If not, is there some other mechanism?
Russ
∂11-Apr-83 0855 grinberg@Diablo Re: Alto Account
Message-Id: <83/04/11 0850.816@Diablo>
In-Reply-To: Russell Greiner's message of 09 Apr 83 1736 PST
We have one account in Mike's name. I assume the password is secret.
But use your own storage media. Don't use IFS because everyone is
using Mike's account.
milt
∂07-Apr-83 2226 FRALEY.hp-hulk at UDel-TCP Arpanet/CSnet miracles
Received: from UDEL-TCP by SU-AI with NCP/FTP; 7 Apr 83 22:26:44 PST; for: greiner
Date: 6 Apr 1983 1232-PST
From: FRALEY.hp-hulk at UDel-TCP
Received: by HP-VENUS via CHAOSNET; 6 Apr 1983 12:37:18-PST
Subject: Arpanet/CSnet miracles
To: greiner at Su-Ai
Message-Id: <418509439.13573.hplabs@HP-VENUS>
Via: HP-Labs; 6 Apr 83 23:45-PDT
Via: rand-relay; 7 Apr 83 15:18-EST
This is an attempt at a miracle: Can the CS-Net, accessed through our
local net, manage to get a message through to Arpa-net?
-------
∂17:29 9-Apr: FRALEY.hp-hulk@UDel-TCP/subj It worked!
I received your msssage addressed (simply)to greiner@su-ai --
which went
Via: HP-Labs; 6 Apr 83 23:45-PDT
Via: rand-relay; 7 Apr 83 15:18-EST,
sent from FRALEY.hp-hulk at UDel-TCP.
Now to try the other direction.
Let me know if this goes through -- if so I'll send along the hike
guide.
Russ
∂11-Apr-83 2146 FRALEY.hp-hulk at UDel-TCP Re: It worked!
Received: from UDEL-TCP by SU-AI with NCP/FTP; 11 Apr 83 21:46:04 PST; for: RDG
Date: 11 Apr 1983 1157-PST
From: FRALEY.hp-hulk at UDel-TCP
Received: by HP-VENUS via CHAOSNET; 11 Apr 1983 12:01:26-PST
Subject: Re: It worked!
To: RDG at Su-Ai
In-Reply-To: Your message of 9-Apr-83 1729-PST
Message-Id: <418939287.17828.hplabs@HP-VENUS>
Via: HP-Labs; 11 Apr 83 20:45-PST
Via: rand-relay; 12 Apr 83 0:16-EST
I received your message. This time I'm trying a reply from my local
mail system. Your message was mailed at 5:30 pm and arrived here at
10 minutes to midnight. The biggest delay was getting out of the
su-ai machine: 3 hours. Much better, though, than my message to you.
-------
∂16:55 12-Apr: FRALEY.hp-hulk@UDel-TCP/su April hikes, etc.
Bob -
Below is the hike directory. I've asked your Chris to add you to
this list.
Also, don't forget to give Arleen the $94 this evening,
assuming she's there, and you get this message in time.
Russ
(w/Hike stuff)
∂15-Apr-83 1215 CSD.FRALEY@SU-SCORE Sunday's hike
Received: from SU-SCORE by SU-AI with PUP; 15-Apr-83 12:15 PST
Date: Fri 15 Apr 83 12:03:07-PST
From: Chris Fraley <CSD.FRALEY@SU-SCORE.ARPA>
Subject: Sunday's hike
To: hike-distrubution-list: ;
Meeting place is in front of the oval at 9, in case anyone was confused when
I said in front of the quad.
-------
∂TO demmel@berkeley 15:44 7-Feb-83
Peter Edberg
is the "Peter" I couldn't think of (who lived in Ricketts thru '75 or so).
Let me know if this message got through -- in this day of NCP and TIP
confusion, nothing is certain anymore. What is your preferred account?
Russ
∂08-Feb-83 0902 The Mailer Daemon <Mailer@USC-ECL> Message of 7-Feb-83 15:45:32
Received: from USC-ECL by SU-AI with NCP/FTP; 8 Feb 83 09:01:51 PST
Date: 8 Feb 1983 0907-PST
From: The Mailer Daemon <Mailer@USC-ECL>
To: <RDG@SU-AI>
Subject: Message of 7-Feb-83 15:45:32
Message failed for the following:
demmel@UCB-ARPA: 550 <demmel@UCB-ARPA>... User unknown
------------
Mail-from: ARPANET site SU-AI rcvd at 7-Feb-83 1545-PST
Date: 07 Feb 83 1544 PST
From: Russell Greiner <RDG@SU-AI>
Subject: Peter Edberg
To: demmel%UCB-ARPA@USC-ECL
is the "Peter" I couldn't think of (who lived in Ricketts thru '75 or so).
Let me know if this message got through -- in this day of NCP and TIP
confusion, nothing is certain anymore. What is your preferred account?
Russ
-------
∂TO lisa@berkeley 10:48 8-Feb-83
Path to Jim
Lisa,
What is the easiest way of sending a message to Jim Demmel?
I just tried Demmel@Berkeley, and got a
Message failed for the following:
demmel@UCB-ARPA: 550 <demmel@UCB-ARPA>... User unknown
error message.
By the way, how are you? (... and other related formalities ...)
Thanks,
Russ
∂08-Feb-83 1721 Kim.lisa@Berkeley Path to Jim
Received: from USC-ECL by SU-AI with NCP/FTP; 8 Feb 83 17:21:06 PST
Received: from UCB-VAX (for UCBVAX.BERKELEY) by USC-ECL; Tue 8 Feb 83 17:18:50-PST
From: Kim.lisa@Berkeley
Date: 8 Feb 83 10:59:36 PST (Tue)
Subject: Path to Jim
Message-Id: <8301081858.1371@UCBVAX.BERKELEY.ARPA>
Received: by UCBVAX.BERKELEY.ARPA (3.300 [1/17/83])
id AA01371; 8 Feb 83 10:58:52 PST (Tue)
To: @USC-ECL:RDG@SU-AI, lisa%UCB-ARPA@USC-ECL
Russ -
demmel@Berkeley should work! Try n:demmel@berkeley or v:demmel@berkeley
It should work! I am fine would you like to go skiing march 20 - 27
at Tahoe City?
I am a litte sore - but gotta go! See you soon,
Lisa
∂TO lisa@berkeley 12:47 15-Feb-83
Msg for Jim
Lisa -
Could you forward this message to Jim -- all of my attempts have failed.
Thanks.
∂TO n:demmel@berkeley,demmel@ucb-vax,demmel%ucb-arpa@usc-ecl 20:57 8-Feb-83
<<failed, at n:">>
∂TO "n:demmel"@berkeley,demmel@ucb-vax, demmel%ucb-arpa@usc-ecl 21:00 8-Feb-83
Subject: Three+ quickies
Jim -
How could both of us have forgotten to exchange our most relevant
(respective) addresses -- on the ARPAnet?!
Anyway, RDG@SU-AI, CSD.GREINER@SCORE, greiner@DIABLO, GREINER@SUMEX, greiner@Rand-Unix, ...
all get to me. I was surprised that the obvious DEMMEL@Berkeley failed.
Lisa's suggestion, of using a "N:" prefix, also failed.
Forgive the possible multiple copies of this message you may receive,
as I'm experimenting with names/machine-ids/...
Anyway, on to the (non-meta) messages:
1) Peter Edberg is the "Peter" I couldn't think of
(who was at CalTech, on and off, from about 72 thru 80; and lived
at times at Ricketts).
2) Jim Celoni says Hi. He can be reached as CSL.JLH.Celoni@Score.
3) What is your preferred ARPAnet address?
Caio,
Russ
∂26-Feb-83 1851 UCBKIM.demmel.berkeley at UDel-TCP (James Demmel) Hello
Received: from UDEL-TCP by SU-AI with NCP/FTP; 26 Feb 83 18:50:50 PST; for: rdg
Date: 25 Feb 83 22:28:07 PST (Fri)
From: UCBKIM.demmel.berkeley at UDel-TCP (James Demmel)
Return-Path: <UCBKIM.demmel@Berkeley>
Subject: Hello
Received: by UCBKIM.ARPA (3.310/3.5)
id AA11635; 25 Feb 83 22:28:07 PST (Fri)
Received: from UCBKIM.ARPA by UCBVAX.ARPA (3.314/3.5)
id AA18318; 25 Feb 83 22:18:09 PST (Fri)
Received: from UCBVAX.ARPA by udel-relay.ARPA ; 26-Feb-83 01:42:16-EST (Sat)
Message-Id: <8302260628.11635@UCBKIM.ARPA>
To: RDG at Su-Ai
Sorry I took so long to get back to you. Life has been pleasantly frantic.
I'm currently planning for Wilkinson's visit to Berkeley from Stanford next
week. He's been working on a problem lately which is a subset of my thesis,
and my results appear to be better than his, so I have to make sure I have
my act together. I have managed to find time to go out with Lydia, the
German woman we both met on the ski trip. She has a boyfriend back in
Germany, but she's fun and the social distraction is welcome.
Say hello to Celoni for me (Father Jim?). I'll be at Stanford March 12
for Numerical Analysis Day (aka Gene Golub's Birthday party) and maybe we
can get together for a while then.
There was just a message today about appropriate net addresses now that
we've changed protocols.
demmel@ucbkim.BERKELEY.ARPA
should be more than enough postage. Hope to talk again soon.
Jim
∂TO csl.jlh.celoni@score 12:49 27-Feb-83
↑Demmel
Jim -
I just got a return msg from Jim Demmel. He claims his best address is
demmel@ucbkim.BERKELEY.ARPA
(I've not tried that one.)
He'll be here at Stanford 12 March -- but warns he may be busy.
Russ
∂27-Feb-83 1706 CSL.JLH.CELONI@SU-SCORE Re: ↑Demmel
Thanks. +j
-------
----still didn't get sent----
∂TO demmel@ucbkim.BERKELEY.ARPA 16:15 1-Mar
∂TO demmel@ucbkim
∂TO demmel@"ucbkim.BERKELEY.ARPA" 16:19
Another day, another address...
Good to hear from you. Let me know your schedule for 12 March
when it becomes more settled; maybe we could get together then.
By the way, "demmel@ucbkim.BERKELEY.ARPA" didn't work.
Nor did the version sans "."s fare better. This time the "."s are
within a quoted string.
Jim Celoni says hi, again.
Russ
NO: demmel%ucbkim@BERKELEY/su This is a test
∂17:42 1-Jan: "ucbkim.demmel"@BERKELEY/su This is a test
Does this make it through?
∂21:11 2-Apr: csd.lenat@score/subj Upcoming talk
Doug -
Could we meet soon, to discuss what I should present to the
Learning Seminar? I just realized I don't have a sense of what you
want me to convey. Perhaps
(1) a synopsis of Winston's and Carbonell's articles,
augmented with a short description of my work.
(2) a more extensive literature review
(mostly of AI works, but including some sorta-relevant philo, educ, ...)
(3) a description of what analogy is and how it can be used,
concentrating on its relation to learning.
Number (3) corresponds to your learning overview, last Thursday.
Also, no one (including the CS library) has a copy of Winston's
latest report. The best I can find is his C-ACM paper, from 12/80.
Is that really worth xeroxing?
Russ
∂03-Apr-83 1455 CSD.LENAT@SU-SCORE Re: Upcoming talk
That all sounds right; use the CACM if no better. Yes, let's
talk about it; maybe Tuesday around 11ish?
Doug
-------
∂15:20 3-Apr: csd.lenat@score/subj Talk, and related (analogous?) things
Doug,
(0) Thanks for your quick response.
(1) Ah, I see. When I mailed the message I had meant for (1), (2) and (3) to
be disjoint. But no matter,...
Current proposed outline:
I. Intro
II. Why useful?
A. In genl
B. wrt Learning
III. Survey
A. PreAnalysis/Comments
B. non-AI Examples
C. AI Examples
D. Others, when contrasting with "primitives"
IV. Common Themes
V. Limitations of Examples
VI. My approach
...if time permits...
VII. Dimensions
VIII. Properties of Analogy
----
My notes are in {SAIL} anlgy.tlk[rdg,dbl]. Feel free to read thru the file,
if you wish.
(2) Tues 11 is ok. See you then.
(3) Should I ask Donna to generate 40 copies of Winston's article?
[If yes, I'll leave that magazine in her office this afternoon.]
Russ
∂03-Apr-83 1548 CSD.LENAT@SU-SCORE Re: Talk, and related (analogous?) things
Yes!
-------
∂14:39 18-Apr: FRALEY.hp-hulk@UDel-TCP, genesereth@sumex/cc/su re:MRS
(Msg from Bob to me, included for Mike's benefit)
I just saw your name attached to a document of MRS. What can you tell
me about MRS? The name came up in a discussion of theorem proving systems.
