Humanlike dialogue agents
The paper deals with an investigation and modeling of the inner mechanisms of human-human dialogue, with the final goal to construct an artificial humanlike dialogue agent. An ontological basis of human dialogue interaction, underlying follow-up reasoning, is obtained on the basis of analysis of hum...
Saved in:
| Published in: | Математичне та комп'ютерне моделювання. Серія: Технічні науки |
|---|---|
| Date: | 2014 |
| Main Author: | |
| Format: | Article |
| Language: | English |
| Published: |
Інститут кібернетики ім. В.М. Глушкова НАН України
2014
|
| Online Access: | https://nasplib.isofts.kiev.ua/handle/123456789/86456 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Journal Title: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Cite this: | Humanlike dialogue agents / I.A. Chimir // Математичне та комп'ютерне моделювання. Серія: Технічні науки: зб. наук. пр. — Кам’янець-Подільський: Кам'янець-Подільськ. нац. ун-т, 2014. — Вип. 11. — С. 163-186. — Бібліогр.: 17 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| _version_ | 1859909587633700864 |
|---|---|
| author | Chimir, I.A. |
| author_facet | Chimir, I.A. |
| citation_txt | Humanlike dialogue agents / I.A. Chimir // Математичне та комп'ютерне моделювання. Серія: Технічні науки: зб. наук. пр. — Кам’янець-Подільський: Кам'янець-Подільськ. нац. ун-т, 2014. — Вип. 11. — С. 163-186. — Бібліогр.: 17 назв. — англ. |
| collection | DSpace DC |
| container_title | Математичне та комп'ютерне моделювання. Серія: Технічні науки |
| description | The paper deals with an investigation and modeling of the inner mechanisms of human-human dialogue, with the final goal to construct an artificial humanlike dialogue agent. An ontological basis of human dialogue interaction, underlying follow-up reasoning, is obtained on the basis of analysis of human-human dialogues and illustrated by one of Plato’s dialogue, Protagoras. The paper proposes two models directed to explain some inner mechanisms of human-human dialogue. Both models are synthesized for the case of one type of human-human dialogues, called erotetic dialogue. The model of the structure of dialogue transaction is created in the context of a declarative knowledge representation, based on Language of Ternary Description. The model for the cognitive cycle of dialogue interaction is in fact a development of Neisser’s cyclical model of perception.
Стаття присвячена дослідженню та моделюванню внутрішніх механізмів діалогу між людьми з метою розробки штучного людиноподібного діалогового агента. Онтологічний базис діалогової взаємодії між людьми, на якому базуються дослідження, одержано шляхом аналізу діалогів між людьми і ілюструється одним з Платонівських діалогів під назвою Протагор. У статті запропоновані дві моделі, які пояснюють внутрішні механізми діалогу між людьми. Обидві моделі синтезовані для винятку одного типу діалогу між людьми під назвою еротетичний діалог. Модель структури діалогової трансакції синтезована у контексті представлення декларативних знань за допомогою Мови Тернарного Опису. Модель когнітивного циклу діалогової взаємодії є розвитком циклічної моделі перцепції Найсера.
|
| first_indexed | 2025-12-07T16:02:25Z |
| format | Article |
| fulltext |
Серія: Технічні науки. Випуск 11
163
UDC 004.94:004.78
I. A. Chimir, Ph. D., professor
Odessa State Environmental University, Odessa
HUMANLIKE DIALOGUE AGENTS
The paper deals with an investigation and modeling of the inner
mechanisms of human-human dialogue, with the final goal to construct
an artificial humanlike dialogue agent. An ontological basis of human
dialogue interaction, underlying follow-up reasoning, is obtained on
the basis of analysis of human-human dialogues and illustrated by one
of Plato’s dialogue, Protagoras. The paper proposes two models di-
rected to explain some inner mechanisms of human-human dialogue.
Both models are synthesized for the case of one type of human-human
dialogues, called erotetic dialogue. The model of the structure of dialo-
gue transaction is created in the context of a declarative knowledge re-
presentation, based on Language of Ternary Description. The model
for the cognitive cycle of dialogue interaction is in fact a development
of Neisser’s cyclical model of perception.
Key words: natural dialogue, erotetic dialogue, dialogue transac-
tion, declarative knowledge, Language of Ternary Description.
1. Introduction. Dialogue is an essential phenomenon of human mental
and intellectual activity and outwardly represented by a chain of interdepen-
dent information transactions generated during knowledge interchange be-
tween partners of dialogue interaction. Purposefulness is one of the fundamen-
tal characteristics of dialogue, and purposefulness necessarily implies a logical
dependency not only between a transaction’s components but between adja-
cent dialogue transactions as well. Analysis of practically any cognitive activi-
ty of humans reveals that its basis is a dialogue process. For instance, reason-
ing and inference, in broad and narrow interpretation, are, in fact, a «conversa-
tion with the self» or a dialogue interaction in which both dialogue partners are
represented by one and the same person.
Dialogue transaction serves as a means of knowledge interchange be-
tween dialogue partners and, therefore we can qualify dialogue processes as
knowledge-based processes and dialogue systems as knowledge-based sys-
tems. The vast majority of researchers explicitly or implicitly presuppose that
knowledge in dialogue transactions is represented in a form of propositions
and hence: (1) verbal examples of transactions and their elements prevail in
relevant publications, and (2) notation of First Order Logic uses, as a rule, for
formal representation of propositions [1; 2]. Such orientation on propositional
understanding and modeling of dialogue transactions determines by a tradi-
tional way of knowledge representation in this area of research and from our
point of view must not be considered as the only possible way.
© I. A. Chimir, 2014
Математичне та комп’ютерне моделювання
164
The bulk of journal publications with the word combinations «dialo-
gue process» and «dialogue agent» reflect a purely pragmatic attitude to-
wards the study of dialogues. Usually the goal of these publications is not
a deeper understanding of the phenomenon, but rather constructing various
artificial systems that are able to work as a verbal interface between end-
users and applied computer systems. For instance, searching the Web ge-
nerates more than seven million references to Web-pages which include
the word combination «dialogue agent» and describe systems realizing
mainly communicative functions.
We believe that formal modeling of the dialogue process must begin
from detailed investigation of natural dialogues in all their diversity. We
also believe that if models of dialogue agents rely on fundamental prin-
ciples of human dialogue, then we can expect that artificial dialogue
agents will inherit rich cognitive abilities of natural human-human dialo-
gue process. An artificial dialogue agent can be considered as such if it is
able to maintain the dialogue, with a degree of complexity commensurable
with the complexity of natural dialogue.
The final goal of this article is to create formal models of dialogue trans-
actions, oriented mainly on more comprehensive understanding of the inner
logical nature of the dialogue, but at the same time definite enough to serve as
a framework for artificial dialogue agents engineering. As a first step towards
achieving this goal we found it necessary to develop an initial ontological
model of natural dialogue. Further speculations completely rest upon this on-
tological model and are oriented towards synthesis of formal models of classes
of declarative knowledge, associated with question-answering pairs based on
ideas of Uyemov’s Language of Ternary Description [3–5].
2. Essential ontology. Investigations related to dialogue interactions
can be found in different and often not close scientific areas. The list,
which is probably incomplete, includes Artificial Intelligence, Cognitive
Psychology, Erotetic Logic, and Epistemology.
One of the well-known confirmations of importance of the dialogue
process, which we find in Artificial Intelligence, is a test, offered by Alan
Turing [6] for operational verification of intelligence in technical systems,
which is completely based on a dialogue between human and a technical
system. Modern exposition of Artificial Intelligence, integrates all main
ideas in an area into a coherent subject, based on the conception of intelli-
gent agents [7]. Intelligent agent is dialogue agent. It can’t exist outside its
task environment, and transactions between the task environment and the
intelligent agent determine behavior of the latter.
Piaget’s theory of cognitive development, which usually ascribes to the
areas of Cognitive Psychology and Epistemology implicitly presupposes that
a «moving force» for schemata development is dialogue [8]. If human be-
Серія: Технічні науки. Випуск 11
165
ings come into the world with a certain number of innate schemata, needed
for survival during the initial period of life, and that acquired knowledge
determines by further development and growth of these schemata, then me-
thod, which realizes evolution of innate schemata is the dialogue process. In
the area of Cognitive Psychology there are some more theories and models
that also make contribution to understanding the essence of the dialogue. For
example, the cyclical model of perception offered by Neisser [9] is, in fact, a
model of the dialogue process if the source of a flow of sensory events is
considered as one of the partners of the dialogue.
Since dialogue transactions, in all their polymorphity, serve as a
means of knowledge interchange between dialogue partners and the ve-
hicle of this interchange is an inquiry-response pair, then results obtained
in Question-Answering Logic [10] are applicable to the study of dialogue
transaction logical structure.
