Knowledge Reduction Algorithm Based on Relative Conditional Partition Granularity

Abstract

In order to solve complex knowledge reduction, the relative conditional partition granularity and new knowledge significance, quantitative representations for the relative classification ability of decision attributes are defined in this paper. And new knowledge partition granularity and new relative conditional partition granularity are constructed to transform inconsistent decision tables into "consistent" decision table. On this basis, common knowledge reduction algorithm is proposed for both consistent and inconsistent decision tables. The algorithm can effectively obtain the optimal or a sub-optimal relative reduction of decision table and its time complexity is relatively low as O(|U| <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> |U|) through theoretical analysis. Finally, we show that this algorithm is effective through an example.