
devices address beliefs presumably entertained by the 
listener which may affect the acquisition of the IM, but, 
unlike the beliefs addressed by Instantiations and 
Descriptions, are not directly instrumental to its 
comprehension \[Zukerman 1990b\]. Our mechanism is 
based on ideas introduced in \[Zukerrnan and Cheong 
1988\], and it follows the tenet that Peripheral RDs are 
generated to invalidate anticipated impairments to a 
listener's comprehension process. To this effect, our 
mechanism departs from the existing discourse genera- 
tion strategies, and adopts a predictive approach, 
whereby the effect of a message is simulated on a shal- 
low model of a listener's beliefs. If an impairment to the 
comprehension process is anticipated by this model, the 
generation of a remedial rhetorical device is called for. 
In the following section, we briefly consider a model 
of a listener's beliefs capable of predicting commonly 
drawn inferences. We then describe our mechanism for 
the generation of Peripheral RDs. 
Model of a Listener's Beliefs 
In order to address beliefs presumably entertained by a 
particular listener, we maintain an epistemological 
model which represents a listener's beliefs as a function 
of the presented material. This function accounts both 
for direct and indirect inferences drawn from presented 
messages \[Zukerman 1990a\]. 
We represent a listener's beliefs by means of a net- 
work whose nodes contain individual information items 
and whose links contain the relationships between the 
nodes (see Figure 2). The information in the network is 
represented at a level of detail which is consistent with 
the level of expertise required to learn the subject at 
hand, e.g., for a high-school student learning algebra, 
well-known concepts, such as numerical addition and 
subtraction, are primitive, whereas relatively new or 
complex concepts, such as bracket simplification, are 
represented in terms of more primitive concepts. The 
links in the network are labeled according to the manner 
in which they were acquired, i.e., they can either be 
Inferred, Told or previously Known, where Inferred 
links are generated by means of generally applicable 
Common-sense Inference Rules. In addition, each link is 
accompanied by a Measure of Belief (MB) between -1 
and 1, akin to Certainty Factors \[Buchanan and Short- 
liffe 1985\], which represents a user's level of expertise. 
The nodes are labeled according to their complexity, p 
for primitive concepts and e for complex ones. Like 
links, nodes may be Inferred, Told or previously 
Known, and each node has a Degree of Expertise (DE) 
between 0 and 1. The DE of a c-node is a function of 
the DEs and MBs of its constituent nodes and links, 
respectively. 
In this paper, we focus on technical domains, where 
the transmitted information typically pertains to pro- 
cedures, objects and goals. We define a context as a tri- 
ple composed of a procedure, an object to which it is 
applied, and the goal accomplished by this procedure 
when applied to this object (labeled cl-c4 in Figure 2), 
to reflect the fact that one procedure may achieve dif- 
ferent goals when applied to different objects. For exam- 
ple, factoring out a common factor will only partially 
factorize a quadratic trinomial such as 3x2+5x-2, while 
it will completely factorize a binomial such as 3x2+5x. 
Newly Interred Links 
.~" Nodes and Links to be added 
Fig. 2: Network Model of a Listener's Beliefs in 
High-School Algebra 3 
Our inference mechanism generates plausible infer- 
ences from links in the network by means of generally 
applicable Common-sense Inference Rules which portray 
reasoning activities such as generalization, specialization 
and similarity-based inference (see Figure 3). These 
roles are inspired by rule adaptations commonly per- 
formed by students which were studied by Matz (1982), 
Brown and Van Lehn (1980), Van Lehn (1983) and 
Sleeman (1984). In order to account for the deductive 
abilities of a particular type of listener, we annotate each 
rule with a measure of uncertainty, denoted p, which 
represents a listener's belief in the validity of a conclu- 
sion given that the evidence is certain. This measure 
resembles the rule strength used in ACT* \[Anderson 
1983\]. The application of the similarity-based rule in 
Figure 3 to the context cl and the link \[Numbers simile 
AT\] in Figure 2 yields the erroneous context e3. 
3 In the actual network each link may have a counterpart 
representing the inverse relationship. However, for clarity of 
presentation, only links which are relevant to our discussion 
are shown here. 