Is that what MRS is? If so, how adaptable is it to different methods of
proof? Also, which documents about the system are the most useful? Can
the software be acquired? What machine/Lisp system does it run on?
MRS is (historically) a representation language for AI systems, used to store
facts and draw inferences. It currently includes various inference schemes --
including backward- and forward- chaining (production systems),
and resolution theorem proving.
It will soon (does now?) have all the major features of Prolog.
(In addition, the MRS user can guide the inference,
to provide a far more efficient performance.)
It may later incorporate other proof techniques as well --
or the user can use the "handles" MRS provides to "roll his own".
Documents -- using the MRS Tutor is probably the best tactic for
getting aquainted with the system. An example dialogue composes one
of the documents. The other documents (esp the manual), designed
for users with some familiarity, are then usable.
Yes, the software is available. If you are interested, I'll send a
copy of the "Request Letter".
Implementation: it currently runs on MacLisp, Franz Lisp, and
(Symbolics) LispMachine Lisp.
(Bringing up an InterLisp version is currently on the (far) back burner.)
Anyway, give me a call if you have other questions; I'd be glad to use the
higher bandwidth to better describe the system.
Russ
∂20 Apr 83 20:49:23-PST <BARR@SUMEX-AIM> friday
To: greiner
Russ,
Can we please reschedule--I am a little confused these days and
Friday I am busy all day. I will be meeting with Ed in Jacks "all day"
tomorrow (Thursday), which means that I will be free most of the time while
he is processing "interruptions", several of which he has already
scheduled. I will look for you then, and if you are free we can talk, else
we can reschedule for next week. Sorry.
.Av
-------
∂15:22 25-Apr: BARR@SUMEX-AIM/su Meeting Time
How about Tuesday (tomorrow) afternoon, at say 1:30PM?
or late-ish Wednesday (~3:30), or ...
.Russ
∂25-Apr-83 1701 BARR@SUMEX-AIM Re: Meeting Time
How about lunch Thursday. I am preparing for a speech Wednesday
evening and I won't be able to pay much attention until it is over!
I have meetings with bruce (11:00) and Bill Clancey (1:30) on
thursday, so if luch is out, maybe 10:00 am?
.Av
-------
∂17:29 25-Apr: BARR@SUMEX-AIM/su Ungood
I've a lunch meeting on Thursday. 10AM is acceptable, but undesirable.
Is Friday before SigLunch ok? If not, I'll be here at 10am on Wednesday.
.Russ
∂25-Apr-83 2058 BARR@SUMEX-AIM Re: Ungood
friday before siglunch it is. see you then
.Av
-------
∂27 Apr 83 17:41:44-PDT <FEIGENBAUM@SUMEX-AIM> phone call
To: greiner
Shmuel just called. Said he was sick. Call him at home.
Ed F.
-------
-------
∂19:58 28-Apr: FEIGENBAUM@SUMEX-AIM/su Thanks...
... for conveying the message. It's nice knowing that, in a pinch like this,
I can be reached. (Although I don't know how/why it went through you.)
Russ
∂16:31 9-Mar-83: dekleer@parc, csd.lenat@score/subj Less Procedural Rules
Johann -
How is your paper coming? Any news on the electrical "rule"
formulation front? (And I'll, of course, ask the anticipated) Did your
bike patch hold?
Do let me know when a draft of your paper is perusable.
(Don't mean to be rushing you; think of this as assurance that I was not
just being polite in requesting that info, but am genuinely interested.)
If you have time, it would be nice to chat in the near future,
perhaps with Doug.
Russ
∂10-Mar-83 1559 dekleer.PA@PARC-MAXC Does this get to you?
Received: from SU-SCORE by SU-AI with PUP; 10-Mar-83 15:59 PST
Received: from PARC-MAXC.ARPA by SU-SCORE.ARPA with TCP; Thu 10 Mar 83 15:56:13-PST
Date: Thu, 10 Mar 83 14:19 PST
From: dekleer.PA@PARC-MAXC.ARPA
Subject: Does this get to you?
To: RDG%SU-AI@SU-SCORE.ARPA
-------
∂14:00 11-Mar-83
dekleer@parc/sub Yep,
I did get your "Does this get to you?" message,
(the one mailed to "RDG%SU-AI@SU-SCORE.ARPA", in case there were several).
FYI, the header included
∂10-Mar-83 1559 dekleer.PA@PARC-MAXC Does this get to you?
Received: from SU-SCORE by SU-AI with PUP; 10-Mar-83 15:59 PST
Received: from PARC-MAXC.ARPA by SU-SCORE.ARPA with TCP; Thu 10 Mar 83 15:56:13-PST
Date: Thu, 10 Mar 83 14:19 PST
(so it took 3:59 - 2:19 = 1 hr 40 minutes to reach me.)
Is this another mail system a viable example of material flow?
What equations govern its behavior, I wonder?
Feel free to send things to CSD.GREINER@SCORE or GREINER@SUMEX
if that simplifies things for you.
(Or I can be reached over the Ma Bell network: 497-1863.)
Russ
∂15:46 30-Mar: deKleer@Parc/subj Any news?
Johann -
Greetings! How are your papers coming? Are any drafts at the
perusal state yet? Any new thoughts on (water) circuits, or analogy?
Waiting with bated breath,
Russ
∂04-Apr-83 1740 dekleer.PA@PARC-MAXC.ARPA Any news?
To: RDG%SU-AI@USC-ECL.ARPA
Delay while we quickly churned out a AAAI paper. It is really an
excised section we decided to chop out of the longer paper. I just
mailed it to you for your comments.
-------
∂07-Apr-83 1905 DEKLEER.PA@PARC-MAXC Paper
I've sent you a version of the longer paper also.
-------
13:02 9-Apr: deKleer@Parc/subj Great!
Johann,
I've received both papers, and look forward to perusing them soon.
I'm currently "swapped out", preparing for an upcoming two-hour presentation,
for this Thursday. But I hope we can meet after that...
Russ
∂28-Mar-83 1349 CSD.FRALEY@SU-SCORE April hikes
From: Chris Fraley <CSD.FRALEY@SU-SCORE.ARPA>
To: hike-distrubution-list: ;
Sunday, April 3
Mt. Diablo (about 1hr drive northeast - off I-680 east of Oakland/Berkeley)
- The proposed hike is the 7 mile trail around the mountain, which has lots
of ups and downs. There are many other trails, so we can split up into
more than one group if we want.
Sunday, April 17
Pinacles Nat'l Monument (about 2hr drive : south via US 101 through Gilroy,
then CA 25 through Hollister)
- If you've only been here in the summer, you probably wonder why this place
is a national monument. The wildflowers in spring are worth the trip. There
are some good views from the High Peaks which can be reached by several not
terribly steep trails. Also there are some caves near the Park HQ, so those
who want to go in them should bring flashlights.
Sunday, April 31
Los Trancos trail, Foothills Park, Palo Alto
- This is the longest trail in Foothill (7.5 miles). We need at least one
person who can "prove" they are a Palo Alto resident in each car.
Future
Skyline trail
- This trail parallels Skyline Drive and runs from King's Mtn Rd to Sky Londa.
Total distance is 8 miles; lots of views. We'll have to figure out some
way to coordinate cars; probably having a group start from each end would
be easiest.
Big Basin revisited
- Some of us may want to make this an overnight and do the entire Skyline
to the Sea trail.
this schedule is tentative and subject to veto
feedback is appreciated : comments, criticism, and suggestions
this info is available in file hike.info on my score account
unless you hear otherwise, meet at 9AM in front of the quad
Ruth Ossher has suggested that there be a sort of yum yum guide for hiking
trails in the area. Anyone who wants to make contributions can send them to
me and I'll start a file.
-------
∂01-Apr-83 0628 CSD.FRALEY@SU-SCORE Mt. Diablo
Received: from SU-SCORE by SU-AI with PUP; 01-Apr-83 06:27 PST
Date: Fri 1 Apr 83 06:27:38-PST
From: Chris Fraley <CSD.FRALEY@SU-SCORE.ARPA>
Subject: Mt. Diablo
To: hike-distrubution-list: ;
Just reminder that the Mt. Diablo hike is this Sunday. In case anyone is
worried about road conditions there, the North gate is closed, but the
South gate (the one we will use is open. Hope the weather will be good
for a change. Chris
Meet at 9 at the oval.
-------
∂05-Apr-83 1412 CSD.FRALEY@SU-SCORE next hike
To: hike-distrubution-list: ;
According to hike.info, the next hike is Sunday April 17 to Pinnacles Nat'l
Monument. I just went to Pinnacles on Saturday, and think we should postpone
the Pinnacles trip to April 31st to catch the wildflower activity at its height.
Perhaps someone else would like to suggest an alternate for the next hike.
Otherwise it will be the Los Trancos trail in Foothill on April 17. Chris
-------
∂11-Apr-83 1505 CSD.FRALEY@SU-SCORE next hike
To: hike-distrubution-list: ;
The next hike will be the Los Trancos trail in Foothill Park (7.5 mi) on
Sunday April 17. Palo Alto residents should be sure to bring ids; we need
to have at least one Palo Alto res. in each car. Those who are allergic to
poison oak are going to have to be on the alert in Foothill at this time
of year. Meet at 9 in front of the quad as usual. Hope the weather is good.
Chris
Those who are new to the hike distr. list may want to have a look at the file
hike.info on my score account.
-------
∂16:57 12-Apr: CSD.FRALEY@SU-SCORE, FRALEY.hp-hulk@UDel-TCP/cc/su New Hike-ee
Chris,
please add (your namesake) Bob Fraley to the
hike-distrubution-list:. His ARPA-net(-ish) address is
FRALEY.hp-hulk@UDel-TCP.
(I tested it - it even seems to go thru, despite the myriad of gateways.)
Thanks,
Russ
∂15-Apr-83 0749 CSD.FRALEY@SU-SCORE Sunday's hike
To: hike-distrubution-list: ;
Just a reminder about Sunday's hike - Los trancos trail in Foothill.
Meet at 9AM in front of quad. No RSVP necessary. Palo Alto residents
please bring ids.. Chris
-------
∂17-Apr-83 2209 FRALEY.hp-hulk at UDel-TCP MRS
Subject: MRS
To: greiner at Su-Ai
I just saw your name attached to a document of MRS. What can you tell
me about MRS? The name came up in a discussion of theorem proving systems.
Is that what MRS is? If so, how adaptable is it to different methods of
proof? Also, which documents about the system are the most useful? Can
the software be acquired? What machine/Lisp system does it run on?
Boy, the questions sure come out fast sometimes.
I have sent your check in the mail, since I forgot to give it to Arleen.
Where do people gather for the Sunday hike? It says in front of the quad;
I assume that is on the Palm Drive side of the walkway that leads to the
Chapel. Bringing a lunch is in order?
-------
∂13:55 18-Apr-83: FRALEY.hp-hulk@UDel-TCP/su non-MRS issues
Bob -
First, I just spoke with Chris -- I'm glad you went on the hike,
and (hopefully) enjoyed it. (Despite my best-laid plans, I didn't quite
make it.) Anyway, you now know when and where to meet, etc, I assume.
As SAIL's connection to HP seems more solid than SCORE's, I'll forward my
copy of further hike messages to you -- if Chris's start going thru,
let me know and I'll stop.
Second, I did receive (and deposit) the check; thanks for
getting it to me.
I'll send you a message, CCed to Mike Genesereth, re: MRS soon.
Russ
∂12:55 20-Apr: FRALEY.hp-hulk@UDel-TCP/su Did you get this message?
By the way, Chris' extension is 497-2902.
∂19-Apr-83 1059 CSD.FRALEY%SU-SCORE.ARPA@SCORE next hike
To: hike-distrubution-list: ;
The next hike will be on Sunday, April 31. The hike was tentaively going to be
to Pinnacles Nat'l Mon., but it seems several people (including myself) have
recently been there; also the Outing Club is sponsoring a trip to Pinnacles this
Saturday April 23. I can get details for anyone who is interested, or you can
find out from the I-center. Right now I am soliciting suggestions from people
as to where they would like to go on the 31st. Chris
-------
∂22-Apr-83 1317 CSD.FRALEY@SCORE mail
To: rdg@SAIL
I'm definitely no able to send any mail to Bob F.; the address I have for him
is FRALEY.hp-hulk@UDel-TCP. Is this what you use? Chris
-------
∂~22:Apr: FRALEY.hp-hulk@UDel-TCP/su
∂22-Apr-83 1139 CSD.FRALEY@SCORE next hike
Received: from SU-SCORE by SU-AI with PUP; 22-Apr-83 11:38 PST
Date: Fri 22 Apr 83 11:40:02-PST
From: Chris Fraley <CSD.FRALEY@SU-SCORE.ARPA>
Subject: next hike
To: hike-distrubution-list: ;
The suggestions for the next hike on April 31 (alias May 1) are:
Muir Woods
Angell Island (anyone into swimming?)
