Українська   Русский
DonNTU   Masters' portal

Abstract

Content

Introduction

Expert systems are complex systems that are designed to help experts in various fields to solve problem situations. They are widely used in many fields of activity. There are two types of knowledge: collective knowledge and personal experience. Expert systems are created so that a specialist can share personal experience with the system, and it, in turn, will help other less qualified users in this area.

1. The relevance of the topic

One of the most promising and dynamically developing areas of artificial intelligence is ontology. This state of Affairs makes it an urgent task to find new ways to develop tools for working with ontologies, and, in particular, ontology editors. An ontology can be understood as:

not so long ago, ontology editors referred to systems that are more intended only for a full description of the concepts of various subject areas, with a rather weak tool base for entering rules or products. This state allowed us to consider two classes of systems: ontology editors, on the one hand, and expert systems and tool systems for creating expert systems, on the other hand. Currently, ontology editors for more acquire the features of instrumental systems for creating expert systems, but built already, unlike the old systems of this class, on the basis of a full-fledged domain model. Modern ontologies perform three tasks:

problem Statement:

There are many different implementations of expert systems. This article discusses the possibility of developing an expert system using logical inference in intelligent patterns constructed as ontologies of objects with physical semantics. The existing pattern tool allows you to select a specific class of some type, but in a very limited form. If you make the pattern intelligent (productive) this would greatly increase its efficiency. The intended implementation assumes that the pattern is a knowledge module with this structure:

I.e. at the input:

the output:

Also, this implementation would allow you to synthesize program code based on the production approach.

2. Purpose and objectives of the study, planned results

The purpose of the study is:

Investigation of methods for organizing logical inference in intelligent patterns constructed as ontologies of objects with physical semantics

Main research objectives:

Development of an expert system using logical inference in intelligent patterns constructed as an ontology of objects with physical semantics. The object of the research: tools for the organization of logical inference, a means of verification knowledge-base, ontology. Subject of the research: effectiveness of means of organizing logical inference, means of verifying the knowledge base, ontologies, the possibility of developing an expert system based on these components. It is planned to prove that:

Problem Statement:

There Are many different implementations of expert systems. This article discusses the possibility of developing an expert system using logical inference of logical inference in intelligent patterns constructed as an ontology of objects with physical semantics. The existing pattern tool allows you to select a specific class of some type, but in a very limited form. If you make the pattern intelligent (productive) this would greatly increase its efficiency. The intended implementation assumes that the pattern is a knowledge module with this structure:

I.e. at the input:

At the output:

This implementation would also make it possible to synthesize program code based on the production approach.

3. Research and development overview

This topic is popular not only in international, but also in national scientific communities.

This section will provide an overview of research in the field of knowledge base verification, tools for organizing logical inference, and ontology.

3.1 Overview of international sources

Rules and ontologies for the semantic web, Thomas E., Giovambattista I., Thomas K., Alex P. [2], discusses rules and ontologies that play a key role in the multi-level architecture of the semantic Internet, since they are used to assign meaning to data in the network and to discuss them. Although the semantic web ontology layer is quite advanced and the web ontology language (OWL) has been a W3C recommendation for several years, the rules layer is much less developed and is an active area of research; to date, a number of initiatives and proposals have been put forward, but no standard has yet been released. There are many implementations of rule mechanisms that work with semantic web data in one way or another. This article provides an exhaustive overview of such systems, describes the languages they support, and establishes their relationships with theoretical approaches to combining rules and ontologies, as provided for in the semantic web architecture. The technical problems underlying the integration of rules and ontologies are considered, and representative proposals of theoretical approaches to integration are classified into various categories.