157 
R2 (Similarity-based Inference) 
; If two objects are identified as similar, the 
; applicability of a procedure to one of the objects 
; can be inferred from its applicability to the other, 
; accomplishing the same goal 
IF { 3 a link \[OBJm simile OBJn\] (MB = kin,) 
AND 
B a context \[PROCa--OBJm--GOALt \] 
with links \[PROCo apply-to OBJm\] (MB = kam) 
and \[PROC= has-goal GOALt\] (MB = ktam) } 
THEN (with certainty P2 ) 
Add a context \[PROCa---OBJ,--GOALt\] 
with links \[PROCa apply-to OBJ,\] of type I 
(MB = k. = p2k~ka~) 
and \[PROCa has-goal GOALt\] of type I 
(MB =ktan = p2kmnktam) 
Fig. 3: Similarity-based 
Common-sense Inference Rule 
Generating Peripheral Rhetorical Devices 
The generation of Peripheral RDs is performed by 
applying a procedure called Impairment-Invalidate. This 
procedure simulates the effect of an IM on our network 
model to anticipate possible impairments which may 
inhibit a listener's acquisition of this IM, and then pro- 
poses Peripheral RDs, such as Contradictions, Revisions 
and Motivations, to invalidate these impairments. To 
this effect, it activates three mechanisms: (1) Propaga- 
tion of the effect of a message, (2) Recognition of 
impairments, and (3) Selection of Peripheral RDs. In the 
following subsections, we describe these mechanisms 
and then discuss procedure Impairment-Invalidate. 
Propagation of a Message 
Propagation simulates the alterations taking effect in a 
network representing a listener's beliefs due to infer- 
ences drawn from a message. These inferences, which 
are drawn by activating applicable Common-sense Infer- 
ence Rules, result in changes in the MBs of existing 
links or addition of new links and nodes to the network. 
For instance, a Contradiction such as "You cannot 
always simplify inside the brackets of algebraic terms," 
presented to weaken a listener's belief in the applicabil- 
ity of bracket simplification to Algebraic Terms, may 
cause the listener to weaken his/her correct belief in the 
applicability of bracket simplification to Like Algebraic 
Terms or even to Numbers, and also to weaken his/her 
incorrect belief in the applicability of bracket simplifica- 
tion to Unlike Algebraic Terms and in the applicability 
of addition and subtraction to Algebraic Terms. This 
effect is simulated by the propagation of the Contradic- 
tion \[BrS ~apply-to AT\] in the sample network in Figure 
2, which produces the inferences \[Br$ -,apply-to L'I'\] and 
\[BrS ~apply-to UT\] through the application of a 
specialization rule, \[BrS ~apply-to Numbers\] through the 
application of the similarity-based rule R2 (see Figure 
3), and \[+/- ~apply-to AT\] through the application of a 
deductive inference rule. These inferences contradict the 
beliefs represented by existing links, thereby lowering 
the MBs associated with these links. However, their 
effect ultimately depends on the strength of the rules in 
question and on the MBs of the links, i.e., the impact of 
an inference on a weakly believed link is more pro- 
nounced than for a strongly believed one. 
Recognition of Impairments 
The Recognition mechanism anticipates possible impair- 
merits to a listener's comprehension process by examin- 
ing changes in a network representing the listener's 
beliefs due to a message or an inference. It does not 
guarantee that a certain impairment will occur, rather, it 
conjectures that an impairment is likely to affect a par- 
titular link, denoted a culprit link. 
We have characterized in terms of our network model 
two main types of impairments which lead to undesir- 
able effects commonly encountered in a knowledge 
acquisition setting: (1) Affect-related impairments which 
elicit negative affective responses such as Confusion and 
Loss of Interest, and are caused by a conflict between a 
message or inference and a belief held by a listener; and 
(2) Belief-related impairments such as Mislearning, 
Insufficient Learning or Insignificant Change in a 
listener's knowledge status, which are caused by a 
discrepancy between a listener's belief in a proposition 
(possibly as a result of an inference) and the belief the 
speaker intends him/her to have in this proposition. Let 
us first consider the Affect-related impairments. 
