Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface
This paper is devoted to the analysis of sources in the field of development and building intelligent user interfaces. Particular attention is paid to presenting an ontology-based approach to constructing the architecture of the interface, the tasks arising during the development, and ways for solvi...
Gespeichert in:
| Veröffentlicht in: | Control systems & computers |
|---|---|
| Datum: | 2020 |
| Hauptverfasser: | , , |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Міжнародний науково-навчальний центр інформаційних технологій і систем НАН та МОН України
2020
|
| Schlagworte: | |
| Online Zugang: | https://nasplib.isofts.kiev.ua/handle/123456789/181216 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Zitieren: | Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface / H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko // Control systems & computers. — 2020. — № 4. — С. 44-55. — Бібліогр.: 17 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
nasplib_isofts_kiev_ua-123456789-181216 |
|---|---|
| record_format |
dspace |
| spelling |
Pidnebesna, H.A. Pavlov, A.V. Stepashko, V.S. 2021-11-06T17:49:10Z 2021-11-06T17:49:10Z 2020 Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface / H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko // Control systems & computers. — 2020. — № 4. — С. 44-55. — Бібліогр.: 17 назв. — англ. 2706-8145 https://nasplib.isofts.kiev.ua/handle/123456789/181216 004.8 DOI https://doi.org/10.15407/usim.2020.04.044 This paper is devoted to the analysis of sources in the field of development and building intelligent user interfaces. Particular attention is paid to presenting an ontology-based approach to constructing the architecture of the interface, the tasks arising during the development, and ways for solving them. An example of construction of the intelligent user interface is given for software tools of inductive modeling based on the detailed analysis of knowledge structures in this domain. Мета статті. У роботі пропонується підхід до "інтелектуалізації" програмних засобів індуктивного моделювання на основі онтологічних концепцій подання знань про предметну галузь для проектування бази знань, обчислювальних інструментів та інтелектуального інтерфейсу користувача. Цілі цього дослідження: виконати огляд джерел про сучасні засади конструювання інтелектуального інтерфейсу користувача (ІІК) , пояснити наш підхід до побудови ІІК, заснований на онтології, застосувати цей підхід до галузі індуктивного моделювання, а також навести приклад, що демонструє подальше прийняття рішень як процес моделювання на основі ІІК та подати відповідні прикінцеві зауваження. Результати. Розглянуто сучасні підходи до організації інтелектуального користувацького інтерфейсу. Описано характеристики і завдання ІІК, принципи онтологічного підходу до його конструювання та функціювання. Данная статья посвящена анализу источников в области разработки и построения интеллектуальных пользовательских интерфейсов. Особое внимание уделяется представлению основанного на онтологии подхода к построению архитектуры интерфейса, задачам, возникающих при разработке, и способов их решения. Приведен пример построения интеллектуального пользовательского интерфейса программных средств индуктивного моделирования на основе детального анализа структур знаний в этой области. en Міжнародний науково-навчальний центр інформаційних технологій і систем НАН та МОН України Control systems & computers Intellectual Informational Technologies and Systems Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface Онтологічний підхід до побудови інтелектуального інтерфейсу користувача в інструментальних засобах індуктивного моделювання Онтологический подход к разработке инструментов индуктивного моделирования с интеллектуальным интерфейсом Article published earlier |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| title |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface |
| spellingShingle |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface Pidnebesna, H.A. Pavlov, A.V. Stepashko, V.S. Intellectual Informational Technologies and Systems |
| title_short |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface |
| title_full |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface |
| title_fullStr |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface |
| title_full_unstemmed |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface |
| title_sort |
ontology application to construct inductive modeling tools with intelligent interface |
| author |
Pidnebesna, H.A. Pavlov, A.V. Stepashko, V.S. |
| author_facet |
Pidnebesna, H.A. Pavlov, A.V. Stepashko, V.S. |
| topic |
Intellectual Informational Technologies and Systems |
| topic_facet |
Intellectual Informational Technologies and Systems |
| publishDate |
2020 |
| language |
English |
| container_title |
Control systems & computers |
| publisher |
Міжнародний науково-навчальний центр інформаційних технологій і систем НАН та МОН України |
| format |
Article |
| title_alt |
Онтологічний підхід до побудови інтелектуального інтерфейсу користувача в інструментальних засобах індуктивного моделювання Онтологический подход к разработке инструментов индуктивного моделирования с интеллектуальным интерфейсом |
| description |
This paper is devoted to the analysis of sources in the field of development and building intelligent user interfaces. Particular attention is paid to presenting an ontology-based approach to constructing the architecture of the interface, the tasks arising during the development, and ways for solving them. An example of construction of the intelligent user interface is given for software tools of inductive modeling based on the detailed analysis of knowledge structures in this domain.