In the Expert system based on ontology for General practitioners for the diagnosis of cardiovascular disease [3] presented an expert system to help General practitioners (GP) to diagnose any disease of the coronary arteries. The system provides experts with diagnostic strategies that can be used, and suggests medications and/or other necessary operations that need to be performed, along with an explanation of the decision. The design of the system relies on ontological knowledge of the patient's symptoms to create a knowledge base, and then it uses the semantic web rule language (SWRL) to determine the appropriate medication and the necessary operation for the patient. The system was tested by several General practitioners using 16 instances to validate and evaluate the system. The reminder and accuracy coefficients were calculated at 0.83 and 0.87, respectively.

3.2 Overview of national sources

In the paper Complex verification of production knowledge bases using vtf logics, Arshinsky L.V., Ermakov A.A., Nitezhuk M.S. [4], the procedure for verifying production knowledge bases using logics with vector semantics in the VTF logics variant with a special representation of facts and rules is considered. We also consider the use of logics with vector semantics for verification of the apparatus, which retain the ability to infer abnormal truth values.

In the article Using production expert systems for analyzing cognitive models, Pesterev D.V. [5], the possibility of applying cognitive modeling in energy security research is considered. The author considers the problem of manual analysis of cognitive models and suggests using the capabilities of production expert systems to automate the analysis.

In the Non-classical logic in the task of verification of the production of knowledge bases, Nitiuk M.S., Arshinsky L.V. [6], the possibility of the applicability of non-classical logical calculus to the problem of verifying production knowledge bases. It is explained that a good approach to verification is to use logics with vector semantics in the form of VTF logics.

In the article Data Verification in task tracking systems using product rules, Katerinenko R.S., Bessmertny I.A.[7], a method for verifying data in task tracking systems using the product rules model is proposed . The application of the developed verification system for a real software development project is described.

In Semantic technology of designing knowledge bases, Ivashenko V.P. [8], describes the technology of designing of knowledge bases based on universal semantic way of coding knowledge. An approach to building knowledge bases on a modular basis is proposed.

In the article Semantic features of technical documentation: an ontological view of correctness problems, Sidorova E.A., Garanina N.O., Kononenko I.S., Borovikova O.I. [9], the features of technical documentation are considered from the point of view of extracting the necessary content information. The main feature of the proposed approach is the use of ontology as a link between the text and formal verification.

In Model inference, Sidorova E.A. [10], the proposed model of logical inference in the form of a bipartite graph where one type of vertices corresponds to the predicate letters and the other divided into two subtypes, the favorable clauses and sets, respectively.

In the article An analysis of ontology editors, in terms of representation of classes and production rules, Kutelyov R.S., Work, M.I., Grigoriev A.V. [11], a comparative analysis of ontology editors. An analysis of the representation of classes and patterns is performed. The possibility of using products for class synthesis in ontology editors is studied.

3.3 Overview of local sources

Consider the work of other masters.

Valery A. Chaika's paper Information system for forming the knowledge base of scientific and technical events [12] deals with the problem of preserving data extracted from documents of scientific and technical events. A way to solve this problem by moving to a higher level of information representation – the semantic level–is proposed.

Nikita O. Bilyk's paper Models and algorithms for updating the knowledge of expert systems based on the ontological approach [13] describes the process of developing an expert system and implementing an ontology into it. The author considers what advantages this approach will bring.

4. Structure of the expert system

4.1 knowledge Base in expert systems

The Central element of any expert system is the knowledge base. Knowledge base – the core of the intelligent system, the combination of domain knowledge, written in machine carrier in a form understandable to the expert and user[15].

As a rule, at the initial stage of development, knowledge bases contain a fairly large number of errors. This is primarily due to the fact that in most cases intelligent training systems are created for complex, poorly formalized subject areas. Another factor complicating the development of knowledge bases is the complexity of obtaining knowledge from an expert. The presence of a large number of errors in the knowledge base significantly worsens the quality of the intelligent system as a whole, which can lead to partial or even complete inactivity. As a result, there is a need for continuous verification of knowledge bases of intelligent systems throughout their entire life cycle. Currently, the most common methods of checking knowledge bases for correctness are manual methods, i.e. methods in which an expert in a subject area, together with a knowledge engineer, consistently checks knowledge for correctness, conducting various kinds of tests to identify all kinds of errors. In this situation, it is obvious that with increasing size and complexity knowledge base increasing greatly the cost of such testing, which leads to the impossibility of complete knowledge test for correctness[16].

