Abstract World for Opportunistic Local Decisions in Multi-Agent Systems
  Using Bayesian Knowledge Bases
A common solution in multi-agent systems is to commit a team of collaborating
agents to a joint plan.
Since any deviation from the plan by an agent is hazardous, these solutions
face up to potential unplanned ``opportunistic'' actions by
ignoring them, or by ad-hoc rules determining whether to
accept such opportunities.
Since neither of these solutions is desirable, we develop AWOL(Abstract World for Opportunistic Local decisions), an abstract framework which attempts a disciplined treatment of opportunistic action, in
the context of an existing joint plan. The idea is to model the (stochastic)
tradeoff of opportunism vs. continued commitment
to the joint plan, while abstracting away from the
state of the world. The abstract model is evaluated using
strict decision-theoretic criteria, with the goal of applying the 
optimal decision on whether to accept an opportunistic action
in the original domain.
When we modeled this abstract domain as MDP, its complexity, even
is lower than in a real or simulated domain, continues being high.
In order to reduce this complexity, we try to implement a compact
representation using Bayesian Knowledge Bases(BKB). It is a rule-based probabilistic model
that allows to simply the transition probabilities in the domain. The BKB extends the well-known Bayes Network.