Мета статті. У роботі пропонується підхід до "інтелектуалізації" програмних засобів індуктивного моделювання на основі онтологічних концепцій подання знань про предметну галузь для проектування бази знань, обчислювальних інструментів та інтелектуального інтерфейсу користувача. Цілі цього дослідження: виконати огляд джерел про сучасні засади конструювання інтелектуального інтерфейсу користувача (ІІК) , пояснити наш підхід до побудови ІІК, заснований на онтології, застосувати цей підхід до галузі індуктивного моделювання, а також навести приклад, що демонструє подальше прийняття рішень як процес моделювання на основі ІІК та подати відповідні прикінцеві зауваження. Результати. Розглянуто сучасні підходи до організації інтелектуального користувацького інтерфейсу. Описано характеристики і завдання ІІК, принципи онтологічного підходу до його конструювання та функціювання.
Данная статья посвящена анализу источников в области разработки и построения интеллектуальных пользовательских интерфейсов. Особое внимание уделяется представлению основанного на онтологии подхода к построению архитектуры интерфейса, задачам, возникающих при разработке, и способов их решения. Приведен пример построения интеллектуального пользовательского интерфейса программных средств индуктивного моделирования на основе детального анализа структур знаний в этой области.
|
| issn |
2706-8145 |
| url |
https://nasplib.isofts.kiev.ua/handle/123456789/181216 |
| citation_txt |
Ontology Application to Construct Inductive Modeling Tools with Intelligent Interface / H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko // Control systems & computers. — 2020. — № 4. — С. 44-55. — Бібліогр.: 17 назв. — англ. |
| work_keys_str_mv |
AT pidnebesnaha ontologyapplicationtoconstructinductivemodelingtoolswithintelligentinterface AT pavlovav ontologyapplicationtoconstructinductivemodelingtoolswithintelligentinterface AT stepashkovs ontologyapplicationtoconstructinductivemodelingtoolswithintelligentinterface AT pidnebesnaha ontologíčniipídhíddopobudoviíntelektualʹnogoínterfeisukoristuvačavínstrumentalʹnihzasobahínduktivnogomodelûvannâ AT pavlovav ontologíčniipídhíddopobudoviíntelektualʹnogoínterfeisukoristuvačavínstrumentalʹnihzasobahínduktivnogomodelûvannâ AT stepashkovs ontologíčniipídhíddopobudoviíntelektualʹnogoínterfeisukoristuvačavínstrumentalʹnihzasobahínduktivnogomodelûvannâ AT pidnebesnaha ontologičeskiipodhodkrazrabotkeinstrumentovinduktivnogomodelirovaniâsintellektualʹnyminterfeisom AT pavlovav ontologičeskiipodhodkrazrabotkeinstrumentovinduktivnogomodelirovaniâsintellektualʹnyminterfeisom AT stepashkovs ontologičeskiipodhodkrazrabotkeinstrumentovinduktivnogomodelirovaniâsintellektualʹnyminterfeisom |
| first_indexed |
2025-11-26T02:29:19Z |
| last_indexed |
2025-11-26T02:29:19Z |
| _version_ |
1850608467450003456 |
| fulltext |
44 iSSN 2706-8145, системи керування та комп'ютери, 2020, № 4
doi https://doi.org/10.15407/usim.2020.04.044
Udk 004.8
H.A. pIdneBesnA, junior researcher,
international research and training centre
for information technologies and Systems of the NaS of Ukraine,
glushkov ave., 40, kyiv, 03187, Ukraine,
pidnebesna@ukr.net
A.V. pAVloV, phd (eng), researcher,
international research and training centre
for information technologies and Systems of the NaS of Ukraine,
glushkov ave., 40, kyiv, 03187, Ukraine,
andriypavlove@gmail.com
V.s. stepAsHko, dSci (eng), professor,
international research and training centre
for information technologies and Systems of the NaS of Ukraine,
glushkov ave., 40, kyiv, 03187, Ukraine,
stepashko@irtc.org.ua
ontology ApplICAtIon to ConstruCtIng
InduCtIVe modelIng tools wItH IntellIgent InterFACe
This paper is devoted to the analysis of sources in the field of development and building intelligent user interfaces. Particular at-
tention is paid to presenting an ontology-based approach to constructing the architecture of the interface, the tasks arising during
the development, and ways for solving them. An example of construction of the intelligent user interface is given for software tools
of inductive modeling based on the detailed analysis of knowledge structures in this domain.
Keywords: intelligent user interface; inductive modelling tools; GMDH; ontology approach; domain metamodel, human-computer
interaction; user modeling.
Introduction
The problem of constructing and implementing
various types of intelligent interface between users
and software means has long history, see e .g . [1–3] .
A series of International Conferences on Intelli-
gent User Interface (IUI) is held yearly; see the
home page [4] . As it is noted in [5], an IUI involves
the computer side having sophisticated knowledge
of the domain, and/or a model of the user . These
allow the interface to better understand the user’s
needs and personalize or guide the interaction .
Another interpretation of the IUI may be done as
follows: a program that has an intelligent interface
and uses intelligent techniques when interacting
with the user . The IUI might be based on some
user models, or on knowledge of the system func-
tionality, or on procedures helping the user [6] .
