Knowledge acquisition using rough sets when membership values are fuzzy sets
A. de Korvin, C. McKeegan, Robert M. Kleyle · Journal of Intelligent & Fuzzy Systems · 1998
In this paper we model uncertainty using the so-called rough set approach in which upper and lower approximations of a set of objects are based on equivalence classes determined by attribute values. However, due to imprecision in the information, both the attributes and the resulting decisions are modeled as fuzzy sets. Furthermore, the membership of these fuzzy sets is also fuzzy, creating fuzzy sets of type II. From information of this type, we construct inference rules of unequal strength. The strength of any rule is determined by both its degree of truth and its degree of belief, each of which are obtained from the fuzzy memberships.