which bring software engineers (broadly termed to include
expert system developers) in contact with students of
those other disciplines. While the computer science students refine their knowledge about particular issues in
computer science, they also reap the benefits of exposure
to other fields such as, psyschology, sociology, or any of
the humanitits. They will extend their rational capabilities
and will have increased sympathy for other fields of study.
Likewise, the students of the domain field will improve
their command of that study, and at the same time they
gain appreciation of computer science.
Learning intensifies.
In Socrates, even the expert claimed to have
benefited through the conflicts that she found in DSM-III.
CITATIONS
Bobrow, Daniel G., Sanjay Mittal, and Mark J. Stefik.
Expert Systems: Perils and Promise," in Communications
of th_e ACM. Volume 29,9 (September, 1986.)
Buchanan, Bruce G., David Barstow, Robert Bechtal, et
al "constructing an Expert System," in Building. E_~P_~
S_ystems.
Hayes-Roth, Waterman, Lenat eds. Reading,
Mass: Wesley Publishing Company, Inc. (1983) 127-67.
Diag.nostic and Statistical Manual of Mental Disorders,
Third Edition. Washington, DC: American Psychiatric Association, (1980).
Fairley, Richard. Software Enci~._eerin_9 Concepts. New
York: McGraw-Hill Book Company, (1985).
Fellers, Jack
W.
"Key
I-actors
in
Knowledge
Acquisition," in Computer Personnel. A quarterly publication of the Special Interest Group on Computer Personnel
Research: ACM Press. Volume II, 1 (May, 1987).
Frank, Edwina D., Phd.D., Professor of Psychology,
Loyola University, New Orleans, LA: informal interviews and
project reviews. (January, 1987 - - August, 1987)
Harmon, Paul, and David King. Expert Systems: Artificial Intelligence in Business.
New York: John Wiley &
Sons. (1985).
Hayes-Roth, Frederick, Donald A. Waterman, and
Douglas B. Lenat. "An Overview of Expert Systems," in
Building Expe~ Systems.
Hayes-Roth, Waterman, Lenat
eds.
Reading, Mass:
Wesley Publishing Company, Inc.
(1983) 3-29.
Page-Jones, Meilir. The Practical Guide to Structured
Systems Desi.gLn
~. Englewood Cliffs, NJ: Yourdon Press.
(1980).
Sagalowicz, Daniel.
Quoted by Edith Meyers in
"Expert Systems: Not for Everyone," in Datamation. (May
15, 1987).
Welburn, Tyler
Advanced Structured Cobol: Batch,
On-line, and Data-Base Concepts. Palo Alto, CA: Mayfield
Publishing Company, 1983.
Zuckerman, Edward, Ph.D., Professor of Psychology,
Wellness Resource Center, University of Pittsburg. Pittsburg, PA:
correspondence and telephone interviews.
(January 1987).
SIGART Newsletter, July 1988, Number 105
AN EXPERT SYSTEM FOR AN
IDIOSYNCRATIC DOMAIN: LOVE, INTIMACY,
AND FRIENDSHIP
Joseph S. Fulda
Hofstra University
(C) 1987 Joseph S. Fulda
Expert systems conceived and developed thus far
have dealt with problems, such as medical diagnosis, circuit design, and mineral exploration, which have objective
solutions. Humans, however, are also experts in reasoning
about domains where the key concepts are not objectively
fixed. We do form subjective, but sound, judgments about
idiosyncratic--i.e., p e r s o n - s p e c i f i c - - domains. One such
domain is that of love, intimacy, and friendship. Despite
the fact that our notions of love, intimacy, and friendship
vary greatly and that these relationships have been the
subject of poets, novelists, biographers, psychologists, and
other students of human nature for centuries, there is
enough of a common consensus to make communication
and, very frequently, c o m m o n judgments about this
domain possible.
We present a rule-based expert system, currently b e ing implemented in PROLOG, which provides a general
mechanism for handling idiosyncratic domains. Our system also provides a thorough example of how the attainment of cognitive-affective states and resulting relationships between people can be modelled using a rule-based
system. Thus this work should interest cognitive scientists as well as computer scientists seeking to endow
computers with truly human-like thought.