Generally, there is an idea in most sources that
an intelligent user-computer interface should pre-
dict the users’ intentions and present information
based on that meeting their current needs . Besides,
the following IUI features are eligible: adaptability
to the needs of a particular user and the ability to
learn new concepts and techniques .
iSSN 2706-8145, control systems and computers, 2020, № 4 45
Ontology Application to Constructing Inductive Modeling Tools with Intelligent Interface
The generation of interfaces, called "intelli-
gent", is designed to provide the user with ad-
ditional capabilities, including adaptability, cus-
tomization and assistance in solving problems .
The implementation of the intelligent user in-
terface should be an intelligent agent or inter-
mediary between the user and a particular com-
puter application that implements approaches
and methods to support communication with
the user [7] .
As it was mentioned in [8], there is an inten-
tion to implement the following more broad
characteristics of the intelligent user interface
when designing the intelligent modeling system:
user-system interactive mode that tracks and sup-
ports all stages the modeling process; acquisition
and active usage of the user’s knowledge when
performing the process; permanent monitoring,
testing and correcting all decisions made by user
at any stage; training the user during interaction
with the system; and others .
Under the user-system interactive mode we
mean the spectrum of possible mutual actions
starting from the fully automated mode for a
novice user to the possibility of planning the
whole modeling process by a skilled user (ex-
pert) . This is possible only when the intelligent
interface have means for the evaluating the le-
vel of user’s skills and automatic adaptation to
them [8] .
This paper offers a way of "intellectualization"
of the inductive modeling software tools by ap-
plying an ontological approach to representing
the knowledge of the domain to design a know-
ledge base, computing tools, and intelligent user
interface .
The aims of this research is to make a survey of
sources on the state-of the art in intelligent user
interface (IUI) field (Section II), to explain our
ontology-based approach to the IUI construction
(Section III), to apply this approach to the domain
of inductive modeling (Section IV), to give an
example demonstrating the consequent decision
making as an IUI-based modeling process (Sec-
tion V) and to present relevant conclusive remarks
(Section VI) .
Intelligent user Interface:
Characteristics and tasks
The user interface combines all the elements and
components of the program that can affect the user’s
interaction with the software . These include:
• a set of user tasks that he solves using the
system;
• system control elements;
• navigation between system units;
• visual (and not only) design of program
screens;
• means of displaying information;
• devices and technologies of data input/
output;
• the procedure for using the program and
documentation for it, etc .
Like traditional interfaces, intelligent user inter-
face should be easy to learn, practical and under-
standable . But, in addition, intelligent interfaces
promise to provide some additional features [9]:
• recognition of inaccurate, ambiguous and/or
partial multimodal input of information (acqui-
ring and processing information in the system us-
ing various devices such as mouse, keyboard, mi-
crophone);
• creating a coherent, unified and understan-
dable multimodal representation;
• fully or partially automatic problem solving;
• interaction management (problem-solving,
customization, interface adaptation) through re-
presentation, inference, and use models of user,
domain, task and context .
Fig . 1 illustrates main differences between IUI
and traditional interfaces .
Fig. 1. IUI Characteristics
adaptability
customization
assistance in solving
problems
practical
understandable
easy to learn
intelligent user
interface (IUI)
traditional
interface
46 iSSN 2706-8145, системи керування та комп'ютери, 2020, № 4
H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko
In other words, "intelligence" of the user inter-
face can be interpreted as some subroutine-agent
(Fig . 2) . On one hand, it monitors the user's actions,
analyzes them and helps the user by suggesting op-
tions for action . On the other hand, it connects to
a given software application and calls its commands
according to the model of dialogue with the user .
Interaction management is the core of the IUI
and is responsible for:
• interpretation and integration of recognized
information from different sources based on a data
transmission model;
• creating and adjusting a real-time dialogue
model, advancing the dialogue with the user ac-
cording to this model based on the domain model,
task model, and user model;
• recognition and formation of an action plan
based on the model of dialogue and past actions of the
user (direct input by the user or tracking his actions);
• presentation of results or formation of clari-
fying questions for the purpose of elimination of
ambiguities in the course of dialogue with the user .
An important characteristic of IUI is the abi-
lity to adapt the system response to the level of
understanding of a particular user . This ability
is based on the construction and use of the user
model and inference in the direction of appropri-
ate adaptations of the interface . The IUI adap-
tability includes adaptability both to the user and
to the context . Adaptability to the user requires the
construction of a user model, and adaptability to
the context - the model of the domain, the model
of the task and the model of dialogue . Ontologies
[10] and rules [11] can be used to build a domain
model and task models .
As well as ontologies cover conventional static
knowledge of a particular field, they are also valu-
able for other areas of research . In natural language
Fig. 2. IUI Functionality
iSSN 2706-8145, control systems and computers, 2020, № 4 47
Ontology Application to Constructing Inductive Modeling Tools with Intelligent Interface
applications, ontologies can be used for natural
language processing or for automatic extraction of
knowledge from (scientific) texts .
