Law-governed Linda as a semantics for agent dialogue protocols

Tuple spaces and the associated Linda language are a popular model for distributed computation, and Law-Governed Linda (LGL) is a variant allowing processes to have differential and secure access to tuple spaces. We propose a form of LGL as a means of implementing a multi-agent dialogue game protocol, such that utterances under the dialogue protocol are interpreted as actions on particular tuple spaces subject to certain laws. In this way, the tuple spaces, their associated law and the actions on them may be viewed as a semantics for the dialogue protocol syntax.