4.2 Logical output

At first glance, the inference process seems quite simple – the same type of operations are performed to iterate through the database entries and compare them with existing facts, until a solution or a certain target fact is found. However, managing the output process, regardless of the context of the problem, is not very effective. When solving problems in the real world people very rarely resorts to brute force the data. Instead, people use heuristic rules that significantly limit the search space for solutions and allow you to quickly and efficiently solve problems. Heuristic knowledge has an empirical nature, that is, it is formed on the basis of an expert's experience and intuition. A striking example of the superiority of the heuristic approach over the algorithmic one (based on full or partial search) is the game of chess [17]

There are two main types of logical inference: forward and reverse. Direct inference corresponds to the usual course of solving the problem–from the initial facts to the target ones. An example of direct inference is the classification problem. The ES performs a gradual generalization of the initial facts describing the properties of the object under study, identifying the most characteristic features of a particular class. Reverse inference corresponds, as the name implies, to the inverse problem of determining what facts are required to confirm a given goal. This type of inference corresponds to the opposite course of the solution: first, the inference machine considers those BZ rules whose effect is to infer the target fact. Then new subgoals are selected from the conditions of these rules, and the process continues from the target facts to the source ones. We can say that when the reverse conclusion is made, the properties of the object under study are specified. This type of logical inference gives ES a new fundamental property – the ability to explain how a solution was obtained, or what is required for a particular fact to take place. Real-world systems usually use a combination of forward and reverse output[18].

4.3 Ontologies in expert systems

An ontology can form the framework of a knowledge base, create a basis for describing the main concepts of a subject area (software), and serve as a basis for integrating databases containing actual knowledge necessary for the full functioning of an expert system. In addition, expert rules can be described in terms of an ontology, which significantly increases their level of description and understanding by expert users.

Decision support System is an interactive automated information and analytical system that helps decision makers use data and models to solve their professional poorly formalized tasks. DSS and ES are systems of almost the same class. often a DSS has several different ES in its composition, so everything that was said above about the role of ontology in ES also applies to DSS. However, there are aspects of using ontologies that are specific to the DSS. So, in view of the weak formalizability of the tasks solved by the DSS, it is very important to have a detailed consistent description of the problem area within which the DSS supports the solution of decision–making tasks. An ontology is an essential tool for creating such a description. Most types of DSS use large amounts of heterogeneous data and knowledge. Due to the fact that the ontology allows you to explicitly describe the semantics of data and knowledge, it provides a basis for their integration and sharing in problem solving.

The scientific knowledge portal is a specialized Internet portal that provides systematization of knowledge and information resources of a given field of knowledge, their integration into a single information space and meaningful access to them. The ontology is used as the core of the knowledge portal information model. By introducing a formal description of the system of concepts that exist in the portal's knowledge domain in the form of object classes and relationships between them, the ontology also defines structures for representing real objects and relationships between them. Accordingly, data in the portal content is represented as a set of different types of information objects-instances of ontology classes that are linked by relationships that are also defined in the ontology. This creates a basis for easy navigation through the portal content and meaningful search in it. When developing POPS, it is possible to use ontologies that were previously developed for the domain of knowledge of these systems. This allows you to reuse proven knowledge that ensures high quality of the systems and their potential to integrate with already deployed systems[19].

Outputs

In this paper, the following research tasks were performed:

Based on the obtained knowledge, the implementation of an expert system using logical inference in intelligent patterns constructed as an ontology of objects with physical semantics.

