perm filename MRG[RDG,DBL] blob
sn#770264 filedate 1984-09-22 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00063 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00008 00002 Mailed to CSD.GENESERETH@SCORE, GENESERETH@SUMEX, MRC@MIT-AI 13:14 23-June-80
C00010 00003 ∂23 Jun 1981 1445-PDT CSD.GREINER Functors, and friends
C00012 00004 ∂ 1 Jul 1981 1005-PDT CSD.GENESERETH things
C00017 00005 ∂TO MRG 15:14 22-Jul
C00022 00006 ∂TO MRG 13:33 29-Jul
C00024 00007 ∂TO MRG 16:39 30-Jul
C00027 00008 ∂TO STT (CC MRG) 18:21 31-Jul-81
C00033 00009 ∂TO MRG 13:23 28-Oct
C00034 00010 ∂29 Oct 1981 1058-PST Russell Greiner <CSD.GREINER> Little MercuriVenereal
C00037 00011 ∂ 6 Nov 1981 1352-PST <CSD.GREINER> "Mycin" and Mycin (NOT Mycin↑ or ...)
C00041 00012 ∂21 Feb 1982 2001-PST <CSD.GREINER at SU-SCORE> Records (note absense of "er")
C00043 00013 ∂TO csd.genesereth@score 18:54 26-Feb
C00044 00014 ∂TO csd.genesereth@score 15:19 10-Mar
C00046 00015 ∂TO csd.genesereth@score, ars, tom@kestrel 18a9:50 6-Apr
C00047 00016 ∂18-Jun-82 1830 <CSD.GREINER at SU-SCORE> Quote from MRG
C00048 00017 ∂13 Jul 1982 2126-PDT <CSD.GREINER at SU-SCORE> Theory Parts
C00053 00018 ∂TO csd.genesereth@score 15:09 26-Aug-82
C00054 00019 ∂ 3 Oct 1982 1624-PDT <CSD.GREINER at SU-SCORE> Ready for Round 1
C00056 00020 ∂TO csd.genesereth@score 13:28 1-Dec-82
C00057 00021 ∂25 Sep 1982 1404-PDT <CSD.GREINER at SU-SCORE> Slave Labor...
C00059 00022 ∂TO csd.genesereth@score 17:47 23-Jan-83
C00062 00023 ∂TO csd.genesereth@score 15:31 2-Feb-83
C00063 00024 ∂TO csd.genesereth@score, freeman@diablo 14:58 14-Feb-83
C00069 00025 ∂04-Mar-83 0002 CSD.GENESERETH@SU-SCORE mrs
C00070 00026 15:50 18-Apr: genesereth@sumex/su IJCAI Paper
C00072 00027 ∂15:56 28-Mar-82 -- dietterich@sumex/su Jack's comment
C00091 00028 ∂17:02 3-May: jdm/su wrt Note from MRG
C00096 00029 ∂16:45 3-May: genesereth@sumex/su Quickies
C00098 00030 ∂15-Jun-83 2234 GENESERETH@SUMEX-AIM Re: Any response to this?
C00099 00031 ∂14:18 30-Jun: genesereth@sumex/su Algebraic Manipulator
C00100 00032 ∂12:40 1-Aug: genesereth@sumex/su Meating Thyme
C00101 00033 ∂11:55 29-Aug: genesereth@sumex/su Representation Issues
C00110 00034 ∂11:59 29-Aug: genesereth@sumex/su NP-completeness and parallelism, humor
C00120 00035 ∂12:02 29-Aug: genesereth@sumex/su NETL semantics
C00126 00036 ∂15:36 29-Aug: genesereth@sumex/su Final Thoughts on Novelty
C00128 00037 ∂13:33 2-Sept: genesereth@sumex/su Your SAIL stuff
C00139 00038 ∂10:54 1-Sep: genesereth@sumex/su Meeting Time
C00140 00039 ∂03-Aug-83 1132 MULLEN@SUMEX-AIM check for proceedings
C00142 00040 ∂15:19 15-Sept: genesereth@sumex/su Another week, another problem
C00143 00041 ∂18-Sep-83 1220 greiner@Diablo SUKAY
C00144 00042 ∂19-Sep-83 1057 GREINER@SUMEX-AIM [norvig%ucbkim@Berkeley (Peter Norvig): OPS5 tutor (or just doc)]
C00146 00043 ∂20:28 22-Sept: genesereth@sumex/su Hofstader & analogy, sorta
C00152 00044 ∂12:16 26-Sep: genesereth@sumex/su Many misc things (some timely)
C00154 00045 ∂Thu 29 Sep 83 21:58:46-PDT <GENESERETH@SUMEX-AIM.ARPA> meeting
C00157 00046 ∂13:48 14-Nov-83: genesereth@sumex/su ... and a time for everything ...
C00158 00047 ∂20:35 20-Feb-84: genesereth@sumex/su Is "undecidable" better than "independent"?
C00161 00048 ∂15:33 22-Mar: genesereth@sumex/su Aha, a new thesis generation program!
C00164 00049 ∂18:06 26-Mar: genesereth@sumex/su Grief, etc
C00167 00050 ∂19:46 19-Apr: genesereth@sumex\su From AIList
C00178 00051 ∂13:27 30-Apr: genesereth@sumex/su Timely (AAAI)
C00179 00052 ∂2 Apr 84 14:39:30-PST Mullen <MULLEN@SUMEX-AIM.ARPA> Appt with Mike on Tues 2:30
C00181 00053 ∂19 Jun 84 17:59:02-PDT <GREINER@SUMEX-AIM.ARPA> Good book
C00182 00054 ∂to MRG 15-Aug
C00185 00055 ∂14:04 21-Aug: genesereth@sumex/su Good News
C00186 00056 ∂11:11 27-Aug: genesereth@sumex/su CSLI Library
C00189 00057 ∂11:14 27-Aug: jfinger@diablo, genesereth@sumex/su from AI-List -- of interest?
C00192 00058 ∂12:38 29-Aug: lenat@score, genesereth@sumex/su Taxonomy Assistant [from AI-List]
C00196 00059 ∂16:37 2-Sep: genesereth@sumex, lenat@score/su Re: Taxonomies
C00199 00060 Return-Path: <greiner@diablo>
C00200 00061 ∂21:12 12-Sep: sjg, genesereth@sumex/su challenge for MRS?
C00205 00062 PROJECT ACTIVITIES FOR PROJECT F-1: THEORIES OF INFORMATION
C00208 00063 What's in a theory?
C00216 ENDMK
C⊗;
Mailed to CSD.GENESERETH@SCORE, GENESERETH@SUMEX, MRC@MIT-AI 13:14 23-June-80
Mike:
Going over my notes from Qual studying, I found a comment that I should
tell you about an article by Tanimob (from U of Wash, @ Seattle) which deals with
min'l cost rep'ns. It appears on page 871 of IJCAI-79. See if you think it's
at all relevant.
Also, (in case you are reading this from afar,) do you want me to pick up
newspaper, mail or whatever from you abode in your absence. I was going to ask
before I left last Saturday, but forgot.
Ciou,
Russ
∂23-Jun-80 1412 CSD.GENESERETH at SU-SCORE Re: Unimportant MISC
Ok I'll look at Tanimob's article. The problem with the IJCAIs is that
there are too many articles for me to read them all. Actually I would
appreciate it if you could drop by my house maybe once each week and collect
the pile of newspapers that will appear. Mail goes inside so that's no
problem. An thnaks. It's incredibly hot here (suppsedly 106
degrees).
-- Mike
-------
∂23 Jun 1981 1445-PDT CSD.GREINER Functors, and friends
To: CSD.GENESERETH
cc: CSD.GREINER
I followed up on your suggestion, and meandered over to the Math/CS library,
intent on finding literature about "function generation". Results:
in a word, blank. (<- or is that Zero Words?) Anyway, have you any
idea of even the category in which to probe? Functors, category theory
and various subspecies of combinatorics are all out -- or at least I
couldn't begin to understand what was being said in the various notes,
books and journals. Under the heading of Functions were hordes of sub-
entries -- none of which seemed pertanent. I consulted a few "organization
of the (math) world" indices; with equally little success.
Any other suggestions? Have you any math friends who owe you
half-a-favor? I "enclose" below a brief statement of the issue, both
to refresh your memory (as you're probably by now "What is he talking about!"
[in your confusion you probably even got the pronoun correct]) and to forward
to that mystery friend, as well.
------
<<blurb>>
Thanks,
Russ
-------
∂ 1 Jul 1981 1005-PDT CSD.GENESERETH things
To: csd.greiner
How's the homestead? Have you found a tenor yet? I had an idea
the other day that may be worth pursuing. I've been
wanting to play my lute for a while. How about if we form a "broken consort"
of 2 recorders, lute, and viola da gamba? Would Elizabeth join us?
mrg
-------
∂ 1 Jul 1981 1425-PDT CSD.GREINER Re: (ANY)things ?
To: CSD.GENESERETH
cc: CSD.GREINER
In-Reply-To: Your message of 1-Jul-81 1005-PDT
I've yet to reside there -- first my bro Seth & friend were there, and now
Elizabeth's bro, Bill, has taken over the residence. (It seems the last
clause of my "Feel free to stay there WHILE YOU SEARCH FOR A PERMANENT
PLACE" was ignored -- and I haven't the guts to play ogre and throw the
usurper out. Oh for some assertiveness...)
Well, what's the great new idea?
Elizabeth would probably consent (sp) to join us -- when are we'all meandering
East (E will be at Julliard starting late Sept...).
How was DIA? And MIT in general? and Arthur? any other Axxx females?
The ACL thing here is just winding down. One or two good talks -- a lot
more not-so-good. Do you know of John Barwise's work, with John Perry?
(They're some local philosphers, attempting to non-Montague-ise semantics.)
Bob Moore seems to really like it -- anyway, I felt there is something,
potentially useful there; and I do want to learn more about his stuff.
Another cute bit of (to me irrelevant)
research dealt with the famous when-are-pronoun-"referents"-resolved problem.
Nice solid work. The conclusion (that, in most cases, they are determined
at "input" (as opposed to later q/a-ing) time) might have some indirect
value -- perhaps as another idication of the sophisticated economy of
cognitive processing of which people are capable.
One paper had "analogy" in the title (should that have been `"analogy"'?
[actually that last quoted phrase (note I hereby refuse to help resolve
this problem by actually typing its extension) probably should have been
embedded in yet another set of quotes - yes?]) -- a bit disappointing.
It said nothing about how analogies are formed, or how they "work" --
only about how people actually generate a new context within an
analogizing clause. Nice, but irrelevant.
Which well summarizes my feel about that conference in general.
You're welcome to borrow my proceedings for a quick look see,
if you've a yen to confirm/deny this perspective.
I yakked with Hofstadter the other day, about analogies. He had little
to say, beyond the superficial stuff presented in his paper. Oh well...
In case you didn't know STT is beginning to MRS -- he called me in frustration
the other day - unable to figure out what happened to a few of the functions
mentioned in OUR manual... Now that DE2 has returned he can probably
begin work in earnest.
Your grass is green, and, at last check, there are no new infestations
(besides your usual warriors). Anything else I can check on for you?
When are you boating?
See you when you return...
Russ
∂TO MRG 15:14 22-Jul
Thesis related stuff
Whew... I came awfully close last Sunday to losing any chance
of ever graduating from Stanford:
I almost missed a little notice placed in your mailbox then,
telling you, in bold terms, to remove those offending blackberries from your
front yard, or suffer all sorts of bad consequences.
Figuring you'd have a hard time signing my thesis from behind prison bars,
I took the liberty of spending Sunday morning stripping in your yard.
(Yes, I agree it seems immoral, but 'dem are 'de rules.)
<I'm tempted now to go into great detail about the pain and anguish
I had to suffer during this arduous task;
and mention what a pathetic, wretched sight I was at its completion,
with blood spurting through each of the thousands of orifices opened
by those sharp, painful thorns on those hideous death-defying plants.
But you might feel I was exaggerating a tad, to get sympathy and perhaps
some unjustified faculty backing during subsequent thesis sorts of things.
So you're spared -- I will not now mention
those 7 long miles, thru snow and sleet, I had to trudge....>
Anyway, U No Hu U'r Friends Really R (← original subject of this message).
Other things:
They came and disconnected your phone the other day. Boo, hiss! I was all
set to do a full day of hacking, whilst recording.
Bill went back east for a while, so the house has been relatively vacant.
(He escaped just in time, to avoid the slow and excruciatingly painful chore
of denuding that obnoxious plant of yours.)
He apparently reglued one of your chairs, and unstopped your sink during
his stay. All I got for playing nice guy and going out of my way
by inviting him to stay is his sister's wrath -- who's now furious at me.
I just don't understand people in general, or women in particular.
Anyway, in addition to those sorts of thesis-promoting stuff I mentioned
above, I've also spent some time assembling the start of what may eventually
become something which sorta resembles a thesis draft.
(I realize that's not the best use of my time, considering the other favors
I could be doing for people like you and Doug, but someone's got to do
this sort of dirty work...)
Hopefully this skeleton will be sufficiently fleshed out by the time you
return to permit you to peruse it...
Hope you enjoyed your sailing, and rest of your vacation.
Doesn't it feel terrible to know it's now behind you -- that you've
used up all the leisure time you're likely to get for at least a year...
(Just thought I'd rub it in...)
Russ
∂TO MRG 13:33 29-Jul
Exploring the Depths of Undying Gratitude
Mike
E's bro Bill called, and asked how best he could express his
gratitude for the use of your "in abstentia" hospitality, and house.
After hanging up it dawned on me there was a nice, inexpensive giftee
he could supply:
Bill seems just the sort of person who would savor the opportunity to
go spelunking about your "basement". So when he calls, if you still
desire that service, you could suggest this possibility...