Confusion occurs when an inference decreases signifi- 
candy a listener's confidence in a previous belief, caus- 
ing a discomforting transition from a self-perception of 
possessing knowledge to one of increased uncertainty. In 
terms of our network model, Confusion takes place 
when the absolute value of an Anticipated Measure of 
Belief (AMB), obtained by combining the MB of a link 
with the MB of an inference regarding this link, is signi- 
ficantly lower than the absolute value of the original 
MB of this link. For instance, the statement "One can- 
not always add algebraic terms," which yields a nega- 
tive value for the MB of the link \[+/- apply-to AI'\] in 
Figure 2, may trigger the erroneous inference \[+/- 
-apply-to LT\], which contradicts the link \[+/- apply-to 
LT\], thereby lowering its MB. The Strength of this 
impairment for a link L is defined as follows: 
Strength (Confusion ,L ) = 
max/( I MB (L ) I -lAMB (L ) I ),0} 
Loss o/Interest occurs when a listener who is initially 
motivated to acquire knowledge is presented with an IM 
s/he considers redundant. In terms of our model, this 
takes place if there exists a node B which subsumes a 
new node A, i.e., new distinguishing links incident upon 
A are connected to the same nodes and have MBs of 
compatible magnitude and sign as the corresponding 
158 
links incident upon B. For this type of impairment, a 
culprit link is an erroneous link incident upon B 
representing belief, whose reversal into disbelief results 
in B no longer subsuming A. This situation is illustrated 
in Figure 2, where we try to add the node DL, represent- 
ing distributive law, and the links \[DL apply-to AT\] and \[DL has-goal BrE\] to the network representing a 
listener's beliefs. However, the existence of the culprit 
link \[BrS apply-to AT\] supports the erroneous belief that 
bracket simplificauon is equivalent to distributive law, 
thereby rendering the new procedure redundant. If all 
the existing links participating in an impainnent due to 
Loss of Interest are correct, no culprit link is identified. 
A characterization of Belief-related impairments must 
take into consideration the difference between a 
listener's level of expertise and a level of expertise con- 
sidered satisfactory. To this effect we define the 
Strength of an impairment I in link L as follows: 
(-1,1) 
:ii 5:i! 
(-I,.i) 
kAMB 
• ~ ~ ~ • ========================="'"'"" 
• • • s s • -.::::::::::::::::: 
s% . s%s%s ...:!: 
,• • • / • s s • 
•ssssss~• 
•ssssssss 
(I ,i) 
SMB 
iT (1 ,-1) 
IT 
Strength (I ,L ) = 
max{(SMB (L )-AMB (L )),O} 
if SMB (L)>O 
I min{(SMB (L)-AMB (L)),O} \[ 
if SMB (L )<0 
where SMB is a Satisfactory Measure of Belief represen- 
tative of an adequate level of expertise with respect to a 
link. Its value may be obtained from a network which 
represents the speaker's beliefs. The relative position of 
Belief-related impairments and Confusion in a listener's 
belief space is graphically represented in Figure 4 which 
depicts these impairments as a function of the AMB and 
the previous MB of a link with SMB >0. The diagram 
for a link with SMB <0 is syrnmelric to the one in Fig- 
ure 4. 
Mislearning takes place when an erroneous belief 
with a relatively high degree of certainty is produced by 
an incorrect inference drawn by a listener. In terms of 
our network model, this takes place when the AMB of a 
link represents a substantial incorrect belief, and if this 
link existed previously, the strength of the impairment 
has increased, i.e., the AMB of this link is farther than 
its previous MB from its SMB. If the absolute value of 
the AMB of the link in question is higher than the abso- 
lute value of its previous MB, the listener will falsely 
perceive him/herself as being more proficient. 
Insufficient Learning occurs when a correct inference 
yields a correct belief with a relatively high degree of 
certainty, but which still falls short of a desired degree 
of certainty representative of proficiency. In terms of 
our network model, this occurs when the AMB of a link 
represents a substantial correct belief, and ff this link 
existed previously, the strength of the impairment has 
decreased, i.e., the AMB of this link is closer than its 
previous MB to its SMB. 
Finally, an Insignificant Change in a listener's 
knowledge status occurs when an inference accom- 
ptishes a rather inconsequential change with respect to a 
previously non-existent link or with respect to a link 
No impairment (NI) Q Confusion (C) 
Q Mislearning (ML) Q Insufficient Learning (IL) 
Q Insignificant Change (IC) 
Fig. 4: Characterization of Impairments for 
a Link with SMB>0 
with an MB representative of insufficient proficiency. 
This MB may represent either a correct belief or an 
incorrect belief. 
