Toward a Theory of Communication and Cooperation for Multiagent Planning

In this paper we develop a formal computational theory of high-level linguistic communication that serves as a foundation for understanding cooperative action in groups of autonomous agents. We do so by examining and describing how messages affect the planning process and thereby relating communication to the intentions of the agents. We start by developing an abstract formal theory of knowledge representation based on the concept of information. We distinguish two types of information: state information, which describes the agent's knowledge about its world (knowing that) and process information, which describes the agent's knowledge of how to achieve some goal (knowing how). These two types of information are then used to formally define the agent's representation of knowledge states including the agent's intentional states. We then show how situations and actions are related to the knowledge states. Using these relations we define a formal situation semantics for a propositional language. Based on this semantics, a formal pragmatic interpretation of the language is defined that formally describes how any given knowledge representational state is modified by a given message. Finally, using this theory of meaning of messages or speech acts, a theory of cooperation by means of communication is described.