Forest of the Nisene Marks (vaguely near Santa Cruz)
Point Reyes
No one has said they're eager to go to Pinnacles, and it seems that the die-
hard hikers aren't too fussy, so we probably won't go there since it's a 2
hour drive. If anyone else has any inclination, they should let me know by
early next week. Thanks, Chris
-------
∂22:44 25-Apr: FRALEY.hp-hulk@UDel-TCP/su
∂25-Apr-83 2101 CSD.FRALEY@SCORE next hike
To: hike-distribution-list: ;
The next hike will be on Sunday April 31 (May 1), and will be to Coe State Park
(off 101) near Morgan Hill. I was just there on Saturday ad got soaked, but I'm
anxious to go back on a sunny day since the wildflowers are great there now.
The park hq has a list of hiking loops in order of increasing length, on top
of their topo map, so we should be able to find something to suit everyone.
The drive is about 1.3 hr, so it isn't as long a distance as Pinnacles or Pt.
Reyes. Chris
It seems that several people are interested in Pt. Reyes and Angell Island, so
they are possibilities for future hikes. I will get info on the ferry to
Angell Island. These places near the sea stay cool, so they are still OK for
late spring / summer hikes, whereas places like Coe or Pinnacles get too hot
for hiking.
Barry Hayes sent me a message asking if anyone knows of a place where people
can hike with dogs off leash. Does anyone know of any?
-------
∂29-Apr-83 1159 CSD.FRALEY@SCORE hike reminder
To: hike-distribution-list: ;
Next hike is this Sunday May 1, to Coe State Park. Hope the weather
will be good. As usual, I'll listen to 7AM weather report on Sunday, so anyone
who wants can cal;l me Sunday AM to find out. Otherwise meet near the
front of the oval at 9 as usual. Chris
-------
∂ 8 February 1983 11:11 est Johnson.RADC←MISu at RADC-MULTICS Test
Received: from RADC-MULTICS.ARPA by SU-SCORE.ARPA with TCP; Tue 8 Feb 83 08:15:12-PST
From: Johnson.RADC←MISu at RADC-MULTICS
Return-Path: Johnson@RADC-MULTICS.ARPA
To: CSD.GREINER at SU-SCORE
How is the weather in california, it is cold in New York and fouinches
of snow. Please reply to Johnson -at RADC-MULTICS thanks
∂14:59 2-May- "johnson.RADC←MISu%RADC-MULTICS"/su Success?
I tried several times in Feb to respond -- mailing to
johnson@radc-multics.arpa [failed because of "." ]
johnson@radc-multics [failed because entry not found ]
"johnson.RADC←MISu"@radc-multics [Some directory in path specified does not exist. ]
johnson.RADC←MISu%RADC-MULTICS@USC-ECL [just failed...]
I then gave up, figured I'd wait until SAIL became a TAC before trying again.
Having been "brought up" yesterday, I tried again.
Johnson@RADC-MULTICS
"johnson.RADC←MISu"@radc-multics
both failed, so I'm now trying
johnson.RADC←MISu%RADC-MULTICS@USC-ECL
Let's see if that works now.
Anyway, if it's not to personal,
and if you'll forgive both my ignorance and my curiousity,
who are you?
I hope the snow has subsided -- the percipitation here has been limited to
rain in form, but seems unlimited in quantity.
Russ
∂15 Apr 83 09:40:42-PST <CSD.TAJNAI@SU-SCORE.ARPA> Sohio recruiting AI PHD students
To: csd.barr@SU-SCORE.ARPA, csd.chelberg@SU-SCORE.ARPA, csd.berlin@SU-SCORE.ARPA,
csd.pcohen@SU-SCORE.ARPA, csd.davidson@SU-SCORE.ARPA,
csd.dietterich@SU-SCORE.ARPA, csd.jjf@SU-SCORE.ARPA, csd.gardner@SU-SCORE.ARPA,
csd.greiner@SU-SCORE.ARPA, ccg@SU-AI.ARPA, csd.jock@SU-SCORE.ARPA,
csd.rosenschein@SU-SCORE.ARPA, csd.rowe@SU-SCORE.ARPA, csd.smith@SU-SCORE.ARPA,
mah@SU-AI.ARPA
cc: csd.Feigenbaum@SU-SCORE.ARPA, csd.tajnai@SU-SCORE.ARPA
Robin Lake and Peter Piro of Sohio will be here Tuesday, May 3, recruiting
AI PHD students. The particular position is in Cleveland, new research
center south of town. He promised to send me more information. Whether
interviewing for this position or for future contacts, you are the students
he chose from the resume booklet.
Please send me a message with suggested times for interview. I don't have
conference rooms arranged yet.
Sohio belongs to the Forum and is considering a second membership. Robin
Lake has been in touch with Prof. Feigenbaum.
Thanks,
Carolyn
-------
∂19 Apr 83 11:06:42-PST <CSD.TAJNAI@SU-SCORE.ARPA> Re: Sohio recruiting AI PHD students
To: csd.barr@SU-SCORE.ARPA, csd.chelberg@SU-SCORE.ARPA,
csd.berlin@SU-SCORE.ARPA, csd.pcohen@SU-SCORE.ARPA,
csd.davidson@SU-SCORE.ARPA, csd.jjf@SU-SCORE.ARPA,
csd.gardner@SU-SCORE.ARPA, csd.greiner@SU-SCORE.ARPA, ccg@SU-AI.ARPA,
csd.jock@SU-SCORE.ARPA, csd.rosenschein@SU-SCORE.ARPA,
csd.rowe@SU-SCORE.ARPA, csd.smith@SU-SCORE.ARPA, mah@SU-AI.ARPA
cc: csd.Feigenbaum@SU-SCORE.ARPA
In-Reply-To: Your message of Fri 15 Apr 83 09:40:50-PST
Please respond to my message regarding signing up to interview with
Sohio reps on Tuesday, May 3. Thanks, Carolyn
-------
∂13:06 20-Apr:
CSD.TAJNAI@SU-SCORE.ARPA/su Re: Re: Sohio recruiting AI PHD students
I'm available an 11AM meeting, on 3-May. Is that a good time?
∂28-Apr-83 1402 CSD.TAJNAI@SCORE Help!! Sohio Interviewing
To: csd.barr@SCORE, jjf@Diablo, dietterich@SUMEX-AIM, RDG@SAIL
cc: csd.tajnai@SCORE
I called Peter Piro and told him the timing was bad and the interest slight,
but he and Robin Lake are going to be in Berkeley on Monday and it won't
be a special trip for them. I didn't want them to make a special trip to
talk to students who really had little interest. Under the embarassing
circumstances, it would be wonderful if you could spend some time with
Peter Piro and Robin Lake. Sohio Research Center is our newest Forum
company and it is an HPP company.
How about this:
252 220 (Chairman's conf. room)
9:00 - 9:30 Dietterich 9:30 to 10 Dietterich
9:30 - 10:00 Greiner 10:00 to 10:30 Greiner
10:00 - 10:30 Finger 10:30 to 11:00 Finger
11:00 - 11:30 Barr 11:30 to 12:00 Barr
????
Carolyn
-------
∂28-Apr-83 1402 CSD.TAJNAI@SCORE Re: Help!! Sohio Interviewing
Sorry to be so vague, but I'm talking about Tuesday, May 3.
I'm wasted from champagne at lunch. We entertained the CSL secretaries!!
I think I was the one entertained. Ah, champagne.....
-------
∂28-Apr-83 1540 jjf@Diablo Re: Help!! Sohio Interviewing
To: Carolyn Tajnai <CSD.TAJNAI at SCORE>
Cc: csd.barr at SCORE, jjf at Diablo, dietterich at SUMEX-AIM,
RDG at SAIL
I replied earlier, but it looks like the System gobbled it up.
Sorry to say that I'll be in Texas on Tuesday. I'm glad you
enjoyed your lunch.
-- Jeff --
∂17:39 28-Apr: CSD.TAJNAI@SCORE/su SOHIO Interview
Carolyn
I didn't understand your message.
What time am I scheduled to meet with them -- at 9:30 in 252,
(and with Tom?) or at 10 in 220, w/JJF.
Actually I would prefer to meet a bit later, at say 11AM, if possible...
Russ
∂28-Apr-83 1805 CSD.TAJNAI@SCORE Re: SOHIO Interview
To: RDG@SAIL
In-Reply-To: Your message of Thu 28 Apr 83 17:39:00-PDT
It was the champagne, my message isn't too clear.
Avron had already requested 11 in his first message, but we could this:
Peter Piro in room 352 from 10 - 10:30
Robin Lake in room 220 from 10:30 to 11. Can you make that?
Carolyn
-------
∂01-May-83 1559 BARR@SUMEX-AIM Re: Help!! Sohio Interviewing
To: CSD.TAJNAI@SU-SCORE.ARPA
cc: csd.barr@SU-SCORE.ARPA, jjf@SU-HNV.ARPA, dietterich@SUMEX-AIM.ARPA,
RDG@SU-AI.ARPA
In-Reply-To: Your message of Thu 28 Apr 83 14:03:13-PDT
Carolyn,
Assuming you read this on a sober Monday morning, perhaps
you could tell us exactly what Sohio would like to hear from us
Tuesday, and why we each have to tell it to them in two half-hour
sessions?
.Av
-------
∂14:32 2-May: TAJNAI@SU-SCORE/su Do I have to?
Anytime on Tuesday is next to impossible. Is it really important?
Russ
∂02-May-83 1514 TAJNAI@SCORE Re: Do I have to?
To: RDG@SAIL
Russ, it is downright embarrassing (how ever you spell it). Next time
their recruiting will be handled by EAF's office since HPP brought
them in. It is down to you and Avron and I would really appreciate your
talking to them. pleeeeaaaaassssseeee.
Carolyn
-------
∂15:29 2-May: TAJNAI@SCORE/su Sympathy, and suggestions?
Will they be around Wednesday at all? or I could meet with them today...
I just have so many things planned for Tuesday
(including two VERY important meetings) that I hate to have
yet one other thing to worry about. Is there someone else who could
speak to them tomorrow?
Russ
∂02-May-83 1622 TAJNAI@SCORE Re: Sympathy, and suggestions?
To: RDG@SAIL
In-Reply-To: Your message of Mon 2 May 83 15:29:00-PDT
They are in Berkeley today and will leave for Cleveland tomorrow afternoon.
I sent the message to Goldman, Chelberg, Berlin, Davidson, Cohen,
Dietterich, Finger, Gardner, Gray, Mackinlay, Rosenscheim, Rowe, Smith
and Holstege.
If you can't, you can't. Do you think you could talk someone into it?
Anne is working on her oral; Paul out of town; Tom out of town; Jeff out of
town; Dave not interested; Jeff R. not interested. Jock said "I don't want
to interview at this time, but will be interested in the future" -- I sent
him a msg today, but no response.
I should call them and tell them not to come until 11 to see Av. sigh,
Carolyn
-------
∂11-Apr-83 1029 DEKLEER.PA@PARC-MAXC paper and talk.
Subject: paper and talk.
To: rdg%su-ai@SU-SCORE.ARPA
Good. Will your talk be on analogy and/or mechanisms?
If so I would like to hear it.
-------
∂14:58 11-Apr: DEKLEER.PA@PARC-MAXC/sub Upcoming presentation
Context: in Doug's "Machine Learning" seminar.
Content: broad overview of analogy, esp wrt (machine) learning.
It'll include some of (sketchy) ideas on
* What is an analogy?
* What is a good analogy?
* Reformulation
Relevance to JdK: ??,
but you are certainly invited/encouraged to attend, anyway.
Background: audience will have read Carbonell's chapter in recent ML book,(*)
and Winston's Dec 80 CACM article
Length: about a hour
[R HayesRoth will then discuss chapters 6-8 of that book, for about 1 hr]
Starting-Time: 2:30pm, Thursday, 14-Apr
Location: <<still unknown - it outgrew MJH 301>>
(*) "Machine Learning: An Artificial Intelligence Approach",
edited by Ryszard S. Michalski, Jaime G. Carbonel, and Tom M. Mitchell.
∂15-Apr-83 2233 DEKLEER.PA@PARC-MAXC.ARPA Upcoming presentation
Thanks for letting me know. I was thinking of coming to hear,
but got too busy. How did it go?
∂15:28 18-Apr: DEKLEER.PA@PARC-MAXC, csd.lenat@score/su/su Talk
Too bad you were busy last Thursday... but you didn't miss anything new
or all that interesting (at least in terms of my presentation).