The immediate invalidation of Affect-related impair- 
ments is essential for the smooth continuation of the 
knowledge acquisition process, since their persistence 
diverts a listener's mental resources from the task of 
acquiring knowledge. On the other hand, the invalidation 
of Belief-related impairments with respect to links which 
are not currently in focus may be temporarily postponed 
due to didactic or stylistic considerations. Therefore, 
although Confusion may take place in conjunction with 
a Belief-related impairment, the recognition and subse- 
quent invalidation of Confusion takes precedence over 
the detection of this impairment (see Figure 4). 
Selection of Rhetorical Devices 
The Selection mechanism proposes a Peripheral RD to 
address a recognized impairment. The type of this rhe- 
torical device, its wording, and its position in the final 
message sequence depend on the type of the impairment, 
the correctness and magnitude of the previous MB of 
the culprit link, and the SMB of this link (see Table I). 
The strength of an impairment affects the need for addi- 
tional explanations, such as Causal explanations and 
Instantiations, to convey a Peripheral RD. 
159 
Table 1: Peripheral RDs as a Function of 
Impair.ment Types and Link Values 
Impairment 
Type 
Loss of 
Interest 
Confusion 
Mislearnin 8 
Insufficient 
Learning 
Insignificant 
Change 
Link Value Peripheral RD 
Correct Motivate (add links', 
Incorrect Contradict link 
Correct Revise link 
Incorrect Contradict link 
I ..... -- Contradict inference 
-- Revise inference 
High MB (Correct) Revise link 
High MB (Incorrect) Contradict link 
\[ Low MB (SMB >0) Revise information 
Low MB (SMB <0) 
The Link Value column in Table 1 contains informa- 
tion pertaining to the previous MB and the SMB of the 
culprit link. This information is unnecessary when 
addressing Mislearning and Insufficient Learning, since 
these impairments are completely characterized by their 
type. However, for the rest of the impairments, the 
correctness of the previous MB is the main factor in the 
determination of the type of a Peripheral RD. Loss of 
Interest is invalidated by a Contradiction of a culprit 
link, if such a link is identified; otherwise, a Motivation 
has to generated by adding nodes and links which render 
the new information non-redundant, e.g., "You already 
know how to solve quadratic equations by completion to 
square. A faster method is ... " (demil~ regarding the 
types of Motivations which are suitable for different 
situations and users appear in \[Zukerman 1987\]). The 
invalidation of an Insignificant Change in a link with an 
MB representative of lack of expertise, i.e., an MB 
whose absolute value is close to 0, is performed in a 
manner similar to the correction of complete ignorance. 
That is, if the SMB of this link is positive, this impair- 
ment is invalidated by a Revision of the relevant infor- 
marion, whereas ff it is negative, i.e., representative of 
disbelief, no Peripheral RD is proposed, since it may be 
superfluous to induce disbelief with respect to a proposi- 
tion which is hardly entertained by a listener. 
As seen in Table 1, we distinguish between two types 
of Peripheral RDs according to the source of the belief 
being addressed, namely Peripheral RDs which address 
previously existing links and Peripheral RDs which 
address current inferences. This distinction is not always 
reflected in the English realization of a rhetorical device, 
rather, it may affect the Meta Comments which accom- 
pany it and its position in the final message sequence. 
For instance, the propagation of the Contradiction \[BrS 
~apply-to AT\] may yield the erroneous inference \[BrS 
--,aBly-to LT\] which in turn may cause an impairment in 
the link \[BrS apply-to LT\]. If the anticipated impairment 
is Confusion, it will be invalidated by a Revision of this 
link, whereas if it is Mislearning, it will be invalidated 
by a Contradiction of the erroneous hfference. The most 
succinct realization of both Peripheral RDs is the 
sentence "You can always simplify bracketed Like 
Terms." However, the Revision of the previous belief 
may appear either before or after the above Contradic- 
tion and include a Meta Comment which states the 
source of this belief, e.g., "As we saw in Section 7, you 
can always simplify bracketed Like Terms"; while the 
Contradiction of the erroneous inference would usually 
appear after the above Contradiction and would be 
accompanied by a Meta Comment which indicates a vio- 
lation of an expectation established by the inference, 
e.g., "Bracket simplification does not always apply to 
algebraic terms, but it always applies to Like Terms." 
The Peripheral RDs proposed by our Selection mechan- 
ism contain sufficient information to support the genera- 
tion of these types of Meta Comments by means of 
mechanisms such as the ones presented in \[Zukerman 
and Pearl 1986\] and \[Zukerman 1989\]. 