The key to our design is, as might be expected, to
abstract out whatever is c o m m o n to most people's understanding of the nature of a particular class of relationships. Nevertheless, there will be a substantial failure rate
if idiosyncratic criteria for the attainment or blocking of
relationships are simply ignored. Therefore, we adopt the
following formalism:
Iff
X (a)
Y (b)
Then Z (c).
This bidirectional rule is to be understood as follows: If X
& Y then Z with probability c; If Z then X with probability
a; If Z then Y with probability b. (Of course, "If Z then X"
is identical to "Only if X then Z.") Thus a, b, and c are what
have been called certainty factors or, more gracefully, attenuation factors (1). (Note that we are nod. associating
certainty factors with the antecedent conditions of the
rules in either direction, though it is trivial to add these.)
The concept of attenuation factors was developed for the
MYCIN system, a rule-based expert system for medical
diagnosis (2). This formalism using bidirectional rules with
attenuation factors oerative in both directions (cf. 3, 4)
neatly handles the problems inherent in reasoning about
an idiosyncratic domain. These are two: somesone may
have idiosyncratic requirements for the attainment or
Page 30
blocking of relationship Z or he may idiosyncratically i g nore some of the elements that most do use in deciding
w h e t h e r to enter relationship Z.
Thus, in the example above, if c is .7 this means that
30% of the instances where X and ¥, nevertheless not Z
because of idiosyncratic requirements for the attainment
or blocking or the relationship. Likewise, if a is .8, then in
20% of the instances where the parties have relationship
Z, they do so despite the prevalent v i e w that X ought to
hold for Z to obtain but does no__t.
We deal here, intentionally, with a small number of
rules and relationships; nuances of these are not treated,
since they are far t o o idiosyncratic for general agreement,
even broadly considered.
The relationships considered
are, in order of increasing c o m m i t m e n t and initial intensity,
companionship, friendship, confidant(e)s, intimacy, love,
marriage, and family.
Before presenting the rules themselves, which are the
ultimate test of the merit of the idea, some perfatory
remarks on the system as a whole are in order.
First, we have designed the system with t w o purposes
in mind: ease of implementation and testing and naturalness of representation, with the latter easily the more i m portant desideratum.
Thus, except for companionship,
each level of interpersonal closeness includes and subsumes the previous (and therefore all prior) level(s). Thus
if X and Y have a family together, they are married, they
love each other, they are intimates, they are confidants,
and they are friends (subject to the usual idiosyncratic
exceptions). This seems natural enough for the attainment
of relationships. However, relationships are d y n a m i c and
when they degrade, as they so often do, they do not do
so nicely and monotonically. So while it may be true that
apart from the idiosyncratic exception friendship always
precedes marriage, after many years it is certainly possible
for X and Y to love each other and miss each o t h e r when
apart; but not to like each other or take any pleasure in
each other's company: An'~odd, but common, state of affairs. So the system presented here should be see either
as a r e l a t i o n s h i p - a t t a i n m e n t model or as an ideal.
Second, the specific attenuation factors selected were
not obtained by empirical research. They are s i m p l y leftbrain estimates of frequency or r i g h t - b r a i n estimates of
probability. They do seem to conform to our (limited) e x perience and our intuition, but a m e t h o d o l o g i c a l cognitive
scientist m i g h t
wish
to use other numbers,
more
r i g o r o u s l y obtained. Tant mieux. Since we are, after all,
dealing
with
and
presenting
a
methodology
for
idiosyncratic domains, the author's o w n p e r s o n - s p e c i f i c
biases or insights may occasionally be visible.
COMPANIONSHIP.
Iff X
X
X
Then
and Y: like each other 7 (.7)
and Y: willing to share time with each o t h e r (1)
and Y: lonely (.8)
X and Y: c o m p a n i o n s (.5)
In the forward direction, there are three a n t e c e d e n t
c o n d i t i o n s which are representative of the three classes of
c o n d i t i o n s on which relationships may be contingent.
First, there is the "engine," the e m o t i o n a l impetus for the
1A rule elaborating this will be presented later.