Epistemology has direct relation to investigation of the nature of the
dialogue process, as this science is directed mainly on study and modeling
of those types of knowledge, which circulated within dialogue transactions
and are essential for dialogue partners. Especially interesting is the version
being called Socratic Epistemology, where separate question-answering
pairs and question-answering dialogue play the key role [11].
To build formal models, designated as a final goal of the article, we be-
gan from an ontological model of natural dialogue, constructed on the basis of
analysis of a certain quantity of protocols of real dialogue processes with all
their logical and operational diversity. There is a great number of such proto-
cols. For instance, protocols of crime suspect interrogation in which an inves-
tigator is trying to prove the guilt of a suspect and the latter is trying to con-
vince the investigator of his innocence. However, a more accessible and rich
source of natural dialogue protocols is dialogical literature, especially from the
areas of Philosophy and Theology. One of the most significant sources of this
kind is Plato’s dialogues [12]. It is reasonable to begin forming an ontological
model of natural dialogue from analysis of a small fragment of one of Plato’s
dialogue called Protagoras. The protocol, given below, presented in the form,
slightly differs from the original text. Insignificant changes and additions have
been included into the dialogue to make it more structured. For example, we
marked beginning and ending of each transaction and definitely indicated
partners of the dialogue. The dialogue occurs between Socrates (S) and Hip-
pocrates (H) before their conversation with Protagoras. Terms «active» and
«reactive» will be explained in the following parts of the article.
Beginning of transaction 1. S is active.
S: Then we are going to pay our money to Protago-
ras in the character of a Sophist?
H: Certainly.
Математичне та комп’ютерне моделювання
166
End of transaction 1.
S: But you should not assume, Hippocrates, that
the instruction of Protagoras is of this na-
ture: may you not learn of him in the same way
that you learned the arts of the grammarian,
musician, or trainer, not with the view of mak-
ing any of them a profession, but only as a
part of education, and because a private gen-
tleman and freeman ought to know them?
H: Just so and that, in my opinion, is a far truer
account of the teaching of Protagoras.
End of transaction 2.
S: I wonder whether you know what you are doing?
H: And what am I doing?
End of transaction 3. Roles interchange. S becomes reactive.
S: You are going to commit your soul to the care of
a man whom you call a Sophist. And yet I hardly
think that you know what a Sophist is; and if
not, then you do not even know to whom you are
committing your soul and whether the thing to
which you commit yourself be good or evil.
H: I certainly think that I do know.
End of transaction 4. Roles interchange. S becomes active.
S: Then tell me, what do you imagine that he is?
H: I take him to be one who knows wise things as
his name implies.
End of transaction 5.
S: And might you not affirm this of the painter and
of the carpenter also: Do not they, too, know
wise things? But suppose a person were to ask us:
In what are the painters wise? We should answer:
In what relates to the making of likenesses, and
similarly of other things. And if he were further
to ask: What is the wisdom of the Sophist, and
what is the manufacture over which he presides? —
how should we answer him?
H: How should we answer him, Socrates? What other
answer could there be but that he presides over
the art which makes men eloquent?
End of transaction 6. Roles interchange. S becomes reactive.
S: Yes that is very likely true, but not enough;
for in the answer a further question is in-
Серія: Технічні науки. Випуск 11
167
volved: Of what does the Sophist make a man
talk eloquently? The player on the lyre may be
supposed to make a man talk eloquently about
that which he makes him understand, that is
about playing the lyre. Is not that true?
H: Yes.
End of transaction 7. Roles interchange. S becomes active.
S: Then about what does the Sophist make him elo-
quent? Must not he make him eloquent in that
which he understands?
H: Yes, that may be assumed.
End of transaction 8.
S: And what is that which the Sophist knows and
makes his disciple know?
H: Indeed I cannot tell.
End of transaction 9.
S: Well, but are you aware of the danger which you
are incurring? If you were going to commit your
body to someone, who might do good or harm to
it, would you not carefully consider and ask the
opinion of your friends and kindred, and delibe-
rate many days as to whether you should give him
the care of your body? But when the soul is in
question, which you hold to be of far more value
than the body, and upon the good or evil of
which depends the well-being of your all,-about
this never consulted either with your father or
with your brother or with any one of us who are
your companions. But no sooner does this fo-
reigner appear, than you instantly commit your
soul to his keeping. In the evening, as you say,
you hear of him, and in the morning you go to
him, never deliberating or taking the opinion of
any one as to whether you ought to intrust your-
self to him or not;-you have quite made up your
mind that you will at all hazards be a pupil of
Protagoras, and are prepared to expend all the
property of yourself and of your friends in car-
rying out at any price this determination, al-
though, as you admit, you do not know him, and
have never spoken with him: and you call him a
Математичне та комп’ютерне моделювання
168
Sophist, but are manifestly ignorant of what a
Sophist is; and yet you are going to commit
yourself to his keeping.
H: No other inference, Socrates, can be drawn from
your words.
End of transaction 10.
S: Is not a Sophist, Hippocrates, one who deals
wholesale or retail in the food of the soul? To
me that appears to be his nature.
H: And what, Socrates, is the food of the soul?
End of transaction 11. Roles interchange. S becomes reactive.
Dialogue agents create dialogue transaction. Although an arbitrary
number of members can take part in the dialogue, dialogue transaction is
always created only by a pair of dialogue agents. In Plato’s dialogue Pro-
tagoras, in different time points, transactions are created by the following
pairs of agents: Socrates-Hippocrates, Socrates-Protagoras, etc.
Dialogue agents, in the process of dialogue interaction, play one of
two roles: the role of the active dialogue agent or the role of the reactive
dialogue agent. An agent plays the active role in cases when he/she needs
some knowledge and supposes to get it from the opposite dialogue agent,
and, correspondingly, an agent plays the reactive role in cases when he/she
grants some knowledge, he/she possesses to the opposite dialogue agent.
In the fragment cited, Socrates, initially, plays the role of active agent
and Hippocrates — the role of reactive agent. They then changed their roles
several times. An analysis of the given fragment allows us to make the prelim-
inary conclusion that, as a rule, the initiator of role switch is the reactive agent.
The active agent is normally satisfied with his/her role and after role-
interchange tries to restore his/her active status at the first chance.
In the case when dialogue interaction occurs between two persons,
we will say that an outer dialogue takes place. However, dialogue transac-
tion can be formed by only one agent. In this case one agent alternately
plays both roles. Such a type of dialogue we will call an inner dialogue.
Verbal inner dialogues are often called monologues, and mental inner di-
alogues are thoughts.
In the case of an outer dialogue, knowledge, requested by the active
agent, is kept in the long-term memory of the reactive agent, and the know-
ledge to which the reactive agent refers in his answer — is in the long-term
memory of active agent. In the case of an inner dialogue, the requested and
returned knowledge are in the long-term memory of the same agent.
Most part of the dialogue given above illustrates the outer type of di-
alogue however it includes examples of inner dialogue as well. In eleventh
transaction Socrates carries on a dialogue with himself, and what he says
can be presented by the following way:
Серія: Технічні науки. Випуск 11
169
S(active): Is not a Sophist, Hippocrates, one
who deals wholesale or retail in the
food of the soul?
S(reactive): To me that appears to be his nature.
The outer manifestation of a dialogue is a chain of dialogue transac-
tions where each transaction is a complete cycle of knowledge interchange
between the active and reactive agents. The distinguishing feature of a di-
alogue, which distinguishes it from other forms of interactive communica-
tion (for instance, from the communication with an applied data base sys-
tem) is the presence of strong logical dependencies, not only within separate
transaction, but, what is more important, between links of the transaction
chain. Precisely because of such logical dependencies, a set of separate
transactions is transformed into a goal-oriented intellectual process.
Every dialogue transaction is represented by two information messages.
An information message from the active agent has the status of a question.
The word «status» here means that an information message from an active
agent is not necessarily a question in grammatical or linguistic sense. It can
take various forms. The form of an information message does not affect either
the goal of a single transaction or the goal of the dialogue as a whole. What is
really important for the active agent is an access to the requested knowledge,
but not the form of the request. So, an active agent can get access to the same
chunk of knowledge by means of a series of different information messages.
In several epistemic publications this natural free choice of the form of access
to the chunk of knowledge by the active agent is treated as a problem and
called «the problem of convergent knowledge» [13].
Analysis of natural dialogues shows that information messages of an
active agent can vary from simple yes/no questions to a long-lasting inner
dialogue. However, even when an information message of an active agent
is a long-lasting reasoning, it all the same has status of a question with
respect to the reactive agent. Therefore, we can consider the degree of in-
terrogativness of the active agent’s information message.
The information message of the active agent can be understood as a
search prescription needed to get access to the knowledge of the reactive
agent, or as a reference which provides an access to the section of the reactive
agent’s long-term memory. Similarly, the information message of the reactive
agent has the status of an answer and can be interpreted as a search prescrip-
tion needed to get access to the active agent’s knowledge, or as a reference to
a section in his/her long-term memory, where the required knowledge resides.