Procedure Impairment-Invalidate 
Impairment-Invalidate is activated with one argument 
which contains an IM. It returns a Message-List com- 
posed of the IM and the Peripheral RDs which were 
proposed to invalidate the impairments anticipated as a 
result of this message. 
Procedure Impairment-Invalidate(Message) 
I Message-List ~- Message 
2 Peripheral-RDsJnferences ~- nil 
3 Firstlmp ~- (Message ,TM ,SM ,LM), 
where (TM ,SM ,LM) ~ Recognize(Message) 
4 If Firsdmp Then 
Message-List 
Append (Message-List ,Select (Firstlmp )) 
5 For each m E Message-List do 
6 Inferences ~- Merge (Inferences ,Propagate (m)) 
7 endfor 
8 Impairments ~ {(i,Ti,Si,Li) I (i Elnferences) ^ 
( (Ti ,Si ,Li )~-Recognize (i ) )} 
9 While Impairments do 
10 Maxlmp ~ (I,T/,St,L/), 
where {(I ,Tt ,St fit )~ Impairments ^ 
Ranking (I ,Tt ,st ,Lt ) = 
max {Ranking(j,Ts,sj,Li)}} q ,Tj ~ l ~ j )e t, nt,ar~,at 
11 Impairments ~ Impairments -- Maxlmp 
12 RDM=t,~, ~- Select (Maxlmp ) 
13 Peripheral-RDs 
Append (Peripheral-RDs ,RDMaa,,~, ) 
14 Secondary-Inferences ~-- Propagate (RD~=a,~,) 
15 Secondary-Effects ~ Secondary-lnferences c~ 
{j I (j ,T j ,S j ,Lj )~ Impairments} 
16 Impairments ~-- {Impairments - 
{(i ,Ti ,si ,Li ) l i E Secondary-Effects }}k. ) 
{(i ,Ti ,si ,Li ) I(i ~ Secondary-Inferences) ^ 
( (Ti ,Si fii )+--Recognize ( i ) )} 
17 endwhile 
18 Message-List 
Append (Message-List ,Peripheral-RDs ) 
160 
Table 2: Inferences and Possible Impairments after Propagation 
Message 
IM \[DL apply-to 
AT has-goal BrE\] 
Contradiction 
\[BrS -~apply-to AT\] 
of the 1M and 
Rule 
Similarity 
Specialization 
I SpecialiT~fion 
Similarity 
Specialization 
Specialization 
Deduction 
the Contradiction in the Sample Network 
Inference \[ Possible Impairment 
\[DL apply-to Numbers has-goal BrE\]\] 
\[DL apply-to LT has-goal BrE\] ~ Insufficient Learning 
\[DL apply-to UT has-goal BrE\] J 
\[BrS -~apply-to Numbers\] 
\[BrS -~apply-to LT\] 
\[BrS -~apply-to UT\] 
\[+/- -~apply-to AT\] 
\] ConfusionlMislearningl 
J Insignificant Change 
\] Confusion/Inst. Learning/ 
J Insignificant Change 
We distinguish between two main stages of this pro- 
cedure: (1) The preliminary stage (lines 1-4) which 
determines the need for a Peripheral RD related to the 
input message, and (2) The iterative stage (lines 5-17) 
which ascertains the need for Peripheral RDs pertaining 
to subsequent inferences. 
In the preliminary stage, procedure Recognize is 
applied in order to determine whether the IM is likely to 
cause an impairment. If this is the case, Recognition 
returns a triple (T,S,L), where T contains the Type of 
the impairment, S its Strength, and L indicates whether 
the belief represented by the link in question is correct 
or incorrect. Otherwise, it returns n/l and no impairment 
is predicted. If an impairment was anticipated, procedure 
Select is activated to propose a Peripheral RD. In our 
example, the Selection mechanism invalidates the 
impairment responsible for the Loss of Interest by 
means of the Contradiction \[BrS -,apply-to AT\] which 
induces disbelief in the link \[BrS apply-to AT\]. Upon 
completion of the preliminary stage, the proposed Peri- 
pheral RD together with the IM form the Message-List, 
which constitutes the input to the next phase of the 
impairment invalidation procedure. 