SIGART Newsletter, July 1988, Number 105
relationship, here loneliness. Second, there is the defining
condition, a condition which holds universallv w h e n the
relationship exists, here "willing to share free time with
each other." Third, there is a precondition, here simple a f fection, w i t h o u t which the relationship is blocked (5),
To the c o m p u t e r scientist interested in the f o r m a l i s m
i n t r o d u c e d and the f r a m e w o r k p r o v i d e d for i d i o s y n c r a t i c
domains, there is no distinction of i m p o r t a n c e b e t w e e n
these three classes of conditions. To the c o g n i t i v e s c i e n tist, the specialist imparting his k n o w l e d g e to the e×pert
system and the k n o w l e d g e e n g i n e e r assisting him, the
user a p p l y i n g the expert system, and the general reader,
on the o t h e r hand, these distinctions are p r e g n a n t w=th
meaning. In presenting the rules, therefore, we shall c o n tinue to draw them.
A rule must have one or m o r e engines, for there is no
relationship which begins with no emotional i m p e t u s of
any sort; usually, but as we shall see not always, there is
just one. When the attenuation factors of the engine c o n ditions sum to less than one, as is the case here, with
c o m p a n i o n s h i p , it is an indication that the e m o t i o n a l i m petus for the relationship class is s o m e t i m e s idiosyncratic.
A rule may or may not have a defining condition, but
if it has m o r e than one, it is an indication of s l o p p y
d e s i g n - - in particular, failure to generalize. Note that a
defining condition does not guarantee a r e l a t i o n s h i p (as
here), although the relationship guarantees it. As we shall
see, however, most a n t e c e d e n t c o n d i t i o n s are p r e c o n ditions.
The l o w attenuation factor for c o m p a n i o n s h i p reflects
the fact that, unlike m o r e c o m m i t t e d relationships, it arises
from a generalized loneliness on the part of, to use the
system's names, X and Y, and many other r a n d o m or
idiosyncratic factors help decide w h e t h e r the t w o will b e c o m e companions.
As we ascend the hierarchy of
relationships, random influences will become i n c r e a s i n g l y
negligible and idiosyncratic factors increasingly visible.
FRIENDSHIP.
Iff X and Y:
X and Y:
X and Y:
Then X and
like each o t h e r 2 (.8)
respect each other. (.7)
w a n t to share free time with each o t h e r (1)
Y: friends (9)
For friendship, w a n t i n g to share free time with each
o t h e r is both the engine and the defining condition. Note
that the p r e c o n d i t i o n s are still necessary to ensure that
the m o t i v a t i o n for wanting each other's c o m p a n y is
friendship and not (say) business. Those w h o attain the
c o g n i t i v e - a f f e c t i v e state c o m m o n to friends w i t h o u t liking
or respecting each other are, of course, accounted for by
the attenuation factors.
For some people, all important
business associates may be friends.
Many will find the p r e c o n d i t i o n s given here inadequate; their notion of friendship is probably akin to our
n o t i o n of confidants, which will be presented next and
a m o u n t s to classical, true friendship ("friendship with a
capital 'F'"). As the w o r d is n o r m a l l y used today, however,
"friend" means (at most) what is given above.
Notice the difference between the defining conditions
for c o m p a n i o n s h i p and friendship: "willing" vs. " w a n t " to
2A rule elaboratin 9 this will be presented later.
Page 31
share free t i m e with each other. A p e r s o n - s p e c i f i c need
has replaced a g e n e r a l i z e d loneliness.
CONFIDANTS.
Iff X and Y: friends (.85)
X and Y: respect each o t h e r highly (.7)
X and Y: w i l l i n g / w a n t to share their t h o u g h t s and
feelings with each o t h e r (1)
X and Y: miss each o t h e r w h e n apart (.5)
X and Y: trust each o t h e r 3 (.8)
Then X and Y: confidant(e)s (.7)
Once again, the e n g i n e and the defining c o n d i t i o n are
the same: sharing t h o u g h t s and feelings.
This is
friendship on a m o r e i n t i m a t e level (it subsumes friendship
w i t h i d i o s y n c r a t i c exceptions) than sharing time alone can
even reach. It is also the l o w e s t level of the h i e r a r c h y of
t h e r e l a t i o n s h i p classes w e c o n s i d e r at w h i c h missing
each other, a c o g n i t i v e - a f f e c t i v e state central to strong
relationships, c o m e s into play. For some, it is a p r e c o n d i t i o n to b e i n g a c o n f i d a n t (or a very quick result); for
s o m e , it m a y also be an engine condition.