In the cited fragment of the Plato’s dialogue, one of the simplest
transactions, from the point of view of the structure of the information
message, is the transaction number one.
S: Then we are going to pay our money to Protago-
ras in the character of a Sophist?
H: Certainly.
Математичне та комп’ютерне моделювання
170
In this transaction the active agent Socrates transmits to the reactive
agent Hippocrates an information message to provide Socrates access to
the knowledge, possessed by Hippocrates and not possessed by Socrates.
Socrates does not know whether Hippocrates is going to pay money to
Protagoras only on the ground that Protagoras is Sophist. This information
message has a high degree of interrogativness because it is represented by
a purely interrogative sentence.
In the tenth transaction, the information message of Socrates, who
plays the role of the active agent, has a more complex structure. It is
represented by a certain preliminary reasoning conducted by Socrates but,
however, has some degree of interrogativness.
S: Well, but are you aware of the danger which you
are incurring? If you were going to commit your
body to someone, who might do good or harm to
it, would you not carefully consider and ask the
opinion of your friends and kindred, and delibe-
rate many days as to whether you should give him
the care of your body? But when the soul is in
question, which you hold to be of far more value
than the body, and upon the good or evil of
which depends the well-being of your all,-about
this never consulted either with your father or
with your brother or with any one of us who are
your companions. But no sooner does this fo-
reigner appear, than you instantly commit your
soul to his keeping. In the evening, as you say,
you hear of him, and in the morning you go to
him, never deliberating or taking the opinion of
any one as to whether you ought to intrust your-
self to him or not;-you have quite made up your
mind that you will at all hazards be a pupil of
Protagoras, and are prepared to expend all the
property of yourself and of your friends in car-
rying out at any price this determination, al-
though, as you admit, you do not know him, and
have never spoken with him: and you call him a
Sophist, but are manifestly ignorant of what a
Sophist is; and yet you are going to commit
yourself to his keeping.
H: No other inference, Socrates, can be drawn from
your words.
Socrates intends to know whether Hippocrates agrees with his conclu-
sion that someone cannot entrust his/her soul to a man with whom he/she is
not familiar and whom people call Sophist if someone doesn't know what this
Серія: Технічні науки. Випуск 11
171
word means, because his/her soul can be damaged just like the body. Socrates’
information message is an inner dialogue beginning with a question. Socrates
poses the question, then explicates and answers it. We can assume that So-
crates’ inner dialogue has the goal to specify requested knowledge and that
without this additional specification, his information message will be exces-
sively ambiguous and won’t get access to the knowledge he really needs.
So, we can state that dialogue transaction is a fundamental structural
element of dialogue interaction and in all cases has interrogative nature.
Dialogue is a goal-driven intellectual process. Each dialogue agent tends
to achieve his/her own goal and therefore generates flows of information mes-
sages, which conforms with their goals. From this point of view dialogue
processes are problem solving processes. It’s easy to notice the purposeful
nature of Socrates’ information messages in the cited fragment of Plato’s di-
alogue. The sequence of his messages directed to achieve the goal, which we
could formulate, for instance, as to convince Hippocrates that his initial wish
to become a pupil of Protagoras by paying him any money, is wrong.
Because of obvious correlation of the natural dialogue process with the
area of problem solving, it is reasonable to introduce the concept of dialogue
behavior, and interpret it as a realization of a problem solving method by
means of the dialogue process. The cited fragment of the Plato’s dialogue is
one of the protocols of Socrates’ dialogue behavior, which implements his
method called, in some modern publications, Socratic Inquiry Method [14].
Consider briefly the essence of this method. Socrates could reach his
goal in several ways. For instance, he could, referring to his authority,
offer Hippocrates his final conclusion at once, saying: «Dear Hippocrates,
you shouldn’t try to become Protagoras’ pupil at any cost without clarify-
ing what subject he will teach you and whether knowledge acquired from
him will be useful for you». In this case Socrates transmits to Hippocrates
his belief, based on his authority, but not knowledge, justified by logical
inference. However, Socrates prefers to enter into a dialogue with Hippo-
crates and offers him a series of interrogative messages with predeter-
mined variants of answers. Socrates' information messages are formulated
in such a way that Hippocrates’s answers establish a chain of reasoning,
which finally generates the above formulated conclusion. This is a more
efficient method, because Hippocrates, guided by Socrates, deduces the
above formulated conclusion by himself. This is the essence of the method
invented by Socrates. Playing the role of the active agent Socrates manag-
es the dialogue in such a manner to induce his opponent to answer in the
way, which is necessary to form a chain of reasoning leading to the target
conclusion. The secret of Socrates’ success is in knowing that each ques-
tion is associated with the set of possible answers, which does not depend
on the reactive agent’s knowledge, but merely on the question itself. Be
aware of this fact Socrates, playing the role of the active agent, constructs
Математичне та комп’ютерне моделювання
172
his information messages in such a way to narrow the number of possible
answers as much as possible, or even reduce it to unity.
The subsequent parts of this article rely on several simplifying as-
sumptions regarding natural dialogue. First of them is that instead of a
full-scale human-human dialogue we will consider a simplified version
called erotetic dialogue [15; 16].
Within erotetic dialogue, information messages from the active agent
have hundred-per-cent degree of interrogativness and in a linguistic sense
are questions, while information messages from reactive agent are answers
on these questions. Erotetic dialogue is a step towards simplification of
real situations, but it appears to be rational for several reasons. Firstly, as a
rule, it is possible to convert protocols of natural dialogues into erotetic
ones keeping initial goal and method; secondly, an investigation of an ero-
tetic dialogue can be based on more or less formal theories, and thirdly,
results obtained from erotetic dialogue research can be a good foundation
for generalization to a full-scale natural dialogue.
In erotetic dialogue, knowledge requested by the active agent and re-
turned by reactive agent, has declarative nature. Therefore, a model of the
logical structure of an erotetic transaction, in the context of knowledge
representation, must be based on certain plausible conceptions of declara-
tive knowledge representation. Although human’s system of knowledge,
seemingly, is one and indivisible, where adjacent parts concatenate with
each other, questions and answers operate with small separate parts or
chunks of a human’s system of declarative knowledge. Therefore, for our
purpose, we’ll need only means for representing chunks of declarative
knowledge associated with erotetic transaction.
Epistemologists usually don’t use terms «declarative knowledge» and
«procedural knowledge» and declarative-procedural dichotomy of know-
ledge. However, from the context of some publications it is clear that
knowledge-wh and knowledge-that are no other than epistemic names of
chunks of declarative knowledge, associated with erotetic transaction [13].
These chunks represented by natural-language propositions and are an-
swers on wh- and that-questions respectively. Wh-question is a class of
questions generalizing six subclasses: (1) who-question; (2) what-
question; (3) when-question; (4) where-question; (5) how-question; and
(6) why-question. Epistemic understanding of knowledge-wh can be illu-
strated by examples of the following six natural-language sentences: (1) «I
know who is Protagoras»; (2) «I know what he will speak about»; (3) «I
know when he will begin his conversation»; (4) «I know where the conver-
sation will take place»; (5) «I know how he will begin his speech»; and (6)
«I know why he is wrong». This classification reflects and illustrates defi-
nite linguistic orientation of the epistemic representation of knowledge
associated with question-answering transactions.
Серія: Технічні науки. Випуск 11
173
In erotetic dialogue we consider a question as a search prescription
which determines the area in the long-term memory of the reactive agent,
which contains requested knowledge. Such an interpretation means that the
reactive agent already possesses requested knowledge. However, it is only one
of several possible cases. It is easy to formulate examples of questions that
request knowledge that is absent in the memory of the reactive agent. Such
questions presuppose certain mental work or mental effort involving attention-
al resources. Therefore, with respect to the participation of attention in the
process of constructing the answer, we can divide questions into two classes:
«search prescription» question type, and «task» question type.
To answer the «task» question type, the reactive agent must activate
his/her attention and solve the task, associated with the question. The an-
swer is a variant of solution, obtained by the agent. Consider the example:
How old would be Socrates today, if he was born in
469 B.C?
It is clear that the reactive agent does not keep in mind the requested
knowledge and needs to activate his/her resource of attention to form the
answer.
The «search prescription» question type presupposes that all possible
answers are already in the memory of reactive agent and that the structural
elements of the question position memory to the required part. To prepare
the answer of the «search prescription» question type the reactive agent
does not need to activate attention. The example is as follows:
What is your name?
The class of «search prescription» questions type is not final and can
be divided into subclasses. When the active agent constructs a «search
prescription» question type he/she doesn’t know, in advance, whether or
not the reactive agent possesses the requested knowledge. The question,
which requests missing knowledge, is able to put the reactive agent into
impasse. For example, the question:
Why are you concealing the fact that you are a man?
can put the normal woman into impasse.
