2. METHOD OF AUTOMATED DEVELOPMENT AND EVALUATION OF ONTOLOGIES’ QUALITIES OF KNOWLEDGE BASES
2.1. INTRODUCTION
Knowledge base (KB) is the main component of intelligent systems, which is formed according to the subject area on which the functionality of operation system is oriented. Traditional knowledge of engineering (receiving knowledge from expert, data analysis, machine learning, etc.) are not based on a system of common and verified standards, that is why knowledge bases, built on this basis, eventually lose their functionality due to the low efficiency of its operation. Ontological engineering is used as the standard of knowledge engineering, applying of which results in receiving the ontology of knowledge base. Ontology – is a detailed formalization of a certain area of knowledge presented by means of a conceptual scheme. This scheme consists of a hierarchical structure of concepts, relationships between them, theorems and constraints, that are accepted in a particular subject area [1]. Considering the foregoing, the formal model of ontology O determines the following , where – finite set of concepts notions (concepts, terms) of subject area, which ontology defines ; – finite set of relations between concepts notions (terms, concepts) in a given subject area; – finite set of interpretation functions (axiomatization, constraints) specified on the concepts or relations of ontology . Using of ontologies as a part of KB helps to solve a number of methodological and technological types of problems that arise during the development of such systems. In particular, for Ukraine, the distinctive problems consist in the lack of conceptual integrity and consistency of certain techniques and methods of knowledge engineering; the lack of qualified professionals in this field; in the stiffness of the developed software tools and their low adaptive capacity; in the complexity of intelligent systems implementation, caused by psychological aspects. All these thesis indicate and confirm the relevance of research problems of using ontologies in the process of intelligent systems development [2-4].
2.2. THE PROBLEM FORMULATION
In order to manually design a complete related ontology for a specific subject area it is necessary to spend a lot of time and resources. It is explained that applied ontologies must contain tens of thousands of items to be suitable for solving a wide range of problems that arise in this subject areas. Manual ontology designing – is a long routine process, which also requires a thorough knowledge of a subject area and understanding of the principles of building ontologies. Therefore, methods and algorithms of automated ontology designing are actively developing. The mathematical software implementation of the automation process of designing ontology will be suggested, or rather, its development, as it is accepted that the human expert introduced the basic terms and relations between them in the ontology manually. Such initial ontology will be called base and will be denoted as . That is, ontology designing starts from the moment when it has already had some data. Therefore, this process is called base designing ontology. Formally, we will write .
Ontology – is the language of science. The language of science, as structured scientific knowledge, sets a hierarchical multilayer formation, in which the following components are distinquished: terminological system; nomenclature; tools and rules for forming conceptual apparatus and terms. So, for designing an ontology, it is necessary to build the terminological system and the nomenclature . Basic ontology necessarily contains some part of terminological system, that is . Encyclopedias, terminological and explanatory dictionaries on the basis of which terminological system of subject area is build, usually have a clear structure and consist of dictionary works. The process of building a nomenclature is more complicated. When in dictionaries terms are singled, then in scientific texts (books, monographs, etc.) they must be allocated and the properties of concepts and relations between them should be searched. Thus, the natural language technology of processing scientific text are required.
The purpose of this work is to develop the method of automated designing base ontology and evaluating its quality [5-7].
2.3. MAIN PART
2.3.1. STRUCTURAL MODEL OF ONTOLOGY CONCEPTS AND RELATIONS
Let the given set of names of relations be suggested. Then the relation in the ontology is given as a reflection from C to C, using the element of set V: . That is, relation – triplet form: . As the ontology forms the taxonomy concepts, then, using the object-oriented approach terminology, each concept represents a class. Let’s define the concept as class with this structure:
| , | (1) |
where N – the name of the concept; – set of relations in which the class C is domain (area of definition); – set of relations in which class C is the set of values; S – superclass C; D – subclasses C; A – axioms definition C, – instances C.
(Для ознайомлення з повним текстом статті необхідно залогінитись)