One m i g h t
s u p p o s e t h a t "trusting each o t h e r " and being " w i l l i n g to
share t h e i r t h o u g h t s and feelings w i t h each o t h e r " w o u l d
be coincident. Experience teaches that this is not always
so, as the need to confide, in m a n y so often o v e r p o w e r s
d o u b t s a b o u t the other's t r u s t w o r t h i n e s s .
It is also a
r e f l e c t i o n of the m e c h a n i s m of trust buildup, w h i c h is not
unlike the b u i l d u p of band credit. One trusts a little; if it
w o r k s out, o n e trusts a little more. Of course, h o w e v e r
natural this s y s t e m m a y be and h o w e v e r prevalent, it
leaves the c o n f i d a n t (or bank) open to disaster in u n f o r t u n a t e instances.
Perhaps that's w h y a n o t h e r m e t h o d - f o r m a l and i n f o r m a l r e f e r e n c e s - - h a s e v o l v e d for the
r e s o l u t i o n of the trust issue (by p o t e n t i a l c r e d i t o r s and
p o t e n t i a l friends alike).
INTIMATES.
Iff X and Y: confidants(e)s (.95)
X and Y: k n o w each other for s o m e t i m e (.5)
X and Y: e m o t i o n a l l y c o m p a t i b l e even in
close quarters (.5)
X and Y: find each o t h e r a t t r a c t i v e (.8)
Then X and Y: i n t i m a t e s (.5)
Intimacy, as w e are c o n s i d e r i n g it, is a serious
relationship one step shy of r o m a n t i c love, not m e r e
physical i n t i m a c y w i t h o u t a very p o w e r f u l e m o t i o n a l c o m ponent.
Obviously, mere physical i n t i m a c y is possible
w h e n X and Y are strangers, hardly i n t i m a t e s as w e are
using t h e term. The i n t i m a c y to w h i c h w e refer is a state
of heart and mind, a c o g n i t i v e - a f f e c t i v e state, joined to a
state of affairs (in the philosopher's sense of the phrase)
which can be r e a c h e d w i t h o u t a c o n f i d a n t - t y p e r e l a t i o n ship i d i o s y n c r a t i c a l l y only one t i m e in t w e n t y .
For the
f o r w a r d - c h a i n i n g a n t e c e d e n t "X and Y: p h y s i c a l l y i n t i m a t e "
the c o n s e q u e n t s w o u l d have v e r y different a t t e n u a t i o n
factors, say, .05, .25, .02, .6, respectively.
A n o t h e r issue that must e v e n t u a l l y be addressed here
is that of s a m e - s e x relationships. For heterosexuals, the
fourth a n t e c e d e n t c o n d i t i o n rules out s a m e - s e x r e l a t i o n ships. For others, it does not, nor does a n y t h i n g else in
the rule.
Studies s h o w that o n e - o n e r e l a t i o n s h i p s are
much m o r e c o m m o n a m o n g heterosexuals, but it is not to
be supposed that this creates a serious o v e r s t a t e m e n t of
the chances of an intimate r e l a t i o n s h i p d e v e l o p i n g (50%)
given all the stated a n t e c e d e n t c o n d i t i o n s .
Having dealt w i t h these t w o side issues, w e return to
an e x p o s i t i o n of the rule. The e n g i n e here is m u t u a l a t traction and as a n y o n e will a d m i t under p e r s i s t e n t q u e s tioning it is not a sure thing; technically, that is, it does
not give "X and Y: intimates" an a t t e n u a t i o n f a c t o r of 1.
What will make t w o p e o p l e decide to u p g r a d e a c o n f i d a n t
relationship
to
an
intimate
relationship
is
highly
idiosyncratic, so m u c h so that fully half the t i m e that the
a n t e c e d e n t s are true, i n t i m a c y will n e v e r t h e l e s s not result.