The classification of questions into «search prescription» type and «task»
type is not absolute and must be considered with respect to the reactive agent’s
knowledge. If the task, caused by the question, is solved by the reactive agent
for the first time or very seldom (so seldom that the process of forgetting pre-
vails), then to solve the task, according to Kahneman’s theory [17], resource of
attention is needed. In the case of repetitive answers to the same or similar
«task» question type, the solution is eventually stored in memory, the mechan-
ism of attention is not needed any more, and the question turns into «search
prescription» question type.
Математичне та комп’ютерне моделювання
174
Classification of questions into two classes, with respect to the participa-
tion of attention in the process of constructing the answer, is not the only one
of its kind. Analysis of examples of natural dialogues allows us to find out that
each question can be characterized by a certain degree of uncertainty. There-
fore, with respect to the degree of uncertainty, questions can be divided at least
into two classes: uncertain questions and certain questions. This classification
is obviously inaccurate because it does not take into account the actual degree
of uncertainty, but it should be enough for the purpose of the article.
Uncertain question means that its answer belongs to one of several
classes of answers. If the reactive agent did not receive special instructions
on how to answer the question, he/she can act according to one of the fol-
lowing strategies: (1) chooses the class of answers, follows his/her own
way; (2) with the purpose to reduce or even remove uncertainty, formu-
lates an additional question. The latter case means that agents interchange
their roles and can be considered as one of several possible reasons for
role interchange in the erotetic dialogue. Apparently, the process of clarifi-
cation of uncertain question can be iterative, when the reactive agent con-
siders the new question as uncertain as well. For example, the question:
Who is living behind this door?
is uncertain, because requested knowledge belongs to one of several
classes: (1) knowledge regarding the passport data of the tenant; (2) know-
ledge regarding the tenant’s appearance, etc. When the reactive agent gets
such a question and wishes to clarify what concrete knowledge the active
agent is requested, he/she could transmit to the active agent a clarifying
question instead of the answer. For instance:
You wish to know the name of the person who is
living behind this door, or something else?
If the active agent returns to the reactive agent the question:
Could you describe me this person?
then we have an uncertain question once again. Certain questions presup-
pose that the requested knowledge belongs to only one class:
Tell me the name of the man living behind this door?
3. Formulas. In the subsequent part of the article we will use the con-
cepts offered in the first part for formal modeling of the logical structure of
dialogue transaction. The peculiarity of our approach to the modeling of dia-
logue transaction is focusing attention on the fact that the main purpose of
dialogue transaction is to operate with declarative knowledge of the reactive
agent. The models, that we construct, give possible variants of the logical
structure of erotetic transaction in the context of representation of knowledge,
associated with the transaction. Our modeling will be limited to questions of
the type «search prescription», and this is our second simplifying assumption.
Серія: Технічні науки. Випуск 11
175
In Epistemology and Erotetic Logic declarative knowledge partici-
pating in question-answering transactions usually represented in the form
of propositions and, hence, an elementary chunk of declarative knowledge
is considered as a sentence. A generally accepted way of formalization of
such a method of knowledge representation is the translation of natural-
language sentences into sentences written in First-Order Logic notation.
We guess that modeling of erotetic transaction, in the context of dec-
larative knowledge representation, can be realized by means of a more
unified ontology than of First-Order Logic. Under unified ontology we
understand ontology based not on language-depended variables (as in
First-Order Logic) but on language-independent fundamental entities of
declarative knowledge. In this regard we considered the Language of Ter-
nary Description (LTD), initially suggested by Uyemov, for the formal
description of parametric variant of the General Systems Theory and then
developed into independent Non-classical Logic [3–5].
3.1. The Language of Ternary Description. A fundamental entity
of LTD is object. An object, depending on its place in the knowledge
structure, exists in one of three forms: object-thing, object-property, and
object-relation.
The categories «thing» and «property» have traditional meanings in
LTD, while the category «relation» differs from the traditional. It is gener-
ally accepted to use the «relation» concept as a name of the mutual influ-
ence between things, i.e. relation between things. For instance, between
two men can exist «fatherhood» relation. In LTD context, a relation is
something that forms a thing, i.e. relation on a thing, or relation that takes
place in a thing. In other words, an LTD-relation is, in some sense, anoth-
er name of the inner structure of the thing.
The binary association of object-thing with object-property generates
two prototypes for representation entities in LTD:
1. The name for the first prototype is «thing, which possesses property»
and formal notation has the following form: (*)*.
2. The name for the second prototype is «property, which attributed to
thing» and formal notation has the form: (*))*.
The binary association of object-thing with object-relation generates
two more prototypes:
3. The name for the third prototype is «thing, in which relation takes
place» and formal notation has the following form: *(*).
4. The name for the fourth prototype is «relation, which takes place in
thing» and formal notation has the form: *((*).
For the formal representation of entities in LTD a specific parenthes-
es notation is used. The symbol «asterisk» indicates a place for the symbol
Математичне та комп’ютерне моделювання
176
of an object. The symbol of an object-thing is written down inside paren-
theses, the symbol of an object-property — outside and on the right side of
the parentheses, and the symbol of an object-relation also is written out-
side parentheses but on the left side.
The association of an object-thing with an object-property or an ob-
ject-relation has direction. If a symbol of an object-thing is in normal (sin-
gle) parentheses, this means that the association is directed from the ob-
ject-thing to the object-property or object-relation. In natural language it
can be expressed as a «thing, which possesses property» or a «thing in
which relation take place». Asymmetric (doubled) parentheses means that
the association is directed from the object-property or the object-relation to
object-thing and expresses as «property, which is attributed to thing» or
«relation, which take place in thing».
It should be noted that in publications, which deal with LTD, a place
inside parentheses is used only for the object-thing symbol. This limitation
excludes from the set of prototypes the following two: «relation, which
possesses property», and «property, which attributed to relation».
An object, depending on the degree of uncertainty of knowledge
about it, exists in one of three alternative forms:
definite (asterisk in the prototype substituted by symbol t);
indefinite (asterisk in the prototype substituted by symbol a);
arbitrary (asterisk in the prototype substituted by symbol A).
The categories «thing, property, and relation» as well as «definite-
ness, indefiniteness, and arbitrariness» are independent and form nine
classes of objects: (1) definite object-thing; (2) indefinite object-thing; (3)
arbitrary object-thing; (4) definite object-property; (5) indefinite object-
property; (6) arbitrary object-property; (7) definite object-relation; (8) in-
definite object-relation; and (9) arbitrary object-relation.
Substitution of the symbol asterisk, in the prototypes, by symbols t,
a, and A, gives us a collection of possible models of chunks of declarative
knowledge in LTD ontological basis.
Model (t)a (1)
represents knowledge regarding certain definite thing which possesses
some sort of (indefinite) property. The direction of association is from
thing to property.
Model (t))a (2)
represents knowledge regarding some sort of (indefinite) property, which
is attributed to certain definite thing. The direction of association is from
property to thing.
Model (a))A (3)
represents knowledge regarding arbitrary property, which is attributed to some
sort of (indefinite) thing. The direction of association is from property to thing.
Серія: Технічні науки. Випуск 11
177
Model a(t) (4)
represents knowledge regarding a certain definite thing, in which some
sort of (indefinite) relation takes place. The direction of association is from
thing to relation.
Since there are four prototypes and each prototype can be filled by ob-
jects from nine classes, the total number of models for elementary chunks of
declarative knowledge in the ontological basis of LTD is thirty six.
3.2. Transaction’s structure. The structure of a question of erotetic
transaction in the context of knowledge representation will be considered as a
further development of the idea of basic interrogative formula of Belnap and
Steel [10], which postulates that the inner logical structure of the question is
determined by two components: question’s subject and question’s request.
Que = <Subj, Req> (5)
An important element of the Belnap and Steel philosophy, underlying
their theory, is the concept of the question’s subject. From this concept it
follows that those questions, which include subject, for instance «search
prescription» questions type, not only predetermine answers, but moreover
include them. This means that we can think a question’s subject as an an-
swer with some degree of uncertainty.
A question’s subject is a key component of erotetic transaction. The
essence of the work, which the reactive agent fulfills while constructing
the answer, is the transformation of uncertain knowledge, on which the
subject points out, into more definite knowledge associated with the an-
swer. Using other words we can say that the reactive agent reduces the
degree of uncertainty of knowledge, corresponding to the subject of the
question, to the level appropriate for the answer.
We will model chunks of declarative knowledge, on which ques-
tion’s subject points out, by the following alternative formulas:
Ksubj = (t)a (6)
Ksubj = (a))t (7)
Ksubj = a(t) (8)
Ksubj = t((a) (9)
Models (6)–(9) obtained from the four, considered earlier, LTD-
prototypes for the case of the association of definite and indefinite objects.
There are several reasons for selection of the models (6)–(9) for
representing declarative knowledge. Firstly, subjects of all instances of
questions, mentioned for example in [2; 10; 13], can be represented by one
of the models (6)–(9). Secondly, these models are exactly the sort of mod-
els (from the thirty six, that are possible), which accord with the idea that
the reactive agent, while constructing an answer, transforms uncertain
knowledge of the subject into certain knowledge of the answer.