The initiative in the interaction between the IUI
and the user can be given to the user, the system,
or be mixed (fig . 3) [12] . Mixed-initiative systems
support the natural alternation of user contribu-
tions and systems aimed at solving tasks . At the
moment, systems with mixed-initiative are consid-
ered the most promising .
This approach declares a significant improve-
ment in human-machine interaction, allowing
the computer to behave more like a partner who
is able to work with users to develop a common
understanding of goals and contribute to prob-
lem-solving in the most appropriate way . Creating
such complex systems requires a number of time-
consuming tasks . In particular, it is necessary to
develop mechanisms for collecting information,
formal presentation of knowledge of the domain,
drawing logical conclusions about the intentions
of the user, his attention, and his abilities in condi-
tions of uncertainty .
A dialogue is built between the user and the
agent, who pursue a common goal that must be
accomplished over the application . The IUI acts
as the driving force behind this dialogue . Its role is
to maintain and monitor the state of the dialogue
and to offer the following options for the develop-
ment of events moving along a certain tree of the
plan, consisting of actions necessary to achieve the
goal or sub-goal . The state of dialogue in the sys-
tem consists of a stack of goals and an action plan
in the form of an ontological model . The ontologi-
cal model of the user's task specifies the objects of
activity, defines the classes of terms, characteris-
tics, and areas of possible values, states, and ne-
cessary to perform the tasks of linking objects . This
is necessary to work out the functionality of ap-
plications, to build a sequence of actions that solve
specific types of tasks,
Note that IUI is abstracted from the specific
ways of transmitting the information, domain,
software, task and the user . One of the modern
approaches that allow solving the above problems
and reducing the effort required for implementa-
tion is Model-Based User Interface Development
(MBUID) [11] . The purpose of this approach is to
identify high-level models that allow you to define
and analyze software from a more semantically
oriented level . This allows you to create a complete
specification of the entire interface model, with the
division into components, modules, reuse, as well
as allow the implementation of automatic coding
of components according to the built specification
of the user interface .
principles
of the ontology-Based Approach
The use of ontologies can facilitate the description
of the IUI design problem as a part of complex sys-IUI design problem as a part of complex sys-
tems from components and implement a program
that makes such a configuration independent of the
product and the components itself, makes it pos-
sible to reuse .
The advantage of an ontological approach is that
ontology defines a conceptual structured environ-
ment in which the process of constructing a model
of an automation object takes place .
Ontology is the exact specification of some field
that contains a glossary of terms and a set of domain
links describing relations between these terms . It
actually is a hierarchical conceptual skeleton of
domain .
Formal ontology model (O) is an ordered triplet
О = <Т, R, F>,
where T is finite set of terms of the domain being
described by the ontology О; R is a finite set of re-
lations between terms of the given domain; F is a
finite set of the interpretation functions given on
the terms and/or relations of the ontology О .
Additional information about classes and their
relationship to each other, as well as restrictions on
the value of attributes for each class are fixed in axi-
oms . Axioms are sometimes segrerated in a sepa-
rate (fourth) set A, then the ontology is given by the
quadruple О = <Т, R, F, A> . In the classical sense,
they are part of a set of interpretation functions .
As a rule, ontologies are classified by levels of
generality [10]:
• Top-level ontologies (meta-ontologies, repre-
sentational ontologies) do not link to any specific
48 iSSN 2706-8145, системи керування та комп'ютери, 2020, № 4
H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko
domain, operate with general concepts and rela-
tionships that do not depend on the subject area .
Such ontologies provide representative entities
without specifying what should be represented .
• Domain ontologies cover the knowledge that
applies to a particular type of domain, consists of
concepts that describe a particular subject area, sig-
nificant relationships in it, the set of these concepts
and relationships . If ontologies operate in several
domains, they are called generic or core ontolo-
gies, as well as super theories .
• Application ontologies (task ontology, applied
ontology) contain all the necessary knowledge to
model a specific process (usually a combination of
domain and method ontologies), where the con-
cepts are the types of problems to be solved and the
relationships are the decomposition of these tasks
into subtasks .
Fig. 3. IUI Architecture
iSSN 2706-8145, control systems and computers, 2020, № 4 49
Ontology Application to Constructing Inductive Modeling Tools with Intelligent Interface
The main principles of the ontological approach
to designing the user interface are [13]:
1 . To separate the development and mo-
dification of the user interface from the application
under the rules of their interaction .
2 . To combine content-homogeneous inter-
face information into separate components . For
example, you can select several basic systems of
concepts in the user interface:
• a system of user concepts through which the
user interacts with the system;
• developer concept system, which should
contain concepts to describe the structure and
means to display information in the interface,
describe the scenario of the dialogue and the
implementation of the interaction between the
application and the user interface .
3 . To present the components of the user
interface in the form of declarative models formed
on the basis of universal ontologies describing each
component .
The user interface is the part of the software
package that, given the wide range of potential us-
ers, can often adapt, taking into account different
levels of training, a wide range of tasks, and per-
sonal preferences . Implementation of the user-
friendly interface, meeting these requirements, is
possible by building layouts of the future interface,
based on the use of universal ontological models of
the user interface .