If it s e e m s to the reader that this is s u r e l y a g r o s s e x aggeration, it is likely that he is u n w i t t i n g l y using i n t i m a c y
in a w e a k e r sense by not being sensible of the full range
of a n t e c e d e n t c o n d i t i o n s for a c o n f i d a n t - t y p e relationship,
which r e l a t i o n s h i p is, in all but a handful of i d i o s y n c r a t i c
cases, p r e r e q u i s i t e to i n t i m a c y as here defined. (And t h e r e
are of course the r e q u i r e m e n t s of the e n g i n e c o n d i t i o n
and the t w o preconditions.)
ROMANTIC LOVE,
X and Y: i n t i m a t e s (.95)
and Y: w i l l i n g / w a n t mutual e m o t i o n a l c o m m i t m e n t ( . 7 5 )
and Y: w i l l i n g / w a n t a b s o r p t i o n in each other's life(.7)
and Y: miss each o t h e r quite i n t e n s e l y
w h e n apart (1)
Then X and Y: love each o t h e r ( r o m a n t i c a l l y ) (.9)
Iff
X
X
X
The e n g i n e s here are the m u t u a l desire for e m o t i o n a l
c o m m i t m e n t , o n e - o n e loyalty, and the m u t u a l desire for
total i n v o l v e m e n t in each other's life ( i n v o l v e m e n t does
n o t mean d o m i n a t i o n , u n w a n t e d interference, etc.) and the
defining c o n d i t i o n for this c o g n i t i v e - a f f e c t i v e state is
g i v e n by the final a n t e c e d e n t c o n d i t i o n . It will be i s t r u c t i r e to c o m p a r e loving and liking (the l a t t e r is required for
c o m p a i o n s h i p and friendship (and, as e x p l a i n e d early in
this paper, by inclusion in all m o r e intense relationships)):
Iff
X and Y: e n j o y each other's c o m p a n y (.95)
X and Y: e m o t i o n a l l y c o m p a t i b l e (.8)
Then X and Y: like each o t h e r (1)
Thus the e m p h a s i s in l o v i n g is on a l o n g i n g w h e n
apart, w h i l e the emphasis in liking is on an a g r e e a b l e n e s s
w h e n t o g e t h e r . Of course, e x c e p t for i d i o s y n c r a t i c cases,
l o v e includes i n t i m a c y w h i c h includes....friendship and,
therefore, liking.
Nevertheless, this n e g a t i v e emphasis
helps explain w h y love w i t h o u t affection as m a y o c c u r
w h e n a r e l a t i o n s h i p d e g r a d e s can be a clearly d e s t r u c t i v e
f o r c e and w h y lectures entitled " H o w to Fall o u t of Love"
d r a w large audiences.
MARIAGE.
Iff X and Y: l o v e each o t h e r (.65)
X and Y: c u l t u r a l l y c o m p a t i b l e (.7)
X and Y: k n o w each o t h e r for a c o n s i d e r a b l e t i m e (.75)
X and Y: w i l l i n g / w a n t m u t u a l i n s t r u m e n t a l c o m m i t m e n t
X and Y: find each other's i d i o s y n c r a t i c
b e h a v i o r s t o l e r a b l e to pleasing (.5)
X and Y: l o n e l y (.7)/X and Y: w a n t kids (.9)
Then X and Y: m a r r i e d (.7)/X and Y: m a r r i e d (.75)
3A rule elaboratm 9 this will be presented later.
SIGART Newsletter, July 1988, Number 105
Page 32
Marriage c o n d i t i o n s are fairly idiosyncratic; hence o n l y
7 0 - 7 5 % of the time that the a n t e c e d e n t c o n d i t i o n s are
satisfied is the c o n s e q u e n t realized. Since no c o n d i t i o n as
to gender is imposed, if a s a m e - s e x r e l a t i o n s h i p m e e t s all
the a n t e c e d e n t conditions, it is a m o n g the 30% (forget
a b o u t w a n t i n g kids as an antecedent!) of such cases
w h e r e the rule is i d i o s y n c r a t i c a l l y not t r i g g e r e d . Also l o w
are the a t t e n u a t i o n factors in the o t h e r direction.
This
m a y indicate w h y so m a n y m a r r i a g e s sour.
In o t h e r
words, if the c o n s e q u e n t w e r e "X and Y: h a p p i l y married"
the a t t e n u a t i o n factors w o u l d be far higher, perhaps .98,
.95, .8, .7, .9, .7/.9 respectively.