Математичне та комп’ютерне моделювання
178
One of the objects in models (6)–(9) is indefinite, and the active
agent is waiting for more concrete knowledge regarding this object from
the reactive agent.
On conceptual level, models (6)–(9) can be elucidated in the follow-
ing way.
Model Ksubj = (t)a represents knowledge regarding a given definite
thing, which possesses an indefinite property. Questions with such a
subject are generated by the active agent when he/she wants to know
which properties have given thing.
Model Ksubj = (a))t represents knowledge regarding some given defi-
nite property, which is attributed to an indefinite thing. Questions with
such a subject are generated by the active agent when he/she wants to
know which things have given property.
Model Ksubj = a(t) represents knowledge regarding some given definite
thing in which an indefinite relation takes place. Questions with such a
subject are generated by the active agent when he/she wants to know
which relations take place in given thing.
Model Ksubj = t((a) represents knowledge regarding some given defi-
nite relation, which occurs in an indefinite thing. Questions with such
subject are generated by an active agent when he/she wants to know in
which things occurs given relation.
The shortcoming of models (6)–(9), from the point of view of engineer-
ing of knowledge-based program systems is their poor suitability for mapping
into relevant data structures. These models could be practically suitable for
software engineering in the case, when we find out the way of their transform-
ing into types or data structures of modern systems of programming. It is
talked, first of all, about datalogical interpretation of an indefinite object.
One of the possible datalogical interpretations of indefiniteness is
multiplicity. An indefinite object can be understood as a set of definite
objects and the cardinality of this set as a degree of indefiniteness. Then
decreasing of the degree of indefiniteness is equal to decreasing of the
cardinal number of the corresponding set. An indefinite object turns into
an absolutely definite one when the cardinality of the corresponding set is
equal to unity, or when the set is represented by one object.
Taking into account such kind of interpretation of indefiniteness, we
may substitute indefinite objects by lists of definite objects and express
models (6)–(9) in the following way.
Ksubj = <object-thing>{list of objects-properties} (10)
Ksubj = <object-property>{list of objects-things} (11)
Ksubj = <object-thing>{list of objects-relations} (12)
Ksubj = <object-relation>{list of objects-things} (13)
Серія: Технічні науки. Випуск 11
179
The models (10)–(13) are datalogical counterparts of models (6)–(9)
and represent declarative knowledge, transmitted to the reactive agent via
the question’s subject. Two additional models, inexpressible within LTD
notation (inside parentheses we may use only object-thing symbols), are
introduced by formulas (14) and (15).
Ksubj = <object-property>{list of objects-relations} (14)
Ksubj = <object-relation>{list of objects-properties} (15)
As the subject of the question is, in fact, an answer with some degree
of uncertainty, and the reactive agent, while constructing the answer, re-
duces this degree of uncertainty to the level appropriating for the answer
(determines by a question’s request Req in interrogative formula (5)), then
models of chunks of declarative knowledge, associated with the answer
should be similar to (10)–(15). The difference is in cardinality of the set of
objects. Hence, we can express them in the following way.
Kans = <thing> possesses properties {properties} (16)
Kans = <property> attributed to things {things} (17)
Kans = in <thing> take place relations {relations} (18)
Kans = <relation> occurs in things {things} (19)
Kans = <property> attributed to relations {relations} (20)
Kans = <relation> take place in properties {properties} (21)
The models (16)–(21) represent declarative knowledge, which the
reactive agent returns to the active one via the answer on the question with
the corresponding subject.
The models (16)–(21) can be relatively easily illustrated by numerous ex-
amples of transactions, built on whether-questions, used, for instance by
Belnap and Steel [10]. Consider the transaction.
Active: What is the freezing point of water, in
degrees Fahrenheit, under standard con-
ditions?
Reactive: The freezing point of water under stan-
dard conditions is 32°F.
The subject of the question in given transaction provides access to
declarative knowledge, represented by the model (10):
Ksubj = <water under standard conditions>
{temperatures of freezing for liquids by Fahrenheit}
The model represents knowledge regarding given the definite thing
«water under standard conditions», which possesses an indefinite proper-
ty — some sort of freezing point. Active agent, posing the question with
such a subject wants to go from uncertainty to certainty and get access to
the reactive agent’s knowledge regarding the concrete property of «water
under standard conditions». Uncertainty in knowledge, which the subject
Математичне та комп’ютерне моделювання
180
points out, expressed by the list with a relatively large number of alterna-
tives. Knowledge, associated with the answer of the reactive agent, is quite
certain and represented by the model (16).
Kans = <water under standard conditions> possesses property {freeze at 32°F}
Consider the question.
Which primes lie between 10 and 20?
The subject of this question provides access to declarative know-
ledge, represented by the model (11):
Ksubj = <be prime number> {10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20}
The model represents knowledge regarding given definite property
«be prime number», which attributed to an indefinite thing — numbers
between 10 and 20. The active agent posing the question with such a sub-
ject wants to go from uncertainty to certainty and learn from the reactive
agent which concrete things have the property «be prime number». Uncer-
tainty in knowledge, which the subject pointed out, expressed by the list of
eleven elements. Knowledge, associated with the answer of the reactive
agent represented by the model (17).
Kans = <be prime number> attributed to things {11, 13, 17, 19}
As our next example consider transaction taken from Plato’s dialogue
Protagoras. The fragment of this dialogue, given at the beginning of the
article, is not purely erotetic, because the information messages of the ac-
tive agent do not always have the form of questions. However, some trans-
actions in this dialogue are erotetic. Consider the first one.
S: Then we are going to pay our money to Protago-
ras in the character of a Sophist?
H: Certainly.
The subject of Socrates’ question in this transaction provides access to
declarative knowledge, represented by the model (10). The model represents
knowledge regarding given definite thing «Socrates& Hippocrates», which
possesses an indefinite property — willingness/unwillingness to pay money to
Protagoras only because he is Sophist. Socrates, posing the question with such
a subject, wants to move from uncertainty to certainty and learn from Hippo-
crates which concrete property is possessed by «Socrates &Hippocrates». So-
crates expects a certain answer and based on this expectation, he constructs the
subject of his question in such a way that uncertainty in knowledge, which the
subject points out, represented by a list of only several alternatives.
Ksubj = <Socrates&Hippocrates>
{to give money to Sophist Protagoras,
not to give money to Sophist Protagoras}
Серія: Технічні науки. Випуск 11
181
Hippocrates’ answer is, in fact, a pointer to a chunk of declarative
knowledge, represented by the model (16).
Kans = <Socrstes&Hippocrates> possesses property
{to give money to Sophist Protagoras}
4. Cognitive cycle of erotetic dialogue interaction. The conclusive
part of the paper is focused on the modeling of dialogue interactions from
the point of view of cognitive processes, which are involved in the percep-
tion and constructing of erotetic transaction elements.
With respect to any agent, the dialogue process is similar to the
process of perceptual interaction between a human and the environment.
The difference is that during the dialogue both main components of the
human sensory system, visual and auditory analyzers, are connected not to
a natural environment, but to an artificial one, formed by the opposite di-
alogue agent. Thus, in the dialogue process, the natural environment is
substituted by artificial one, but, it’s obvious, that the perception and fur-
ther processing of stimuli, generated either by the dialogue agent or the
natural environment are realized under the same rules and by means of the
same chain of human cognitive subsystems.
When we are building models of anthropomorphic dialogue agents, it
is essential to ensure the adequacy of these models to the models of per-
ception and information processing in humans. In this case we can expect
that artificial dialogue agents will inherit flexibility and universality of
human abilities to realize a task-oriented dialogue process.
Among a number of models, offered by cognitive psychologists and
related to the erotetic dialogue process, we focus our attention on models
describing dialogue on the level, which is invariant to the process of syn-
thesis of erotetic transaction, because generating of questions and answers
depends to a considerable degree on the domain of concrete dialogue, its
goal and the task, solving during the dialogue. From our point of view the
model of dialogue interaction, which could be helpful in applied aspect,
has to describe certain «dialogue machine», which functions invariantly to
the dialogue interaction domain.
We will develop a model of the cognitive cycle of erotetic dialogue inte-
raction, starting from cyclical model of perception [9]. In the context of erotet-
ic dialogue, and with respect to the active agent, the perceived environment in
our model will be substituted by a flow of answers from reactive agent, and a
set of anticipatory schemata by a set of anticipatory answers.
Although, an active agent expects to receive only one answer, a ques-
tion’s subject and request predetermine a list of answers. The number of
elements in this list range from two, for yes/no type of questions, to a sub-
stantial number for other types of questions. In [10] there is an example of
Математичне та комп’ютерне моделювання
182
a question regarding tobacco smoking, which was supposedly articulated
by King James I.