There are several types of ontological models,
and each type plays a different role in the process
of software interaction with the user (Fig . 4) .
The IUI architecture includes next ontological
models:
•The domain model D that represents the cha-
racteristics of the specific area, defines concepts,
characteristics, and areas of possible values .
•The user model V (ontology of presentation) that
collects, stores and updates knowledge about the
user so that the IUI effectively meets his require-
ments .
•The dialog script ontology model S that:
1 . stores a dynamic representation of the user-
system dialogue;
2 . defines: abstract terms to describe reactions
to events; the set of actions that occur when events
occur; event sources; types of window transitions;
and how to select window instances etc .
The task model T• that:
1 . stores the representation of the solution of a
particular task in a formal form;
Fig. 4. Types of ontological models used for IUI
50 iSSN 2706-8145, системи керування та комп'ютери, 2020, № 4
H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko
2 . determines: the set of tools and capabilities
provided by the application; the list of provided
functions that are characterized by the type of va-
lues, a set of parameters, their types, etc .
Then, the generation of the interface model
component is reduced to setting values of the cor-
responding model concepts of universal ontology
{D, V, S, T} . This representation allows imple-
menting one of the most important features of the
IUI, namely adaptability .
ontology for the domain of
the gmdH-based Inductive modeling
The task model and the domain model of the sys-
tem can be developed using ontologies of different
levels . The automation of tasks for intelligent mod-
eling systems requires simulation of the modeling
process, which is a meta-modeling . A metamodel
is a model that describes the structure and princi-
ples of other mo-dels’ operation .
A metamodel provides a logical level of the do-
main and is interpreted dynamically at the applica-
tion level . This adds additional flexibility to the sys-
tem, since the domain logic can be changed without
modifying the code . To allow or disallow a type of
communication at the logical level, it is enough to
assign it only to the meta-model formal terms .
In fact, the metamodel is defined by a high-level
ontology, in terms of concepts of solution methods,
key stages, and constraints . The ontological model
of the domain at the lower level describes the al-
gorithmic components of each particular mode-
ling method in more details . To solve a practical
problem, an ontological model of the task is used .
This model has its own parameters, specific cha-
racteristics and tolerance ranges .
Designing an ontology for a specific domain re-
quires in-depth analysis, identification of relevant
concepts, attributes, relationships, constraints, in-
stances and axioms of this domain . Such analysis of
knowledge leads to systematization, construction
of a hierarchy of concepts with their attributes, va-
lues and relations .
The results of the analysis and structuring of the
inductive modeling area help to design the GMDH
meta-ontology [14] . It contains the basic compo-
nents of the modelling process . The main principles
of generating its characteristics are determined .
When formalizing knowledge, different types of
ontologies are used to solve problems – these are the
so-called ontologies of methods and tasks ([15]) .
Ontologies of problems defining specific problems
contain terms and relations for specific tasks, and
ontologies of methods do terms that are specific to
specific methods of solving problems . Ontologies
of methods and tasks allow to clearly defining the
interaction between problem solving and know-
ledge of domains .
Inductive modeling is a model generation pro-
cess based on the analysis and generalization of
given statistical data obtained through observations
or experiments . The inductive modeling methods
provide building mathematical models of objects
and processes to solve a range of tasks:
•time series forecasting;
•classification (supervised learning);
•clusterization (unsupervised learning or self-
training: identification of effective features, forms and
rules of distinction); this problem is called “Objective
Computer Cauterization” (OCC) in GMDH field;
•determining the set of independent (inputs),
dependent (outputs), and irrelevant (uninforma-
tive) variables among the measured ones with the
aim to build an adequate model of the system; this
problem is called “Objective System Analysis”
(OSA) in GMDH field .
A number of methods are used to solve different
types of problems . In cases where it is necessary to
model objects, parametric GMDH algorithms have
proved to be effective . In cases of modeling objects
with fuzzy characteristics, it is more effective to
use nonparametric GMDH algorithms, in which
polynomial models are replaced by a sample of data
divided into intervals or clusters . Algorithms of this
type completely solve the problem of eliminating
the bias of coefficient estimates [16] .
The ontology of methods specifies a set of
GMDH algorithms .
Parametric:
•the basic Combinatorial (COMBI) algo-
rithm is based on full or reduced sorting-out of
iSSN 2706-8145, control systems and computers, 2020, № 4 51
Ontology Application to Constructing Inductive Modeling Tools with Intelligent Interface
gradually complicated models and evaluation of
them by external criterion on separate part of
data sample;
• Multilayered Iteration algorithm (MIA)
uses at each layer of sorting procedure the same
partial description (iteration rule) . It should be
used when it is needed to handle a big number
of variables;
•Objective System Analysis (OSA) algorithm .
