Beliefs and Communication, The Basis of Collaborative Agents

Nowadays, simulation scenarios require enhanced agents being capable of communication, learning and having beliefs about their self or about their world. The distinction between facts and (possible false) beliefs is part of the requirements in order to describe scenarios closer to real world. This work proposes a solution which allows information exchange between agents, making use of generic/standard and ad-hoc message types, involving their self beliefs and their beliefs about the world where they act. Standard agent types being capable of automatic management of standard message types are defined. SPARQL, a standard RDF based query language, is used as part of the communication protocol. The paper is build around a social simulation scenario that provides the possibility of testing and describing the benefices of this work.