Laboratoire d'Informatique de Grenoble

Abstract

Agents may use their own, distinct vocabularies to reason and talk about the world, structured into knowledge representations, also called ontologies. In order to communicate, they use alignments: translations between terms of their ontologies. However, ontologies may be subjected to change, requiring their alignments to evolve accordingly. Experimental cultural evolution offers a framework to study the mechanisms of their knowledge evolution. It has been applied to the evolution of alignments in the Alignment Repair Game (ARG). Experiments have shown that, through ARG, agents improve their alignments and reach successful communication. Yet, these experiments are not sufficient to understand the formal properties of cultural knowledge evolution.

This thesis presents a modeling of ARG in Dynamic Epistemic Logic to assess its formal properties by (a) encoding the ontologies, (b) translating agents ontologies and alignment into knowledge and belief, (c) translating adaptation operators as announcements and conservative upgrades. With this modeling, it is shown that all but one adaptation operator are correct, they are incomplete and some are partially redundant.

These properties must, of course, be considered with respect to the game and the logical modeling. Three differences will be discussed, leading to an independent model of awareness based on partial valuations and weakly reflexive relations. An alternative modeling of ARG is then defined under which the formal properties are re-examined, showing that this modeling is closer to the original game. This paves the way for defining a theoretical model of cultural knowledge evolution.