论文部分内容阅读
从知识区分能力角度提出同可区分度的概念,并对其性质进行分析.利用同可区分度来刻画信息系统中属性的相对重要性,设计了一种基于信息论观点的启发式约简算法.该算法直接对原信息系统进行约简,不需要预处理,且对完备和不完备信息系统都适用,在保证较高约简率的同时使得信息论观点的约简算法在完备信息系统中的最坏时间复杂度降为O(│A│2│U│).最后用实例说明该算法的高效性.
From the perspective of the ability of knowledge differentiation, this paper puts forward the concept of the same degree of discrimination and analyzes its nature.Using the same degree of discrimination to characterize the relative importance of attributes in information systems, a heuristic reduction algorithm based on information theory is proposed. The algorithm directly reduces the original information system, does not require preprocessing, and applies to both complete and incomplete information systems. While ensuring a high reduction rate, it also makes the information theory viewpoint reduction algorithm in the complete information system The bad time complexity reduces to O (│A│2│U│). Finally, an example is given to illustrate the efficiency of the algorithm.