期刊名称：Eastern-European Journal of Enterprise Technologies
印刷版ISSN：1729-3774
电子版ISSN：1729-4061
出版年度：2017
卷号：3
期号：2
页码：11-17
DOI：10.15587/1729-4061.2017.103630
语种：English
出版社：PC Technology Center
摘要：The problem of operation of intelligent agents of action planning with the use of ontological approach was studied. Operation of intelligent agents is possible based on the knowledge of the subject-area, in other words, the knowledge base is used. Ontologies became the standard of knowledge base. Therefore, there arises the problem of development of methods and means of operation of intelligent systems based on ontologies, in particular intelligent agents of action planning.The method of functioning of intelligent agents of action planning based on ontologies was developed. For this purpose, weights of importance of concepts and relationships were introduced to the structure of ontology. These weights are used for finding a path in the space of states. The space of states itself is built by using the language of requests to ontology. Optimization problem, which assigns the rational behavior of an intelligent agent, is two-criterial. To solve it, we chose the method of the main component, if objective functions may be evaluated, or the method of complex criterion, if these functions are impossible to evaluate.Dimensionality of the space of states depends on the completeness of the ontology, and behavior effectiveness of an intelligent agent depends on the relevance of ontology. With this aim, in the course of automated development of ontology, we developed a method for evaluation of reliability of information sources that are used for developing ontologies. As a result of the studies, it was found that this approach allows us to increase operational efficiency of intelligent agents, if the process of ontology development is relevant to the needs of a subject domain.The developed approach may serve as a base for constructing a unified methodology for development of intelligent agents of action planning if ontology of a subject domain is the central component of this software complex.
关键词：ontology;intelligent agent;natural language processing;concept;space of states;action planning
