一种基于粒计算的不完备序决策表约简算法  

AN INCOMPLETE ORDER DECISION TABLE REDUCTION ALGORITHM BASED ON GRANULAR COMPUTING

作　　者：史进玲[1] 杜根远[1] 熊德兰[1] 

机构地区：[1]许昌学院国际教育学院,河南许昌461000

出　　处：《计算机应用与软件》2012年第10期113-116,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(60873104;61040037);河南省基础与前沿技术研究计划项目(102300410060);河南省科技攻关计划项目(112102210079);河南省高等学校青年骨干教师资助计划项目(2010GGJS-177);河南省教育厅自然科学研究计划项目(2010A520035)

摘　　要：为有效地处理具有遗漏型未知属性值的不完备序信息系统,提出扩展序关系的概念,将不完备序信息系统转换为扩展序值矩阵。分析讨论扩展序值矩阵中粒计算模型和粒度度量方法,尝试在不改变不完备序信息系统中扩展序关系的前提下,从核属性集出发,结合启发式约简思路提出一种新的基于粒计算的不完备序决策表约简算法。实例验证表明,该方法能有效地从不完备序决策表中获取最小相对约简。To effectively deal with incomplete order information system which contains missing unknown values, the concept of extended order relation is proposed to transform the incomplete order information system to the extended order matrix. Then the granular computing model and granularity metric in extended order matrix are analysed and discussed. Attempt has been made in proposing a new granular com putingbased reduction algorithm of incomplete order decision table in conjunction with heuristic reduction idea proceeding from core attributes set and in condition that the extended order relation of incomplete order information system keeps unchanged. Example validation shows that the algorithm can effectively extract minimum relative reduction from incomplete order decision table.

关 键 词：不完备序信息系统 扩展序关系 粒度量 属性约简 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]