Qualitative Simulation and Reasoning with Feature Reduction Based on Boundary Conditional Entropy of Knowledge

The present paper discusses a new definition of knowledge rough entropy based on boundary region from the aspect of Pawlak topology. This definition accurately reflects an idea that the uncertainty of set can be described by boundary region. It thus proves an important conclusion that boundary conditional entropy of knowledge monotonously reduces with the diminishing of information granularity. Combining qualitative reasoning technology with knowledge information entropy based on rough sets theory, a heuristic algorithm for feature reduction is proposed which can be used to eliminate the redundancy in the qualitative description and the qualitative differential equations are obtained. The result shows that the rough sets theory (RST) is of good reliability and prospect in qualitative reasoning and simulation.