There are t w o or p o s s i b l y three e n g i n e s here w h i c h
overlap; w a n t i n g i n s t r u m e n t a l c o m m i t m e n t , w a n t i n g kids,
and being lonely. As the a t t e n u a t i o n factors for m a r r i a g e
show, w a n t i n g kids is a s t r o n g e r spur to m a r r i a g e than
loneliness, w h i c h can m o r e easily be satisfied by means
short of marriage. As i m p l e m e n t e d , the rules for m a r r i a g e
in the system n u m b e r three. One with "X and Y: w a n t kids
(.9)" leading to "X and Y: "married (.75)"; one w i t h "X and
Y: lonely (.7)" leading to "X and Y: m a r r i e d (.7)"; and one
w i t h both "X and Y: l o n e l y (.7)" and "X and Y: w a n t kids
(.9)" leading t o "X and Y: m a r r i e d (.75)."
There is no
r e d u n d a n c y in the presence of the third rule, since all
rules are bidirectional.
F u r t h e r m o r e , that t w o rules say
that 75% of couples will end up m a r r i e d does not
g u a r a n t e e that they will be the same 75%, so f r o m an intellectual p o i n t of view, it is well to include it, even if it
makes no different probabilistic prediction.
There are three degrees of e m o t i o n a l c o m p a t i b i l i t y
required by the system: finding each other's idiosyncratic
behaviors tolerable to p l e a s i n g - - f o r marriage, e m o t i o n a l
c o m p a t i b i l i t y even in close q u a r t e r s - - f o r intimacy, and
(basic) e m o t i o n a l c o m p a t i b i l i t y - - f o r
c o m p a n i o n s h i p (via
liking). The levels of c o m p a t i b i l i t y required may vary on a
p e r s o n - s p e c i f i c basis and that, of course, is dealt w i t h by
the a t t e n u a t i o n factors.
A word a b o u t loneliness, a r e q u i r e m e n t for c o m panionship and m a r r i a g e - - b u t n o t h i n g in b e t w e e n , is in
order.
The s y s t e m reflects the idea that either c o m panionship will satisfy loneliness or if it does not, n e i t h e r
will friendship, a confidant, or even i n t i m a c y and love.
These i n - b e t w e e n relationships satisfy p e r s o n - s p e c i f i c and
t y p e - s p e c i f i c (to confide, to be intimate, etc.) needs, rather
than loneliness in general. Marriage, on the o t h e r hand,
adds to love (with its satisfaction of a wide v a r i e t y of
p e r s o n - s p e c i f i c and t y p e - s p e c i f i c needs) c o n s t a n t c o m panionship.
FAMILY
Iff X and Y: married (1)
X and Y: w a n t kids (.9)
X and Y: sufficient means and space (.5)
Then × and Y: have a f a m i l y t o g e t h e r (.95)
The engine here, obviously, is w a n t i n g kids. Virtually
all m a r r i e d couples w h o w a n t kids do have or a d o p t them.
The only s o m e w h a t general c o n s t r a i n t seems t o be sufficiency of means and space and, as the a t t e n u a t i o n f a c t o r
shows (50% of families don't have sufficient m e a n s and
space), it is not terribly constraining.
It is instructive to c o m p a r e the final three rules of the
hierarchy: family g i v e n marriage, m a r r i a g e given love, love
given intimacy. The c o m p a r i s o n s u g g e s t s that the c o n ditions for m a r r i a g e are far m o r e i d i o s y n c r a t i c than are
t h o s e for a f a m i l y g i v e n marriage and l o v e given intimacy.