Both the IM and the proposed Peripheral RD cause 
modifications in a listener's degree of belief in the 
addressed links. These modifications in turn may lead to 
changes in his/her beliefs in other links. Hence, in order 
to prevent impairments due to inferences drawn from 
these messages, additional Peripheral RDs may be called 
for. Furthermore, Peripheral RDs may be required in 
order to invalidate an Insignificant Change in links 
related to the IM which have MBs representative of 
insufficient proficiency, i.e., to attain a satisfactory 
degree of expertise with respect to these links. To deter- 
mine the need for additional Peripheral RDs, Propagate 
produces inferences from each message in Message-List. 
During this process, inferences from different messages 
which affect the same link are merged into one inference 
with a combined effect (lines 5-7). Recognition then 
ascertains the attributes of the impairments which are 
likely to be caused by these inferences (line 8). For 
instance, the Propagation of the input message \[DL 
apply-to AT has-goal BrE\] and the Contradiction \[BrS 
-,apply-to AT\] in the sample network in Figure 2 may 
yield the inferences in Table 2. In principle, each of 
these inferences may be responsible for an impairment. 
However, as stated above, the effect of these inferences 
ultimately depends on the p-s of the rules applied in the 
propagation process and on the MBs of the affected 
links. In our present discussion, we assume that Confu- 
sion was recognized in the links \[BrS apply-to LT\] and 
\[BrS apply-to Numbers\]. 
Based on Gricean maxims of cooperative conversation 
(Grice 1975), we propose to generate a minimal number 
of rhetorical devices to invalidate impairments occurring 
concurrently in a number of links. To this end, we 
iterate over the set of impairments, selecting at each 
stage the culprit link with the highest ranking impair- 
ment (line 10). The impairments are ranked according to 
their type and strength, where impairments causing Con- 
fusion are ranked higher than Belief-related impairments. 
A Peripheral RD is then proposed to invalidate the 
impairment in the selected link, and its effect is pro- 
pagated (lines 12-14). Once a Peripheral RD has been 
generated to address a particular link, further inferences 
drawn during the same activation of Impairment- 
Invalidate no longer affect this link. For each iterauon, 
the set of impairments is updated by merging the infer- 
enccs responsible for the previously recognized impair- 
ments with the inferences resulting from the most recent 
propagation, and reapplying the Recognition process 
0ines 15-16). In this manner, a low-ranking impairment 
in a given link may be spontaneously invalidated by an 
inference resulting from a Peripheral RD generated to 
invalidate an impairment in another link. For example, 
ff the impairment in the link \[BrS apply-to Numbers\] 
ranks higher than the impairment in the link \[BrS apply- 
to LT\], a Revision is generated to invalidate the impair- 
ment in the former link. The propagation of this Revi- 
sion may invalidate the Confusion with respect to the 
latter link. If this Revision does not cause further 
impairments, the procedure terminates returning the fol- 
lowing messages: 
IM \[DL apply4o AT has-goal BrE\] 
Contradiction \[BrS -~apply-to AT\] 
Revision \[BrS apply-to Numbers has-goal BrE\] 
The generation of belief Revisions tends to inhibit 
further impairments, since they reinforce beliefs which 
are likely to be consistent with the rest of a listener's 
beliefs, while the generation of belief Contradictions 
tends to foster impairments, since they disagree with 
161 
existing beliefs. However, in a knowledge acquisition 
setting, misconceptions are generally not allowed to pile 
up, hence Contradictions to existing beliefs are expected 
only during the initial stages of the propagation process. 
Therefore, although in principle our algorithm may 
iterate indefinitely, in practice, impairments should no 
longer be detected after a few iteraations, and the process 
should halt after proposing a few Peripheral RDs. This 
expectation is confirmed by tests run with our sample 
network (see next section). However, this network is 
rather small, and there may be situations in which the 
number of Peripheral RDs proposed by our mechanism 
will have to be restricted due to stylistic or pedagogical 
considerations. In such cases, our mechanism will have 
to be adjusted to invalidate only a subset of the recog- 
nized impairments such as the highest ranking impair- 
ments or impairments in links which are closest to the 
IM. 