Tobacco smoking: a vice, a virtue, a vagary,
an extravagance, a cure for all ills?
The declarative knowledge, which the subject of this question points
out, contains certain degree of indefiniteness and is represented by a set of
five chunks, which we can associate with propositions:
(1) Tobacco smoking is a vice.
...
(5) Tobacco smoking is a cure for all ills.
The question’s subject and request predetermine list of answers:
(1) Tobacco smoking is a vice.
(2) Tobacco smoking is a virtue.
(3) Tobacco smoking is a vagary.
(4) Tobacco smoking is an extravagance.
(5) Tobacco smoking is a cure for all ills.
(6) Tobacco smoking is a vice and virtue at the
same time.
(7) Tobacco smoking is a vice and vagary at the
same time.
...
The active agent, being in the state of awaiting the answer and being un-
aware of what answer will be returned by reactive agent, however, can predict
the set of possible answers, or a set of anticipatory answers. It should be noted
that the «set of anticipatory answers» concept is more constructive than the
«true answer» concept because truthfulness is relative with respect to reactive
agent. For example, reactive agent is not always honest and constructs the
answer in accordance with his/her own goals and motivation.
The set of anticipatory answers is a part of more extensive cognitive
structure, or an active agent’s script. The active agent’s script operates
with references to declarative knowledge in the form of questions, asso-
ciated with anticipatory answers. From this point of view, the script of an
active agent is certain dialogue access method to his declarative know-
ledge. A script determines the dialogue behavior of an active agent and
reflects his/her method. Thus, we suppose that the Socratic Inquiry Me-
thod [14], for instance, can be represented by a script. From the point of
view of procedural-declarative dichotomy of knowledge, the active agent’s
script represents his/her procedural knowledge.
After perceiving and categorizing the actual answer, obtained from
the reactive agent, an active agent modifies the set of anticipatory answers,
Серія: Технічні науки. Випуск 11
183
preparing, thereby, the next transaction. The modified set includes the an-
swer which active agent anticipates in the next transaction.
Perceiving and categorizing the actual answer causes two cognitive
processes. One of them is the process of modification of the set of anticipa-
tory answers just described, and another one is the process of constructing
the question next in turn. It is clear that answers are determined by the ques-
tions and which actual answer is received from reactive agent in a given
transaction depends on the question formed during previous transaction.
One of the distinguishing features of erotetic dialogue interaction is
its goal-driven nature. Behind the outward process of question-answers
interchange we can always recognize the process of achieving a goal. In
the case of the active agent the goal could be a certain target answer, re-
ceived from reactive agent, or the realization of the script along a certain
«target trajectory». An example is the fragment of dialogue between So-
crates and Hippocrates, cited at the beginning of the paper.
The goal-driven nature of erotetic dialogue means that the active agent is
not randomly construct questions, but rather constructs them in compliance
with the «direction» towards of the goal. The active agent must manage this
movement and he/she does it by formation of an interrogative hypothesis,
regarding reactive agent's knowledge. In this hypothesis, the assumptions of
the active agent regarding the knowledge of the reactive agent are determined
by the question’s subject. The interrogative hypothesis is transmitted to the
reactive agent for testing in the form of the question.
A description of the cognitive processes that are components of an
erotetic dialogue interaction can be represented in the form of a repeated
cyclical process. Figure1depicts the main elements of the erotetic dialogue
cognitive cycle.
Modification of
the set of
anticipatory
Set of anticipatory
answers
Interrogative hypothesis
in the form of the
question
Transmission
of the question
for testing
hypothesis
Formation of
interrogative
hypothesis
ACTIVE AGENT REACTIVE AGENT
Que
Selection and
transmission of a
single answer
Formation of
possible answers
Ans
Set of possible
answers
Fig. 1. Cognitive cycle of erotetic dialogue interaction
Математичне та комп’ютерне моделювання
184
The cognitive cycle of erotetic dialogue interaction, depicted in fig-
ure 1, reflects, mainly, the cognitive processes of the active agent. Howev-
er, a multiplicity of possible answers, predetermined by question’s subject
and request give us grounds to divide the mental activity of the reactive
agent into two stages.
In the first stage the set of possible answers is generated. This stage
fully depends on question’s structure, and, probably, doesn’t depend on
the current goal of the reactive agent. A question’s subject and request
predetermine the set of possible answers.
In the second stage, from the set of possible answers the reactive
agent selects just one. This selection presupposes the possibility of as-
sessment of each of the possible answers on its «proximity» to the goal.
This stage is determined by the knowledge system and the current goal of
the reactive agent. In the case of the King James’ question, the knowledge
system and the current goal of certain reactive agent can suggest him/her
to generate the following answer:
Tobacco smoking is a virtue.
Conclusions. Erotetic dialogue is one of several types of natural dialo-
gue, which can be employed by artificial dialogue agents, demonstrating anth-
ropomorphic behavior. Erotetic dialogue transactions are limited to question-
answering pairs and therefore, externally, erotetic dialogue is not so various as
a freewheeling natural dialogue, but it can achieve the same goals and solve
the same problems as a full-scale dialogue between humans.
It is rational to construct models of logical structure of erotetic dialo-
gue transaction based on the structure of knowledge associated with the
transaction. In this case models obtained not only assist better understand-
ing of the inner nature of a dialogue transaction, but also have a pragmatic
value, because they can serve as a theoretical foundation for engineering
of the dialogue agent knowledge base.
In one of our previous publications [16] we introduced the idea of a di-
alogue knowledge base in the form of composition of the memory of ques-
tions (QueMem) with direct access to its elements, and the dialogue access
method (DiAM). From the point of view of procedural-declarative dichoto-
my of knowledge, DiAM is a repository of procedural knowledge of an ac-
tive agent, and converts each answer of the reactive agent into QueMem
address. The present article develops the idea of the dialogue knowledge
base in the light of representation of chunks of declarative knowledge, asso-
ciated with dialogue transactions. It seems more rational to consider a mem-
ory of declarative knowledge associated with questions’ subjects (KsubjMem)
instead of the memory of questions. In this case: (1) KsubjMem stores all
subjects, which are necessary for synthesis of those questions that the active
agent constructs within the concrete dialogue process; and (2) DiAM con-
verts Kans of the current answer into the address of Ksubj of the following
Серія: Технічні науки. Випуск 11
185
question. If, for example, we represent the fragment of the dialogue, given at
the preamble of the article, in the form of erotetic dialogue, then Socrates’
knowledge needed for computer synthesis of this dialogue, can be stored in
the dialogue knowledge base. In this case, DiAM should store knowledge of
Socrates’ dialogue behavior (Socratic Inquiry Method) with regard to his
dialogue with Hippocrates, and KsubjMem — subjects of questions needed
for synthesis of Socrates’ questions.
The goal of the article and it size do not allow us to develop the idea
of applicability of models of logical structures of dialogue transactions,
offered in the article, to the architecture of the dialogue knowledge base.
Authors suppose to introduce the results of this investigation in their sub-
sequent publication.
References:
1. Hintikka J. Questions about questions / J. Hintikka ; eds. M. K. Munitz,
P. K. Unger // Semantics and Philosophy. — New York : New York University
Press, 1974. — P. 103–158.
2. Wisniewski A. The Posing of Questions: Logical Foundations of Erotetic Infe-
rence / A. Wisniewski. – Kluwer : Academic Publishers, 1995.
3. Uyemov A. The Ternary Description Language as a Formalism for the Para-
metric General Systems Theory / A. Uyemov // Int. J. of General Systems. —
1999. — Vol. 28 (4-5). — Part I. — P. 351–366.
4. Uyemov A. The Ternary Description Language as a Formalism for the Para-
metric General Systems Theory / A. Uyemov // Int. J. of General Systems. —
1999. — Vol. 31 (2). — Part II. — P. 131–155.
5. Uyemov A. The Ternary Description Language as a Formalism for the Para-
metric General Systems Theory / A. Uyemov // Int. J. of General Systems. —
2003. — Vol. 32 (6). — Part III. — P. 583–623.
6. Turing A. Computing machinery and intelligence / A. Turing. — Mind, 1950. —
№ 59. — P. 433–460.
7. Russel S. Artificial Intelligence: A Modern Approach, second edition / S. Russel,
P. Norvig. — New Jersey : Pearson Education, Inc., Upper Saddle River, 2003.
8. Brainerd C. Piaget’s Theory of Intelligence / C. Brainerd. — New Jersey :
Prentice-Hall, Inc., 1978.
9. Neisser U. Cognition and Reality / U. Neisser. — San Francisco : W.H. Free-
man and Company, 1976.
10. Belnap N. B. The Logic of Questions and Answers / N. B. Belnap, T. Steel. —
New Haven ; London : Yale University Press, 1976.
11. Hintikka J. Socratic Epistemology: Explorations of Knowledge Seeking by
Questions / J. Hintikka. — Cambridge University Press, 2007.
