基于属性重要性的逐步约简算法

粗糙集理论研究的重要内容之一是知识约简的有效性计算问题，目前求解知识约简的算法主要有两种：一种是利用辨识矩阵构造区分函数，另外一种是基于属性重要性的启发式算法．这两种算法均能求得决策系统的最小或次小约简，但由于计算的复杂度高，所以当数据量增大时这些算法的计算性能是不能令人满意的．本文在对后一种算法充分研究的基础上设计了基于属性重要性的逐步约简算法，利用在决策系统中己获得的正区域逐步缩小数据处理范围，减少求解时间．本文将该算法与基于属性重要性的算法进行了实验比较并对结果进行了分析．

An Improved Algorithm for Reduction of Knowledge Based on Significance of Attributes

The fast algorithm for reduction of knowledge is one of the important topics in the research on rough sets theory. The main two existing algorithms can find the minimal reduction for most information system, however, they are both of time consuming computation which is not suitable for data analysis in practice. In this paper here, a improved algorithm is proposed, and the complexity of this algorithm is analyzed; then the experiments comparing this algorithm to the old algorithm based on the significance of attributes is summarized.