People w h o are m a r r i e d do a l m o s t a l w a y s have families;
p e o p l e w h o are i n t i m a t e and m e e t t h e r e m a i n i n g c o n -
SIGART Newsletter, July 1988, Number 105
ditions do fall in love. But even if l o v e is tried and true,
m a r r i a g e is often n o t the result. (By often, w e mean, of
course, 2 5 - 3 0 % of the time.) The n o n - i d i o s y n c r a t i c c o n ditions for m a r r i a g e are also a g o o d deal m o r e d e m a n d i n g
than t h o s e for love g i v e n i n t i m a c y and f a m i l y g i v e n m a r riage. Thus w i t h such a m o d e l of r e l a t i o n s h i p a t t a i n m e n t ,
it is no surprise that marriages are hard to maintain. That
c o n c l u s i o n f o l l o w s f r o m the single p r e m i s e that the harder
a relationship is t o form, the easier it is to damage. While
we could see an intellectual case e i t h e r w a y on the
premise, empirical e v i d e n c e supports b o t h the p r e m i s e and
the conclusion. This is only one of m a n y d o m a i n - specific
c o m m e n t s made in this work. It is natural when designing
a s y s t e m to be an e x p e r t and discussing one's ideas with
c o l l e a g u e s and associates to learn a c o n s i d e r a b l e a m o u n t
a b o u t the d o m a i n as well as a b o u t e x p e r t s y s t e m design
for idiosyncratic d o m a i n s in general.
Returning to the latter, an a t t e m p t was m a d e
t h r o u g h o u t the system to restrict c o n s i d e r a t i o n to c o n ditions with a t t e n u a t i o n factors that w o u l d make at least
.5. Thus, in the case of the familial relationship, the 5% of
couples w h o don't have or a d o p t children despite that
desire, t h e i r being married to each other, and t h e i r h a v i n g
sufficient means and space have "reasons of their o w n " - i d i o s y n c r a t i c or p e r s o n - s p e c i f i c factors which influence
the relationships t h e y form. Likewise, the 10% of couples
w h o have a family but do not w a n t children m u s t also
have "reasons of t h e i r o w n " - - p e r h a p s religious beliefs or
concern with w h a t the n e i g h b o r s think, etc. It is i m p o r t a n t
in an expert s y s t e m designed for an idiosyncratic d o m a i n
not to try to d i s c o v e r e v e r y i d i o s y n c r a s y and assign it an
a p p r o p r i a t e a t t e n u a t i o n factor (less than .5), or the s y s t e m
will rapidly be t r a n s f o r m e d f r o m one which handles
idiosyncrasies in its d o m a i n to one w h i c h m e r e l y reflects
the idiosyncrasies of its author. The latter is to s o m e e x tent i m p o s s i b l e to avoid in any d o m a i n w h e r e the key
c o n c e p t s are not o b j e c t i v e l y fixed; w e h o p e w e have d o n e
a credible job.
REFERENCES
1. Winston, P.H., Artificial Intelligence, 1984, Reading,
MA; Addison Wesley.
2. Shortliffe, E. H. and Buchanan, B.G., "A M o d e l of Inexact
Reasoning
in
Medicine,"
Mathematical
Bio__sciences 23(1975): 351-379.
3. Duda, R.O., et al., " D e v e l o p m e n t of the P r o s p e c t o r
C o n s u l t a t i o n S y s t e m for Mineral Exploration," S e p t e m b e r 1978, M e n l o Park, CA: SRI.
4. Duda, R.O, et al., "A C o m p u t e r - B a s e d C o n s u l t a n t for
Mineral Exploration," S e p t e m b e r 1979, M e n l o Park,
CA: SRI.
5. Vere, S.A., "Relational P r o d u c t i o n Systems," Artificial
Intelligence 8(1977): 47-68.
APPENDIX
A. DEFAULT RULE.
Iff t r u e (1) Then X and Y: n o t h i n g (.8)
The system's universe of discourse is all couples, X
and Y, w h o k n o w each other, h o w e v e r slightly. This rule
states t h a t 80% of such couples are neither c o m p a n i o n s
nor friends (nor a n y t h i n g closer); it serves as a default.
B. ADOPTION RULES (UNDIRECTIONAL).
1. If
X and Y: married
Y and Z: friends (or more)
Then X and Z: c o m p a n i o n s (.6)
Page 33
2. If
X and Y: married
Y and Z: confidants (more would
probably lead to divorce)
Then X and Z: friends (.4)
3. If
X and Y: married
Y and Z: confidants
Then X and Z: companions (.7)
There is little point in writing rules for × and Y: less
than married, since the attenuation factors would be very
low (very few such "adoptions"). it is possible that for X
and Y: have a family together, the attenuation factors
would be lower, since people with families have less time.
Finally, subject to idiosyncratic exceptions, adoptions never
result in more than a friend or, if they do, they cease being an adoption and meet the standard antecedent c o n ditions.
