Each agent can be decomposed
into following components:
(1) A network interface: It simply couples the agents
to the network.
(2) A communication interface: It is composed of
several methods or functions for communicating with other agents.
(3) A local knowledge model: Resources in this model
include Java classes, other agent addresses, messaging languages
(KQML), community ontologies
and Strategies. Strategies are the general plans for coordination behavior
that balances
the agent's local interests
with global (community) interests.
(4) A domain knowledge model: The model comprises
expertise that is required for an agent to perform functional tasks,
and skills that may be methods
for activating actions corresponding to the received requests.