Representing and Revising Norms and Values Within an Adaptive Agent Architecture

摘要

This paper presents an adaptive agent architecture which is capable of representing and manipulating social notions such as norms and values. The architecture stems from a coherence-based approach to perception and belief revision, in which high level explanations and constraints are used to guide the search process to the most coherent hypothesis. The addition of norms and values to this architecture will lead to the creation of agents which can form, update, and revise complex beliefs concerning the intentions and probable behaviour of other agents.