Multi-agent belief revision
Noa Kfir-Dahav, Moshe Tennenholtz · 1996
We introduce a basic framework for multi-agent belief revision in heterogeneous societies where agents are required to be consistent in their beliefs on shared variables. We identify several properties one may require a general multi-agent belief revision operator to satisfy, and show several basic implications of these requirements. Our work reveals the connection between multi-agent belief revision and the theory of social choice, and attempts to provide some initial understanding of the multi-agent belief revision process. 1 Introduction Consider an agent, a set of beliefs it adopts, and a new observation it makes. Taking the new observation as a fact that should be incorporated to its beliefs, the agent may need to revise its beliefs. Belief revision is a basic research topic in AI [Gar88, KM91, FH94, dS92, Gro88, Doy91], and research on the theory of belief revision seems to converge to a well-established theory [Gar92]. Most of the theory developed in this area has concentrated ...