Protocols structure interactions among communicating agents. A commitment machine models a protocol in terms of how the commitments of the various parties evolve. Commitment machines thus support flexible behavior while providing a meaningful basis for compliance with a protocol. Unfortunately, current formulations of commitment machines are not sufficiently general or rigorous.
This paper develops generalized commitment machines (GCMs) whose elements are described generically in terms of inferences, and whose computations are infinite (thus supporting nonterminating protocols). This paper shows how a GCM can be understood as a nondeterministic Buchi automaton (BA), whose acceptance condition accommodates infinite as well as finite executions.
Deterministic BA are readily emulated by conventional software, e.g., a script running in a Web browser. In general, nondeterministic BA may have no equivalent deterministic BA. However, under well-motivated technical conditions, a GCM yields a deterministic Buchi automaton that, although not necessarily equivalent in automata theory terms, is sound (produces only computations allowed by the GCM) and complete (produces the effect of any computation allowed by the GCM).
[1]
Munindar P. Singh,et al.
Nonmonotonic Commitment Machines
,
2003,
Workshop on Agent Communication Languages.
[2]
Nicolas Maudet,et al.
Protocol Conformance for Logic-based Agents
,
2003,
IJCAI.
[3]
Wolfgang Thomas,et al.
Automata on Infinite Objects
,
1991,
Handbook of Theoretical Computer Science, Volume B: Formal Models and Sematics.
[4]
Munindar P. Singh,et al.
Commitment Machines
,
2001,
ATAL.
[5]
Yves Lespérance,et al.
On the semantics of conditional commitment
,
2006,
AAMAS '06.
[6]
Timothy W. Finin,et al.
Semantics and Conversations for an Agent Communication Language
,
1997,
IJCAI.
[7]
Michael Winikoff,et al.
Enhancing Commitment Machines
,
2004,
DALT.
[8]
Peter McBurney,et al.
Posit spaces: a performative model of e-commerce
,
2003,
AAMAS '03.