Evaluation 
Procedure Impairment-Invalidate has been implemented 
in a system called WISHFUL, which was run on several 
instances of the network in Figure 2. These instances 
featured MBs which represented three types of students: 
(1) Competent -- with high MBs associated with correct 
beliefs, (2) Average- containing some incorrect beliefs 
and medium-range MBs associated with correct beliefs, 
and (3) Mediocre -- with very low MBs associated with 
most beliefs. The initial values assigned to the MBs of 
the finks and the p-s of the Common-sense Inference 
Rules yielded Peripheral RDs which were compatible 
with rhetorical devices people would generate under 
similar circumstances, and the response time was instan- 
taneous for these rather small networks. As expected, 
changes in the p-s of the Common-sense Inference 
Rules resulted in variations in the generated RDs, with 
additional Contradictions due to Misleaming being gen- 
erated as the p-s of unsound rules increased, and addi- 
tional Revisions due to Insufficient Learning as the p-s 
of sound rules decreased. In addition, the number and 
strength of the recognized impairments increased as the 
ability of the students being represented in a network 
decreased, iadicating that additional explanations would 
be required to convey a message to the poorer students. 
In particular, for the networks representing competent 
students, either Revisions of inferences due to Insuffi- 
cient Learning or no Peripheral RDs were proposed 
(depending on the p-s of the Common-sense Inference 
Rules); for the networks representing average students, 
Contradictions and Revisions were proposed in a manner 
similar to the explanations in this paper, and for the net- 
works representing mediocre students most of the pro- 
posed Peripheral RDs were Revisions of information. 
Our mechanism has also proven successful as an 
analytical tool. It accounts for the presence of Peripheral 
RDs in over twenty texts in a variety of domains, rang- 
ing from expert domains (Telecommunications, Cogni- 
tive Science and Linguistics) through intermediate ones 
(high-school Algebra, Data Structures, Lisp and intro- 
ductory Chess) to novice domains (Childcraft Encyclo- 
pedia and Dr. Spock's Baby and Child Care). 
Limitations and Future Research 
Our mechanism proposes rhetorical devices under the 
assumption that after it has done "its best," the 
listener's beliefs addressed by the discourse will be 
modified in the desired direction, although not neces- 
sarily to the desired extent. This is a valid assumption 
for discourse generation, since one can not say more 
than one knows. However, for this mechanism to gen- 
erate effective discourse on a continued basis, the model 
of the fistener's beliefs must be updated by an indepen- 
dent assessment of the his/her understanding. 
In addition, our mechanism must be implemented on 
different domains and larger networks to test both its 
response time and the adequacy and number of the pro- 
posed rhetorical devices. As stated above, this may 
reveal the need to adjust procedure Impairment- 
Invalidate to enable it to control the number of the pro- 
posed Peripheral RDs. 
At present, research is in progress to extend the 
impairment invalidation paradigm to the generation of 
other types of rhetorical devices, such as Descriptions, 
Instantiations and Causal explanations, and to devise an 
algorithm to sort the generated messages according to 
rhetorical considerations (Zukerman 1990b). In addition, 
an alternative representation for MBs which keeps track 
of the sources of an inference is being considered. 
Further research is required to recognize and rectify pos- 
sible misconceptions in the Common-sense Inference 
Rules, and to characterize conditions for the generation 
of rhetorical devices which satisfy a number of com- 
municative goals. Finally, the effect of the knowledge 
representation and the rules of inference on the types of 
the proposed rhetorical devices needs to be further 
investigated. 
Conclusion 
This paper offers a text planning mechanism which sup- 
ports the generation of explanations tailored to particular 
types of users. Our mechanism generates Peripheral RDs 
which help convey an intended message by anticipating 
and preventing potential impairments to a listener's 
comprehension process. To this effect, it characterizes 
these impairments in terms of a model of a fistener's 
beliefs and inferences, and simulates a fistener's 
comprehension process on this model. Clearly, this pro- 
tess does not dispense with the need to interact with a 
listener, but it addresses commonly occurring impair- 
ments, thereby focusing the interaction. Furthermore, it 
is envisioned that the impairment invalidation mechan- 
ism will become a useful tool to guide the generation of 
cogent responses to user follow-up queries, as it can 
point to issues which are potentially troublesome. 
