Inference Based on Notifications: A Holonic Metamodel Applied to Control Issues

The ability to infer is necessary to achieve desirable behavior within many systems. In software systems, this ability can be achieved through a number of decision-making solutions. This paper proposes an alternative inference solution inspired in rule-based systems (RBSs). This solution considers classical RBS elements, such as rules and fact-base sets, as actual computational entities. The innovative aspect of the proposed solution is that computational entities carry out inference by means of collaborations based upon notifications instead of search methods. The notification approach has advantages, such as high reactivity, easy distribution, and good tradeoff between generality and applicability. The solution consists of a holonic metamodel whose features are highlighted, in this paper, by applying it to a collaborative control of an agile manufacturing system.