I read thru your short paper this weekend, and have some quick comments.
I plan to read your larger work soon; perhaps we could get together then,
to discuss both them and my program(me).
When is a good time in general, for you?
Russ
∂19-Apr-83 2256 DEKLEER.PA@PARC-MAXC.ARPA
<RDG%SU-AI@USC-ECL.ARPA>
After lunch any day this week would be fine.
∂12:53 20-Apr: DEKLEER.PA@PARC-MAXC, csd.lenat@score/cc/su Meeting
How about this Friday, at say 2PM, at your office?
(Or, if you are going to Masinter's SigLunch talk, in my office
at Stanford at that time.)
Russ
∂20-Apr-83 1349 DEKLEER.PA@PARC-MAXC.ARPA Meeting.
To: rdg%su-ai@USC-ECL.ARPA
Someone just scheduled himself there. How about later that afternoon, say 4.
∂21-Apr-83 1032 DEKLEER.PA@PARC-MAXC.ARPA Visit.
<RDG%SU-AI@USC-ECL.ARPA>
Did you get mmy msg that 4pm was better. I now have trhee things scheduled
for that same time (1pm) so juggling won't even work.
∂11:34 21-Apr: DEKLEER.PA@PARC-MAXC, csd.lenat@score/cc/su Meeting time
Ok - I'll be in your (Johann's) office, PARC, at 4PM Friday.
I have 2 strong anti-recommendations:
First, the ?Eber? HiFi (on ?Forest) is of questionable character:
[Asked to align and re-adjust the heads on my cassette player --
ok, $45 once. 4 months later (after 3 month warranty) same minor problem --
not only didn't they fix it, but found the heads needed to be replaced as
well.]
Second, L&S ?Repair -- ?1011 Alma?
My brother & I brought 2 different cassette players there, with a variety
of ailements. Even after bringing each one back 3 times, none of the
problems were adequately fixed -- those which seemed to work did so only
for a short time.
∂12:42 13-May: Lenat@score/su Statistics
Doug -
The data below verifies my claim that >1/2 of the IJCAI
*SHORT* papers in L&KA were accepted -- 22/35. I'd feel a lot
better about my paper if I thought the referees were more selective.
Is Saul postponing his talk, or cancelling it? Did he return
to Rutgers, or ... (I told many people, including Johan, the schedule;
and wanted to update them.)
Finally, is there any truth to the rumor that Peter Hart has left
Fairchild?
See you Tuesday, thesis-time-schedule in hand.
Russ
--------
∂12:42 13-May: genesereth@sumex/su Will AAAI Have something like this?
Date: Thursday, 21-Apr-83 15:23:45-BST
From: BUNDY HPS (on ERCC DEC-10) <bundy@edxa>
Subject: Statistics on IJCAI-83 Papers
[I don't think Alan Bundy will mind my passing along these
statistics. I have edited the table slightly to fit the 70-column
capacity of the digesting software made available by Mel Pleasant,
the Human-Nets moderator. The digester was developed by James
McGrath at SCORE. -- KIL]
PAPER STATISTICS - IJCAI-83
Submitted Accepted Moved
Subfield Long Short Long Short L -> S
Miscellaneous - 3 - 1
Automatic Prog. 8 11 1 7 4
Cognitive Modelling 9 32 2 12 3
Expert Systems 31 56 8 31 9
Knowledge Repn. 28 40 7 24 6
Learning & Know. Acq. 14 35 1 22 5
Logic Prog. 14 17 4 9 4
Natural Language 23 74 2 39 7
Planning & Search 11 20 3 11 5
Robotics 11 8 5 7 2
System Support 4 9 - 5 2
Theorem Proving 7 16 5 8 -
Vision 32 38 10 31 14
TOTAL 192 359 48 207 61
COMPARISON WITH PREVIOUS IJCAI CONFERENCES
LONG SHORT TOTAL
IJCAI-79 Submitted Total unk unk 428
IJCAI-79 Accepted Total 83 145 228
IJCAI-79 Acceptance Rate 53%
IJCAI-81 Submitted Total unk unk 576
IJCAI-81 Accepted Total 127 74 201
IJCAI-81 Acceptance Rate 35%
IJCAI-83 Submitted Total 192 359 551
IJCAI-83 Accepted Total 48 207 255
IJCAI-83 Acceptance Rate 46%
REMARKS
You will see that I succeeded in my aim of shifting the burden
of papers from the long to the short categories. This enabled us to
apply high standards to the long papers without decreasing the overall
acceptance rate.
Alan
------------------------------
∂16-May-83 1156 mccune@aids-unix Future
Russ,
When are you graduating? I'd like to talk with you about research positions
at AI&DS. We're employee-owned, have 43 employees now, and do a range
of basic research to prototype development in a number of AI areas and
other technologies.
Brian
∂13:52 17-May: mccune@aids-unix/su Future Shock
Brian -
I'm at that "just 6 to 9 more months" stage, and have been for
more years than I'd like to admit. I now figure that I'll be seriously
interviewing by about Jan/84. Do you have a "prospectus" on AI&DS;
or other description you could send? Or would you rather wait until
then?
Russ
∂17-May-83 1414 mccune@aids-unix Future Shock
To: RDG@SU-AI, mccune@AIDS-UNIX
Russ,
There will be a new brochure in time for the AAAI conference. If you
don't attend the conference, I'll mail you a copy.
Brian
∂15:00 13-May: deKleer@Parc/su TEX "Answers"
Johan -
I explained your situation to a TEX wizard (wrt correctly
producing 4 output pages for each conference proceding page);
who ventured several solutions.
The quick and dirty (and not-quite-right) solution is to include
\clubpenalty 0
\widowpenalty 0
at the start of your document -- this tells TEX not to worry about
widow lines or club lines. (Each of these is now 150.)
There are some other misc penalty values to reset (like beginning a
page with a hyphenated word etc), but they're all 2nd order.
Anyway, these parameters are all used (as opposed to mentioned) in
{?TEX database director?}PLAIN.TEX . They are all documented in the
still being written "TEX Book".
If you are willing to muck with various output routines, you could
tell TEX the page height is 11.5 inches, and have it generate "logical
pages" of that height. But when it comes time to output it, that output
process will chunk it nicely into "physical pages".
(This is still an "hour or two" task to implement, I'm told.)
Answer three involves waiting. Eventually you'll be able to tell
TEX82 (or maybe a decendant...) the overall page dimensions, including
column specifications, etc; and IT will do the output stuff. Or so
I'm told. [This will facilitate n-th order problems, like figures
which take both columns, etc.]
Anyway, I'm sending off this puzzle to the local TEXmasters, hoping
that their interest in this (and hence attention, expenditure of
resources, ...) will be elevated according.
Hope this helped. Enjoy your move,
Russ
PS Apparently Saul Amarel will NOT be speaking this next Thursday.
I'll let you know if he intends to speak later.
∂13:59 16-May: dek,drf,jdh/cc,dekleer@parc/su TEX Question/Challenge
Gentlemen,
Does the following TEX question already have a solution?
If not, what half-way measures would you propose, towards
solving it? (I'll mention the ones which John Hobby and I
considered at the end.)
Thank you,
Russ
-------
One standard conference paper format is a two-column page,
where each column is 11.5" x 4.25".
It would nice if we could simply tell TEX this information,
and watch it print out dover pages, each about 5.75" high and
(exactly) 4.25" wide, where each of these physical pages would be
one half of a column (hence 1/4th of the logical conference paper
page.)
The 0th level answer, of simply telling TEX to print out 5 3/4 x 4 1/4
pages, is woefully inadequate, for the following reasons:
(1) The same amount of interline and inter-paragraph spacing glue
should be applied throughout the column; and the difference between
the two columns of the same page should be as small as possible.
(2) Widow and club lines (and related issues of hyphenation, ...)
should only matter (i.e. be penalized) every other page
-- i.e. NOT at the top of a page which will end up in the middle of a
column.
(3) Footnotes should be accomulated(sp), and only appear on "even" pages.
[As a secondary issue: perhaps a footnote could appear under the PREVIOUS
COLUMN of the same page, when necessary.]
(4) Figures, includes those which require the page's full 8.5+" width,
should be handled correctly.
-----
Partial solutions:
Just turning off widow- and club- penalties is a first measure.
Writing some hairy output-function was the next proposal; which would
still require several hours of hacking.
The 3rd idea was to wait: that TEX8? would eventually (already?) have
the facility of handling arbitrary page descriptions, and would automatically
format the (physical) pages output appropriately.
∂16-May-83 1447 DRF Russ' long pages
To: RDG@SU-AI, DEK@SU-AI, JDH@SU-AI, dekleer@PARC-MAXC
The proper solution is to run TeX with 11.5 x 4.25 pages, and make your
DVI->Dover program smart enough to print pages on two sheets of paper when
they are too long to print on one sheet. Or you can print them magnified
down to the point where they fit on a single sheet.
-david
∂15:26 16-May: DRF, DEK@SU-AI/cc, JDH@SU-AI, dekleer@PARC-MAXC/su re: Russ' long pages
David -
Thanks for you quick answer. Can you elaborate on this
DVI->DOVER program:
Where is it documented?
Do all versions of the current TEX employ this (in particular, XEROX Parc's)?
Will this handle the two-column figure problem? or prior-column footnote?
etc.
----
The second proposal (of (de)magnifying) is of no use,
as the goal is to fit the output on particular sheets of paper...
-russ
∂16-May-83 1635 DRF
To: RDG@SU-AI, JDH@SU-AI, dekleer@PARC-MAXC
We use DVIDOV for TeX82 at Sail and Score. Sources are DVIDOV.SAI[TEX,SYS]
at Sail. For old TeX, at Sail we use a version from Parc that makes Press
files directly, but at Score we use an older version of DVIDOV called DVIPRE.
My suggestion works for anything you may have on the pages in question, since
TeX is actually making pages of just the size you eventually want. Since
the output device is limited in the size of paper it can print on, the
output driver is responsible for cutting each page in two. When you paste
the two halves together, you should get a `perfect' page. Unfortunatly,
I don't think that the DVIDOV program supports this feature. It might
not be too hard to put it in (you just have to check each character's
Y coordinate, and don't print it if it's not within the bounds specified
by a run-time parameter. Then the user just has to run DVIDOV twice, once
setting the parameter for the top half of the page, once for the bottom).
-david
∂16 MAY 83 14:29 PDT DEKLEER.PA@PARC-MAXC.ARPA TEX
To: GREINER@SU-SCORE.ARPA
My solution: tell TEX to 11.5 in. Change output routine to shift
page up 2in, print and run paper through tex tiwce since output
routines do not do ejects. (It would help if I could
figure out how to eject a page in an output routine. i.e.,
\save1\page\copy1\eject\vskip-3in\eject fails to eject and
just overprints on the same page.
-------
∂23-May-83 1325 DEK 11.5-inch pages
To: RDG@SU-AI, JDH@SU-AI, dekleer@PARC-MAXC
CC: DRF@SU-AI
Another way, before DVIdover is rewritten, or with the old version
of TeX that you may be using, is to run your program twice.
First, you get pages with stuff missing at the bottom.
On the second run, change your \output routine so that it has
\vskip-5in at the very beginning (or just inside the \vbox being
shipped out, in TeX82). Then you'll get the output with stuff
missing at the top, because it's off the page.
∂Send to DEK - 23/V/83
Return-Path: DEKLEER.PA%PARC-MAXC.ARPA@SU-SCORE.ARPA
Received: from SU-SCORE.ARPA (SU-SCORE.ARPA) by SUMEX-AIM.ARPA with TCP; Mon 23 May 83 15:46:29-PDT
Received: from PARC-MAXC.ARPA by SU-SCORE.ARPA with TCP; Mon 23 May 83 15:45:09-PDT
Date: 23 MAY 83 15:37 PDT
From: DEKLEER.PA@PARC-MAXC.ARPA
Subject: How do I do this?
To: greiner@su-score.ARPA
Subject: Undelivered mail
From: Cabernet.MS (a Grapevine mail server)
To: DEKLEER.PA
Date: 23-May-83 15:34:25 PDT
The message sent by DEKLEER.PA at 23-May-83 15:34:03 PDT could not be
delivered to the following recipients because they were rejected by the
MTP server "Maxc". The reason given was: Sorry, I never heard of an
ARPA domain named "su-ai%su-score"
rdg@su-ai%su-score.arpagateway, dek@su-ai%su-score.arpagateway
----------------
Date: 23 MAY 1983 1533-PDT
From: DEKLEER at PARC-MAXC
Subject: 11.5 in pages
To: dek at SU-AI%SU-SCORE
cc: rdg at SU-AI%SU-SCORE
You describe exactly what I did. It would be easier however if there
were some way to output the same page twice in the output routine, one
normal one with \vskip-5in.