(Remember this would really count as a favor FROM ME, as it was really
my magninimity(sp) he is now paying back; and by the transitivity of
favors ...)
Russ
∂TO MRG 16:39 30-Jul
"x is a variable"
Mike -
First, your house is once again yours -- I have cleared out;
leaving only messages in my wake. (Sounds poetic, no?)
Second, I have a question about various degrees of meta.
--- Begin Story ---
Let's say I have a variable, x, which is a member of the set, S.
I think we'll agree that x:Isa = (S)
[or, in MRG-ise, ($ASSERT '(MEMB x S))].
Now we want to state that x is a variable -- ie it makes sense to store
things like what is may be "skolemed on" on x. So we should be able to
make a meta statement, which is
x:MyIsa = (Variables)
[translation: ($ASSERT '(MEMB ↑x Variables))].
True so far? After all, x doesn't know whether it's a constant or not,
but the unit representing it had better, to know certain things about
how to handle dealings with this x...
Anyway, now some joker comes along and asks us how many properties does
this x unit have. When we say 31, he says "Aha, so x is a verbose unit",
and proceeds to make such a declaration:
x:MyIsa = (VerboseUnit)
[or, before he was enlightened: ($ASSERT '(MEMB ↑x VerboseUnits))].
--- End of Story ---
(Don't cry -- I never said it would be a happy story...)
Doesn't this seem wrong to you? Saying "x is a variable" seems clearly
different from "the x unit has many facts"; should both of these be lumped
together?
The 3rd assertion seems right to me, but I question the second one --
So now the question:
how do I represent the fact that x is a variable?
Russ
∂TO STT (CC MRG) 18:21 31-Jul-81
NonTerminal Terminology
I just yakked with Mike, and found that his terminology is quite
different from what I had thought.
He uses the term "abstraction" to refer to the mapping from an object
to a theory, and "generalization" to relate a pair of theories.
His justification was a little obscure, and had to be teased out:
he claims that any theory we might write about Fido would necessarily
contain only a subset of the possible facts about this entity --
i.e. many (significant but extraneous to the task) facts would be omitted.
Hence this "representation", T, would be a mere subtheory of the full
theory of Fido -- where that full encoding holds every possible fact
about Fido.
Now any subset of that theory, T, (call it S) is a "generalization"
of T -- ie every model of T is also a model of S.
As best I understand, it appears that Abstraction is inverse of Satisfies --
ie if the theory A is an abstraction for some object M, then M satisfies A;
and vice versa.
Comments:
1) To avoid problems of logical implication, we define theory
to refer to the deductive close of the initial axioms.
Realize that a proper generalization will therefore require removing
an infinite number of propositions to be.
2) Any Generalization of an Abstraction is still an Abstraction.
3) Note we need not, at this "epistlemological level", worry about isomorphism...
Two theories are equivalent iff they have the same models.
Hence the actual symbols used are irrelevant
-- this mapping from object to constant, etc,
is in the jurisdiction of the interpretation of the model's elements
in the theory.
[Heuristically this will clearly the bulk of the analogizing work --
finding the method of mapping from symbols in one theory into the
corresponding symbols in the other.]
The related issue of how we can "default" some relations, become irrelevant --
it just means the interpretation of this symbol is slightly
different in this theory than in the other one.
-----
I still liked using the term abstraction to refer to the relation
connecting a pair of theories -- if
(Abstraction A B) <=> (ForEvery (M in Models) (Satisfies M B) => (Satisfies M A)).
However this seems quite a bit harder to pin down, as this quantification over
models is a tricky business. Comments, Mike?
Russ
-- To MRG again 20:18 1-Aug ---
∂01-Aug-81 1358 SYS
Queued mail to csd.genesereth at SU-SCORE -- ok
The following message was received from the remote host:
Mail to csd.genesereth at SU-SCORE ready
∂TO STT, MRG 10:57 1-Aug
FollowUp
I'd even be content to consider something like
(Abstraction A B) <=> (ForEvery (M in Models).
(Exists (I, I' in Interpretations).
(Satisfies M B I) =>
(Related I I') & (Satisfies M A I'))),
where this Satisfies relation considers the Interpretation used; and
the Related relation insists that the pair of interpretations are closely
related to one another: preferably identical, but it may allow things like
the teriary R relation symbol in I to correspond to the binary relation symbol
R' in I'. (I'm considering the interpretation I as a mapping of objects
in the domain of M into symbols in (the language of) B, such that every
relation ...)
Has anyone defined such a type of satisfaction; or dealt with
this sort of similar interpretations?
Russ
∂TO MRG 13:23 28-Oct
Cuteness for MRS manual
(all x (if (pair-of-sentences x)
(exist x1 x2 (if (equal x (stack-sentences x1 x2))
(equivalent x1 x2)))))
(if (pair-of-sentences $x)
(if (equal $x (stack-sentences ?x1 ?x2))
(equivalent ?x1 ?x2)))
[You may want:
(all x x1 x2 (if (and (pair-of-sentences x)
(equal x (stack-sentences x1 x2)))
(equivalent x1 x2)))
(if (and (pair-of-sentences $x)
(equal $x (stack-sentences $x1 $x2)))
(equivalent $x1 $x2))
...]
∂29 Oct 1981 1058-PST Russell Greiner <CSD.GREINER> Little MercuriVenereal
To: CSD.GENESERETH
cc: CSD.GREINER
How could we both have forgotten Venus? Any way, the Greek Hermaphrodite-
culus sound like a combination between a poisoning and a disease. No
wonder the Roman's overtook them...
Jeanne, Tom and I were all able to cancel our Bostonian reservations, you'll
be happy to know -- and we decided, instead to practice in our home turf.
Can you make it next Thursday? (Nov 5th I think.) It'll be at your
house, in case you're interested.
Did you read my Analogy.Bbd message yet? Do you agree with its contends,
especially those ideas attributed to you?
Next, after I sent that Cuteness to you, I realize the connective within
the existential quantifier should, of course, be an AND, not the IF I used.
Finally, I've got a grumble with you, which I've found myself unable to
express verbally over the last few months. Hence this written communique:
I felt rather slighted back at IJCAI. Correct me if I'm wrong, but
I really believe that I have made significant contributions to your thinking
about modifiability in general, and with respect to representations
in particular. In addition, I have clearly spent a non-trivial amount
of time and energy I on MRS -- both on its early design and on publicity.
I feel I was entitled to some recognition for these efforts, which you failed
to provide.
While I did mention this, in passing, at that dinner, I don't think I
adequately expressed my deep disappointment and betrayal.
Comments?
Isn't it amazing what I pack into these little notes? Never a dull moment...
Russ
∂ 6 Nov 1981 1352-PST <CSD.GREINER> "Mycin" and Mycin (NOT Mycin↑ or ...)
To: CSD.GENESERETH
cc: CSD.GREINER
I'd a quick, obvious comment I wanted to convey, regarding your comment
on sociological/legal aspects of Mycin.
1. I couldn't find Mycin in that book of pre-/suf-fixes (...ixen?). Next
time I see Ted I'll ask him.
2. Your 95% to 80% accuracy, with different areas of discrepancies, does NOT
seem to be a problem. There are two subcases, depending on how well known
Mycin's limitations are:
(i) If it is well-known (in which areas Mycin is limited), the physician should
not consult Mycin for these cases... and if he does, Mycin should defer to
the judgement of some human expert. (Take a limiting case: what if someone
asked Mycin about Cancer? It should know enough NOT to office any advice,
or, if it does, to clearly preface these suggestions with a
"I don't know as much about this as X, Y or Z, but if you're desperate.."
How is this situation different?)
(ii) No one knows, a priori, what Mycin's weak areas are. I, for one, would
much rather have a 95% chance of cure than an 80% chance. If there is no
reason for the doctor to think anyone/thing is superior to Mycin, I can't
imagine why he should not consult it, and follow any reasonable advice it
gives.
----
This was too easy -- I must not have understood your point. What was it?
Russ
-------
∂27 Dec 1981 1621-PST <CSD.GREINER at SU-SCORE> MRS = ...
To: CSD.GENESERETH
cc: CSD.GREINER
I finally realized what MRS "really" stands for:
MisNamed Representation System.
(or possibly Mis-Abbreviated R.S. ?)
There's only one obvious problem:
If we name it that, that name will no longer apply!
Where are Betrand and Wilhelm when we need them?
Russ
PS. If you'll forgive this minor trivia:
When can we meet to discuss my thesis ideas?
It's been a full month; and I really am waiting on you...
Could you, perhaps, devote, say, one hour to perform a quick and
obviously superficial reading? (Note this would still represent the largest
time commitment of any of the readers thus far.)
After this we could meet briefly
-- or even electronically if that would be less (time) expensive.
There is another option: I've been assuming you wanted to be on my reading
committee. Let me know if this is a faulty belief...
Russ
-------
-------
∂21 Feb 1982 2001-PST <CSD.GREINER at SU-SCORE> Records (note absense of "er")
To: CSD.GENESERETH
cc: CSD.GREINER
Can I borrow some of your records for a recording session, at my house,
tomorrow (22/Feb/82)? In particular, some of the Clannad stuff, and
who knows what else?
If so, how can I procure (sp) them? Perhaps I could simply pilfer(sp)
one of your keys, and meander over there in your absense, and
take the desired records... Other ideas?
Thanks,
Russ
-------
∂22 Feb 1982 0949-PST <CSD.GENESERETH at SU-SCORE> Re: Records (note absense of "er")
To: CSD.GREINER
no. You can record them at my house, but I have a long standing
policy of not loaning records to be played on other people's
equipment.
mrg
-------
∂TO csd.genesereth@score 18:54 26-Feb
Times when even MRS can't help...
Q: Where, of where, is your briefcase?
A: You left it by the ascending unenclosed staircase on the 2nd floor.
I stuck it in your office.
Remember this next Gray Tuesday/Black Friday, ...,
Russ
∂27-Feb-82 1150 Mike Genesereth <CSD.GENESERETH at SU-SCORE> Re: Times when even MRS can't help...
While I was lying in bed this morning, it occurred to me that I hadn't
brought it home with me. I spent 15 minutes trying to decide whether
o not I knew its whereabouts. Now you've spoiled all the fun.
mrg
-------
∂TO csd.genesereth@score 15:19 10-Mar
Noteworthy events, and others
Mike -
First, circle 18/March/82 on your calendar. On that date you
will have the honor of our presence at your humble abode, en-instrumented.
Second, I know of a bike which will be available, to the highest
bidder, from 21-March thru 26-March.
Over this interval its owner, having taken leave of his senses,
will be in the bowels of the east coast, visiting relatives in NY and Boston.
Poor fellow. Any recommendations to proffer to this poor misguided wanderer --
things an un-bicycled person can do or see in that area?
Finally, I'm curious about (on?) your thoughts/reflections wrt
that retreat. (I.e. should we just have surrendered?)
(I have to have something to report back to the JGF on the activities
of his missonary.)
Russ
∂TO csd.genesereth@score, ars, tom@kestrel 18a9:50 6-Apr
The Taxman Cometh
To whom it may concern:
There will be a recorder rehearsal this Thursday WEEK -- that is,
the 15th of April -- at Mike's house. All are welcome.
(By the way, we met yesterday evening -- apparently no one "leaked" this
information.)
Russ
∂18-Jun-82 1830 <CSD.GREINER at SU-SCORE> Quote from MRG
"I don't care about people; you can quote me on that"
(17/VI/82)
-------
∂13 Jul 1982 2126-PDT <CSD.GREINER at SU-SCORE> Theory Parts
To: CSD.GENESERETH at SU-SCORE
cc: CSD.GREINER at SU-SCORE
Mike -
This is an excerpt from a message I sent to Dave the other day.
...
6) A half-baked request:
(This is a more formal version of the request I issued verbally a while back.
This doubles as a justification for the "veracity" of Gordon's "fire only
certain rules" method, and proposes a different mechanism for achieving
this result.)
I would like to add additional structure to the internals of a theory.
For my applications, it is often useful to divide the contents of a theory
into several parts, perhaps annotated as
MATH:FACTS MATH:RULE.
These two would partition the set of assertions associated with the MATH
theory. The advantage is that each of these could themselves be considered
a theory; and as such could be readily activated or de-activated.
(This could lead to a considerably faster searching time --
the various FI and BC processes would need to only scan rules, and would
not be slowed down by the presence of additional facts.)
The next step is to subdivide this MATH:RULE theory-part into
MATH:RULE:ForwardOnly, MATH:RULE:BackwardOnly, and MATH:RULE:BothWays.
Now when we are forward chaining, we would de-activate
MATH:RULE:BackwardOnly, and consider only the other rules.
As I'm using already accepted operators for theories, all of these processes
are "legal". The only issues are (1) proliferation of names, and
(2) actual organization.
The first seems strictly an implementation issue -- there is no reason
a theory could not have, for its name, an arbitrary list -- e.g.
(MATH RULE ForwardOnly).
It would be easy to state the part-whole dependencies -- that the
($A $*B $*C) theory-part is contained within the ($A $*B).
This notation also simplifies our (well, my) wish to de-activate all
theories whose name matches ($*A ForwaredOnly).
Amused? Realize this gives a fairly well-founded justification to GSN's work.
There still are some problems (why, otherwise would I be writing this)?
We know that (INCLUDES SET-THEORY MATH). Should this mean that
(MATH RULE) includes (SET-THEORY RULE)? That is, what is the relation
of INCLUDES to this containment/partitioning?
There are some subtle problems associated with asserting a new fact --
if the current theory is MATH, should (IF ....) be stored in MATH,
or the more relevant (MATH RULE)? How much of this type of inferences
should go on before making the assertion?
There are, of course, other problems with this approach.
Thanks for your comments, hints, corrections, etc.
