Specification of ontologies and a support of agents’ reasoning in a multi-agent world

Abstract

The digital age has ushered in a vast reservoir of accessible information through online textual repositories. However, agents in a multi-agent environment often encounter frustration when searching for sources of information, as traditional keyword-based searches yield unsatisfactory results. This underscores the need for an advanced "intelligent question-answering system" capable of deriving inferable knowledge and providing semantically correlated ramifications in responses. A limitation inherent in this system pertains to the treatment of semantically simple expressions. Hence, there is a requirement for methodologies capable of transforming semantically simple expressions into compound expressions that can function as their definitions. This thesis comprises algorithms and procedures designed to refine elementary expressions into their respective explications, and it also provides recommendations for relevant textual information sources to assist the agent in its decision-making process by means of logical analysis of natural-language texts, machine learning and data mining methods.