Uncertain knowledge management is a key issuein the expert system. By analysis of the reasons forknowledge’s uncertainty in complete universe anduncompleted universe, two methods based on set-theoreticgranular computing are presented to deal with theuncertainty knowledge which is the rough observablemethod and rough inclusion method respectively. The roughset can be approximately expressed as a union of some basicsets on the given universe using the concepts of roughmembership function and rough observable set.Experiments on the information system show that theoriginal indefinite sets can be described with itsapproximate sets, which have the certain knowledge. So theuncertainty sets have the certainty at some degree, and fromwhich one can get more deterministic information.
展开▼