-------
-------
∂13:49 17-May: rod/su Welcome Back
Rod -
It's amazing who(m) Stanford is letting in now.
I needed motivation like this to inspire me to finish up and leave...
This just shows what having friends in the right places
(e.g. faculty search committee) will do for you...
Seriously though, congratulations!
When will you be moving out here?
... and when will there be three of you?
(I just heard about that as well.)
Looking forward to seeing you,
Russ
∂18-May-83 1628 @MIT-MC:BROOKS%MIT-OZ%MIT-MC@SU-DSN Re: Welcome Back
Received: from MIT-MC by SU-AI with TCP/SMTP; 18 May 83 16:28:28 PDT
Date: 18 May 1983 1820-EDT
From: BROOKS%MIT-OZ%MIT-MC@SU-DSN
Subject: Re: Welcome Back
To: RDG@SU-AI
In-Reply-To: Your message of 17-May-83 1349-EDT
Don't know exectly when we'll be moving out. July or August or
sometime. We are due to be three on Sept 20.
-------
∂14:24 6-May: sleeman@sumex/su ML Workshop
Derek -
Do you have more information about the upcoming Machine Learning
workshop? (esp when, and an overview of format)
Thanks,
Russ
PS I look forward to hearing your talk in Doug's class.
∂06-May-83 2143 SLEEMAN@SUMEX-AIM Re: ML Workshop
Russ, It's to be held in a country house off the Urbana campus on
22,23,24 of June. There are to be long and short paper presentations and I
HOPE LOTS of discussion!
I have send a note to Carbonell and co and will pass on the response
when I hear from them.
Alternatively you could send them mail directly.
Best wishes,
D
-------
∂13:27 24-May: tgd/su UofIll,C-U
Tom -
Is U of Ill, C-U on the ARPAnet? What is the best address
for Dr Michalski?
Thanks,
Russ
∂24-May-83 1517 DIETTERICH@SUMEX-AIM Re: UofIll,C-U
Received: from SUMEX-AIM by SU-AI with PUP; 24-May-83 15:17 PDT
Date: Tue 24 May 83 15:18:03-PDT
From: Tom Dietterich <DIETTERICH@SUMEX-AIM>
Subject: Re: UofIll,C-U
To: RDG@SAIL
cc: DIETTERICH@SUMEX-AIM
In-Reply-To: Your message of Tue 24 May 83 13:27:00-PDT
He's not on the net. The best address is
--Tom
-------
∂20:10 24-May: Carbonell@CMU,Mitchell@Rutgers,sleeman@sumex/cc/su ML #2
Drs Michalski, Carbonell, and Mitchell,
Thank you for sending me information about the upcoming
International Workshop on Machine Learning, and for issuing an
invitation. Unfortunately, I will not be able to attend --
my time constraints are just too limiting.
I look forward to seeing the preceedings, and hope to hear
(unfortunately only second-hand) of the discussions which will take
place. Please do keep me in mind for subsequent meetings on this
topic, as I am quite interested and would like to participate.
Sincerely,
Russ Greiner
PS Please forward this to Dr Michalski if possible. I couldn't find
his net address.
∂13 Oct 1982 1633-EDT PAIGE at RUTGERS
To: Greiner at SU-SCORE
Hi Russ,
Nieba and I finally moved into our NYC apt (a bargain at $1000/month),
a modest 2 bedroom apt. in a safe Italian neighborhood on a quiet
residential block. The transition from country to city has not been
as smooth as expected. Finding an apt. for both of us and a job for
Nieba were headaches. The first apartment we liked was in a charming
historical building 250 years old, and a mere $1400/month. We would
have taken it had we not encountered an angry tenant, who complained
that mail was only delivered 3 times a week (the eccentric owner
didn't give the gate key to the mailmail). We also observed that
there was no buzzer system at the gate (The building only had a
rear entrance.). So we searched on until we found our current
place, good for an oct. 1 occupancy. After two weeks of waiting,
we finally got partial telephone service, and are now preparing to
make our 'space' livable with the help of an architect and carpenter.
Nieba got a job working for Lehman Brothers, a bank investment firm on
Wall Street. She's doing database work, and likes it so far.
My NSF grant came through, and POPL accepted one of the papers I submitted
in a frenzy at the end of my stay at Stanford.
Good luck with your work, and best regards. Please say hello to Jock and
Ben when you see them. Also, how do you spell Ben's last name?
Bob
PS. Our address is
18 Leroy St. Apt. 4
NY, NY 10014
phone: 212-675-3697
-------
∂14:17 2-May: PAIGE@RUTGERS/su Finally...
Greetings!
As Ben probably mentioned, I tried several times responding to your
13-Oct-1982 message, but always had it bounced back.
Now that SAIL is a (more or less official) TAC, let's see if it works now...
My thesis work is going incredably slowly --
all sorts are things are turning out to be on my critical path,
and have to pushed-aside/climbed-over/documented-around.
Very annoying!
Ben is finishing up, and preparing for his venture to Cambridge.
Do you think you'll see him before he leaves -- at any meetings, or whatever?
I may be back east this summer, either in route to Europe, or attending a meeting
in Washington, depending on which conference(s) I decide to attend.
Have you and Nieba made summer plans yet? Will you be attending any
conferences then? Have you been to any interesting ones recently?
How does it feel to be back at Rutgers? ... and how is your work going?
Any new and interesting results?
Let's see, what else? Jock and I have moved to a new office
-- on the 3rd floor, sandwiched in with various NA types.
(MJH 311, 497-1863)
Saul Amarel was in our old office, with two other people;
but he also moved elsewhere.
Nothing else to report from here -- except that Jock and I have been
Tell Nieba "hi".
Russ
∂02-May-83 2035 PAIGE%RUTGERS.ARPA@SCORE Re: Finally...
To: RDG@SU-AI, PAIGE%RUTGERS@SU-SCORE
Hi from Nieba and me,
Sorry to hear about your thesis blues. I remember times like that too.
Ben will visit us June 11 and stay a few days to do some joint work.
I just returned from the STOC conference in Boston, and enjoyed it very much;
it's now my favo;rite ;conference. Although not presenting anything there,
I met three people who I'm do;ing jo;int wo;rk with, and we made some progress.
The work is in the area of tree pattern matching, scheduling, and
new language constructs.
Hi, Russ, it's Nieba now...
We have many summer fantasies hatched up, but no summer plans just yet: but
we would love to see you if you come by this way, and better yet, we would love
to offer you our roof, futon, company, and protection while you are sight-seeing
in the city!
I have my first four-star programming job!!! Every job I've had in the past
proved to be a mixed blessing, so I'm really enthused.... It is with a compiler
writing company, and my first project is doing the back-end (machine-dependent
code generation) for a COBOL (boohiss) compiler for the IBM 370. Thank heaven
that compilers don't have to be coded int the language they are compiling!!!
We write all our code in either IBM PASCAL or IBM assembler. It's great fun,
technically the most challenging and interesting of my computer jobs so far,
my peers are really nice, interesting, unsnobby and creative, and the management
is the best I've ever had (they not only treat us like people rather than
objects, but actually seem to feel and act like just one of the gang. And they
are very good at setting reasonable deadlines and dividing a task up into doable
and unintimidating subtasks, etc., etc., etc., ...)
If I sound rapturous it's because I am....
I love the city, love living with 3 Italian bakeries, 2 fresh fish markets,
2 green grocers, 3 butchers, an Indian, a Japanese, a Middle Eastern, a Creole,
about 5 top-notch Italian, a Spanish, and a good French restaurant, as well as
4 espresso cafes and brunch/bistro joints where guitarists and poets play or
recite at odd hours, all this within a block and a half of where we live. When
you expand the radius to ten blocks, well, I can't begin to list the galleries,
off-broadway theatres, concert halls, and you name it that just wait to be
discovered and experienced!!! Well, I've gotten over feeling guilty about
the good life I'm leading, and have taken up having company whenever possible,
so that my friends can share it with me! As a native Californian, I think
this is a fine adaptation, with the best of both worlds (coasts)!! So you
should come and enjoy it with us too!
Say hi to Ben and Jock from both of us, and we wish you the best of luck
through the slow-going stages of your work!
Love,
Nieba and Bob
-------
∂13:04 16-May: PAIGE@RUTGERS/su Still no solid plans, but ...
Dear Nieba and Bob
Such a cheery message!
It's good hearing things are going well there, for the two of you.
As you know Ben is REALLY finishing up, which is beginning to inspire me
to similarly move along. I'm now saying December 83 as my official
"thesis draft in hand" interview time; we'll see.
It looks like I will go to Germany for IJCAI this year -- as it appears
that HPP will pay essentially all my expenses [all for a 15 minute
presentation!]
That's 8-12 Aug; by 22-Aug I want to be in Washington DC. I'm just now
beginning to make plans, and will definately keep your enthusiastic offer
in mind. Will you be around in August?
Take care,
Russ
∂18-May-83 1429 PAIGE@RUTGERS Re: Still no solid plans, but ...
Yes, we'll be around in August. Hope to see you then.
Bob and Nieba
-------
∂04-May-83 1209 DIETTERICH@SUMEX-AIM Paper on EMACS and typewriting
To: rdg@SAIL
cc: DIETTERICH@SUMEX-AIM
Russ,
Have you finished reading that chapter of the disseration on EMACS and
typewriting? I'd like to have it back soon.
Thanks,
--Tom
-------
∂13:48 9-May: DIETTERICH@SUMEX-AIM/su Paper on EMACS and typewriting, etc.
Tom -
I finally finished that paper, and returned it to your desk.
Were you able to get a (more complete) copy of her thesis?
I'd much like an opportunity to examine it ...
alternatively, does she have an ARPAnet address (so I could ask her myself)?
How is your thesis work developing? I'm again finding myself
shifting domains -- toward programming again. Sigh.
Would you like to get together and talk soon?
Russ
∂13:46 31-May: dekleer@parc/su Confession
Johan -
I just now spooled off a (2nd) copy of John's thesis;
and obviously haven't yet begun to read it. Should we still plan
to meet tomorrow, or postpone our get-together by a week?
(Certainly by then I'll have "cracked" it.)
Russ
PS Remember that Saul Amarel is scheduled to speak this Thursday.
Perhaps we could meet here (at MJH) before that?
∂31 MAY 83 13:53 PDT DEKLEER.PA@PARC-MAXC.ARPA Confession.
To: greiner@su-score.ARPA
I have read it believe it or not. In any case, will you be finished
by Thursday noon --- it is a little bit excessive a read.
Do you happen to know if Amarel is talking 1st?
-------
∂17:44 2-Jan: dekleer@parc/su (Continuously) Upcoming Amarel Talk
Johan -
You did the right thing missing today's talk --
Saul never made it to the podium.
He'll talk, for sure, this coming week - 9/Jun/83.
(I'll also give a run of my 15 minute IJCAI talk on novelty.)
See you then - 2:30, in History 205?
Russ
∂23-May-83 1633 FRALEY@SU-SCORE.ARPA next hike
To: hike-distribution-list: ;
A lot of people are going to be away this weekend; however maybe there are some
people who are interested in hiking. Monday is out, but I think a hike in the
Santa Cruz Mountains - either Big Basin or Skyline - on Sunday or Saturday
would be in order. What do people think of this? Chris
-------
∂25-May-83 1523 FRALEY@SU-SCORE.ARPA backpack postponed
To: hike-distribution-list: ;
Lia is going East for a few weeks, so the Big Basin backpacking trip is going
to be postponed until sometime in June at least. There will be a hike this
Saturday on Skyline trail. I'll be away June 6 - 16, but will continue trying
to arrange hikes through summer as long as people are interested. If anyone
has any specific ideas about places they would like to go, they should let me
know. Chris
-------
∂26-May-83 1648 FRALEY@SU-SCORE.ARPA Saturday hike
Received: from SU-SCORE by SU-AI with TCP/SMTP; 26 May 83 16:47:59 PDT
Date: Thu 26 May 83 16:08:08-PDT
From: Chris Fraley <FRALEY@SU-SCORE.ARPA>
Subject: Saturday hike
To: hike-distribution-list: ;
Hike this Saturday; Skyline trail from Saratoga gap to Huddart Park (about 8 mi)
We'll have to figure out a way to arrange cars, but people that only want to go
part way can easily be accommodated. Meet at 9 AM near the front of the oval.
Chris
-------
∂12:20 30-May: FRALEY@SU-SCORE/su ↑Bob
Chris -
This worked for me:
"FRALEY%HP-HULK"@Rand-Relay
Russ
∂31-May-83 1013 FRALEY@SU-SCORE.ARPA Re: ↑Bob
In-Reply-To: Message from "Russell Greiner <RDG@SU-AI>" of Mon 30 May 83 12:20:00-PDT
I also sent mail to Bob via the new address & nothing has come back so far.
