Article 5414

Title of the article

ARCHITECTURE AND PRINCIPLES OF SELF-LEARNING OF ENGINEERING AND DESIGN EXPERT SYSTEMS BASED ON KNOWLEDGE USING MONITORING OF INTERNET

Authors

Averchenkov Andrey Vladimirovich, Doctor of engineering sciences, associate professor, senior staff scientist, Institute of engineering and design informatics of the Russian Academy of Sciences (building 1a, 18 Vadkovsky lane, Moscow, Russia), mahar@mail.com
Leonov Evgeniy Alekseevich, Candidate of engineering sciences, senior staff scientist, Institute of engineering and design informatics of the Russian Academy of Sciences (building 1a, 18 Vadkovsky lane, Moscow, Russia), johnleonov@gmail.com

Index UDK

004.93'12

Abstract

Background. Most of modern intelligent systems based on knowledge use ontologies. The quality of such systems depends on the quality of the ontologies used - their relevance, integrity and consistency. Creation and extension of ontologies are an extremely time-consuming process requiring maximum automation. The article discusses approaches to building of engineering and design expert systems based on knowledge that are capable of building their own domain knowledge using information from the Internet. Under the approach it is proposed to use the closed cycle in which the system automatically learns to find high quality documents within the domain by using ontology and preferences of the experts, as well as extends ontology by knowledge extracted from the retrieved documents.
Materials and methods. The proposed concept of the system is based on using the approaches to creation of multi-agent systems. In order to describe the learning algorithms of metasearch subsystems and features of using thereof at ontologies learning the authors used the general theory of sets and tuple mathematics.
Results. The researchers developed a conceptual architecture of self-learning systems based on knowledge taken from the Internet. The authors offered a learning algorithm for metasearch subsystems using active control and considered the application features of ontology learning methods within the self-learning systems that interact with the Internet through metasearch subsystems, as well as proposed a multi- agent architecture of intelligent expert systems based on knowledge from the Internet.
Conclusions. The proposed approach will allow to create a system that will be capable to constantly increase its own knowledge by obtaining the documents from the Internet and solve target problems by using the current actual state of a knowledge area.

Key words

expert systems, databases, extention of ontologies, monitoring of Internet, learning of ontologies.

Download PDF
References

1. Averchenkov V. I. Vestnik Bryanskogo gosudarstvennogo tekhnicheskogo universiteta [Bulletin of Bryansk State Technical University]. 2011, no. 2 (30), pp. 101–110.
2. Averchenkov V. I., Averchenkov A. V., Leonov E. A. Vestnik komp'yuternykh i informatsionnykh tekhnologiy [Bulletin of computing and information technologies]. 2012, no. 1, pp. 38–45.
3. Averchenkov A. V., Leonov E. A. Izvestiya Volgogradskogo gosudarstvennogo tekhnicheskogo universiteta [Proceedings of Volgograd State Technical University]. 2011, no. 11 (84), pp. 30–35.
4. Niazi M., Hussain A. Scientometrics. 2011, vol. 89, iss. 2, pp. 479–499.

 

Дата создания: 03.03.2015 09:22
Дата обновления: 03.03.2015 11:48