Russ
----
∂TO csd.genesereth@score 15:09 26-Aug-82
Grrrr
...umble. I was here at 2:30PM, et tu? Anyway, I've
1) a few quick questions, and
2) a comprehensive outline of that paper.
When can we meet?
Russ
-----
Q1: Do you have Bill Mark's Master Thesis? (It has words like
"Reformulation" and "Expertise" in the title.)
Q2: What do you know about deBono? (Someone suggested his articles
might be relevant.)
Q3: What do you know about Byron Davies? (he's that aforementioned
"someone".)
Q4: Where is a description of that EXPERS system(s)?
∂ 3 Oct 1982 1624-PDT <CSD.GREINER at SU-SCORE> Ready for Round 1
To: CSD.GENESERETH at SU-SCORE
cc: CSD.GREINER at SU-SCORE
Mike -
The G box of the Dover Room now has YOUR VERY OWN COPY of the
preliminary draft of the soon-to-exist proto-"What's in an Analogy" paper,
printed by RDG@SAIL. There are still many holes,
(those I know are bracketed by "<<Here" and ">>",)
but it seemed time to give you a chance to lampoon it.
Should I figure on meeting this Thursday to discuss it?
----
Also, what do you know of Greismer, of IBM's Yorktown Heights
research group? (He's somehow associated with symbolic algebra and/or
the SketchPad project, or something like that.)
----
Sorry this message is so straightforward, and non-cutesy (excluding, possibly,
this current run-on sentence). I'm clearly slipping in my old age.
Russ
-------
∂TO csd.genesereth@score 13:28 1-Dec-82
Warning
Mike -
This fellow called me today (at 13:20) asking how to get
a copy of RLL. I indoctrinated him on the marvels, reputation and splendor
of MRS, told him I'd send him relevant documents ASAP, and pointed him to you.
He's a graduate student at University of Illinois at Champaigne-Urbana,
considering the use of this system for some dictionary -> lexicon scheme.
Russ.
Data:
-----
Rick Dinitz
Coordinate Science Laboratory
1101 West Springfield Ave,
Urbana, Illinois 61801
(217) 333-6446
∂25 Sep 1982 1404-PDT <CSD.GREINER at SU-SCORE> Slave Labor...
To: CSD.GENESERETH at SU-SCORE
cc: CSD.GREINER at SU-SCORE
Mike -
Steve Tepper dropped by yesterday, looking for thesis projects
with an AI bent. I mentioned that MRS was hurting for people, at all sorts
of levels; from software aspects through deep AI-ish areas, and added that
there were many projects which could use MRS as a useful tool.
Finding him intrigued, I suggested that he yak with you.
He, however, wanted to give you a neutral opportunity to hear (and,
if necessary, to reject) his offer, and so asked me to ask you.
Anyway, he seems pretty good; and I recommend grabbing him.
Let me know if you like to meet with me to discuss my evaluation...
Otherwise send him a message, to G.GREEP @ SCORE (feel free to CC me).
This situation does seem pretty similar to .... oops, that
goes in that other paper.
Russ
-------
-------
∂TO csd.genesereth@score 17:47 23-Jan-83
E/h [recall E = h*(nu) ]
Mike -
I placed a copy of the current iteration on your desk.
I'm quite happy with the propositional/predicate-calculus split --
as it allows us to present the issues in a more incremental fashion.
I still need to "put to paper" the bulk of the predicate calculus
chapter -- expanding the nucleus there. Hopefully by tomorrow...
Can you write the motivation section; and (the subset of) the conclusion
(now possible)?
I'll send off an electronic request to Bob Moore soon;
did you want to attend that session as well?
If so, what times should be considered?
Caio,
Russ
PS Hope you are over your flu/cold/aches/whatever.
∂TO csd.dietterich@score, dietterich@sumex 14:38 27-Jan-83
Newness, etc.
Tom -
I'll be depositing the "What's New" paper at SRI tomorrow (Friday)
at about 2PM. Will you be able to spare an hour (or less?) to
peruse/correct/modify the paper before then?
(As MRG is gone until Sunday, the words are, unfortunately, all mine.)
thanks,
Russ
∂TO csd.genesereth@score 15:31 2-Feb-83
MRS Meta-Queries
Mike -
Just what is the status of XMRS, especially on Diablo?
Are star variables really going away? Has the GETVAL problem gone
away (where GETVAL, assuming it is getting a non-atomic term, asks
for the CAR of its form... which means the ($GETVAL 4) blows up...)?
Who is really in charge of MRS these days?
Russ
∂TO csd.genesereth@score, freeman@diablo 14:58 14-Feb-83
MRS Grumbles & MetaGrumbles
Mike, Andy -
This MRS experience is becoming increasngly frustrating!
I of course realize there are many features not yet in place --
while annoying, this is understandable and expected, and so doesn't
particularly disturb me. What does bother me are the variety of things
which don't work.
This is confounded by the realization that the current MRS is not available;
meaning that fixing these bugs may be a total waste of time!
Which leads to particular complaints/questions:
(1) Just when will the current MRS be up on Diablo?
Is the "↑" issue the only problem?
If so, can't it be solved by just seperating the assertions from the
procedures -- placing them in a seperate (uncompiled) file?
(Mike: Andy was hesitant to do this, fearing that you would ignore
this change during your next MRS re-write. Can I assure him that this
wouldn't happen?)
That that new file should probably contain pointers back to the first
file, from which those assertions were lifted; as well as the corresponding
backpointers.
--- body of questions, related to example given below ---
(2) TOTRUEP seems broken. That is, ($totruep '(= 3 3)) calls BACKCHAIN,
rather than TRUEP-=, and then returns NIL. Something isn't working.
(See below for one theory.)
(3) How exactly are defaults handled? As best I can tell, the "default
statements" are simply universals, which happened to be placed at the end
of the searched corpus of facts.
(At least that's what the TOTRUEP situation looks like.)
Is that true? Not only is this inelegant, but it also means that subsequent
rules may not work correctly.
(4) What of modal operators --
I'd like to quickly determine if something has not been proven or not.
Should I write these "Known", "Unknown" (or THNOT) functions?
Or do they already exist?
<<<Motivating Situation>>>
It seems a shame that TRUEP knows nothing about GETVAL --
($TRUEP '(+ 2 3 5)) returns NIL,
even though ($GETVAL '(+ 2 3)) returns 5.
How can this be fixed? Desired behavior:
After typing ($TRUEP '(+ 2 3 $A)), some bottle (QB or FB, or whatever)
should search for a TRUEP method corresponding to (+ 2 3 $A), and use any such
method found. Otherwise, BEFORE FALLING TO THE backchain DEFAULT, it should
consider trying a GETVAL on the term-part of that proposition, (+ 2 3),
and unify its result with the final position, here $A.
(I had considered the meta-rule,
(IF (FUNPROP $X) (TOTRUEP $X getval-truep).
which of course did not work as desired.)
Suggestions/comments/...?
---
(5) Another problem (well, inadequacy) is with $FACTS:
As it only lists facts in the PR representation, I don't think it deserved
that "$" prefix.
(It would be nice to have a procedure which produced a list of those facts
(or at least the datum(s),) by the way.)
(6) Final grumble: MIKE, PLEASE DOCUMENT YOUR CODE!!! It's hard enough
understanding this sysem without struggling with your one-letter variable
names, embedding in your myriad of un-commented procedure. Especially if
you want me to be able to contribute to MRS's development I need some help
in decyphering what the various programs do (or are supposed to do).
'Nuff said.
****----****----****
All for now,
Russ
∂04-Mar-83 0002 CSD.GENESERETH@SU-SCORE mrs
Received: from SU-SCORE by SU-AI with PUP; 04-Mar-83 00:02 PST
Date: Thu 3 Mar 83 18:34:11-PST
Milt said he would get me the latest version by tomorrow. I'll start
looking at it then.
mrg
-------
15:50 18-Apr: genesereth@sumex/su IJCAI Paper
Mike --
Should I send this? If so, to whom?
Does bundy have an arpa-net address? ...
------
Dr Mitchell & Dr Bundy,
This message is to check on the status of our IJCAI submission,
LKA.21.RG, "What's New? A Semantic Definition of Novelty".
Is there some reason we have not been notified whether it been accepted
or not, so long after 1-April?
(If it is relevant, we did not receive the "Thank you for submitting..."
card until 25-Mar.)
It would be nice to know soon, so we can begin to make plans for this summer.
Also, if the paper was accepted, we may need some time to provide those corrections
the referees expect.
Thank you,
Russ Greiner & Michael Genesereth
∂18-Apr-83 1654 GENESERETH@SUMEX-AIM Re: IJCAI Paper
Russ, it's a good idea, but I suggest you wait another week. Then
you should probably send it to Bundy. I'll give you his net address
if you need it.
mrg
-------
∂16:18 18-Apr: genesereth@sumex/su Ignore prior message, (I think)
The reviews are finally in!
While they seemed fairly favorable,
the forms do NOT tell whether or not the paper was accepted!
Now what?
Russ
∂15:56 28-Mar-82 -- dietterich@sumex/su Jack's comment
∂18 Mar 1983 1555-PST Jack Mostow <MOSTOW@USC-ISIF> a proposition
To: csd.greiner@SU-SCORE, csd.genesereth@SU-SCORE
Russ and Michael -
I just sent in my favorable review of your IJCAI paper LKA.21.RG, "What's New?
A Semantic Definition of Novelty". In the course of reviewing it I came up
with some additional ideas that I think add quite a bit to the interesting
ideas you have formulated -- enough to warrant making me a co-author if you
decide to adopt them. This put me in a bit of a dilemma as regards my role as
reviewer -- on the one hand, I didn't want to make adoption of my ideas (and
co-authorship) a condition for accepting the paper; on the other hand, I
didn't want to exploit my privileged knowledge of your paper to improve on it
without giving you an opportunity to include the improvements in your paper.
Accordingly, I sent the following description of proposed improvements to the
subfield chairman, who advised me to send in my review and then contact you
separately (as I am now doing):
---------
I believe I can replace Conjectures #2 and #3 (which comprise almost half the
paper) with a much simpler definition.
The paper seeks to formalize the notion of when a fact about an object is new
with respect to what one already knows. It defines a sentence, s, as new with
respect to a consistent and deductively closed theory, Th, if neither s nor
its negation is in Th, and names this condition N(Th,s). As the conclusion
points out, the deductive closure requirement is a serious limitation; I
prefer the definition N'(Th,s) = NOT Th |- s AND NOT Th |- ~s, where "Th |- s"
means "s is provable from Th." This definition is equivalent to the paper's
when Th is deductively closed but extends it in a natural way when it is not
-- one might interpret "|-" as provability by a particular resource-limited
inference procedure. However, the main change is yet to come.
Next the paper defines Nsyn(Th,s) = N(Th,s) AND LexInclude(A,s) and shows that
this condition is neither necessary nor sufficient for s to be a new fact
about the symbol A with respect to theory Th. It is not necessary because a
new fact about A need not mention A: if Th = {A<=>B}, then B is certainly a
new fact about A. It is not sufficient because a fact may mention A and not
be in Th and yet say nothing new about A: if Th = {A OR B}, then adding the
fact A=>B implies B but says nothing new about A.
From this point the paper proposes Conjecture #2, a model-theoretic definition
of novelty: s is a new fact about A if it restricts the range of possible
interpretations of A. It illustrates the inadequacy of this definition and
repairs it in Conjecture #3: s is a new fact about A if the set of possible
values of A consistent with some "partial interpretation" (set of assignments
to some subset of the symbols in the language) decreases when s is asserted
but stays non-empty. The statement of this definition shows off many of the
nice mathematical characters possible on a Dover, but many readers will be
intimidated by the notation. I just felt grateful for my college logic class.
In place of the model-theoretic definitions I propose a simple extension of
the definition Nsyn given in Conjecture #1: s is a new fact about A wrt
Th if together with some additional facts Q it implies some proposition P(A)
that is not implied by Q without s:
New(Th,s,A) =
(1) N'(Th,s) [s is new wrt Th]
AND (2) LexInclude(P,A) [P is some fact that mentions the symbol A]
AND (3) N'(Th,P) [P is new wrt Th]
AND (4) NOT Th |- ~Q [Q is consistent with Th]
AND (5) N'(Th U Q, P) [adding Q is not enough to prove P]
AND (6) Th U Q U {s} |- P [unless s is asserted as well]
Note that (3) is subsumed by (5) if Th |- P implies Th U Q |- P, i.e., if P is
provable from Th, it is still provable if facts Q are added. This property is
not always enjoyed by resource-limited inference procedures that bog down when
new knowledge is added.
This definition is not quite right, because it is satisfied trivially by
letting Q = {C AND s=>P}, where C is any fact new wrt Th and consistent with s
and P. That is, it says that *any* fact s is a new fact about A in that if
s=>P is asserted subsequently we will then know P. This shortcoming is fixed
by imposing a suitable limitation on the form of Q. Q corresponds to the
additional information incorporated in the paper's "partial interpretations",
which are assignments of symbols to objects in the world. If the language L
of the theory Th is purely propositional, as in the examples of the paper, we
need only require every fact in Q to be an atomic propositional symbol (or its
negation). Thus Q corresponds directly to a partial interpretation: every
symbol included in Q is mapped to T, every symbol whose negation is included
in Q is mapped to F, and all other symbols whose truth value is not implied by
Th U Q U s has a "don't-care" interpretation. If n-ary predicate and function
symbols (and constants) are allowed, things get more complex. I think the
solution is to add the condition
(7) Every fact in Q (or its negation) must have one of the following forms:
d=<exp>, where d is a constant symbol in L
C(<exp>, ...), where C is an n-ary predicate symbol in L
f(<exp>, ...)=<exp>, where f is a function symbol in L
I suspect that definition works, at least for finite languages, if the
expressions <exp> are constrained to contain only constants and function
applications, i.e., no quantifiers. Some problems may arise in the case of
infinite universes, but in any event the only examples discussed in the paper
are finite propositional languages, and for those Q can be constructed from
the assignments of the partial interpretation in the paper's definition.
