THE INFERENCE BASED ON ROUGH SETS IN MULTI AGENT SYSTEMS

In this paper, a reasoning model in multi agent systems is defined. The rough sets in the models, and the rough inclusion calculus of sets or set connection computing based on rough set approach are established. The equality between sets can’t be given in knowledge discovery and data mining; but only the rough inclusion between them is discussed, because it is often not easy to obtain the exact and consistent interpretation between sets in different agent models. In general, an association rule based on decision table often describes, by the support value and confidence coefficients, the degree by which the set of objects satisfying the formula on the left hand side of the association rule is included in the set of objects satisfying the formula on the right hand side of the association rule. Reasoning in distributed environment often requires that a learning interface between multi agents be offered. The interface is an approximate space AS or inference model IM of multi agent defined in this paper. All operations are performed on data table or decision table. Hence, a composite and included operation is defined on the interface（AS or IM）. Thus it is possible to obtain the approximation of definable set performed by cooperation between multi agents and the rough included or connection computing between sets in a distributed environment. It is shown with real examples that it is feasible to achieve some goal by actions of multi agents and cooperation between agents in a distributed environment.