162 

References 
\[Anderson, 1983\] J.R. Anderson. The Architecture of 
Cognition, Harvard University Press, Cambridge, 
Massachusetts. 
\[Appelt, 1982\] D.E. Appelt. Planning Natural Language 
Utterances to Satisfy Multiple Goals. Technical Note 
259, SRI International, March 1982. 
\[Brown and Van Lehn, 1980\] J.S. Brown and K. Van 
Lehn. Repair Theory: A Generative Theory of Bugs 
in Procedural Skills. In Cognitive Science 4, pp. 379- 
426. 
\[Buchanan and Shortliffe, 1985\] B.G. Buchanan and 
E.H. Shortliffe. Rule-Based Expert Systems ~ The 
MYCIN Experiments of the Stanford Heuristic Pro- 
gramming Project, Addison-Wesley Publishing Com- 
pany. 
\[Grice, 1975\] H.P. Grice. Logic and Conversation. In 
PJ. Cole and J.L. Morgan (Eds.), Syntax and Seman- 
tics, Volume 3: Speech Acts, Academic Press, pp. 41- 
58. 
\[Hovy, 1987\] E.H. Hovy. Generating Natural Language 
under Pragmatic Constraints. Doctoral Dissertation, 
Computer Science Department, Yale University, New 
Haven, Connecticut. 
\[Hovy, 1988\] E.H. Hovy. Planning Coherent Multisen- 
tential Text. In Proceedings of the Twenty-Sixth 
Annual Meeting of the Association for Computational 
Linguistics, State University of New York, Buffalo, 
New York. 
\[Kukieh, 1983\] K. Kukieh. Knowledge-Based Report 
Generation: A Knowledge-Engineering Approach to 
Natural Language Report Generation. Doctoral 
Dissertation, The Interdisciplinary Department of 
Information Science, University of Pittsburgh, 
Pennsylvania. 
\[Lynch et al., 1979\] B.J. Lynch, R.E. Parr and H.M. 
Keating. Maths 8, Sorrett Publishing. 
\[Matz, 1982\] M. Matz. Towards a Process Model for 
High School Algebra Errors. In D. Sleeman and J.S. 
Brown (Eds.), Intelligent Tutoring Systems, London: 
Academic Press, 1982, pp. 25-50. 
\[MeKeown, 1985\] K.R. MeKeown. Discourse Strategies 
for Generating Natural Language Text. In Artificial 
Intelligence 27, pp. 1-41. 
\[Moore and Swartout, 1989\] J.D. Moore and W.R. 
Swartout. A Reactive Approach to Explanation. In 
IJCAI-11 Proceedings, International Joint Conference 
on Artificial Intelligence, pp. 1504-1510. 
\[Paris, 1988\] C.L. Paris. Tailoring Object Descriptions 
to a User's Level of Expertise. In Computational 
Linguistics, Volume 14, Number 3, September 1988, 
pp. 64-78. 
\[Sleeman, 1984\] D. Sleeman. Mis-Generalizafion: An 
Explanation of Observed Mal-rules. In Proceedings of 
the Sixth Annual Conference of the Cognitive Science 
Society, pp. 51-56. 
\[Van Lehn, 1983\] K. Van Lehn. Human Procedural Skill 
Acquisition: Theory, Model and Psychological Valida- 
tion. In AAAI-83 Proceedings, American Association 
for Artificial Intelligence, August 1983, pp. 420-423. 
\[Zukerrnan and Pearl, 1986\] I. Zukerman and J. Pearl. 
Comprehension-Driven Generation of Meta-technical 
Utterances in Math Tutoring. In AAAI-86 Proceed- 
ings, American Association for Artificial Intelligence, 
pp. 606-611. 
\[Zukerman, 1987\] I. Zukerman, Goal-based Generation 
of Motivational Expressions in a Learning Environ- 
merit. In AAAI-87 Proceedings, American Association 
for Artificial Intelligence, July 1987, pp. 327-331. 
\[Zukerman and Cheong, 1988\] I. Zukerman and Y.H. 
Cheong. Impairment Invalidation: A Computational 
Model for the Generation of Rhetorical Devices. In 
Proceedings of the International Computer Science 
Conference '88: Artificial Intelligence, Theory and 
Applications, pp. 294-300. 
\[Zukerman, 1989\] I. Zukerman. Expectation Verifica- 
tion: A Mechanism for the Generation of Meta Com- 
ments. In Proceedings of the 11th Conference of the 
Cognitive Science Society, Ann Arbor, Michigan, pp. 
498-505. 
\[Zukerman, 1990a\] I. Zukerman. A Predictive Approach 
to the Generation of Rhetorical Devices. To appear in 
Computational Intelligence, Vol 6, No. 1. 
\[Zukerman, 1990b\] I. Zukerman. Anticipating a 
Listener's Response in Text Planning. To appear in 
Advances in Artificial Intelligence, Natural Language 
and Knowledge-based Systems, Springer Verlag. 