In fact, I came up with my definition in the course of trying to explain the
paper's example to myself in English: "if A<=>B and C are known (and B is
not), then adding A<=>C means that if C is ever asserted, A will be known."
Thus for Th = {A<=>B}, s = "A<=>C", my definition of New is satisfied by
letting Q = {C}, P = "A", since one can prove A from {A<=>B, C, A<=>C} but not
from {A<=>B, C}. As this formulation of the example shows, A<=>B is a red
herring, since one can prove A from {C, A<=>C}. Of course, one can also
satisfy the definition by letting Q = {B}, in which case one does use A<=>B.
My definition corresponds to the "assertional novelty" mentioned at the end of
the paper: asserting Q U {s} need not restrict the referent of A, only imply
some fact P(A) not previously known. I think you can get the former
definition by adding a constraint on the form of P:
if A is a constant, require P (or ~P) to be of the form A=<exp>
if A is a predicate, require P (or ~P) to be A(<exp>, ...)
if A is a function, require P (or ~P) to be A(<exp>...)=<exp>
Another problem with the definition given in the paper is its computability,
i.e., given a theory, can you effectively use the definition to tell whether s
is a new fact about A? If not, the definition is not terribly useful. The
paper fails to address this important issue, and should be augmented to at
least mention it. The definition it uses appears to require enumerating the
partial interpretations of the language. This is effectively defined for
finite propositional languages, for which one can set up tableaus like those
in the paper, but is not effective for more complex cases, e.g., theories
whose only models are infinite, or where the referent of an expression may not
be a named constant in the language.
My syntactic definition suggests a practical approach to the problem of
deciding whether s is a new fact about A:
Try the following procedure for both P and ~P, where:
If A is a propositional symbol, P is "A".
If A is a predicate symbol, P is "A(x1, ..., xn)" for variables x1,...,xn.
If A is a function symbol, P is "A(x1, ..., xn)=y" for variables x1,...,xn,y.
1. Construct the implication I = "Th AND s => P".
2. Use simplification to reduce the implication as much as possible.
3. Let Q be a consistent set of irreducible propositions after simplifying,
including assertions of the form <var>=<exp>, so that Th AND s AND Q |- P.
4. Try to prove Th AND Q => P.
If the procedure fails for both P and ~P, for all sets Q (there may be more
than one way to reduce I), conclude that s is a new fact about A.
The correctness of this procedure depends on certain properties of the
simplification algorithm, but it appears to be a more practical method for
realistic problems than tableau enumeration. In any event, my definition,
which is existentially quantified over fact sets Q, seems no more ineffective
in the general (non-propositional) case than the one in the paper, which is
existentially quantified over partial interpretations.
-------
So here's the deal: if you want to incorporate the above in your paper and
make me a co-author, fine. Otherwise I plan to write it up separately as a
follow-up to your paper, in which case I will try to find a suitably clever
title, e.g., "So What Else is New: A Syntactic Definition of Novelty".
Let me know which you prefer. Regards. - Jack
-------
∂22 Mar 1983 1404-PST Jack Mostow <MOSTOW@USC-ISIF> p.s.
To: csd.greiner@SU-SCORE
Russ - Please send me a copy of the extended HPP tech report you cited in
your IJCAI paper. Have you discussed my proposition with Mike yet? - Jack
-------
∂3 May 1983 1448-PDT Jack Mostow <MOSTOW@USC-ISIF> Please acknowledge
To: csd.greiner@SU-SCORE, csd.genesereth@SU-SCORE
Gentlemen - I never heard from you about your paper. Was it accepted? What
gives? - Jack
-------
-------
∂17:23 3-May: genesereth@sumex/su Response to Jack?
Mike -
Tell me what you think of the following:
<MOSTOW@USC-ISIF, genesereth@sumex/cc/su Response>
Jack -
<<changed>>
∂03-May-83 2221 GENESERETH@SUMEX-AIM Re: Response to Jack?
It's pretty good, but you might try to make thetone a little
friendlier and less defensive. I like the comment about how
the longer version is still open.
mrg
-------
∂18:04 5-May: MOSTOW@USC-ISIF, genesereth@sumex/cc/su Response
Jack -
Greetings! Thank you for your positive recommendation -- the paper
was indeed accepted. As both reviewers noted, the paper sorely missed its
future-work section, which we (unfortunately) cut to meet space
requirements. Here we had stated our goal of mechanizing this New
relation, which clearly begins by finding a syntactic way of expressing
this semantic criterion. Indeed, our initial formulation was fairly
similar to yours; but was seriously questioned when we could not prove it
would scale up to the full predicate calculus case -- expressability in PC
is a non-trivial issue.
We therefore decided not to incorporate these still-unresolved issues
in the submitted paper. There is still the longer report. When that
surfaces we will assuredly contact you. (This has yet to leave the
starting blocks: After submitting the paper, both Mike and I pushed this
novelty stuff way down on our priority lists -- and have done little since
then beyond collecting and recording our various thoughts.)
You are, of course, free to write up any ideas you wish, as you will
certainly acknowledge the source of your interest, and of the ideas, as
appropriate.
By the way, when we got your first message 1-April seemed so close
that we decided to wait until then, until we heard from IJCAI, before
responding. Needless to say there were several subsequent delays as well.
Sorry you were left in limbo all the while.
Russ and Mike
∂05-May-83 2309 GENESERETH@SUMEX-AIM Re: Response
To: RDG@SAIL
Russ,
That letter is a paragon of tact. Well done.
mrg
-------
∂17:02 3-May: jdm/su wrt Note from MRG
Jock -
Do you have this note?
Russ
------
∂23 Apr 83 22:56:08-PST <GENESERETH@SUMEX-AIM> jock
To: greiner
Russ,
Jock and I have been talking about mappings between presentations
of information that very closely resemble what we've been calling
analogy. You might want to chat with him about it. I'll send you
a recent note from me to him on the subject that illustrates one
such mapping.
mrg
-------
∂03-May-83 2020 jock@Diablo Here is that note.
------- Forwarded Message
Received: from SUMEX-AIM by Diablo with PUP; Sat, 23 Apr 83 15:02 PST
Date: Sat 23 Apr 83 14:56:49-PST
From: Michael Genesereth <GENESERETH@SUMEX-AIM.ARPA>
Subject: thesis thoughts
To: mackinlay@SUMEX-AIM.ARPA
cc: genesereth@SUMEX-AIM.ARPA
Jock,
On reflection, it iccurs to me that I didn't emphasize what I consider
to be one important point about rule-based design. The standard approach
in building such a system is to write rules that allow it to decide whether
a GIVEN design meets GIVEN functional requirements (e.g. by using TRUEP).
If the rules are adequate to solve this problem, the next step is to use
RESIDUE and GIVEN functional requirements and PRODUCE a design. In your
case this translates into starting with a theory to be presented and
a possible presentation and writing rules to decide whether the candidate
is actually a presenttaion of the theory. Only after this question is
answered correctly should you begin to worry about producing the
presentation. For example, you should write rules that can determine
form the following data base whetehr theory2 is a good presentation
of theory1.
(relsymbol > theory1)
(objsymbol A theory1)
(objsymbol B theory1)
(objsymbol C theory1)
(intheory p23 theory1) where p23 is (> A B)
(intheory p24 theory1) where p24 is (> B C)
(relsymbol ARC theory2)
(objsymbol N1 theory2)
(objsymbol N2 theory2)
(objsymbol N3 theory2)
(intheory p25 theory2) where p25 is (ARC N1 N2)
(intheory p26 theory2) where p26 is (ARC N2 N3)
(F14 > ARC)
(F14 A N1)
(F14 B N2)
(F14 C N3)
??? (presentation F14 theory1 theory2) ???
i.e. is theory2 a presentation of theory1 under mapping F14
The importance of this point is that it is probably easier to
write the rules about what makes a proper presentation without
worrying about how that presentation will be discovered and only after
this is doen worrying about the design process itself. Does this
make sense to you?
mrg
-------
------- End of Forwarded Message
∂16:45 3-May: genesereth@sumex/su Quickies
Mike -
1) "WN:ASDofN" is HPP-83-26.
2) Saul Amarel told me that Mike Sims, at Rutgers, wants to use MRS
to do a Eurisko-ish exploration of Conway Numbers, etc.
He has access to Sumex -- should I just mail him the name of
the directory and have him snarf it; or does he have to pay
some $$? (No, I did not mean a variable amount...)
3) I want to talk with you a moment before adding REPN -- so I'll wait
until next week.
Russ
∂03-May-83 2218 GENESERETH@SUMEX-AIM Re: Quickies
In-Reply-To: Your message of Tue 3 May 83 16:45:00-PDT
(1) Tnx
(2) Get him to sign a non proliferation agreement first. Not necessary
for him to pay any bucks.
(3) Will talk next week.
mrg
-------
∂18:07 5-May: genesereth@sumex/su non proliferation ...
Mike -
Where is that "non proliferation agreement" form? Can you
(get Juanita to) send it MSIMS@Rutgers, or should I?
Russ
∂05-May-83 2304 GENESERETH@SUMEX-AIM Re: non proliferation ...
To: RDG@SAIL
In-Reply-To: Your message of Thu 5 May 83 18:07:00-PDT
I think there's a copy on line but I know that Juanita has hard
copies. If you send her an address, she will mail it out. Thanks.
mrg
-------
∂15-Jun-83 2234 GENESERETH@SUMEX-AIM Re: Any response to this?
Received: from SUMEX-AIM by SU-AI with PUP; 15-Jun-83 22:34 PDT
Date: Wed 15 Jun 83 22:33:57-PDT
From: Michael Genesereth <GENESERETH@SUMEX-AIM>
Subject: Re: Any response to this?
To: RDG@SAIL
In-Reply-To: Your message of Wed 15 Jun 83 20:42:00-PDT
Russ,
Thanks t
for the pointer to that squabble about maple. However, I have
nothing to reply. You might forward a copy to Fateman at Berkeley.
mrg
-------
∂14:18 30-Jun: genesereth@sumex/su Algebraic Manipulator
Mike -
What does/did Minima do? I need an algebraic problem solver
to handle even the simple problem we discussed, of two resistors
in series. (There we have 4 equations and 4 unknowns: solve for Vd in
Vd = Vc - Ic*R1
Vd = Ve + Ie*R2
Ic = Id
Id = Ie
where Vc, Ve, R1 and R2 are all known.)
If Minima doesn't exist/is unavailable/is inadequate,
can you point me to some other system?
Russ
∂12:40 1-Aug: genesereth@sumex/su Meating Thyme
Mike -
Will we be able to meet before I leave for parts-east
(this Saturday)? I've elaborated considerably on that scenario,
and want to get your opinion before beginning to implement it.
I tentatively signed up for Tuesday at 4:30PM, as all of your
usual slots were taken. That's a bad time for me too... perhaps
we could meet on Wednesday or Thursday?
Russ
∂11:55 29-Aug: genesereth@sumex/su Representation Issues
Mike:
I thought you you might be interested in this discussion-ette,
taken from the AIList BBoard, beginning with # 39.
∂#39
Date: 10 Aug 83 16:36:29-PDT (Wed)
From: harpo!floyd!vax135!cornell!uw-beaver!ssc-vax!sts @ Ucb-Vax
Subject: A Real AI Topic
Article-I.D.: ssc-vax.398
First let me get in a one last (?) remark about where the Japanese are
in AI - pattern recognition and robotics are useful but marginal in
the AI world. Some of the pattern recognition work seems to be making
the same conclusions now that real AI workers made ten years ago
(those who don't know history are doomed to repeat it!).
Now on to the good stuff. I have been thinking about knowledge
representation (KR) recently and made some interesting (to me, anyway)
observations.
1. Certain KRs tend to show up again and again, though perhaps in
well-disguised forms.
2. All the existing KRs can be cast into something like an
attribute-value representation.
Space does not permit going into all the details, but as an example,
the PHRAN language analyzer from Berkeley is actually a specialized
production rule system, although its origins were elsewhere (in
parsers using demons). Semantic nets are considered obsolete and ad
hoc, but predicate logic reps end up looking an awful lot like a net
(so does a sizeable frame system). A production rule has two
attributes: the condition and the action. Object-oriented programming
(smalltalk and flavors) uses the concept of attributes (instance
variables) attached to objects. There are other examples.
Question: is there something fundamentally important and inescapable
about attribute-value pairs attached to symbols? (ordinary program
code is a representation of knowledge, but doesn't look like av-pairs
- is it a valid counterexample?)
What other possible KRs are there?
Certain KRs (such as RLL (which is really a very interesting system))
claim to be universal and capable of representing anything. Are there
any particularly difficult concepts that *no* KR has been able to
represent (even in a crude way)? What is so difficult about those
concepts, if any such exist?
Just stirring up the mud,
stan the leprechaun hacker
ssc-vax!sts (soon utah-cs)
[I believe that planning systems still have difficulties in
representing continuous time, hypothetical worlds, beliefs, and
intentions, among other things. In vision, robotics, geology, and
medicine, there are difficulties in representing shape, texture, and
spatial relationships. Attribute-value pairs are just not very
useful for representing continuous quantities. -- KIL]
------------------------------
∂#40
Date: Friday, 12 Aug 1983 15:28-PDT
From: narain@rand-unix
Subject: Reply to stan the leprechaun hacker
I am responding to two of the points you raised.
Attribute value pairs are hopeless for any area (including AI areas)
where your "cognitive chunks" are complex structures (like trees). An
example is symbolic algebraic manipulation, where it is natural to
think in terms of general forms of algebaraic expressions. Try writing
a symbolic differentiation program in terms of attribute-value pairs.
