Abstract:

One of the most important issues in artificial intelligence is uncertainty. Many uncertainty measuring methods have been put forward and widely used in information systems, such as entropy theory and information granularity, which are two main approaches to study the uncertainty of an information system. In this paper, the physical meaning of entropy and the entropy increase principle in information systems are presented firstly, then the axiom definitions of entropy and information granularity are provided and two new uncertainty measureing functions αentropy and αgranularity using this method are developed. Analysis shows that some of the existing definitions of entropy and information granularity become the special forms of axiom definitions. The results unify, standardize and develop the theory of uncertainty measureing in complete and incomplete information systems.

Key words: information system;entropy;information granularity;rough sets;uncertainty