Note that the antecedents of (2) and (3) overlap as do
those of (1) and (2)/(3) to some extent (because, with
idiosyncratic exceptions, confidants are friends).
One
might try to rigorously use available m e t h o d s (see refs.
2,3,4) to obtain better numbers. But the above will do.
C. TRUST RULE.
IFF X and Y: believe t h e y can repose their thoughts
and feelings in each other w i t h o u t facing
ridicule, rejection, etc. (.85)
X and Y: believe ~hey can repose their thoughts
and feelings in each other w i t h o u t facing
exposure leading to mterpersonal a n d / o r
instrumental harm (.85)
X and Y: believe t h e y can repose their thoughts
and feelings in each other w i t h o u t fearing
that their confidences will eventually be
used against them e m o t i o n a l l y i.e., t h r o w n
up at them, etc. (.7)
X and Y: trust each other (1'.)
The difference in attenuation factors reflects the fact
that nearly everyone w o u l d sooner risk l o n g - t e r m e m o tional hurt (condition 3) than s h o r t - t e r m e m o t i o n a l hurt
(condition 1) or exposure (condition 2).
SELECTED AI-RELATED DISSERTATIONS
Assembled by:
Susanne M. Humphrey
National Library of Medicine
Bethesda, MD 20894
and
Bob Krovetz
University of Massachusetts
Amherst, MA 01002
The following are citations selected by title and
abstract as being related to AI, resulting f r o m a c o m p u t e r
search, using the BRS Information Technologies retrieval
service, of the Dissertation Abstracts International (DAI)
database produced by University Microfilms International.
The online file includes abstracts, which are not
published in this listing, but the citations b e l o w do include
the DAI reference for finding the abstract in the published
DAI. Other elements of the citation are author; university,
degree, and, if available, number of pages; title; UM order
number and y e a r - m o n t h of DAI; and DAI subject category
chosen by the author of the dissertation. References are
sorted first by the initial DAI subject c a t e g o r v and second
by the author. In addition the database includes masters
abstracts, denoted by MAI (Masters Abstracts International)
instead of DAI as the reference to the abstract in the
published MAI.
Unless otherwise specified, paper or m i c r o f o r m copies
of dissertations may be ordered from University Microfilms
International, Dissertation Copies, Post Office Box 1764,
Ann Arbor, MI 48106; telephone for U.S. (except Michigan,
Hawaii,
Alaska):
1-800-521-3042,
for
Canada:
1-800-268-6090. Price lists and other ordering and shipping information are in the introduction to the published
DAI.
DIAZ, JULIAN, III. Georgia State University - College of
Business Administration
Ph.D
1987,
310
pages.
PROCESS TRACING INVESTIGATION
INTO PROBLEMSOLVING WITHIN RESIDENTIAL REAL ESTATE APPRAISAL.
DAI V48(09), SecA, pp2378.
University
Microfilms Order Number ADG87-26477. Business A d ministration, General.
MATHIESON, KIERAN DONALDI.
Indiana University Ph.D
1987, 159 pages. A MACHINE LEARNING APPROACH TO
FLEXIBLE MANUFACTURING SYSTEM SCHEDULING. DAI
V48(09), SecA, pp2378.
University Microfilms Order
Number ADG87-27515.
Business
Administration,
General.
BOCK, DOUGLAS BRIAN. Indiana University Ph.D 1987, 201
pages. SETTING DUE DATES FOR COMPUTER BASED
SYSTEM DEVELOPMENT PROJECTS. DAI V48(09), SecA,
pp2384.
University
Microfilms
Order
Number
ADG87-27484. Business Administration, M a n a g e m e n t
ABRAMS, DAVID HOWARD. Georgia Institute of T e c h n o l o g y
D.B. 1987, 220 pages.
A REFERENCE MODEL FOR
HETEROGENEOUS DATA BASE MANIPULATION AND AN
EXPERT SYSTEM PROTOTYPE FOR HOMOGENEOUS
DATA BASE MANIPULATION.
DAI V48(09), SecB.
University Microfilms Order Number ADG87-27184.
Computer Science.
SIGART N e w s l e t t e r , July 1988, N u m b e r 105
Page 34