-------
∂03-Jun-83 0958 FRALEY@SU-SCORE.ARPA future hikes
To: hike-distribution-list: ;
As I mentioned before, I'll be away June 6-16. Someone else might want to use
the hiking list to organize something while I'm gone.
-------
∂12-Jun-83 2329 JF@SU-SCORE.ARPA partitions
Subject: partitions
To: klc@SU-AI.ARPA, rdg@SU-AI.ARPA
you CAN partition the positive integers into an infinite number of
classes each with an infinite number of members such that in each class
there is one number with property p but the number of numbers less than
n with property p is not little-oh(n). let p be "odd". in class i,
put the odd number (2*i)-1. then put the even numbers in any old way such
that there are an infinite number in each class. (simplest way--put
numbers 2,4,6,8,10,12,14,16,18,20,etc. in classes 1,1,2,1,2,3,1,2,3,4,etc.)
this counter-example is due to lane hemachandra.
it looks like my prime-number theorem solution to russell's question is the
simplest one proposed so far; although it still seems that there should be
a more elementary way to see why most numbers are not square-free.
joan
-------
∂06-Jul-83 1436 JF@SU-SCORE.ARPA square-free integers
Received: from SU-SCORE by SU-AI with TCP/SMTP; 6 Jul 83 14:35:46 PDT
Date: Wed 6 Jul 83 14:23:38-PDT
From: Joan Feigenbaum <JF@SU-SCORE.ARPA>
Subject: square-free integers
To: grosof@SU-SCORE.ARPA, rdg@SU-AI.ARPA, klc@SU-AI.ARPA,
pressburger@KESTREL.ARPA
well folks, my flippant use of the prime number theorem to prove that
"most" numbers (i.e. little-oh(n) in the interval [1 .. n]) are square-free
was wrong. and a damned fine thing, too, because it is probably not true
that they are. i inverted the statement of the prime number theorem:
the number of primes less than n is O(n / log(n)), not O(log(n) / n) as i
so stupidly said (and you all so stupidly believed). amazing what non-truths
one resorts to in order to prove still more.
anyway, russell's original question was "how many numbers are square-free";
not how many in the interval [1 .. n]. by sheer coincidence, i read the
answer to that question in "The Mathematical Experience" by Davis and
Hirsch recently. This time, i'll quote so that i don't screw it up:
....................
What is the chance that n has no repeated factor? ...
This will happen if n is not a multiple of 4 or a multiple of 9 or a
multiple of 25 or any other square of a prime. Now, the probability
that a number chosen at random is not a multiple of 4 is 3/4, the probability
that it is not a multiple of 9 is 8/9, the probability that it is not
a multiple of 25 is 24/25, and so on. Moreover, these conditions are all
independent--knowing that n is not a multiple of 4 tells us nothing about
whether it is a multiple of 9. So according to the basic probabilistic
law that the probabiliy of occurrence of two independent events is the
product of their separate probabilities, we conclude that the probability
that [a number is square-free] is the product
3/4 x 8/9 x 24/25 x 48/49 x ...
Even though this product has an infinite number of factors, it can be
evaluated analytically, and it is known that it is equal to
6/((pi)↑2)
........................................
this section of the book is about "heuristic arguments" in favor of the
truth of the riemann hypothesis and is quite interesting. note that we
still haven't said anything about the number of square-free integers in
[1 .. n]. anyway, i'm glad i finally cleared this up.
joan
-------
∂06-Jul-83 1456 JF@SU-SCORE.ARPA square-free numbers
Received: from SU-SCORE by SU-AI with TCP/SMTP; 6 Jul 83 14:56:34 PDT
Date: Wed 6 Jul 83 14:57:57-PDT
From: Joan Feigenbaum <JF@SU-SCORE.ARPA>
Subject: square-free numbers
To: grosof@SU-SCORE.ARPA, klc@SU-AI.ARPA, rdg@SU-AI.ARPA,
pressburger@KESTREL.ARPA
in my previous message, i said that i had argued erroneously that most
numbers are square free. that was a typo: i had argued erroneously that
most numbers are not square free, where "most" meant what i said in the
message.
joan
-------
∂14:57 15-Jun: dkanerva@sumex/su FYI - from ISI
∂14 Jun 1983 1735-PDT Editing/Writing tools Jim Melancon <MELANCON@USC-ISIB>
To: PROJECT-LIST: ;
cc: Project-Assistants: ;, MELANCON, bboard, SCoyazo, Nish
UNIX offers a few prose analysis programs that are useful in evaluating
writing style. The descriptions of the programs come from the [Berkeley]
VAX VMUNIX Reference Manual.
STYLE "analyzes the surface characteristics of the writing
style of a document. It reports on readability, sentence length and
structure, word length and usage, verb type, and sentence openers."
DICTION "finds all sentences in a document that contain phrases from a data
base of bad or wordy diction. Each phrase is bracketed with [ ]."
EXPLAIN "is an interactive thesaurus for the phrases found by DICTION."
These programs have limited usefulness, but they may satisfy some needs at
ISI. If you would like to have a document you're working on analyzed by
STYLE, criticized by DICTION, and possibly improved by EXPLAIN, let me know.
If you send me a file with the text you would like analyzed, I will send the
results of the analysis back to you. Sheila Coyazo and I are available for
assistance in interpreting the analysis and in revising the text.
As a side note, STYLE, DICTION, and EXPLAIN appear to have been incorporated
in Bell Laboratories' Writer's Workbench, which offers a number of other
programs that proofread, check spelling, correct punctuation--and even
detect sexist language. Unfortunately, the Writer's Workbench seems not to
be available on our version of Berkeley UNIX. I'm trying to find out if
later versions of Berkeley UNIX will include the Writer's Workbench.
If anyone is interested in reading more about these programs, I can send you
copies of articles about the writing tools and Writer's Workbench that
appeared in the January 1982 issue of IEEE Transactions on Communications.
Jim
-------
-------
∂15-Jun-83 1517 DKANERVA@SUMEX-AIM Re: FYI - from ISI
Russ,
Thanks for forwarding the message about STYLE, DICTION, and EXPLAIN.
They look interesting--I'm a bit skeptical, of course. I appreciate
your keeping me posted on these things. -- Dianne
-------
∂12:50 12-May: HART@SRI-KL/su Get Together
Peter -
Sorry I had to rush off the other day -- the conversation was
quite interesting, and potentially useful. Unfortunately I was already
15 minutes late when I ran into you,...
Anyway, I would really like to talk some more, to get your ideas
and insights (possibly augmented with advice from that 1912 K.E. Handbook
you found a while back...).
My schedule is relatively open. When is a good time for you?
Perhaps we could meet when you are on campus anyway; otherwise I could
meander over to Fairchild.
Russ
∂17-May-83 1238 Hart%SRI-KL.ARPA@SCORE Re: Get Together
To: RDG at SU-AI
cc: Duda at SRI-KL
Russ,
As you probably have heard by now, I have resigned from Fairchild
and am in what might best be called a transition mode. I'll be out of town
until about May 26, and then life will be pretty hectic for a little bit.
Notwithstanding local logistics problems, I am interested in continuing the
conversation, and will follow up when I'm able. Incidentally, my new location
will be in Menlo Park (at least for a while), within easy bike range of the
Tuesday night folk dancing. The new phone number is 325-9339. Give
me a call if a couple of weeks go by and you don't hear from me. You
might also get in touch with Dick Duda, if you're looking for some sage
comments and a few good ideas...
Cheers,
Peter
-------
∂12:43 6-Jun: HART@SRI-KL/su News, etc
Greetings!
Like everyone else, I was surprised to hear the news;
and curious about the reasons -- especially as Fairchild was so high on
my list of places to seriously consider... though that was due largely
to your presence there.
Anyway, how is your new job going? Has life's "hecticity index"
begun to decline? And will you have a few moments in the near future
to administer "advice to a young scientist" (with apologies to Meduare(sp) )?
If so, I can be reached here, RDG@SAIL, or contacted at 497-1863 [MJH]
or 326-1730.
Cheers,
Russ
∂
He called ~21/Jun - we met 4:30PM 23-Jun-83
∂17:19 14-Jun: jf, pressburger@kestrel/su Problem #2
Express any positive integer, n, as product of primes --
n = ||(pi ↑ ai),
where pi is prime.
For any n, define lp(n) as the largest prime, pm, which divides n.
(Alternatively it is largest of the pi's above, whose associated ai>0.)
Let le(n) be the largest of the exponents associated with n in the above
prime decomposition -- i.e. it is the largest ai above.
Claim that n is "Tom-like" is lp(n)<le(n).
How many Tom-like numbers are there?
Define TL(n) as the number of Tom-like numbers less than n.
Describe the behaviour of TL(n)/n for large n.
∂15-Jun-83 1312 JF@SU-SCORE.ARPA Re: Problem #2
To: RDG@SU-AI.ARPA
you are merciless. i just finished that program and have to write the
documentation. now i will have an excuse to procrastinate.
thanks anyway,
joan
-------
Partition |N (natural numbers) according to their prime divisors --
i.e. each
Take any set of primes, {pi}
∂Mon 13 Jun 83 15:29:05-PDT <BUCHANAN@SUMEX-AIM.ARPA> summer support
To: greiner@SUMEX-AIM.ARPA
cc: rindfleisch@SUMEX-AIM.ARPA, feigenbaum@SUMEX-AIM.ARPA,
lenat@SUMEX-AIM.ARPA, genesereth@Score, bgb@SUMEX-AIM.ARPA
Russ,
for a number of reasons the HPP is in a financial bind this
summer. We have been unable to determine a source of summer support
for your work, much as we have tried. We have been delayed in
bringing this up because we had hoped that "something" would materialize,
but it has not.
Between Ed, Doug, Mike & me we should be able to help
you find interesting work this summer. Please let us know if you need
help. Needless to say, I am very sorry that we can't let things ride
with you on the payroll, but we have no discretionary funds.
bgb
-------
∂19 Jun 83 23:14:28-PDT <LENAT@SU-SCORE.ARPA> [Edith Wang <EWANG@SUMEX-AIM>: Summer support - Greiner]
To: GREINER@SU-SCORE.ARPA
Russ,
As you probably know, I have told Edith to pay for you out of my unrestricted
funds for the summer. This is unprecedented, and I hope you appreciate
the effort we're making to keep you from further delays. So get to work!
Seriously, let me know how things are going; I'll be back from ML-83 in a week.
Doug
---------------
Received: from SUMEX-AIM by Score with Pup; Thu 16 Jun 83 15:44:16-PDT
Date: Thu 16 Jun 83 15:41:40-PDT
From: Edith Wang <EWANG@SUMEX-AIM>
Subject: Summer support - Greiner
To: lenat@SUMEX-AIM
cc: buchanan@SUMEX-AIM, rindfleisch@SUMEX-AIM, ewang@SUMEX-AIM,
genesereth@SUMEX-AIM
Doug,
We have arranged RA support for Greiner -- 50% from your unrestricted
account and another 50% from DARPA. -- Edith
-------
----
∂13:26 27-Jun: lenat@SUMEX-AIM/su Support - Financial and Emotional
First, thanks much for finding financial support for me for the summer.
It will indeed make things easier; and I do appreciate your efforts.
(I am curious what the unprecedented-ness is, though:
I was under the impression it was HPP's financial straits, rather than
summer support coming from unrestricted funds. But no matter...)
Second, I really have to talk with you, soon. While I have felt for a while
that I've been sorta floundering, I've recently been sickened by the realization
that I may have absolutely nothing new to say about analogy -- despite several
years of work!
Anyway, I've scheduled a meeting with Mike tomorrow (Tuesday) at 3:30PM --
if you think it appropriate, could you join us then? Otherwise, my schedule
is fairly open; a Wednesday noon meeting with Johan and possibly John Brown
is my only constraint.
Russ
∂28-Jun-83 1041 LENAT@SU-SCORE.ARPA Re: Support - Financial and Emotional
To: RDG@SU-AI.ARPA
In-Reply-To: Message from "Russell Greiner <RDG@SU-AI>" of Mon 27 Jun 83 13:26:00-PDT
My car is down, so I'll be at PARC all day today.
Tomorrow 10-11:30 at PARC is good, then you can meet with J&J.
OK?
Doug
-------
∂28 Jun 83 08:46:03-PDT <LENAT@SU-SCORE.ARPA> change
To: GREINER@SU-SCORE.ARPA
Russ,
Let's meet 10-11 at MJH, not PARC after all, OK? That still gives
you plenty of time to get to PARC after our mtg.
