Construction of Meanings in Living and Artificial Agents

Dissertation thesis

Comenius University, Bratislava

Submitted in August 2007, defended June 2008.


Author: RNDr. Martin Takáč

Supervisor: Doc. RNDr. Ľubica Beňušková, PhD.

Abstract

This thesis addresses the issues of the nature and origin of understanding and meanings in language with the computational modeling methodology. It reviews formal semantic theories, studies meanings in living organisms within their evolutionary context and takes into account empirical findings about language acquisition by children. It analyzes meanings in existing computational systems with respect to grounding in interaction with real or simulated environments. The main contributions of the thesis follow:

First, we define a new original semantics based on so-called identification criteria. The semantics allows for representation of objects, properties, relations, changes, complex situations and events. All meanings can be constructed by extracting cross-situational similarities among instances of a category. Both the theory and mechanisms of meaning construction are specified rigorously enough to allow for implementation in computational models.

Second, we present two computational models of interaction-grounded meaning construction. In the model of individual category construction, the instances are grouped to categories by common motor programs (affordances), while in the model of social learning, focused on the influence of naming on category formation, entities are considered members of the same category, if they are labeled with the same word by an external teacher. Results of experimenting with both models validate the proposed meaning-formation mechanisms.

Third, we report and analyze simulation results of an experiment focused on the dynamics of meanings in iterated intergenerational transmission.

Keywords: Computational modeling. Cognitive semantics. Symbol grounding. Automated concept formation. Language acquisition.


Full text (pdf)

available for download here


Last modified Oct 1, 2008.