The key feature of it is that it examine not single
equations but systems of algebraic or difference
equations obtained by implicit templates (with-
out goal function) . An advantage of the algorithm
is that the number of regressors is increased con-
sequently, the information embedded in the data
sample is utilized better;
•Two-level (ARIMAD) algorithm for mode-
ling of long-term cyclic (such as stock or weather)
processes . There are used systems of polynomial or
difference equations for identification of models
on two time scales and then choice of the best pair
of models by external criterion value .
Non-parametric algorithms are exemplified by:
•Objective Computer Clusterization (OCC) al-
gorithm that operates with pairs of closely spaced
sample points [5] . It finds physical clusterization
that would as possible be the same on two sub-
samples;
•“Pointing Finger” (PF) algorithm for the
search of physical clusterization . It is implemented
by construction of two hierarchical clustering trees
and estimation by the balance criterion;
•Analogues Complexing (AC) algorithm, which
use the set of analogues instead of models and clus-
terizations [8] . It is recommended for the fuzziest
objects .
Inductive modeling is a process of directed
model generation and selection, consisting of a fin-
ite set of successive stages . All methods of inductive
modeling have standard components . Those can be
used as a basis to form a metamodel of inductive
modeling .
The principles of algorithmic modules formati-
on for solving a specific problem can be determined
as a result of the domain structuring . Depending on
the task type, a relevant method of solving it can be
selected .
structuring the successive
process of Inductive modeling
A step-by-step approach is applied to solve prob- is applied to solve prob-
lems of modeling processes and objects from the
given data set . This approach is a consecutive se-
lection of «the best» solutions from a set of possible
options under condition of existence of the target
solution . An algorithm builds a tree of actions and
puts questions based on the knowledge about the
current action and a given goal . The peculiarity of
the algorithm is that it does not produce the whole
plan of modeling in detail but only a part of it . The
plan is formed in more details in the course of the
modeling process (Fig . 5) .
The process of any real-world problem solution
can be characterized by the following stages:
1 . Data preparation: creation and maintenance
of data files or databases, and their pre-processing
(smoothing, gaps handling, outlier filtering, sca-
ling, centering, normalization) .
2 . Task preparation: determination of the prob-
lem type (static data or real-time data modeling),
the purpose of modeling (approximation, inter-
polation, extrapolation, trend selection, etc .),
and primary data analysis (determining statistical
characteristics, correlation, trend and oscillatory
properties) .
3 . Task definition: determination of the object
class (linear or nonlinear, static or dynamic, sta-
tionary or nonstationary etc .) and selection of the
basic functions class (polynomials, trigonometric
series, difference equations etc .) .
4 . Modeling problem statement: choosing the
model quality criterion, the type of the model
structure generator, and the parameter estimating
method .
5 . Solving the stated problem: the actual modeling
procedure is performed based on the given data .
Inductive modeling is a process of sequential
decision making consisting of a certain set of
successive stages . Therefore, we can decompose
the solution of the relevant modeling method
selection problem into certain sub-problems
to solve them individually in a sequential man-
ner (Fig . 6) [17] .
52 iSSN 2706-8145, системи керування та комп'ютери, 2020, № 4
H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko
Suppose, for example, that the data is pre-pro-
cessed and does not contain gaps, outliers, and so
on . Then, the appropriate sequence of decision-
making stages, taking into account the results of
this report, will be as follows:
1 . Choice of the modeling purpose (approxi-
mation, interpolation, extrapolation, trend de-
termination, construction of an input-output
model etc .) .
2 . Determination of the process type (linear static
object, non-linear static object, linear time series,
non-linear time series, linear dynamic process,
nonlinear dynamic process) .
3 . Determination of process stationarity (sta-
tionary, with increasing, decreasing, or oscillatory
trend, sum of trends) .
4 . Choice of the model class (linear regression,
autoregression, autoregression with the trend, har-
monic, logarithmic, polynomial or exponential
functions of time, difference equations etc .) .
5 . Choice of the external criterion for the mo-
del selection (Fisher criterion, Akaike criterion,
Mallows C
p
-statistics, jack-knife, unbiasedness,
regularity criterion etc .) .
6 . Choice of the parameter estimation method
(LSM, LMM, ridge regression etc .)
7 . Choice of the structure generation method (a
given structure, nested structures, inclusion, ex-
clusion, stepwise, branch and bound, exhaustive
search, combinatorial, combinatorial-selective,
multilayered, relaxation, genetic etc .) .
Fig. 5. Fragment of the ontology describing main stages and
methods used throughout the inductive modeling process
Fig. 6. Decomposition of the process of sequential
decision making
iSSN 2706-8145, control systems and computers, 2020, № 4 53
Ontology Application to Constructing Inductive Modeling Tools with Intelligent Interface
8 . The model validation (Fisher statistics, R2 cri-
terion, accuracy on a validation data set etc .)
At each of these stages, one of a number of pos-
sible decisions (done by experts, based on the lite-
rature or experience) is made, following which one
can move to the next stage . If one uses the decom-
position of the proposed stages, then, depending on
the decisions made at the previous stages, the set
of potential solutions will be significantly reduced
further . This is because the decision at each stage
imposes implicit restrictions on the application of
certain methods required at the subsequent stages .