12. Plato, Complete works / Edited by J.M. Cooper, Associate Editor D.S. Hut-
chinson. — Indianapolis : Hackett Publishing Co., 1997.
13. Schaffer J. Knowing the answer / J. Schaffer // Philosophy and Phenomenolog-
ical Research. — 2007. — Vol. LXXV, № 2. — P. 383–403.
14. Wenger E. Artificial Intelligence and Tutoring Systems / E. Wenger. — Los
Altos, California : Morgan Kaufman, 1987.
Математичне та комп’ютерне моделювання
186
15. Chemyr I. A. SmartText: Using Agents Supporting Personalized Reading
Comprehension / I. A. Chemyr, M. A. Horney, L. Anderson-Inman // Personal
Technologies. — 1998. — Vol.2, № 3. — P. 152–161.
16. Chimir I. Modeling of a Procedural Knowledge by a Dialogue Knowledge
Base / I. Chimir, W. Abu-Dawwas // The International Arab Journal of Infor-
mation Technology. — 2004. — Vol.1, № 1. — P. 128–134.
17. Kahneman D. Attention and Effort / D. Kahneman. — New Jersey : Prentice-
Hall, Inc., Englewood Cliffs, 1973.
Стаття присвячена дослідженню та моделюванню внутрішніх ме-
ханізмів діалогу між людьми з метою розробки штучного людинопо-
дібного діалогового агента. Онтологічний базис діалогової взаємодії
між людьми, на якому базуються дослідження, одержано шляхом
аналізу діалогів між людьми і ілюструється одним з Платонівських
діалогів під назвою Протагор. У статті запропоновані дві моделі, які
пояснюють внутрішні механізми діалогу між людьми. Обидві моделі
синтезовані для винятку одного типу діалогу між людьми під назвою
еротетичний діалог. Модель структури діалогової трансакції синтезо-
вана у контексті представлення декларативних знань за допомогою
Мови Тернарного Опису. Модель когнітивного циклу діалогової вза-
ємодії є розвитком циклічної моделі перцепції Найсера.
Ключові слова: природний діалог, еротетичний діалог, діалогова
трансакція, декларативні знання, мова тернарного опису.
Отримано: 23.06.2014
УДК 004.832.2
C. І. Шаповалова, канд. техн. наук
Национальный технический университет Украины
«Киевский политехнический институт», г. Киев
МОДЕЛИРОВАНИЕ ФРАГМЕНТА ПРОСТРАНСТВА
СОСТОЯНИЙ ДЛЯ ТЕСТИРОВАНИЯ ЭВРИСТИЧЕСКИХ
СТРАТЕГИЙ ПОИСКА
В статье представлена разметка графа пространства со-
стояний для формирования тестовых примеров по задаваемой
эвристической стратегии поиска и характеристикам допусти-
мости оценочной функции.
Ключевые слова: поиск в пространстве состояний, эври-
стический поиск, допустимость оценочной функции.
Введение. Поиск в пространстве состояний является одним из
фундаментальных подходов искусственного интеллекта. Основной
проблемой является слишком большой размер пространства поиска,
© С. І. Шаповалова, 2014
<<
/ASCII85EncodePages false
/AllowTransparency false
/AutoPositionEPSFiles true
/AutoRotatePages /All
/Binding /Left
/CalGrayProfile (Gray Gamma 2.2)
/CalRGBProfile (sRGB IEC61966-2.1)
/CalCMYKProfile (Coated FOGRA27 \050ISO 12647-2:2004\051)
/sRGBProfile (sRGB IEC61966-2.1)
/CannotEmbedFontPolicy /Warning
/CompatibilityLevel 1.3
/CompressObjects /Tags
/CompressPages true
/ConvertImagesToIndexed true
/PassThroughJPEGImages true
/CreateJobTicket false
/DefaultRenderingIntent /Default
/DetectBlends true
/DetectCurves 0.1000
/ColorConversionStrategy /sRGB
/DoThumbnails false
/EmbedAllFonts true
/EmbedOpenType false
/ParseICCProfilesInComments true
/EmbedJobOptions true
/DSCReportingLevel 0
/EmitDSCWarnings false
/EndPage -1
/ImageMemory 1048576
/LockDistillerParams false
/MaxSubsetPct 100
/Optimize true
/OPM 1
/ParseDSCComments true
/ParseDSCCommentsForDocInfo true
/PreserveCopyPage true
/PreserveDICMYKValues true
/PreserveEPSInfo false
/PreserveFlatness false
/PreserveHalftoneInfo false
/PreserveOPIComments false
/PreserveOverprintSettings true
/StartPage 1
/SubsetFonts true
/TransferFunctionInfo /Apply
/UCRandBGInfo /Remove
/UsePrologue false
/ColorSettingsFile ()
/AlwaysEmbed [ true
]
/NeverEmbed [ true
/Arial-Black
/Arial-BlackItalic
/Arial-BoldItalicMT
/Arial-BoldMT
/Arial-ItalicMT
/ArialMT
/ArialNarrow
/ArialNarrow-Bold
/ArialNarrow-BoldItalic
/ArialNarrow-Italic
/ArialUnicodeMS
/CenturyGothic
/CenturyGothic-Bold
/CenturyGothic-BoldItalic
/CenturyGothic-Italic
/CourierNewPS-BoldItalicMT
/CourierNewPS-BoldMT
/CourierNewPS-ItalicMT
/CourierNewPSMT
/Georgia
/Georgia-Bold
/Georgia-BoldItalic
/Georgia-Italic
/Impact
/LucidaConsole
/Tahoma
/Tahoma-Bold
/TimesNewRomanMT-ExtraBold
/TimesNewRomanPS-BoldItalicMT
/TimesNewRomanPS-BoldMT
/TimesNewRomanPS-ItalicMT
/TimesNewRomanPSMT
/Trebuchet-BoldItalic
/TrebuchetMS
/TrebuchetMS-Bold
/TrebuchetMS-Italic
/Verdana
/Verdana-Bold
/Verdana-BoldItalic
/Verdana-Italic
]
/AntiAliasColorImages false
/CropColorImages false
/ColorImageMinResolution 150
/ColorImageMinResolutionPolicy /OK
/DownsampleColorImages true
/ColorImageDownsampleType /Bicubic
/ColorImageResolution 150
/ColorImageDepth -1
/ColorImageMinDownsampleDepth 1
/ColorImageDownsampleThreshold 1.50000
/EncodeColorImages true
/ColorImageFilter /DCTEncode
/AutoFilterColorImages true
/ColorImageAutoFilterStrategy /JPEG
/ColorACSImageDict <<
/QFactor 0.76
/HSamples [2 1 1 2] /VSamples [2 1 1 2]
>>
/ColorImageDict <<
/QFactor 0.76
/HSamples [2 1 1 2] /VSamples [2 1 1 2]
>>
/JPEG2000ColorACSImageDict <<
/TileWidth 256
/TileHeight 256
/Quality 15
>>
/JPEG2000ColorImageDict <<
/TileWidth 256
/TileHeight 256
/Quality 15
>>
/AntiAliasGrayImages false
/CropGrayImages false
/GrayImageMinResolution 150
/GrayImageMinResolutionPolicy /OK
/DownsampleGrayImages true
/GrayImageDownsampleType /Bicubic
/GrayImageResolution 150
/GrayImageDepth -1
/GrayImageMinDownsampleDepth 2
/GrayImageDownsampleThreshold 1.50000
/EncodeGrayImages true
/GrayImageFilter /DCTEncode
/AutoFilterGrayImages true
/GrayImageAutoFilterStrategy /JPEG
/GrayACSImageDict <<
/QFactor 0.76
/HSamples [2 1 1 2] /VSamples [2 1 1 2]
>>
/GrayImageDict <<
/QFactor 0.76
/HSamples [2 1 1 2] /VSamples [2 1 1 2]
>>
/JPEG2000GrayACSImageDict <<
/TileWidth 256
/TileHeight 256
/Quality 15
>>
/JPEG2000GrayImageDict <<
/TileWidth 256
/TileHeight 256
/Quality 15
>>
/AntiAliasMonoImages false
/CropMonoImages false
/MonoImageMinResolution 1200
/MonoImageMinResolutionPolicy /OK
/DownsampleMonoImages true
/MonoImageDownsampleType /Bicubic
/MonoImageResolution 1200
/MonoImageDepth -1
/MonoImageDownsampleThreshold 1.50000
/EncodeMonoImages true
/MonoImageFilter /CCITTFaxEncode
/MonoImageDict <<
/K -1
>>
/AllowPSXObjects true
/CheckCompliance [
/PDFX1a:2001
]
/PDFX1aCheck false
/PDFX3Check false
/PDFXCompliantPDFOnly false
/PDFXNoTrimBoxError true
/PDFXTrimBoxToMediaBoxOffset [
0.00000
0.00000
0.00000
0.00000
]
/PDFXSetBleedBoxToMediaBox true
/PDFXBleedBoxToTrimBoxOffset [
0.00000
0.00000
0.00000
0.00000
]
/PDFXOutputIntentProfile (None)
/PDFXOutputConditionIdentifier ()
/PDFXOutputCondition ()
/PDFXRegistryName ()
/PDFXTrapped /False
/CreateJDFFile false
/Description <<
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
/BGR <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>
/CHS <FEFF4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e55464e1a65876863768467e5770b548c62535370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002>
/CHT <FEFF4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc666e901a554652d965874ef6768467e5770b548c52175370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002>
/CZE <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>
/DAN <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>
/DEU <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>
/ENU (Use these settings to create Adobe PDF documents suitable for reliable viewing and printing of business documents. Created PDF documents can be opened with Acrobat and Adobe Reader 5.0 and later.)