DOug
-------
-------
∂16:29 28-Jun: LENAT@SU-SCORE/su re: change
That time's ok, but sub-desirable. Perhaps at MJH AFTER my meeting w/J&J --
at say 3PM? Let me know soon -- else I'll be MJH tomorrow at 10AM.
Russ
∂17-Jun-83 1034 WESCOURT@USC-ISI Party
From: Keith Wescourt
To: rdg@SU-AI
Reply-To: Wescourt@USC-ISI
Russ,
We are having a party on July 2 to belatedly warm our new abode,
which we transferred to from our previous abode next door a couple
of months ago. I don't have your address, so I couldn't mail you
an invitation (with a map--too bad for you). Anyway, you and
a friend, if you like, are invited to arrive after 4pm. Bring
something to barbeque (and special beverages if you are choosy)
and we'll supply other victuals.
Please RSVP (by phone, 851-0934, or netmail). We hope to see
you there. (By the way, I'm leaving on vacation and won't be back
until late June, so if you need further info it will have to wait
till then.)
Keith
-------
∂15:21 3-Jul: WESCOURT@USC-ISI/su wrt Party
Sorry I couldn't attend -- and that I didn't respond until now.
(It wasn't until Friday that I knew one way or the other...)
Anyway, I hope it went well; and please do invite me again to such things.
Thanks,
Russ
∂05-Jul-83 1709 CAB MacArthur
Thanks for the news "clipping".
There is a great film called "If I had a Million"
which stars, among others, W.C. Fields, George Raft, and
Charles Laughton. It was the ancestor of the Tipton Millionaire
series.
Perhaps the MacArthur Foundation should get a print.
Or maybe I should.
yrs,
CAB
∂11:11 29-Jun: dietterich@sumex/su Meetings - past and future
Tom -
I curious to hear what sorts of things you are up to these days,
and about the recent ML workshop. Could we get together in the near
future to discuss this, and anything else which seems pertanent (e.g. my
thesis stuff).
Russ
∂29-Jun-83 1435 DIETTERICH@SUMEX-AIM Re: Meetings - past and future
To: RDG@SAIL
cc: DIETTERICH@SUMEX-AIM
Russ,
I'll be out of town for the rest of this week and the 4rth weekend.
How about next week sometime, Wed-Fri? Doug, Bruce, and I will
probably give a siglunch on 8-july about the workshop.
Lots of interesting analogy papers (ie., at least 2).
--Tom
-------
∂16:29 29-Jun: dietterich@sumex/su Sure,
How about Wednesday, 6-July (ohmigod - is it JULY already!), at 11AM?
Russ
∂10:02 5-Jul: dietterich@sumex/su Damn -
Ellie scheduled me to meet with Wm Gale at 11Am tomorrow (6-July)!
Maybe after his talk; ca 3:00PM?
Russ
∂05-Jul-83 2318 DIETTERICH@SUMEX-AIM Re: Damn -
To: RDG@SAIL
cc: DIETTERICH@SUMEX-AIM
In-Reply-To: Your message of Tue 5 Jul 83 10:02:00-PDT
3:00 is fine. see you then.
--Tom
-------
∂11:00 6-Jul: DIETTERICH@SUMEX-AIM/su Of course,
I just scheduled that slot. How about tomorrow -- either 11-12,
or anytime after 1pm?
Sorry,
Russ
∂06-Jul-83 1103 DIETTERICH@SUMEX-AIM Re: Of course,
To: RDG@SAIL
cc: DIETTERICH@SUMEX-AIM
Friday would be better for me, say 3:00pm? This would be after the siglunch,
too, so there would be less duplication.
--Tom
-------
∂12:23 6-Jul: DIETTERICH@SUMEX-AIM/su This time, for sure...
Ok then, Friday at 3PM. (I've a meeting with DBL at 1:30, which should be
well over by then.) By the way, there has still been no announcement of
that SigLunch.
∂20:12 5-Jul: dekleer@parc/su Sandwiched in
Johan -
I won't be able to get to PARC tomorrow -- I'm scheduled to
yak with some fellow from Bell at 11AM, and then attend his talk at 1:30
that afternoon. Thursday and Friday lunch are both out as well.
(Misc people have scheduled a Machine Learning Workshop discussion
for then -- you might want to attend that.)
Maybe some other time -- or just push our meeting until
sometime next week?
Russ
∂06-Jul-83 1010 DEKLEER.PA@PARC-MAXC.ARPA OK.
Say next week then, what is the Machine Learning Workshop topic?
∂12:36 6-Jul: DEKLEER@PARC-MAXC/su stay tuned
I think it will be a summary of the sessions there. There should be
an announcement soon.
Russ
∂Sun 24 Jul 83 03:11:07-PDT Janet Coursey <JVC@SU-SCORE.ARPA> ai-list
To: greiner@SU-SCORE.ARPA
Russell,
Do you know the address to request being added to the AI-list mailing
list? I thought it was sri-ai but that didn't work. sri-csl maybe? Thanks,
Janet
-------
∂25-Jul-83 1743 JVC@SU-SCORE.ARPA Re: Volume 1 -- or did you have this already?
Thanks, I did not read the first issue. I'll send a note now
and get myself added to the mailing list.
Janet
-------
∂10:59 28-Jul: MULLEN@SUMEX-AIM, genesereth@sumex/cc/su Re: Report
∂27-Jul-83 1136 MULLEN@SUMEX-AIM Report
To: rdg@SAIL
The copies of your report are now in the hpp-lib.
-------
[RDG] Thanks. I looked over one copy, and it looks (of course) perfect.
Russ
∂28-Jul-83 1336 JED jobs
To: RDG@SU-AI, JDM@SU-AI
There's a fellow named Mike Nelson, an executive recruiter, who's
looking for someone with knowledge-based system background, for
a start-up in the LA area. Since I've already taken a job, I
gave him your names, in case you want to talk to him.
(The company seems to be doing both knowledge-based systems and database
work. It's small, growing rapidly, and well-financed. The job pays big
buck$. What the hell, it's gratifying to find out how much the outside
world is interested in you.)
Jim
∂13:16 26-May: bcm/su Prolog hereabouts
Ben -
Can you help here?
Russ
∂16:20 27-May: jjw/su Prolog hereabouts
Joe -
Ben M. tells me you might be able to help here.
Russ
∂25-May-83 0436 FRALEY%HP-HULK.HP-Labs@Rand-Relay [David Warren <WARREN@SRI-AI.ARPA>: Re: Prolog]
Received: from RAND-RELAY by SU-AI with TCP/SMTP; 25 May 83 04:36:45 PDT
Date: 24 May 1983 0812-PDT
From: FRALEY.HP-HULK@Rand-Relay
Return-Path: <FRALEY%HP-HULK.HP-Labs@Rand-Relay>
Subject: [David Warren <WARREN@SRI-AI.ARPA>: Re: Prolog]
Received: by HP-VENUS via CHAOSNET; 24 May 1983 08:12:47-PDT
To: greiner@SU-AI
Message-Id: <422637168.3121.hplabs@HP-VENUS>
Via: HP-Labs; 25 May 83 4:27-PDT
Any chance you could pull a copy of Prolog off of Arpanet for me?
---------------
Mail-From: HP-VENUS received 23-May-83 23:13:24
Date: Mon 23 May 83 15:09:10-PDT
From: David Warren <WARREN@SRI-AI.ARPA>
Received: by HP-VENUS via CSNET; 23 May 1983 23:13:46-PDT (Mon)
Received: from SRI-AI.ARPA by rand-relay.ARPA ; 23 May 83 15:17:23 PDT (Mon)
To: FRALEY.HP-HULK@RAND-RELAY.ARPA
Via: UDel; 23 May 83 23:13-PDT
Subject: Re: Prolog
In-Reply-To: Your message of Wed 18 May 83 17:45:00-PDT
Message-Id: <422604827.25363.hplabs@HP-VENUS>
Officially, we're not supposed to distribute Prolog from SRI.
(SRI is worried about getting sued). So if you could get it
over the Arpanet, that would not require any overt action on my
part, but dealing with tapes would be a different matter.
However, things are looking brighter for sorting out a pukkah
arrangement for disseminating Prolog with British Technology Group.
In the mean time, I'm afraid I can't help, it seems.
David.
-------
-------
FRALEY%HP-HULK.HP-Labs@Rand-Relay/su Prolog
FRALEY%HP-HULK@Rand-Relay/su Prolog
∂13:59 28-May: "FRALEY%HP-HULK"@Rand-Relay/su Prolog
Bob -
I got this response on my second try. Do you have access to
SCORE, or should I try poking around for you?
Russ
∂28-May-83 1155 JJW Prolog
If you're looking for TOPS-20 Prolog, see if what you want is in the
directory <PROLOG> at Score.
Joe
∂29-May-83 1917 FRALEY%HP-HULK.HP-Labs@Rand-Relay Re: Prolog
Received: by HP-VENUS via CHAOSNET; 29 May 1983 10:41:42-PDT
In-Reply-To: Your message of 28-May-83 1400-PDT
Message-Id: <423078103.17352.hplabs@HP-VENUS>
Via: HP-Labs; 29 May 83 19:13-PDT
I am interested in the TOPS-20 version of Prolog, but I don't have access to
SCORE. I don't know if we have any network access that programs can be
transferred over, but I'll check.
-------
∂12:06 1-Jun: "FRALEY%HP-HULK"@Rand-Relay/su {SCORE}<Prolog>PROLOG.TAPE
HOW TO PREPARE A PROLOG TAPE
The tape should be prepared with DUMPER in INTERCHANGE mode at 1600
bpi. The files to be included on the tape are those enumerated in
<PROLOG>TAPE-FILES, namely:
READ.ME PROLOG.EXE
PLOG10.EXE PLCOMP.EXE
PROLOG.HLP PROLOG.DOC
TUTORI.LPT GUIDE3.MEM
TUTORI.PL DEBUG.MEM
CAI.PL
For safety, the files should be duplicated in two separate save sets.
To assist the recipient of the tape, a DUMPER listing of the tape
should be made, and also a listing of READ.ME should be provided.
Here are the DUMPER commands which achieve the above:
@ASSIGN MTA0: ! assign tape drive 0 (or whatever drive
@DUMPER ! the tape is mounted on)
DUMPER> TAPE MTA0:
DUMPER> REWIND
DUMPER> DENSITY 1600
DUMPER> INTERCHANGE
DUMPER> FILES ! just to see what's happening
DUMPER> SSNAME PROLOG1
DUMPER> SAVE @<PROLOG>TAPE-FILES
! Dumper will now proceed to dump 11 files
! and advise their names on the terminal
DUMPER> SSNAME PROLOG2
DUMPER> SAVE @<PROLOG>TAPE-FILES
! same again for luck
DUMPER> REWIND
DUMPER> CHECK ! verify the first save set is correct
DUMPER> CHECK ! verify the second save set is correct
DUMPER> REWIND
DUMPER> PRINT TAPE.DIR ! produce a listing of the tape on TAPE.DIR
DUMPER> EXIT
@UNLOAD MTA0:
@DEASSIGN MTA0:
@PR TAPE.DIR ! print the listing
@PR <PROLOG>READ.ME ! print this to help the tape recipient
-------
∂14:52 27-Jul: appelt@sri, de2smith@sumex, pressburger@kestrel/su Internal Frame Backpack
I'm now in the market for a reasonable (read "not cheap, but not
top of the line") internal frame backpack -- basically for my upcoming
trip to Europe.
I'd be interested in any information any of youse guys could volunteer;
or pointers to others who might have relevant data.
(Mail to RDG@SAIL, drop by MJH 311, or call me at 326-1730 or 497-1863.)
Thanks,
Russ
∂27-Jul-83 1655 APPELT@SRI-AI.ARPA Re: Internal Frame Backpack
To: RDG@SU-AI.ARPA
In-Reply-To: Message from "Russell Greiner <RDG@SU-AI>" of Wed 27 Jul 83 14:52:00-PDT
Russ,
My recommendations for top of the line internal frame gear goes
to Lowe Alpine Systems, who makes a whole line of internal frame backpacks
suitable for anything from overnighters to Everest expeditions. I think
they're among the best on the market, although rather expensive.
Probably one of the best in the price/performance ratio is the
large Wolf Pack by Camp Trails. I know a person who owns one, and thinks
it is a great buy and a great pack. It's available at Dealin' Down.
Also, a number of companies (REI? North Face?) make a "combination"
pack that might be just what you're looking for. They have straps to be
used as a pack, but they can be contained within a zip-off compartment
to convert the pack into a suitcase for airport and train travel.
- Doug
-------
∂28-Jul-83 0942 PRESSBURGER@KESTREL Internal Frame Backpack
To: Russell Greiner <RDG at SU-AI>
In-reply-to: The message of 27 Jul 83 1452 PDT from Russell Greiner <RDG at SU-AI>
ASK dick gabriel, check out NORTH FACE, Dealing Down, maybe Eddie Bauer.
