Logika dynamického diskursu

Abstract

The aim of this master thesis was to implement an application in a form of a software agent, which is able to identify anaphoric references in a sentence and substitute for them to obtain a sentence with complete meaning. The theoretical part explains representation of natural language sentences in computational version of transparent intensional logic named TIL-Script. In this logic, sentences are expressed as constructions. An algorithm was designed and then realized in C# language in the implementation part. Algorithm automatically fills a database of analysed constructions. These constructions are afterwards substituted for anaphoric references. The algorithm is injected in a form of a plugin into an agent, which is able to communicate with other agents in a multi-agent system. The plugin enables the agent to substitute an appropriate construction based on the current context.