/ESP <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>
/ETI <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>
/FRA <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>
/GRE <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>
/HEB <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>
/HRV (Za stvaranje Adobe PDF dokumenata pogodnih za pouzdani prikaz i ispis poslovnih dokumenata koristite ove postavke. Stvoreni PDF dokumenti mogu se otvoriti Acrobat i Adobe Reader 5.0 i kasnijim verzijama.)
/HUN <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>
/ITA (Utilizzare queste impostazioni per creare documenti Adobe PDF adatti per visualizzare e stampare documenti aziendali in modo affidabile. I documenti PDF creati possono essere aperti con Acrobat e Adobe Reader 5.0 e versioni successive.)
/JPN <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>
/KOR <FEFFc7740020c124c815c7440020c0acc6a9d558c5ec0020be44c988b2c8c2a40020bb38c11cb97c0020c548c815c801c73cb85c0020bcf4ace00020c778c1c4d558b2940020b3700020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e>
/LTH <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>
/LVI <FEFF0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020007000690065006d01130072006f00740069002000640072006f016100610069002000620069007a006e00650073006100200064006f006b0075006d0065006e007400750020006100700073006b006100740065006900200075006e0020006400720075006b010101610061006e00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e>
/NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken waarmee zakelijke documenten betrouwbaar kunnen worden weergegeven en afgedrukt. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.)
/NOR <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>
/POL <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>
/PTB <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>
/RUM <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>
/SKY <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>
/SLV <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>
/SUO <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>
/SVE <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>
/TUR <FEFF005400690063006100720069002000620065006c00670065006c006500720069006e0020006700fc00760065006e0069006c0069007200200062006900720020015f0065006b0069006c006400650020006700f6007200fc006e007400fc006c0065006e006d006500730069002000760065002000790061007a0064013100720131006c006d006100730131006e006100200075007900670075006e002000410064006f006200650020005000440046002000620065006c00670065006c0065007200690020006f006c0075015f007400750072006d0061006b0020006900e70069006e00200062007500200061007900610072006c0061007201310020006b0075006c006c0061006e0131006e002e00200020004f006c0075015f0074007500720075006c0061006e0020005000440046002000620065006c00670065006c0065007200690020004100630072006f006200610074002000760065002000410064006f00620065002000520065006100640065007200200035002e003000200076006500200073006f006e0072006100730131006e00640061006b00690020007300fc007200fc006d006c00650072006c00650020006100e70131006c006100620069006c00690072002e>
/UKR <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>
/RUS <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>
>>
/Namespace [
(Adobe)
(Common)
(1.0)
]
/OtherNamespaces [
<<
/AsReaderSpreads false
/CropImagesToFrames true
/ErrorControl /WarnAndContinue
/FlattenerIgnoreSpreadOverrides false
/IncludeGuidesGrids false
/IncludeNonPrinting false
/IncludeSlug false
/Namespace [
(Adobe)
(InDesign)
(4.0)
]
/OmitPlacedBitmaps false
/OmitPlacedEPS false
/OmitPlacedPDF false
/SimulateOverprint /Legacy
>>
<<
/AllowImageBreaks true
/AllowTableBreaks true
/ExpandPage false
/HonorBaseURL true
/HonorRolloverEffect false
/IgnoreHTMLPageBreaks false
/IncludeHeaderFooter false
/MarginOffset [
0
0
0
0
]
/MetadataAuthor ()
/MetadataKeywords ()
/MetadataSubject ()
/MetadataTitle ()
/MetricPageSize [
0
0
]
/MetricUnit /inch
/MobileCompatible 0
/Namespace [
(Adobe)
(GoLive)
(8.0)
]
/OpenZoomToHTMLFontSize false
/PageOrientation /Portrait
/RemoveBackground false
/ShrinkContent true
/TreatColorsAs /MainMonitorColors
/UseEmbeddedProfiles false
/UseHTMLTitleAsMetadata true
>>
<<
/AddBleedMarks false
/AddColorBars false
/AddCropMarks false
/AddPageInfo false
/AddRegMarks false
/BleedOffset [
0
0
0
0
]
/ConvertColors /ConvertToRGB
/DestinationProfileName (sRGB IEC61966-2.1)
/DestinationProfileSelector /UseName
/Downsample16BitImages true
/FlattenerPreset <<
/PresetSelector /MediumResolution
>>
/FormElements true
/GenerateStructure false
/IncludeBookmarks false
/IncludeHyperlinks false
/IncludeInteractive false
/IncludeLayers false
/IncludeProfiles true
/MarksOffset 6
/MarksWeight 0.250000
/MultimediaHandling /UseObjectSettings
/Namespace [
(Adobe)
(CreativeSuite)
(2.0)
]
/PDFXOutputIntentProfileSelector /DocumentCMYK
/PageMarksFile /RomanDefault
/PreserveEditing true
/UntaggedCMYKHandling /UseDocumentProfile
/UntaggedRGBHandling /LeaveUntagged
/UseDocumentBleed false
>>
]
>> setdistillerparams
<<
/HWResolution [600 600]
/PageSize [419.528 595.276]
>> setpagedevice
|
| id | nasplib_isofts_kiev_ua-123456789-86456 |
| institution | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| issn | 2308-5916 |
| language | English |
| last_indexed | 2025-12-07T16:02:25Z |
| publishDate | 2014 |
| publisher | Інститут кібернетики ім. В.М. Глушкова НАН України |
| record_format | dspace |
| spelling | Chimir, I.A. 2015-09-17T18:15:25Z 2015-09-17T18:15:25Z 2014 Humanlike dialogue agents / I.A. Chimir // Математичне та комп'ютерне моделювання. Серія: Технічні науки: зб. наук. пр. — Кам’янець-Подільський: Кам'янець-Подільськ. нац. ун-т, 2014. — Вип. 11. — С. 163-186. — Бібліогр.: 17 назв. — англ. 2308-5916 https://nasplib.isofts.kiev.ua/handle/123456789/86456 004.94:004.78 The paper deals with an investigation and modeling of the inner mechanisms of human-human dialogue, with the final goal to construct an artificial humanlike dialogue agent. An ontological basis of human dialogue interaction, underlying follow-up reasoning, is obtained on the basis of analysis of human-human dialogues and illustrated by one of Plato’s dialogue, Protagoras. The paper proposes two models directed to explain some inner mechanisms of human-human dialogue. Both models are synthesized for the case of one type of human-human dialogues, called erotetic dialogue. The model of the structure of dialogue transaction is created in the context of a declarative knowledge representation, based on Language of Ternary Description. The model for the cognitive cycle of dialogue interaction is in fact a development of Neisser’s cyclical model of perception. Стаття присвячена дослідженню та моделюванню внутрішніх механізмів діалогу між людьми з метою розробки штучного людиноподібного діалогового агента. Онтологічний базис діалогової взаємодії між людьми, на якому базуються дослідження, одержано шляхом аналізу діалогів між людьми і ілюструється одним з Платонівських діалогів під назвою Протагор. У статті запропоновані дві моделі, які пояснюють внутрішні механізми діалогу між людьми. Обидві моделі синтезовані для винятку одного типу діалогу між людьми під назвою еротетичний діалог. Модель структури діалогової трансакції синтезована у контексті представлення декларативних знань за допомогою Мови Тернарного Опису. Модель когнітивного циклу діалогової взаємодії є розвитком циклічної моделі перцепції Найсера. en Інститут кібернетики ім. В.М. Глушкова НАН України Математичне та комп'ютерне моделювання. Серія: Технічні науки Humanlike dialogue agents Article published earlier |
| spellingShingle | Humanlike dialogue agents Chimir, I.A. |
| title | Humanlike dialogue agents |
| title_full | Humanlike dialogue agents |
| title_fullStr | Humanlike dialogue agents |
| title_full_unstemmed | Humanlike dialogue agents |
| title_short | Humanlike dialogue agents |
| title_sort | humanlike dialogue agents |
| url | https://nasplib.isofts.kiev.ua/handle/123456789/86456 |
| work_keys_str_mv | AT chimiria humanlikedialogueagents |