-tom
∂28-Jul-83 0901 DE2SMITH@SUMEX-AIM Re: Internal Frame Backpack
To: RDG@SAIL
In-Reply-To: Your message of Wed 27 Jul 83 14:52:00-PDT
There is a good article on internal frame packs in backpacker magazine
October/November 1981.
Last time I looked, the Kelty's seemed to be more reasonably priced than
many others.
The Sierra Designs seconds department both here and in Berkeley is a good
place to look for bargains.
-- de2
-------
∂28-Jul-83 1109 GREINER@SUMEX-AIM That article ...
To: de2SMITH@SUMEX-AIM
cc: rdg@SAIL
Thanks for the pointer. Do you have a copy of that article,
which I could copy? (I've another friend interested as well.)
Russ
-------
∂28 Jul 83 13:07:00-PDT <DE2SMITH@SUMEX-AIM.ARPA> Re: That article ...
To: GREINER@SUMEX-AIM.ARPA
In-Reply-To: Your message of Thu 28 Jul 83 11:10:04-PDT
Yes, I'll bring it in tomorrow.
-- de2
-------
∂15:56 29-Aug: Frawle@mit-mc/su Re: IJCAI Msg
Dr Frawley -
I did get the message you left for me at IJCAI, but hadn't
had time to respond until now. While I am finishing up,
I still have many months of (research) work to do on my thesis;
I'll certainly not be at an interviewing state before February 1984.
Your job decription does sound intriguing, and I would definitely be
interesting in hearing more about it -- now, and/or when I formally
interview.
Russ Greiner
RDG@SAIL
(415) 497-1863
∂21:02 14-Sept: vanLehn@parc, brown@parc/cc/su In case you didn't see it...
Kurt -
The current issue of "Science" (Vol 221, No. 4615, 9/Sep/83)
reviews the book "The Development of Mathematic Thinking" on pages 1042-43;
and has a lot to say, favorably, of your work.
Congratulations,
Russ
∂15-Sep-83 0920 VanLehn.PA@PARC-MAXC.ARPA Re: In case you didn't see it...
To: Russell Greiner <RDG@SU-AI.ARPA>
cc: vanLehn.PA@PARC-MAXC.ARPA, brown.PA@PARC-MAXC.ARPA
Thanks, Russ. I keep waiting for the Sept. Science to appear in the
bookstore so I can send a copy to my parents. It's not the cover of
Rolling Stone, but I'll take it!
-Kurt
∂20-Sep-83 0707 PETTY@RUTGERS.ARPA 1983 abstract mailing
To: Lerman@SRI-KL.ARPA, SHORTLIFFE@SUMEX-AIM.ARPA, DREIFUS@WHARTON-10.ARPA,
bennett@SU-SCORE.ARPA, MITTAL@PARC-MAXC.ARPA, CHANDRASEKARAN@RUTGERS.ARPA,
JSMITH@RUTGERS.ARPA, WILKINS@SRI-KL.ARPA, BRUCE@BBNA.ARPA,
FRIEDLAND@SUMEX-AIM.ARPA, PLONDON@USC-ISIB.ARPA, RDG@SU-AI.ARPA,
PRESSBURGER@KESTREL.ARPA, CSD.GARDNER@SU-SCORE.ARPA, FAGAN@SUMEX-AIM.ARPA,
FIKES@PARC-MAXC.ARPA, JMC@SU-AI.ARPA, CLANCEY@SUMEX-AIM.ARPA,
KRD%MIT-OZ@MIT-ML.ARPA, HAMILTON.ES@PARC-MAXC.ARPA, AMSLER@SRI-AI.ARPA,
ChiNguyen.ES@PARC-MAXC.ARPA, lisa@UTEXAS-11.ARPA, kwh%MIT-OZ@MIT-ML.ARPA,
UTGOFF@RUTGERS.ARPA, TUROCK@RUTGERS.ARPA, ECG.RICH@DEC-MARLBORO.ARPA
cc: petty@RUTGERS.ARPA, loungo@RUTGERS.ARPA
Below is a list of our newest technical reports.
The abstracts for these are available for access via FTP with user account
<anonymous> with any password. The file name is:
<library>tecrpts-online.doc
If you wish to order copies of any of these reports please send mail via the
ARPANET to LOUNGO@RUTGERS or PETTY@RUTGERS. Thank you!!
[ ] CBM-TR-128 EVOLUTION OF A PLAN GENERATION SYSTEM, N.S. Sridharan, J.L.
Bresina and C.F. Schmidt.
[ ] CBM-TR-133 KNOWLEDGE STRUCTURES FOR A MODULAR PLANNING SYSTEM, N.S.
Sridharan and J.L. Bresina.
[ ] CBM-TR-134 A MECHANISM FOR THE MANAGEMENT OF PARTIAL AND INDEFINITE
DESCRIPTIONS, N.S. Sridharan and J.L. Bresina.
[ ] DCS-TR-126 HEURISTICS FOR FINDING A MAXIMUM NUMBER OF DISJOINT BOUNDED
BATHS, D. Ronen and Y. Perl.
[ ] DCS-TR-127 THE BALANCED SORTING NETWORK,M. Dowd, Y. Perl, L. Rudolph and
M. Saks.
[ ] DCS-TR-128 SOLVING THE GENERAL CONSISTENT LABELING (OR CONSTRAINT
SATISFACTION) PROBLEM: TWO ALGORITHMS AND THEIR EXPECTED
COMPLEXITIES, B. Nudel.
[ ] DCS-TR-129 FOURIER METHODS IN COMPUTATIONAL FLUID AND FIELD DYNAMICS,
R. Vichnevetsky.
[ ] DCS-TR-130 DESIGN AND ANALYSIS OF PROTECTION SCHEMES BASED ON THE
SEND-RECEIVE TRANSPORT MECHANISM, (Thesis) R.S. Sandhu. (If
you wish to order this thesis, a pre-payment of $15.00 is
required.)
[ ] DCS-TR-131 INCREMENTAL DATA FLOW ANALYSIS ALGORITHMS, M.C. Paull and B.G.
Ryder.
[ ] DCS-TR-132 HIGH ORDER NUMERICAL SOMMERFELD BOUNDARY CONDITIONS: THEORY AND
EXPERIMENTS, R. Vichnevetsky and E.C. Pariser.
[ ] LCSR-TR-43 NUMERICAL METHODS FOR BASIC SOLUTIONS OF GENERALIZED FLOW
NETWORKS, M. Grigoriadis and T. Hsu.
[ ] LCSR-TR-44 LEARNING BY RE-EXPRESSING CONCEPTS FOR EFFICIENT RECOGNITION,
R. Keller.
[ ] LCSR-TR-45 LEARNING AND PROBLEM SOLVING, T.M. Mitchell.
[ ] LRP-TR-15 CONCEPT LEARNING BY BUILDING AND APPLYING TRANSFORMATIONS
BETWEEN OBJECT DESCRIPTIONS, D. Nagel.
-------
∂12:03 20-Sept: PETTY@RUTGERS/su Request for Reports
Please mail the following reports to
Russ Greiner
Margaret Jacks Hall
Computer Science Department
Stanford University
Stanford, CA 94305-2085
Thank you.
[x] CBM-TR-128 EVOLUTION OF A PLAN GENERATION SYSTEM, N.S. Sridharan, J.L.
Bresina and C.F. Schmidt.
[x] CBM-TR-133 KNOWLEDGE STRUCTURES FOR A MODULAR PLANNING SYSTEM, N.S.
Sridharan and J.L. Bresina.
[x] CBM-TR-134 A MECHANISM FOR THE MANAGEMENT OF PARTIAL AND INDEFINITE
DESCRIPTIONS, N.S. Sridharan and J.L. Bresina.
[x] DCS-TR-128 SOLVING THE GENERAL CONSISTENT LABELING (OR CONSTRAINT
SATISFACTION) PROBLEM: TWO ALGORITHMS AND THEIR EXPECTED
COMPLEXITIES, B. Nudel.
[x] LCSR-TR-44 LEARNING BY RE-EXPRESSING CONCEPTS FOR EFFICIENT RECOGNITION,
R. Keller.
[x] LCSR-TR-45 LEARNING AND PROBLEM SOLVING, T.M. Mitchell.
[x] LRP-TR-15 CONCEPT LEARNING BY BUILDING AND APPLYING TRANSFORMATIONS
BETWEEN OBJECT DESCRIPTIONS, D. Nagel.
∂20-Sep-83 1806 HOLTZMAN@SUMEX-AIM Copy of message sent to SUMEX.
Subject: Copy of message sent to SUMEX.
To: rdg@SAIL
Russ,
As I told you on Sunday, I would like to meet with you to talk
about MRS. In particular I am interested in the artihmetic-handling routines
I believe you wrote. I would not take much of your time, and lunch is a
possibility. Thursdays are out for me, but the rest of the week is usually
open. Suggest a time and place.
Sam
-------
∂11:02 21-Sept: HOLTZMAN@SUMEX/su Get together
Sam,
Noon today (Wednesday) is OK by me. Shall we meet here in my office
[MJH 311 -- in the NA section]? My only constraint today is a meeting from
2-3PM.
Russ
To: greiner@SAIL
cc: rdg@SAIL
Russ,
I meant he doesn't carry (bear ?) his hame right : the coomputer
doesn't : SAIL .AI. stands for Artificial Intelligence, right ? or
does it stand for Absence of Intelligence ?
Rene
---------------
Return-Path: RDG@SU-AI
Received: from SU-AI.ARPA by SUMEX-AIM.ARPA with TCP; Wed 12 Oct 83 18:00:29-PDT
Date: 12 Oct 83 1756 PDT
From: Russell Greiner <RDG@SU-AI>
Subject: re: re: Fortran/lisp benchmark
To: BACH@SUMEX-AIM
∂11-Oct-83 1958 BACH@SUMEX-AIM re: Fortran/lisp benchmark
To: RDG@SAIL
I am sorry for the inconvenience. Can I send a message to Gabriel@sail
and expect it to arrive ? Why is this computer so stupid ?
It doesn't carry (bear ?) his name right !
Thanks for forwarding the message.
Rene
-------
No problem.
Try it; maybe.
It just is.
Ah, but it does -- he just happens to be RPG.
My pleasure.
RDG
PS. I tried both addresses, let me know if you receive two messages.
-------
∂21:39 12-Oct: BACH@SUMEX/su What a confusing message!
Yes, I did receive two messages --
apparently both RDG@SAIL and GREINER@SAIL will reach me.
∂12-Oct-83 1938 BACH@SUMEX-AIM [Russell Greiner <RDG@SU-AI>: re: re: Fortran/lisp benchmark ]
∂15:46 29-Aug: jc50@cmua/su FollowUp-ette
Jaime -
It was interesting talking to you last week -- do keep me
abreast of your work. (I also thought your presentation was quite good.)
As for my activities: I will forward the results to you soon.
(If only there weren't so many times before I've said that...)
In case you want to contact the fellow in Cambridge who is busily
implementing some of your "analogy to guide a MEA problem solver" stuff
in the blocks world domain, he is
Mr. John Carroll
University of Cambridge,
Computer Laboratory
Corn Exchange Street
Cambridge CB2 3QG
England
Telephone: (0223) 35.2435 x216
Telex: via 81240 CAMSPL G
Russ
For Peter Hart's Benefit
Date: Mon 31 Oct 83 22:57:03-CST
From: John Hartman <CS.HARTMAN@UTEXAS-20.ARPA>
Subject: Fri. Grad Lunch - Understanding and Translating LISP
[Reprinted from the UTEXAS-20 bboard.]
GRADUATE BROWN BAG LUNCH - Friday 11/4/83, PAI 5.60 at noon:
I will talk about how programming knowledge contributes to
understanding programs and translating between high level languages.
The problems of translating between LISP and MIRROR (= HLAMBDA) will
be introduced. Then we'll look at the translation of A* (Best First
Search) and see some examples of how recognizing programming cliches
contributes to the result.
I'll try to keep it fairly short with the hope of getting critical
questions and discussion.
Old blurb:
I am investigating how a library of standard programming constructs
may be used to assist understanding and translating LISP programs.
A programmer reads a program differently than a compiler because she
has knowledge about computational concepts such as "fail/succeed loop"
and can recognize them by knowing standard implementations. This
recognition benefits program reasoning by creating useful abstractions and
connections between program syntax and the domain.
The value of cliche recognition is being tested for the problem of
high level translation. Rich and Temin's MIRROR language is designed
to give a very explicit, static expression of program information
useful for automatically answering questions about the program. I am
building an advisor for LISP to MIRROR translation which will exploit
recognition to extract implicit program information and guide
transformation.
------------------------------