Another example is the "logic grammars" for natural language, whose
implementation in Prolog is extremely clear and efficient.
As to whether FP or more generally applicative languages are useful to
AI depends upon the point of view you take of AI. A useful view is to
consider it as "advanced programming" where you wish to develop
intelligent computer programs, and so develop powerful computational
methods for them, even if humans do not use those methods. From this
point of view Backus's comments about the "von Neumann bottleneck"
apply equally to AI programming as they do to conventional
programming. Hence applicative languages may have ideas that could
solve the "software crisis" in AI as well.
This is not just surmise; the Prolog applications to date and underway
are evidence in favor of the power of applicative languages. You may
debate about the "applicativeness" of practical Prolog programming,
but in my opinion the best and (also the most efficient) Prolog
programs are in essence "applicative".
-- Sanjai Narain
------------------------------
∂#42 ?
Date: 12 Aug 1983 1208-PDT
From: FC01@USC-ECL
Subject: Knowledge Representation, Fifth Generation
About knowledge representation---
Although many are new to this ballgame, the fundamentals of
the field are well established. Look in the dictionary of information
science a few years back (5-10?) for an article on the representation
of knowledge by Irwin Marin. The (M,R) pair mentioned is indeed a
general structure for representation. In fact, you may recal 10 or 20
years ago there was talk that the most efficient programs on computers
would eventually consist of many many pointers (Rs) that pointed
between datums (Ms) in may different ways - kinda like the brain!!! It
has gone well beyond the (M,R) pair stage and Marin has developed a
structure for representation that allows top down knowledge
engineering to proceed in a systematic fashion. I guess many of us
forsake history in many ways, both social and technical.
...
Russ
∂11:59 29-Aug: genesereth@sumex/su NP-completeness and parallelism, humor
Mike -
Recall your "deep" insight, with Mark Stefik, about "constant
time sorts"? Here it is, correctly labelled as humor:
(Again from the AIList BB.)
Russ
----
Date: 16 Aug 1983 2016-MDT
From: William Galway <Galway@UTAH-20>
Subject: NP-completeness and parallelism, humor
Perhaps AI-digest readers will be amused by the following
article. I believe it's by Danny Cohen, and appears in the
proceedings of the CMU Conference on VLSI Systems and
Computations, pages 124-125, but this copy was dug out of local
archives.
..................................................................
The VLSI Approach to
Computational Complexity
Professor J. Finnegan
University of Oceanview, Kansas
(Formerly with the DP department of the
First National Bank of Oceanview)]
The rapid advance of VLSI and the trend toward the decrease of
the geometrical feature size, through the submicron and the
subnano to the subpico, and beyond, have dramatically reduced the
cost of VLSI circuitry. As a result, many traditionally
unsolvable problems can now (or will in the near future) be
easily implemented using VLSI technology.
For example, consider the traveling salesman problem, where the
optimal sequence of N nodes ("cities") has to be found. Instead
of applying sophisticated mathematical tools that require
investment in human thinking, which because of the rising cost of
labor is economically unattractive, VLSI technology can be
applied to construct a simple machine that will solve the
problem!
The traveling salesman problem is considered difficult because of
the requirement of finding the best route out of N! possible
ones. A conventional single processor would require O(N!) time,
but with clever use of VLSI technology this problem can easily be
solved in polynomial time!!
The solution is obtained with a simple VLSI array having only N!
processors. Each processor is dedicated to a single possible
route that corresponds to a certain permutation of the set
[1,2,3,..N]. The time to load the distance matrix and to select
the shortest route(s) is only polynomial in N. Since the
evaluation of each route is linear in N, the entire system
solves the problem in just polynomial time! Q.E.D.
Readers familiar only with conventional computer architecture may
wrongly suspect that the communication between all of these
processors is too expensive (in area). However, with the use of
wireless communication this problem is easily solved without the
traditional, conventional area penalty. If the system fails to
obtain from the FCC the required permit to operate in a
reasonable domain of the frequency spectrum, it is always
possible to use microlasers and picolasers for communicating
either through a light-conducting substrate (e.g. sapphire) or
through a convex light-reflecting surface mounted parallel to the
device. The CSMA/CD (Carrier Sense Multiple Access, with
Collision Detection) communication technology, developed in the
early seventies, may be found to be most helpful for these
applications.
If it is necessary to solve a problem with a larger N than the
one for which the system was initially designed, one can simply
design another system for that particular value of N, or even a
larger one, in anticipation of future requirements. The
advancement of VLSI technology makes this iterative process
feasible and attractive.
This approach is not new. In the early eighties many researchers
discovered the possibility of accelerating the solution of many
NP-complete problems by a simple application of systems with an
exponential number of processors.
Even earlier, in the late seventies many scientists discovered
that problems with polynomial complexity could also be solved in
lower time (than the complexity) by using number of processors
which is also a polynomial function of the problem size,
typically of a lower degree. NxN matrix multiplication by
systems with N↑2 processors used to be a very popular topic for
conversations and conference papers, even though less popular
among system builders. The requirement of dealing the variable N
was (we believe) handled by the simple P/O technique, namely,
buying a new system for any other value of N, whenever needed.
According to the most popular model of those days, the cost of
VLSI processors decreases exponentially. Hence the application
of an exponential number of processors does not cause any cost
increase, and the application of only a polynomial number of
processors results in a substantial cost saving!! The fact that
the former exponential decrease refers to calendar time and the
latter to problem size probably has no bearing on this discussion
and should be ignored.
The famous Moore model of exponential cost decrease was based on
plotting the time trend (as has been observed in the past) on
semilogarithmic scale. For that reason this model failed to
predict the present as seen today. Had the same observations
been plotted on a simple linear scale, it would be obvious that
the cost of VLSI processors is already (or about to be) negative.
This must be the case, or else there is no way to explain why so
many researchers design systems with an exponential number of
processors and compete for solving the same problem with more
processors.
CONCLUSIONS
- With the rapid advances of VLSI technology anything is
possible.
- The more VLSI processors in a system, the better the paper.
------------------------------
∂12:02 29-Aug: genesereth@sumex/su NETL semantics
Final AIList msg of relevance.
(That issue, #44, contains other comments on this theme as well --
from Scott, among others.)
---
Date: 18 Aug 83 0516 EDT
From: Dave.Touretzky@CMU-CS-A
Subject: NETL
I am a graduate student of Scott Fahlman's, and I've been working on
NETL for the last five years. There are some interesting lessons to
be learned from the history of the NETL project. NETL was a
combination of a parallel computer architecture, called a parallel
marker propagation machine, and a representation language that
appeared to fit well on this architecture. There will probably never
be a hardware implementation of the NETL Machine, although it is
certainly feasible. Here's why...
The first problem with NETL is its radical semantics: no one
completely understands their implications. We (Scott Fahlman, Walter
van Roggen, and I) wrote a paper in IJCAI-81 describing the problems
we had figuring out how exceptions should interact with multiple
inheritance in the IS-A hierarchy and why the original NETL system
handled exceptions incorrectly. We offered a solution in our paper,
but the solution turned out to be wrong. When you consider that NETL
contains many features besides exceptions and inheritance, e.g.
contexts, roles, propositional statements, quantifiers, and so on, and
all of these features can interact (!!), so that a role (a "slot" in
frame lingo) may only exist within certain contexts, and have
exceptions to its existence (not its value, which is another matter)
in certain sub-contexts, and may be mapped multiple times because of
the multiple inheritance feature, it becomes clear just how
complicated the semantics of NETL really is. KLONE is in a similar
position, although its semantics are less radical than NETL's.
Fahlman's book contains many simple examples of network notation
coupled with appeals to the reader's intuition; what it doesn't
contain is a precise mathematical definition of the meaning of a NETL
network because no such definition existed at that time. It wasn't
even clear that a formal definition was necessary, until we began to
appreciate the complexity of the semantic problems. NETL's operators
are *very* nonstandard; NETL is the best evidence I know of that
semantic networks need not be simply notational variants of logic,
even modal or nonmonotonic logics.
In my thesis (forthcoming) I develop a formal semantics for multiple
inheritance with exceptions in semantic network languages such as
NETL. This brings us to the second problem. If we choose a
reasonable formal semantics for inheritance, then inheritance cannot
be computed on a marker propagation machine, because we need to pass
around more information than is possible on such a limited
architecture. The algorithms that were supposed to implement NETL on
a marker propagation machine were wrong: they suffered from race
conditions and other nasty behavior when run on nontrivial networks.
There is a solution called "conditioning" in which the network is
pre-processed on a serial machine by adding enough extra links to
ensure that the marker propagation algorithms always produce correct
results. But the need for serial preprocessing removes much of the
attractiveness of the parallel architecture.
I think the NETL language design stands on its own as a major
contribution to knowledge representation. It raises fascinating
semantic problems, most of which remain to be solved. The marker
propagation part doesn't look too promising, though. Systems with
NETL-like semantics will almost certainly be built in the future, but
I predict they will be built on top of different parallel
architectures.
-- Dave Touretzky
∂15:36 29-Aug: genesereth@sumex/su Final Thoughts on Novelty
Mike -
Before burying the newness ideas, I wanted to chronicle
both (interesting) comments I received at IJCAI:
(1) Sri mentioned that the novelty-aboutness connection seems almost
right, but not quite: (at least) for the indexing application, it is
appropriate (essential?) to consider the GOAL associated with the
(intended) retrieval of the fact.
That makes sense to me.
In fact, one may want to consider novelty, in general, with respect
to some purpose or goal. But maybe this should be the Newness-1 relation,
which takes this 4th argument...
(2) Some logician mentioned a few papers, to consider readings:
Nelson Goodman wrote an article titled (ready for this): "About About",
circa 1950.
Hintikka's "Semantics of Questions" may be relevant as well
(but given my estimation of Jaako, I doubt it.)
-----
Anyway, I've no interest in pursuing novelty, in and of itself, any
longer; have you? If not, perhaps I should just send this message to
Dr Mostow, giving him our blessings and permission to carry on...
Comments?
Russ
∂13:33 2-Sept: genesereth@sumex/su Your SAIL stuff
.fing mrg
Last logout
MRG UNKNOWN 13:27 on 5 Aug 1983.
No new mail exists, last written at 13:26 on 5 Aug 1983. No plan.
↑C
.mail genesereth@sumex @∂mrg/su Your SAIL Mail
genesereth@SUMEX-AIM -- (36.45.0.87) -- ? Message too big to be queued!!
queued
Exit
↑C
.er ∂mrg
Invalid or undesired directory kept as part of /R text.
;;** MAIL Command Executed OK, some mail Queued ** MAIL genesereth@sumex/su SAIL Mai
l #1 **
;;** MAIL Command Executed OK, some mail Queued ** MAIL genesereth@sumex/su SAIL Mai
l #2 **
Bye
↑C
.del ∂mrg
Deleted: MRG MSG 2,2
Total space reclaimed = 6 Blk 5.3 Blk 87% Free blocks = 221
Exit
↑C
.ds mrg
DSKSIZ: Disk utilization 13:30 2-Sep-83
MRG is not an authorized user.
