摘要
属性约简是粗糙集理论中的重要问题.为了满足用户对属性的偏好,人们研究了属性序下的属性约简,然而对一些问题却很难给出完整的属性序.针对该问题,比较分析了属性组序下的约简子集的优劣,并提出代价敏感下的属性组序约简的算法.该算法通过属性组序的特点考虑用户偏好并结合属性代价以及属性重要度加权的方式选择局部属性,可以得到更符合用户偏好的约简.理论分析和实验结果验证了该算法的可行性和有效性,并且能在一般情形下找到满足用户偏好的约简.
Attribute reduction is an important issue in rough set theory.In order to satisfy users'preference for attributes,people have studied attribute reduction under attribute order.In some problems,however,it is difficult to give a complete attribute order.Aiming at this problem,this paper compares and analyzes the pros and cons of reduction subsets under attribute group order,and proposes an algorithm of attribute group order reduction under cost sensitivity.This algorithm considers user preferences based on the characteristics of the attribute group order,and selects local attributes by combining the attribute cost and attribute importance weight.Therefore,a reduction more in line with user preferences can be obtained.The theoretical analysis and experimental results verify the feasibility and effectiveness of the proposed algorithm,and we can find the reduction that satisfies the users'preference in the general case.
作者
刘鑫
胡军
张清华
Liu Xin;Hu Jun;Zhang Qinghua(Chongqing Key Laboratory of Computational Intelligence,Chongqing University of Posts and Telecommunications,Chongqing,400065,China)
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2020年第4期469-479,共11页
Journal of Nanjing University（Natural Science）
基金
国家自然科学基金(61876201,61876027)
重庆市人工智能创新重大主题专项(cstc2017rgzn‐zdyfX0040)。
关键词
属性约简
用户偏好
属性组序
代价敏感
attribute reduction
user preference
attribute group order
cost sensitive
