Implicit Commitments through Protocol-Level Semantics

Conversation policies often incorporate the notion of commitment as an elementary building block of agent interactions. However, high-level commitment semantics rely on heavy infrastructure and advanced reasoning capabilities which then become inseparable from agent interactions, a lower-level concept. We show that it is possible to express the notion of commitment as a part of protocols themselves. We present a simple mechanism for implicit commitments based on nite state machines, and reason about actions an agent is required to ful ll to follow the state machine. We also demonstrate the utility of implicit commitments within an interactionoriented programming environment and present a toolkit that implements this scheme, allowing for dramatic savings in MAS development time.