There are 2 UFD's
[1,MRG] 59 Files. 61 Blk. 51369 DW. 36% 23-Jun-82
[MUS,MRG] 0 Files. 0 Blk. 0 DW. 0%
[*,MRG] 59 Files. 61 Blk. 51369 DW. 36%
** unknown user MRG is using 61 blks
Monthly cost of this disk usage would be: $30.50
Exit
↑C
.r charge;mrg
Month ? (type <return> for September) aug
SAIL charges for MRG during August 1983:
A job time: .05 hours $ .09
B job time: .00 .00
C job time: .00 .00
A cpu time: .08 minutes .13
B cpu time: .00 .00
C cpu time: .00 .00
Disk usage: 67.00 blocks 33.50
Total cost: $ 33.72
Exit
↑C
.r ppsav
∂13:36 2-Sep: genesereth@sumex/su Your SAIL Files
All on [1,mrg]:
Filnam Ext PPN Blks Written Time Pro Writer Reference--% Dumped Off
MRSRES LSP 1MRG 0.2 20-Jan-82 1242 000 1MRG FTP 22-Jan-82 00 P2192>
INFERE LSP 1MRG 0.7 20-Jan-82 1242 000 1MRG FTP 22-Jan-82 00 P2192>
MRSREP LSP 1MRG 0.7 20-Jan-82 1243 000 1MRG FTP 22-Jan-82 00 P2192>
INFERE L 1MRG 0.7 22-Jan-82 2030 000 1MRG PUPFTP 22-Jan-82 00 P2196>
RESIDU L 1MRG 0.1 22-Jan-82 2030 000 1MRG PUPFTP 22-Jan-82 00 P2196>
PROPRE L 1MRG 0.7 22-Jan-82 2030 000 1MRG PUPFTP 22-Jan-82 00 P2196>
NEWMRS L 1MRG 0.2 22-Jan-82 2030 000 1MRG PUPFTP 22-Jan-82 00 P2196>
NEWMRS LOA 1MRG 0.1 22-Jan-82 2031 000 1MRG PUPFTP 22-Jan-82 00 P2196>
KEW PRO 1MRG 0.2 23-Feb-82 2125 000 1MRG PUPFTP 23-Feb-82 00 P2214>
FAIRCH PRO 1MRG 0.6 26-Feb-82 1134 000 NETSYS FTPSER 26-Feb-82 00 P2220>
MRS REV 1MRG 0.6 24-Mar-82 1300 000 1MRG PUPFTP 24-Mar-82 00 P2232>
IA OVE 1MRG 1.0 24-Mar-82 1301 000 1MRG PUPFTP 24-Mar-82 00 P2232>
MECHAN BRA 1MRG 0.7 24-Mar-82 1301 000 1MRG PUPFTP 24-Mar-82 00 P2232>
KEW BRA 1MRG 2.8 23-Jun-82 1942 000 NETSYS FTPSER 23-Jun-82 00 P2287>
MRSREP L 1MRG 0.8 02-Oct-81 2116 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MRSS L 1MRG 0.5 02-Oct-81 2116 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MRSMET L 1MRG 0.2 02-Oct-81 2117 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MRSWHY L 1MRG 0.4 02-Oct-81 2117 000 1MRG PUPFTP 02-Oct-81 00 P2118>
CS223 CHA 1MRG 0.2 02-Oct-81 1433 000 1MRG PUPFTP 24-Jan-83 00 P2118>
CS223 LIS 1MRG 0.2 02-Oct-81 1433 000 1MRG PUPFTP 24-Jan-83 00 P2118>
CS223 NOT 1MRG 1.0 02-Oct-81 1433 000 1MRG PUPFTP 24-Jan-83 00 P2118>
CS223 REP 1MRG 1.0 02-Oct-81 1434 000 1MRG PUPFTP 02-Oct-81 00 P2118>
CS223 RES 1MRG 0.8 02-Oct-81 1434 000 1MRG PUPFTP 02-Oct-81 00 P2118>
CS223 SEA 1MRG 0.2 02-Oct-81 1434 000 1MRG PUPFTP 02-Oct-81 00 P2118>
CS223 TOP 1MRG 0.5 02-Oct-81 1434 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MRSBF L 1MRG 0.2 02-Oct-81 2042 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MACROS L 1MRG 1.0 02-Oct-81 2115 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MRSARI L 1MRG 0.3 02-Oct-81 2117 000 1MRG PUPFTP 02-Oct-81 00 P2118>
MRSTRU L 1MRG 0.1 02-Oct-81 2117 000 1MRG PUPFTP 02-Oct-81 00 P2118>
SAGE LSP 1MRG 0.4 08-Oct-81 1120 000 NETSYS FTPSER 22-Jan-82 00 P2125>
SAGE TES 1MRG 0.1 08-Oct-81 1120 000 NETSYS FTPSER 08-Oct-81 00 P2125>
DEMO LSP 1MRG 0.1 08-Oct-81 1306 000 1MRG FTP 22-Jan-82 00 P2125>
MRS DEM 1MRG 0.2 08-Oct-81 1307 000 1MRG FTP 22-Jan-82 00 P2125>
SAGE L 1MRG 0.4 08-Oct-81 1308 000 1MRG PUPFTP 08-Oct-81 00 P2125>
SIMP LSP 1MRG 0.4 10-Oct-81 2211 000 1MRG PUPFTP 11-Oct-81 00 P2125>
RAT LSP 1MRG 0.1 13-Oct-81 2003 000 1MRG PUPFTP 13-Oct-81 00 P2125>
CS223 SYL 1MRG 0.4 18-Oct-81 1955 000 NETSYS FTPSER 18-Oct-81 00 P2132>
TUTORI DES 1MRG 0.5 18-Oct-81 2008 000 1MRG FTP 18-Oct-81 00 P2132>
M74 MRS 1MRG 0.4 31-Oct-81 1945 000 NETSYS FTPSER 31-Oct-81 00 P2146>
MRS PHO 1MRG 0.4 31-Oct-81 1945 000 NETSYS FTPSER 31-Oct-81 00 P2146>
FULL 1MRG 0.1 01-Dec-81 1115 000 1MRG PUPFTP 01-Dec-81 00 P2169>
GATES DDL 1MRG 0.3 31-Oct-81 2046 000 1MRG PUPFTP 31-Oct-81 00 P2146>
DDL L 1MRG 0.1 31-Oct-81 2046 000 1MRG PUPFTP 31-Oct-81 00 P2146>
SAGE SCH 1MRG 0.2 09-Nov-81 1959 000 1MRG PUPFTP 09-Nov-81 00 P2146>
NOCORL LSP 1MRG 0.1 23-Nov-81 2056 000 NETSYS FTPSER 23-Nov-81 00 P2157>
NOMRSR LSP 1MRG 0.7 23-Nov-81 2057 000 NETSYS FTPSER 23-Nov-81 00 P2157>
NOMRS LSP 1MRG 1.0 23-Nov-81 2057 000 NETSYS FTPSER 22-Jan-82 00 P2157>
NOMRSS LSP 1MRG 0.4 23-Nov-81 2057 000 NETSYS FTPSER 22-Jan-82 00 P2157>
MRSPL LSP 1MRG 0.2 23-Nov-81 2057 000 NETSYS FTPSER 22-Jan-82 00 P2157>
NOMRSM LSP 1MRG 0.2 23-Nov-81 2057 000 NETSYS FTPSER 22-Jan-82 00 P2157>
NOMRSL LSP 1MRG 0.6 23-Nov-81 2058 000 NETSYS FTPSER 23-Nov-81 00 P2157>
NOMRSA LSP 1MRG 0.3 23-Nov-81 2058 000 NETSYS FTPSER 23-Nov-81 00 P2157>
NOMRST LSP 1MRG 0.1 23-Nov-81 2058 000 NETSYS FTPSER 23-Nov-81 00 P2157>
NOMRSW LSP 1MRG 0.5 23-Nov-81 2058 000 NETSYS FTPSER 23-Nov-81 00 P2157>
DART LSP 1MRG 0.1 30-Nov-81 2015 000 1MRG FTP 27-Sep-82 00 P2169>
DART L 1MRG 0.3 01-Dec-81 1114 000 1MRG PUPFTP 27-Sep-82 00 P2169>
2X4 1MRG 0.1 01-Dec-81 1115 000 1MRG PUPFTP 24-Jan-83 00 P2169>
MANO L 1MRG 0.1 01-Dec-81 1115 000 1MRG PUPFTP 01-Dec-81 00 P2169>
DART DEM 1MRG 0.1 01-Dec-81 1115 000 1MRG PUPFTP 27-Sep-82 00 P2169>
Total=61 Blk
∂10:54 1-Sep: genesereth@sumex/su Meeting Time
Mike -
Will you be here Friday AM? Could we meet then, or at noon,
rather than 3:30PM? [My brother, (genere)Seth will be here, and wants
my company about then. Such is the price of fame...]
Russ
∂03-Aug-83 1132 MULLEN@SUMEX-AIM check for proceedings
Received: from SUMEX-AIM by SU-AI with PUP; 03-Aug-83 11:32 PDT
Date: Wed 3 Aug 83 11:33:45-PDT
From: Juanita Mullen <MULLEN@SUMEX-AIM>
Subject: check for proceedings
To: rdg@SAIL
I asked Tom for an account number to charge for your request. He asked
Doug about it and Doug said he has copies of those proceedings in his
office (or at least he owns some). Tom thought perhaps you should check
this out.
-------
∂03-Aug-83 1250 MULLEN@SUMEX-AIM proceedings
Received: from SUMEX-AIM by SU-AI with PUP; 03-Aug-83 12:49 PDT
Date: Wed 3 Aug 83 12:48:18-PDT
From: Juanita Mullen <MULLEN@SUMEX-AIM>
Subject: proceedings
To: rdg@SAIL
Now the problem seems to be why do you need two copies??
-------
This should probably be on Mike's account -- he was the one who
suggested
∂13-Sep-83 1217 MULLEN@SUMEX-AIM proceedings
To: rdg@SAIL
The two copies of the Machine Learning workshop Proceedings have arrived.
Come and get'em.
-------
∂15:19 15-Sept: genesereth@sumex/su Another week, another problem
Mike -
Once again some one wants my body (well, my car's body)
at about 4:30PM this Friday. Sooo, could we move the meeting time?
Would noon be ok?
(Then you could meet with that-fellow-who-called at my 3:30 slot.)
Clearly 3:30PM Friday sucks! Could we plan to meet weekly (weAkly)
at noon, until SigLunch returns?
Thanks,
Russ
If I don't hear from you soon I'll give a call.
∂18-Sep-83 1220 greiner@Diablo SUKAY
To: genesereth@sumex
Cc: rdg@sail
Message-Id: <83/09/18 1218.733@Diablo>
Mike -
The Andes group 'Sukay' will be performing on Saturday, 1-Oct,
at "In Toto". Let me know if you want more information.
Russ
∂19-Sep-83 1057 GREINER@SUMEX-AIM [norvig%ucbkim@Berkeley (Peter Norvig): OPS5 tutor (or just doc)]
To: genesereth@SUMEX-AIM
cc: greiner@SUMEX-AIM, rdg@SAIL, norvig%ucbkin@UCB-VAX.ARPA
Mike -
Know anyone around here who knows OPS5 [?6?] -- and who want $$?
Russ
---------------
Return-Path: norvig%ucbkim@Berkeley
Received: from ucbvax.ARPA (UCB-VAX.ARPA) by SUMEX-AIM.ARPA with TCP; Sun 18 Sep 83 18:46:45-PDT
Received: from ucbkim.ARPA by ucbvax.ARPA (4.12/4.7)
id AA03412; Sun, 18 Sep 83 18:45:13 PDT
Received: by ucbkim.ARPA (4.6/4.2)
id AA07757; Sun, 18 Sep 83 18:44:34 PDT
Date: Sun, 18 Sep 83 18:44:34 PDT
From: norvig%ucbkim@Berkeley (Peter Norvig)
Message-Id: <8309190144.AA07757@ucbkim.ARPA>
To: greiner@sumex-aim
Subject: OPS5 tutor (or just doc)
Russ,
I'm trying to get information on the OPS5 production system. Lockheed
(Sunnyvale) is starting an expert system group, but they haven't quite got
it together yet. Anyway, a prof. here ``volunteered'' me to help out this
woman who's trying to learn it. If you know of anyone around there who
would like to earn some money tutoring her, it would be great.
By the way, I got the documentation for MRS that you sent me. Thanks a
lot.
Peter
-------
∂20:28 22-Sept: genesereth@sumex/su Hofstader & analogy, sorta
<<from AIList #61>>
Date: Thu 22 Sep 00:23
From: Jeff Shrager
Subject: Hofstader seminar at MIT
[Reprinted from the CMU-AI bboard.]
Douglas Hofstader is giving a course this semester at MIT. I thought
that the abstract would interest some of you. The first session takes
place today.
------
"Perception, Semanticity, and Statistically Emergent Mentality"
A seminar to be given fall semester by Douglas Hofstader
In this seminar, I will present my viewpoint about the nature
of mind and the goals of AI. I will try to explain (and thereby
develop) my vision of how we perceive the essence of things, filtering
out the details and getting at their conceptual core. I call this
"deep perception", or "recognition".
We will review some earlier projects that attacked some
related problems, but primarily we will be focussing on my own
research projects, specifically: Seek-Whence (perception of sequential
patterns), Letter Spirit (perception of the style of letters), Jumbo
(reshuffling of parts to make "well-chunked" wholes), and Deep Sea
(analogical perception). These tightly related projects share a
central philosophy: that cognition (mentality) cannot be programmed
explicitly but must emerge "epiphenomenally", i.e., as a consequence
of the nondeterministic interaction of many independent "subcognitive"
pieces. Thus the overall "mentality" of such a system is not directly
programmed; rather, it EMERGES as an observable (but onnprogrammed)
phenomenon -- a statistical consequence of many tiny semi-cooperating
(and of course programmed) pieces. My projects all involve certain
notions under development, such as:
-- "activation level": a measure of the estimated relevance of a given
Platonic concept at a given time;
-- "happiness": a measure of how easy it is to accomodate a structure
and its currently accepted Platonic class to each other;
-- "nondeterministic terraced scan": a method of homing in to the best
category to which to assign something;
-- "semanticity": the measure of how abstractly rooted (intensional) a
perception is;
-- "slippability": the ease of mutability of intensional
representational structures into "semantically close" structures;
-- "system temprature": a number measuring how chaotically active the
whole system is.
This strategy for AI is permeated by probabilistic or
statistical ideas. The main idea is that things need not happen in
any fixed order; in fact, that chaos is often the best path to follow
in building up order. One puts faith in the reliability of
statistics: a sensible, coherent total behavior will emerge when there
are enouh small independent events being influenced by high-level
parameters such as temperature, activation levels, happinesses. A
challange is to develop ways such a system can watch its own
activities and use those observations ot evaluate its own progress, to
detect and pull itself out of ruts it chances to fall into, and to
guide itself toward a satisfying outcome.
... Prerequisits: an ability to program well, preferably in
Lisp, and an interest in philosophy of mind and artificial
intelligence.
∂12:16 26-Sep: genesereth@sumex/su Many misc things (some timely)
Mike -
First, when should we plan to meet? (Having still not heard
from DBL, I assume we will NOT meet, en masse, on Tuesday; yes?)
Second, I read a few of the articles in ML83 book. I strongly
recommend perusing Burnstein's article, (p 19-25), and also
consider Carbonell's paper (p12-18) worth reading. [I never thought
I'd be recommending two Yale-derived papers! Oye!]
Third, what are your plans for this Saturday? Some friends and I
plan to meet for dinner, and then hear Sukay. Like to join us?
Russ
∂14:56 26-Sep: mullen@sumex, genesereth@sumex/cc/su Mtg w/MRG
Juanita -
I'd like to schedule a meeting with Mike for 11 AM this Friday.
(As I understood it) Vineet and Jeff R. will meet with MRG in
some other slot.
Russ
∂Thu 29 Sep 83 21:58:46-PDT <GENESERETH@SUMEX-AIM.ARPA> meeting
To: greiner@SUMEX-AIM.ARPA
Russ,
I am sorry to have to ask you to let me abbreviate our meeting
tomorrow morning. My schedule is full and I got a reqquest to
spend a hlaf hour with a vistor from Europe who can't make it at
any other time. I'll try to compress my other meetings to minimize
the impact, but I can't guarantee that I will be free before 11:30.