References

  1. Инструментальные средства проектирования онтологий, 2020 Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  2. Thomas E., Giovambattista I.,Thomas K., Axel P., Rules and Ontologies for the Semantic Web, Institut f`ur Informationssysteme, Technische Universit`at Wien Favoritenstrae, Department of Mathematics, Universit`a della Calabria, Digital Enterprise Research Institute, National University of Ireland, 2008 Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  3. Baydaa Taha Al-Hamadani, Raad Fadhil AlwanAn, Department of Computer Science, Zarqa University., Department of Computer Science, Philadelphia University., Ontology-Based Expert System for General Practitioners to Diagnose Cardiovascular Diseases, 2015 Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  4. Аршинский Л.В., Ермаков А.А., Нитежук М.С., Комплексная верификация продукционных баз знаний с использованием vtf-логик, иркутский государственный университет путей сообщения, 2020 режим доступа:[Ссылка] (дата обращения: 18.11.2020).
  5. Пестерев Д.В. Использование продукционных экспертных систем для анализа когнитивных моделей, Институт систем энергетики им. Л.А. Мелентьева, 2018, Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  6. Нитежук М.С., Аршинский Л.В., Неклассические логики в задаче верификации продукционных баз знаний, Иркутский государственный университет путей сообщения, 2020, Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  7. Катериненко Р.С., Бессмертный И.А., Верификация данных в системах отслеживания задач с помощью продукционных правил, Научно–технический вестник информационных технологий, механики и оптики, 2013 [Ссылка на сборник] (дата обращения: 18.11.2020).
  8. Ивашенко В.П. Семантическая технология проектирования баз знаний, Белорусский государственный университет информатики и радиоэлектроники, 2009 Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  9. Сидорова Е.А., Гаранина Н.О., Кононенко И.С., Боровикова О. И. Семантические особенности технической документации: онтологический взгляд на проблемы корректности, Институт систем информатики им А.П. Ершова СО РАН, 2018 Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  10. Хорошавин Л.О., Шишмарев М.С., Диев А.Н., Шайкин А.Н. Модель логического вывода, Российский химико-технический университет им Д. И. Менделеева, 2010 Режим доступа:[Ссылка] (дата обращения: 18.11.2020).
  11. Кутелёв Р.С., Пахота М.И., Григорьев А.В., Анализ редакторов онтологий с точки зрения представления классов и продукционных правил, Донецкий национальный технический университет, 2020 [Ссылка на сборник] (дата обращения: 18.11.2020).
  12. Чайка В.А. Информационная система формирования базы знаний научно-технических мероприятий, Донецкий национальный технический университет, 2020 Режим доступа: [Ссылка] (дата обращения: 18.11.2020).
  13. Билык Н.О. Модели и алгоритмы обновления знаний экспертных систем на основе онтологического подхода, Донецкий национальный технический университет, 2013 Режим доступа:[Ссылка] (дата обращения: 18.11.2020).
  14. Экспертные системы // [Ссылка]– Загл. с экрана; (дата обращения: 18.11.2020).
  15. Гаврилова Т.А., Хорошевский В.Ф. Базы знаний интеллектуальных систем. Учебник – СПб.:Изд-во «Питер». 2001 – 384 с.
  16. Рыбина Г.В. Автоматизированное построение баз знаний для интегрированных экспертных систем // Изв. РАН. Теория и системы управления. №5,1998, С.152-166.
  17. Элти Дж., Кумбис М., Экспертные системы: концепции о примеры. М.: Финансы и статистика, 1987.
  18. Муромцев Д.И. Введение в технологию экспертных систем. – СПб: СПб ГУ ИТМО, 2005. – 93 с.
  19. Загорулько Ю.А., Загорулько Г.Б., Институт систем информатики им. А.П.Ершова СО РАН, Онтологии и их практическое применение в системах, основанных на знаниях, Режим доступа: [Ссылка] (дата обращения: 18.11.2020).