In general, it is clear that the proposed decom-
position of the process into stages and, in turn, the
stages into basic and auxiliary solutions are also the
structure of an intellectual dialogue . Its application
in the IUI decision-making system is a condition
that the user could get a model of the studied object
using the modeling system, regardless of the level
of his/her training, a priori knowledge of the object
and the initial data . Evidently, the more prepared
is the user and the more information he/she holds
on the object or process, the more adequate model
will be built .
It might be noted that even a user who firstly try
to solve a modeling task using such system can al-
ways get maybe a simple but still satisfactory model
applying the automatic mode of the system .
REFERENCES
1 . Intelligent Interfaces Theory, Research, and Design . P .A . Hancock and M .H . Chignell (eds), North Holland,
New York, 1989 .
2 . Kolski, C., Strugeon, E.Le., 1998 . A review of "intelligent" human-machine interfaces in the light of the ARCH
model, Int . J . of Human-Computer Interaction, 10 (3), pp . 193–231 .
3 . Rogers, Y., Sharp, H., Preece, J., 2011 . Interaction Design: Beyond Human-Computer Interaction (3rd ed), Wiley,
Chichester .
4 . Welcome to ACM IUI 2020! [online] Available at: <https://iui .acm .org/2020>[Accessed 20 Feb . 2020] .
5 . Sonntag, D., 2012 . Collaborative Multimodality, Kunstliche Intelligenz, Springer, 26 (2), pp . 161-168, DOI:
10 .1007/s13218-012-0169-4, http://iui .acm .org/2018/ Last accessed 2019/19/22 .
6 . Intelligent User Interfaces . [online] Available at: <https://web .cs .wpi .edu/Research/airg/IntInt/intint-outline .
html>[Accessed 29 Dec 2019] .
Conclusion
In recent decades, we can distinguish two dif-
ferent views on the progress and development
of intelligent user interfaces . The first group of
scientists is passionate with the development of
methods of presentation and mechanisms of logi-
cal inference evaluation and prediction of the
attention and intentions of the user . The second
group of researchers focuses on the development
of automated tools and models aimed at impro-
ving the user's ability to work directly with the
application . We believe that their symbiosis can
offer a significantly new level of use of systems,
which is characterized by deeper and more natu-
ral human-computer cooperation .
The paper presents the principles of building
a user interface based on ontology models in the
development of inductive modeling systems . This
approach makes it possible to simplify the design of
efficient, user-friendly interfaces of different levels
of training, which have the characteristics of an in-
telligent interface . We believe that the proposed ap-
proach provides an advantage due to the high level
of the formalism of ontologies, their independence
from languages and programming tools, the ability
to make changes in the structure of the interface in-
dependently of other parts of an applied program,
automatic coding based on ontologies .
54 iSSN 2706-8145, системи керування та комп'ютери, 2020, № 4
H.A. Pidnebesna, A.V. Pavlov, V.S. Stepashko
7 . Introduction to Model-Based User Interfaces. [online] Available at: <https://www .w3 .org/TR/2014/
NOTE-mbui-intro-20140107>[Accessed 22 March 2020] .
8 . Stepashko, V., 2019 . On the Self-organizing Induction-Based Intelligent Modeling, Advances in Intelligent
Systems and Computing III / N . Shakhovska, M .O . Medykovskyy, Editors, AISC book series, Cham: Springer,
871, pp . 433–448 .
9 . Intelligent User Interfaces: An Introduction, Morgan Kaufmann, RUIU, San Francisco, 1998, pp . 1–13 .
10 . Gruber, T., 1995 . Toward principles for the design of ontologies used for knowledge sharing, Int . J . Human-
Computer Studies, 43(5-6), pp . 907–928 .
11 . Ahmad, A.-R., Basir, O., Hassanein, Kh., 2004 . Adaptive User Interfaces for Intelligent E-Learning: Issues and
Trends, Proc . ICEB’2004, pp . 925–934 .
12 . M. Maybury, 1998, Intelligent User Interfaces: An Introduction, RUIU, San Francisco: Morgan Kaufmann,
pp . 1- 13 .
13 . Pidnebesna, H., 2014 . An ontological approach to user interface design for inductive modeling systems, Inductive
modeling of complex systems, Kyiv: IRTC ITS NASU, 6, pp . 117–125 (In Ukrainian) .
14 . Pidnebesna, H., Stepashko, V., 2018 . On Construction of Inductive Modeling Ontology as a Metamodel of the
Subject Field, Int . Conf . Advanced Computer Information Technologies (ACIT-2018), Ceske Budejovice,
University of South Bohemia, pp . 71–74 .
15 . Studer, R., Benjamins, V.R., Fensel, D., 1998 . Knowledge Engineering: Principles and methods, Data & Knowledge
Engineering, 25, pp . 161–197 .