My apologies for the late notice, but I just found out today.
mrg
-------
∂Thu 29 Sep 83 22:25:00-PDT <GREINER@SUMEX-AIM.ARPA> Re: meeting
To: GENESERETH@SUMEX-AIM.ARPA
cc: GREINER@SUMEX-AIM.ARPA
In-Reply-To: Your message of Thu 29 Sep 83 21:58:48-PDT
Grrrr... there are many things I wanted to discuss, and was already
worried about how to get to Ed's SigLunch in time to get a chair...
oh well. I'll try to mail some of my recent thoughts about abstractions --
using groups as a show case. I've also assembled a jumbled collection
of thesis ideas, following your "put together some slides" suggestion.
I'll forward that as well.
Russ
-------
-------
∂13:48 14-Nov-83: genesereth@sumex/su ... and a time for everything ...
Mike -
I tentatively pencilled myself in for Friday morning at 10 --
unless you'd rather meet at some other time, e.g. some (nonMRS) time
on Thursday.
[Actually, I guess I DEFINITIVELY made a tentative appointment...]
Russ
∂20:35 20-Feb-84: genesereth@sumex/su Is "undecidable" better than "independent"?
....
Date: Wed, 15 Feb 84 11:15:21 EST
From: John McLean <mclean@NRL-CSS>
Subject: G. Spencer-Brown and undecidable propositions
G. Spencer-Brown is very much alive. He spent several months at NRL a couple
of years ago and presented lectures on his purported proof of the four color
theorem. Having heard him lecture on several topics previously, I did not feel
motivated to attend his lectures on the four color theorem so I can't comment
on them first hand. Those who knew him better than I believe that he is
currently at Oxford or Cambridge. By the way, he was not a friend of Russell's
as far as I know. Russell merely said something somewhat positive about LAWS
OF FORM.
With respect to undecidability, I can't figure out what Charlie Crummer means
by "undecidable proposition". The definition I have always seen is that a
proposition is undecidable with respect to a set of axioms if it is
independent, i.e,. neither the proposition nor its negation is provable.
(An undecidable theory is a different kettle of fish altogether.) Examples are
Euclid's 5th postulate with respect to the other 4, Goedel's sentence with
respect to first order number theory, the continuum hypothesis with respect to
set theory, etc. I can't figure out the claim that one can't decide whether
an undecidable proposition is decidable or not. Euclid's 5th postulate,
Goedel's sentence, and the continuum hypothesis have been proven to be
undecidable. For simple theories, such as sentential logic (i.e., no
quantifiers), there are even algorithms for detecting undecidability.
John McLean
∂15:33 22-Mar: genesereth@sumex/su Aha, a new thesis generation program!
∂22-Mar-84 1526 RDG Computer Manipulated Novel
Date: 19 Mar 84 11:36:10 CST (Mon)
From: ihnp4!houxa!homxa!rem
Subject: Computer Manipulated Novel
For people interested in computer-aided art, I manipulated a small,
unpublished novel of mine called ABRACADABRA a few years ago. The
book is a mystery derived from childhood experiences in St. Louis.
I call the manipulated book ABRACADABRA CADAVER. Chapter-by-chapter
I wrote UNIX shell programs to alter the text according to its con-
tents: for example, in an early chapter I misspelled all words as
a child might do. In another I inserted German proverbs appropriate
to my father's speech in all of his conversations. Another repeats
key phrases again and again, in a minimalist way; another puts all
dialog into footnotes; another, where the mystery unfolds, cryptically
reverses the sentences throughout--and so on. After editing the
end results, I came up with a Joycean-like book that is quite
readable and interesting as a literary document. I no longer
have it on-line, but if anyone is interested, I can provide
more details. And, of course, if anyone knows of a publisher
crazy enough.....
Bob Mueller
BELLCORE
Holmdel, NJ
∂18:06 26-Mar: genesereth@sumex/su Grief, etc
Mike -
Hope your trip has been less frustrating than the environment here has
been: Diablo has been down since 12:01AM Sunday, rendering all my files --
including the entire AAAI paper of course, inaccessible. I did some
related work on Sumex; but lost this 1/2 hour when the local tip crashed!
Moving over to another tip, I managed to work on Sumex about 20 minutes when,
of course, Sumex died!
Having been told that Diablo would be up "within the hour",
I did some other busy work for another few hours.
When I checked again, I was told there was REAL trouble, as in
"*maybe* it'll be up tomorrow (Tues), LATE EVENING -- but
realize everything you've done in the last dump will be gone." [I've put
in about ten hours a day for the last several days, almost all on this!]
I figured I'd restore my stuff from the Thursday AM tape (onto Sumex), and
edit in the difference. And now for the punchline: That tape labelled
23-Mar was really for 20-Jan! Basically this means there is no recent
backup!
This message is basically to vent my frustrations, before actually typing
in the full paper. Can you suggest whom else should see this?
-----
On another theme:
Perhaps I could ArpaNET mail you my file, as it shapes up.
I could place a NLAG.DVI file on your SUMEX directory, which you
could then print out on a Dover-ish printer there.
This will require a DVIDOV (or related) program, which Dick Gabriel
tells me that Alan Bawden [Alan@MIT-MC] was able to find.
I will also try to Federal Express you a draft of the paper, ASAP.
If it's late, you'll know why...
Damn, this is frustrating!
Russ
---
By the way, DBL tells me he too will be in MIT starting this Thursday,
again c/o Randy Davis. Small world, huh...
∂19:46 19-Apr: genesereth@sumex\su From AIList
Mike -
Should MRS be represented here?
(I didn't think so, but thought I'd call this to your attention.)
Russ
----
Date: 12 Apr 84 12:31:31-PST (Thu)
From: harpo!ulysses!allegra!carlo @ Ucb-Vax
Subject: Expert Database Systems Workshop (long msg)
Article-I.D.: allegra.2406
Call for Papers and Participation
FIRST INTERNATIONAL WORKSHOP ON EXPERT DATABASE SYSTEMS
October 25-27, 1984, Kiawah Island, South Carolina
Sponsored by
The Institute of Information Management, Technology, and Policy,
College of Business Administration,
University of South Carolina
In Cooperation With
Association for Computing Machinery - SIGMOD and SIGART
IEEE Technical Committee on Data Base Engineering
Workshop Program
This workshop will address the theoretical and practical issues
involved in making databases more knowledgeable and supportive of AI
applications. The tools and techniques of database management are
being used to represent and manage more complex types of data and
applications environments.
The rapid growth of online systems containing text, bibliographic, and
videotex databases with their specialized knowledge, and the develop-
ment of expert systems for scientific, engineering and business appli-
cations indicate the need for intelligent database interfaces and new
database system architectures.
The workshop will bring together researchers and practitioners from
academia and industry to discuss these issues in Plenary Sessions and
specialized Working Groups. The Program Committee will invite 40 to
80 people, based on submitted research and application papers (5000
words) and issue-oriented position papers (2000-3000 words).
Topics of Interest
The Program Committee invites papers addressing (but not limited to)
the following areas:
Knowledge Base Systems Knowledge Engineering
environments acquisition
architectures representation
languages design
hardware learning
Database Specification Methodologies Constraint and Rule Management
object-oriented models metadata management
temporal logic data dictionaries
enterprise models constraint specification
transactional databases verification, and enforcement
Reasoning on Large Databases Expert Database Systems
fuzzy reasoning natural language access
deductive databases domain experts
semantic query optimization database design tools
knowledge gateways
industrial applications
Please send five (5) copies of full papers or position papers by June
1, 1984 to:
Larry Kerschberg, Program Chairperson
College of Business Administration
University of South Carolina
Columbia, SC, 29208
(803) 777-7159 / (803) 777-5766 (messages)
USENET: ucbvax!allegra!usceast!kersch
CSNET: kersch@scarolina
Submissions will be considered by the Program Committee:
Bruce Berra, Syracuse University Sham Navathe, Univ. of Florida
James Bezdek, Univ. of South Carolina Erich Neuhold, Hewlett-Packard
Michael Brodie, Computer Corp. of America Stott Parker, UCLA
Janis Bubenko, Univ. of Stockholm Michael Stonebraker, UC-Berkeley
Peter Buneman, Univ. of Pennsylvania Yannis Vassiliou, New York Univ.
Antonio L. Furtado, PUC-Rio de Janeiro Adrian Walker, IBM Research Lab.
Jonathan King, Symantec Bonnie L. Webber, U. of Penn.
John L. McCarthy, Lawrence Berkeley Lab. Gio Wiederhold, Stanford Univ.
John Mylopoulos, University of Toronto Carlo Zaniolo, AT&T Bell Labs
Authors will be notified of acceptance or rejection by July 16, 1984.
Preprints of accepted papers will be available at the workshop.
Workshop presentations, discussions, and working group reports will be
published in book form.
Workshop General Chairman Local Arrangements Chairperson
Donald A. Marchand Cathie Hughes-Johnson
Institute of Information Management, Technology and Policy
(803) 777-5766
Working Group Coordinator Industrial Liaison
Sham Navathe Mas Tsuchiya
Computer and Information Sciences TRW 119/1842
University of Florida One Space Park Drive
512 Weil Hall Redondo Beach, CA 90278
Gainesville, FL 32611 (213) 217-6114
(904) 392-7442
_________________________________________________________________________
Response Card (Please mail to address on below)
Name ___________________________________________ Telephone _____________
Organization ___________________________________________________________
Address ________________________________________________________________
City, State,
ZIP, and Country ________________________________________________________
Please check all that apply:
_____ I intend to submit a research paper.
_____ I intend to submit an issue-oriented position paper.
_____ I would like to participate in a working group.
General Topic Areas _________________________________________
_____ Not sure I can participate, but please keep me informed.
Subject of paper ______________________________________________________
_______________________________________________________________________
Cathie Hughes-Johnson
Institute of Information Management
Technology and Policy
College of Business Administration
University of South Carolina
Columbia, SC 29208
------------------------------
End of AIList Digest
********************
∂13:27 30-Apr: genesereth@sumex/su Timely (AAAI)
Mike -
I would have badgered you more yesterday had
I remember that this is your week in LA... In particular,
is there anything I should do, in the immediate future?
Also, I'm really curious to get you quick summary on the
AAAI deliberations -- both in general and wrt our paper.
Russ
∂2 Apr 84 14:39:30-PST Mullen <MULLEN@SUMEX-AIM.ARPA> Appt with Mike on Tues 2:30
To: greiner@SUMEX-AIM.ARPA, treitel@SUMEX-AIM.ARPA, de2smith@SUMEX-AIM.ARPA
Mike forwarded a msg to me regarding tomorrow afternoons Faculty mtg
at 2:30..without actually coming out and saying it, I am presuming
he will not be here from 2:30 on, so you will not be meeting with
Mike...however, if I am wrong, I will immediately let you know.,
-------
-------
∂Fri 18 May 84 13:47:30-PDT <GREINER@SUMEX-AIM.ARPA> <MULLEN@SUMEX-AIM.ARPA>: your appts with Mike]
To: genesERETH@SUMEX-AIM.ARPA
cc: greiner@SUMEX-AIM.ARPA
Booo!!! You already eliminated 1/2 of my other scheduled meeting with
you for this month! and I still have list of things to discuss!
Can we reschedule?
Impatiently,
Russ
---------------
Mail-From: MULLEN created at 18-May-84 13:30:22
Date: Fri 18 May 84 13:30:22-PDT
From: Juanita Mullen <MULLEN@SUMEX-AIM.ARPA>
Subject: your appts with Mike
To: grosof@SU-SCORE.ARPA, greiner@SUMEX-AIM.ARPA
Your appts with Mike for May 22, have been cancelled..something has come up.
-------
∂19 Jun 84 17:59:02-PDT <GREINER@SUMEX-AIM.ARPA> Good book
To: genesERETH@SUMEX-AIM.ARPA
cc: greiner@SUMEX-AIM.ARPA
Mike -
I was just in the Math/Science library, where I saw an intriguing
book: Some one has assembled a compendium(sp) of all (well, many) ESs
around, as of early 1984 into a single book. The appendices include
a list of the companies actively working in this area, as well as a break-
down of application domains, etc. (The former was why I thought of
mentioning this to you.) Anyway, if you're interested, contact
Harry Llull.
Russ
∂to MRG 15-Aug
Date: 13 Aug 84 14:48:35-PDT (Mon)
From: ihnp4!houxm!hou2d!wbp @ Ucb-Vax.arpa
Subject: On having virtually no crime rate.
Article-I.D.: hou2d.472
"Saudi Arabia has virtually no crime rate," is what the commercial
told me about 30 times before I realized what they are really saying.
I understand what having virtually no crime is, and also a very
low crime rate is within my grasp. But virtually no crime rate is a very
odd construction.
If a place has no crime rate then this means that the statistics
are not gathered and that's O.K too.
If the crime rate is virtually non-existent then it indeed exists,
but is in a state of "almost non-being" which may mean that for all practical
purposes it does not exist, but is known to a select few who will tell
no-one. (Or may be a reflection of their different system of justice!)
Are virtual rates calculated on virtual machines, and does one
need either transcendental or imaginary numbers to express them?
Seriously, what would a program do with such a sentence?
And even more interesting, would a sophisticated program have any
problem with it, and could it not even see a problem with it as I am
sure millons of people did not see one!
Submitted for your approval,
Wayne Pineault (hou2d!wbp)
∂14:04 21-Aug: genesereth@sumex/su Good News
Round trip cost, SF -> LA -> SF, is $178 !
[No, not factorial -- that's why I left a space.]
Are you sure you don't want to go to LA next week - Wed PM -> Sat PM?
Russ
∂11:11 27-Aug: genesereth@sumex/su CSLI Library
TECHNICAL REPORTS FROM OVERSEAS
The Center's Reading Room has received a series of technical reports from
the Institute for New Generation Computer Technology of Tokyo, Japan. The
reports will become part of the Reading Room collection.
