An Incremental Algorithm Based on Discernibility Matriz for Reducts of Incomplete Decision Tables

摘要：

Attribute reduction is an important issue of data mining. In this paper an incremental algorithm for computing reducts of an incomplete decision table is proposed based on an improved discernibility matrix, which can obtain all reducts through some simple operations on original discernibility matrix when a new object is added to the incomplete decision table. Firstly an improved discernibility function used to generate reducts is presented based on the improved discernibility matrix. And then an incremental method of computing reducts is introduced by analyzing the different cases of the new object. Example shows that the proposed algorithm is very efficient because it can avoid recomputing a new discernibility matrix.

作者: Dedong Zhang Renpu Li Fuzeng Zhang Yongsheng Zhao

作者单位: School of Computer Science & Technology, Ludong University, Yantai 264025, China

会议类型: 国际会议

会议名称: Third International Symposium on Intelligence Computation and Applications(ISICA 2008)(第三届智能自动化、计算与制造国际研讨会)

会议地点: 武汉

会议语种:英文

页码: 223-227

在线出版日期: 2008-12-19（万方平台首次上网日期，不代表论文的发表时间）