16 . Spectrum of GMDH algorithms. [online] Available at: <http://www .gmdh .net/GMDH_alg .htm> [Accessed
22 Sept . 2020] .
17 . Stepashko, V.S., Zvorygina, T.F. 2003 . On an approach to the decision inference problem in an intricate task, CSC
№ 6, C . 82-87 (In Russian) .
Received 17 .08 .2020
iSSN 2706-8145, control systems and computers, 2020, № 4 55
Ontology Application to Constructing Inductive Modeling Tools with Intelligent Interface
Г.А. Піднебесна, молодший науковий співробітник,
Міжнародний науково-навчальний центр
інформаційних технологій та систем НАН та МОН України,
просп . Глушкова, 40, Київ 03187, Україна,
pidnebesna@ukr .net
А.В. Павлов, к .т .н ., науковий співробітник,
міжнародний науково-навчальний центр
інформаційних технологій та систем НАН та МОН України,
просп . Глушкова, 40, Київ 03187, Україна,
andriypavlove@gmail .com
В.С. Степашко, д .т .н ., професор,
Міжнародний науково-навчальний центр
інформаційних технологій та систем НАН та МОН України,
просп . Глушкова, 40, Київ 03187, Україна,
stepashko@irtc .org .ua
ОнТОЛОгІчний ПІдхІд дО ПОБУдОВи ІнТЕЛЕкТУАЛьнОгО ІнТЕРфЕйСУ
кОРиСТУВАчА В ІнСТРУмЕнТАЛьних ЗАСОБАх ІндУкТиВнОгО мОдЕЛюВАння
Вступ. Генерація інтерфейсів, яка називається "інтелектуальною", призначена для надання користувачеві додаткових
можливостей, включаючи адаптивність, налаштування на конкретного користувача та інтерактивну допомогу у вирі-
шенні проблем . Інтелектуальний інтерфейс користувача має бути інтелектуальним агентом або посередником між ко-
ристувачем і певним комп’ютерним додатком, що реалізує підходи та методи підтримки комунікації з користувачем .
мета статті . У роботі пропонується підхід до "інтелектуалізації" програмних засобів індуктивного моделюван-
ня на основі онтологічних концепцій подання знань про предметну галузь для проектування бази знань, обчис-
лювальних інструментів та інтелектуального інтерфейсу користувача . Цілі цього дослідження: виконати огляд
джерел про сучасні засади конструювання інтелектуального інтерфейсу користувача (ІІК) , пояснити наш підхід
до побудови ІІК, заснований на онтології, застосувати цей підхід до галузі індуктивного моделювання, а також
навести приклад, що демонструє подальше прийняття рішень як процес моделювання на основі ІІК та подати
відповідні прикінцеві зауваження .
Результати. Розглянуто сучасні підходи до організації інтелектуального користувацького інтерфейсу . Описано
характеристики і завдання ІІК, принципи онтологічного підходу до його конструювання та функціювання .
Стан діалогу в системі визначається набором цілей та планом дій у формі онтологічної моделі завдань користувача,
яка визначає об’єкти діяльності, класи термінів і характеристик та області їхніх можливих значень і станів, необхідних
для виконання завдань зв’язування об’єктів . Це необхідно для відпрацювання функціональних можливостей додат-
ків, побудови послідовності дій, що вирішують конкретні типи завдань . Важливо брати до уваги, що IUI абстрагується
від конкретних способів передачі інформації, домену, програмного забезпечення, завдання та користувача .
Результати аналізу і структурування знань загалом у галузі індуктивного моделювання є базисом для роз-
роблення метаонтології конкретної предметної галузі індуктивного моделювання на основі МГУА . Така онтологія
містить основні компоненти процесу моделювання . Визначено основні принципи формування його характерис-
тик . Представлено фрагмент онтології домену індуктивного моделювання .
Показано, що розкладання процесу розв'язання будь-якої реальної задачі на етапи і, у свою чергу, етапів на осно-
вні та допоміжні рішення сприяє формуванню структури інтелектуального діалогу . Застосування його в системі при-
йняття рішень ІІК є умовою того, щоб користувач міг отримати модель об'єкта дослідження за допомогою відповід-
ної системи моделювання, незалежно від рівня його/її підготовки, апріорних знань про об'єкт та початкових даних .
Висновки . У статті представлено принципи побудови інтерфейсу користувача на основі онтологічних моделей
при конструюванні систем індуктивного моделювання . Такий підхід дозволяє спростити розроблення ефектив-
них і зручних інтерфейсів для різних рівнів підготовки користувача, які мають інтелектуальні характеристики . Ми
вважаємо, що запропонований підхід забезпечує перевагу завдяки високому рівню формалізму онтологій, неза-
лежності від мов і засобів програмування, можливості вносити зміни в структуру інтерфейсу незалежно від інших
частин прикладних програм, автоматичного кодування на основі онтологій .
Ключові слова: інтелектуальний інтерфейс користувача; засоби індуктивного моделювання; МГУА; онтологічний
підхід; метамодель предметної галузі, взаємодія людина-комп’ютер .
|