ISSN 0474-8662. Information Extraction and Processing. 2018. Issue 46 (122)
Home Back to issue

Pertinence evaluation system architecture on a basis of learning ontology with planning in a certain domain

Dosyn D. H.
Lviv Polytechnic National University, Lviv

https://doi.org/10.15407/vidbir2018.46.061

Keywords: pertinence evaluation, ontology learning, automated planning, hierarchical task network, expected value of perfect information

Cite as: Dosyn D. H. Pertinence evaluation system architecture on a basis of learning ontology with planning in a certain domain. Information Extraction and Processing. 2018, 46(122), 61-67. DOI:https://doi.org/10.15407/vidbir2018.46.061


Abstract

The method of text document pertinence estimation is proposed. It is based on agent approach, expected value of perfect information analysis, hierarchical task network structure of a knowledge base and automated planning algorithms. Use of Markov decision process approach allows us to estimate expected utility of the strategy built in the framework of agent knowledge base with the aim to evaluate gain of expected utility caused by account of information extracted from the text document. For this purposes the text document is considered as a message with a two-part structure which should help us to supplement information contained in this document by relevant context information from the knowledge base.


References