Included in the collection are ``Qute: A Prolog/Lisp Type Language for
Logic Programming'' (No. TR-016), ``Logic Programming--Past, Present and
Future'' (No. TR-015), and ``LFG in Prolog -Toward a Formal System for
Representing Grammatical Relations'' (No. TR-019). In addition to the
reports from Japan, `` Research in Knowledge Representation For Natural
Language Understanding'' and `` Plain Speaking: A Theory and Grammar
of Spontaneous Discourse'' have been received from Bolt Beranek &
Newman Inc.
A complete lsiting of these titles may be found in the <csli> directory on
Turing in file ``ReadingRoom.Reports.'' This file will be updated as the
Center receives other reports. Also in the same directory, a listing of all
published CSLI reports is available in the file ``Catalog.Reports.''
Because of practical considreations, the Lounge in Ventura Hall will serve
as a Reading Room. All publications will be transferred there early next
week after the necessary bookcases have been delivered.
Any suggestions regarding acquisitions or anything else related to the
Reading Room may be sent to the Newsletter.
∂11:14 27-Aug: jfinger@diablo, genesereth@sumex/su from AI-List -- of interest?
Date: 22 Aug 84 15:05:41-PDT (Wed)
From: ihnp4!drutx!druxx!jlmalito @ Ucb-Vax.arpa
Subject: expert system ``planner'' wanted
Article-I.D.: druxx.604
We are currently planning on building a UNIX-based expert system to
handle ``system administration'' on a computer system. My plan is to
build a knowledge base containing facts about the computer system, as
well as descriptions of possible actions and their consequences. The
expert system would be presented with a description of the current world,
and a goal. The planner (``inference engine'') will determine what
actions are needed to get from the current world to a goal state.
Due to time constraints, we are trying to find a planner that can
accomplish the task described above. If anyone knows of such a planner
(or a system containing one), please contact me. We need source,
preferably for a UNIX system. Anything close will do. We would
consider any purchase agreement and gladly accept freebies.
Quick responses would be appreciated. (Also, any ``Have you
checked...?'' would be great.)
Any responses of general interest will be posted.
thanks,
Jeanine L. Malito
{ihnp4,allegra}!druxx!jlmalito
AT&T Information Systems
Rm. 30G73
11900 N. Pecos St.
Denver CO 80234
303-538-3859
∂12:38 29-Aug: lenat@score, genesereth@sumex/su Taxonomy Assistant [from AI-List]
Date: Fri 24 Aug 84 22:26:52-EDT
From: Wayne McGuire <MDC.WAYNE%MIT-OZ@MIT-MC.ARPA>
Subject: Taxonomy Assistant
The problems of systematically representing the conceptual
relations among a set of abstract objects in any knowledge domain go
right to the heart of much leading-edge AI research. All inferencing
is based, among other things, on implicit taxonomic understanding.
It seems to me that in the knowledgebase management systems which
I hope we will see developed in the near future will be embedded rich
resources for evoking and representing taxonomies. Semantic nets
provide an ideal scheme with which to do just that.
The most useful thinking about taxonomies and classification
theory appears not in the computer science literature, but at the
interface of library science, information science, and philosophy. The
leading journal in the field is ←International Classification← (which
should be available in any world class humanities library). It is
published (as I recall) three times a year, and is chockfull of
pointers to articles, books, dissertations, etc. in the world
literature on all aspects of classification theory.
You might want to scan the following subject headings in some
recent editions of the index ←Library Literature← (published by H. W.
Wilson): Classification analysis, Subject headings, Thesauri. File 57
(Philosopher's Index) and File 61 (LISA -- Library and Information
Science Abstracts) on Dialog are also fertile sources of information
on the literature about taxonomies and classification theory. There
are many insights in the theoretical writings on classification theory
in the library science literature which could be handily transferred
to advanced AI research and systems.
It occurs to me that what we need is a ←meta-taxonomy←, that is,
a thorough inventory of all the fundamental conceptual structures by
which objects in ←any← domain can be taxonomically related.
One way a taxonomy assistant might operate is to combine each and
every significant term in a knowledge domain with every other term,
and offer a list of possible relations with which to tag each offered
matching set. Someday, perhaps, we will be able to buy off the shelf
"taxonomy packs" (dynamic thesauri) in many domains.
-- Wayne --
∂16:37 2-Sep: genesereth@sumex, lenat@score/su Re: Taxonomies
Date: Wed 29 Aug 84 09:31:47-MDT
From: Stan Shebs <SHEBS@UTAH-20.ARPA>
Subject: Re: Taxonomies
Some of the most recent KR systems attempt to provide meta-taxonomies;
I know of RLL/Eurisko, MRS, and AGE, all Stanford products. Am not
sure what LOOPS provides in the way of knowledge about representation
schemes (although one could build something to recommend whether a given
piece of information should be a logical assertion, an object, an instance
variable of an object, Lisp code, etc).
Meta-taxonomies are HARD. The ability to create a taxonomy of some body of
knowledge implies that one has both a deep and broad understanding of that
body. The creation of a meta-taxonomy implies that there is a similar
level of understanding for many issues in knowledge representation, which
is definitely *not* the case. We're still lacking adequate theories of
multiple inheritance, nor have we plumbed the depths of strange logical
systems. Looking at library science is an interesting idea; while I
imagine that many of the classification schemes are informal (probably
relying on human judgement), librarians have been classifying massive
databases (books) for a long time.
Moving farther afield, taxonomies in other AI areas are lacking. I asked
a while back about taxonomies for rule systems, and found that there was
about one paper, by Davis and King in a ca. 1976 MI. This, however, was
an informal taxonomy, and not particularly susceptible to mechanization.
Am still waiting for a tree that puts OPS5, Emycin, and Prolog on
different leaves...
stan
Return-Path: <greiner@diablo>
Received: from diablo by SUMEX-AIM.ARPA with TCP; Tue 21 Aug 84 15:43:37-PDT
Date: 21 Aug 1984 15:42-PDT
From: Russ Greiner <greiner@diablo>
Subject: Flush left enumerate-like environment
To: genesereth@sumex
Cc: greiner@sumex
Message-Id: <84/08/21 1542.200@diablo>
Mike -
Try this. It may not work, as it is taken from a larger
macro, which does yet other things...
-----
\newlist{noindentenum}{{\arabic{list}.\ }}%
{\setlength{\labelwidth}{0em}%
\setlength{\leftmargin}{0em}%
\setlength{\labelsep}{0em}}%
{}}
-----
Russ
-------
∂21:12 12-Sep: sjg, genesereth@sumex/su challenge for MRS?
[from AIList]
....
Date: Saturday, 8-Sep-84 18:35:50-BST
From: O'KEEFE HPS (on ERCC DEC-10) <okeefe.r.a.@EDXA>
Subject: OPS5 problem
An MSc student came to me with a problem. He had a pile of OPS5 rules
and was a bit unhappy about the means he had adopted to stop them
looping. Each rule looked rather like
(p pain77
(task ↑name cert)
(injury ↑name injury6 ↑cert <C>)
(symptom ↑name symptom9 ↑present yes)
-(done pain77)
-->
(make done pain77)
(modify 2 ↑cert (compute ....))
)
There were dozens of them. The conflict resolution rule of never
firing the same rule on the same data more than once didn't help, as
modify is equivalent to a delete and a make. What he actually wanted
can be expressed quite neatly in Prolog:
candidates(BestToWorst) :-
setof(W/Injury, weight(Injury, W), BestToWorst).
weight(Injury, MinusCertainty) :-
prior←certainty(Injury, Prior),
findall(P, pro(Injury, P), Ps),
product(Ps, 1.0, P),
findall(C, con(Injury, C), Cs),
product(Cs, 1.0, C),
MinusCertainty is -(1 - P + P*C*Prior).
pro(Injury, Wt) :-
evidence←for(Injury, Symptom, Wt),
present(Symptom).
con(Injury, Wt) :-
evidence←against(Injury, Symptom, Wt),
present(Symptom).
product([], A, A).
product([W|Ws], A, R) :-
B is A*W,
product(Ws, B, R).
We managed to produce something intermediate between these two, it
used evidence-for and evidence-against tables in working memory, and
had just two hacky rules instead of the scores originally present.
I did spot a way of stopping the loop without using negation, and
that is to make the "certainty" held in the (injury ↑name ↑cert)
WM elements a gensym whose value is the desired number, then as far
as OPS5 is concerned the working memory hasn't been changed. Of
course that makes patterns that use the number harder to write, and
seems rather hacky itself.
To come to the point, I have two questions about OPS5.
1) Is there a clean way of coding this in OPS5? Or should I have
told him to use EXPERT?
2) As I mentioned, we did manage to do considerably better than his
first attempt. But the thing that bothered me was that it hadn't
occurred to him to use the WM for tables. The course he's in
uses the Teknowledge(??) "OPS5 Tutorial" (the one with the Wine
Advisor) and students seem to copy the Wine Advisor more or less
blindly. Is there any generally available GOOD course material on
OPS5, and if so who do we write to? Are there any moderate-size
examples available?
PROJECT ACTIVITIES FOR PROJECT F-1: THEORIES OF INFORMATION
The notions information and of informational content are central to much
of the work done at CSLI and are emerging as central notions in philosophy,
computer science, and other disciplines. Thus we need mathematically
precise and philosophically cogent accounts of information and the forms
it takes. The F-1 project will hold a series of meetings on various CSLI
researchers' approach to the notion of information. The emphasis will be
on gaining a detailed understanding of the theories that are being developed
and discussing issues in ways that will be helpful in making further
progress. Those interested should attend the meetings regularly to help
develop a working group with a shared body of knowledge. For this reason,
we will not make it a practice to announce individual meetings, which will
occur approximately bi-weekly, Tuesdays at 3:15, in the Ventura Seminar
Room. The first meeting will be on October 2, when Jon Barwise will speak
for a bit about the nature and prospects for a theory of information,
followed by Fernando Pereira and/or Stan Rosenschein who will talk about
the current state of situated automata theory.
---John Perry
←←←←←←←←←←←←
What's in a theory?
Date: Wed 5 Sep 84 10:54:11-PDT
From: BARNARD@SRI-AI.ARPA
Subject: induction vs. deduction
Tony Hasemer's comments on causality and its relationship to inductive
versus deductive logic are very well-taken. It's time for people in
AI to realize that deduction is quite limited as a mode of reasoning.
Compared to induction, the mathematical foundations of deduction are
well-understood, and deductive systems are relatively easy to
implement on computers. This no doubt explains its popularity in AI.
The problem arises when one tries to extend the deductive paradigm
from toy problems to real problems, and must confront exceptions,
borderline cases, and, in general, the boggling complexity of the
state space.
While deduction proceeds from the general (axioms) to the specific
(propositions), induction proceeds from the specific to the general.
This seems to be a more natural view of human intelligence. By
observing events, one recognizes correlations, and infers causality
and other relationships. To be sure, the inferences may be wrong, but
that's tough. People make mistakes. In fact, one of the weaknesses
of deduction is that it does not permit one to draw conclusions that
may be in error (assuming the axioms are correct), but that represent
the best conclusions under the circumstances.
Visual illusions provide good examples. Have you ever wondered why
you see a Necker Cube as a cube (one of the two reversals), and not as
one of the other infinite number of possiblities? Perhaps we learn of
cubes through experience (an inductive explanation), but the effect
also occurs with totally unfamiliar figures. A more general inductive
explanation holds that we see the simplest possible figure (the
Gestalt principle of Pragnanz). A cube, which has right angles and
equal-length sides, is simpler than any of the other possiblilities.
The concept of "simple" can be made precise: one description is
simpler than another if it can be encoded more economically. This is
sometimes called the principle of Occam's Razor or the principle of
Minimum Entropy.
Steve Barnard
∂13:11 22-Sep: genesereth@sumex/su
sounds like your meta-epistemology: "what's in a theory", sorta
* * * * * For Steve Tappel * * * * *
Date: Fri 14 Sep 84 14:30:14-PDT
From: BARNARD@SRI-AI.ARPA
Subject: induction vs. deduction
The discussion of induction vs. deduction has taken a curious turn.
Normally, when we speak of induction, we don't mean *mathematical
induction*, which is a formally adequate proof technique. We mean
instead the inductive mode of reasoning, which is quite different.
Inductive reasoning can never be equated to deductive reasoning
because it begins with totally different premises. Inductive
reasoning involves two principles:
(1) The principle of insufficient reason, which holds that in the
absence of other information, the expectation over an ensemble of
possibilities is uniform (heads and tails are equally probable).
(2) The principle of Occam's razor, which hold that given a variety of
theories about some data, the one that is "simplest" is preferred.
(We prefer the Copernican model of the solar system to the Ptolemaic
one, even though they both account for the astronomical data.)
The relationship of time, causality, and induction has been
investigated by the Nobel Laureate, Ilya Prigogine. The laws of
classical physics, with one exception, are neutral with respect to the
direction of time. The exception is the Second Law of Thermodynamics,
which states that the entropy of a closed systems must increase, or
equivalently, that a closed system will tend toward more and more
disordered states. For a long time, physicists tried to prove the
Second Law in terms of Newtonian principles, but with no success.
Eventually, Boltzman and Gibbs explained the Second Law
satisfactorily by using inductive principles to show that the
probability of a system entering a disordered, high-entropy state is
far higher than the converse. Prigogine proposes that random,
microscopic events cause macroscopic events to unfold in a
fundamentally unpredictable way. He extends thermodynamics to open
systems, and particularly to "dissipative systems" that, through
entropy exchange, evolve toward or maintain orderly, low-entropy
states.
Inductive reasoning is also closely connected with information theory.
Recall that Shannon uses entropy as the measure of information.
Brillouin, Carnap, and Jaynes have shown that these two meanings of
entropy (information in a message and disorder of a physical system)
are equivalent.
Steve Barnard