摘要
现实生活中大量数据以区间值形式存在,此时区间值决策表并不是基于等价关系,传统的决策方法并不能解决这一问题.为此,本文在区间值决策表中引入相邻关系、相邻类的定义,进而由相邻类建立了区间决策表的相对优势邻域粒度,拓展了经典决策信息系统的相关方法,并利用相对优势邻域粒度研究了区间决策表属性约简的启发式算法,通过具体案例将得到的属性约简结果与代数约简进行了有效性验证,进一步丰富和完善了信息系统属性约简理论.
In real life,a large number of data exists in the form of interval values.The interval value decision table is not based on the equivalence relation,and the traditional decision-making method cannot solve this problem.For this reason,this paper introduced the definitions of the adjacent relation and adjacent classes in interval valued decision tables.Moreover,the relative dominant neighborhood granularity was established in interval valued decision tables from adjacent classes,which expanded the relevant methods of classical decision information systems.Furthermore,the heuristic algorithm of attribute reduction was established by using the relative advantage neighborhood granularity for the interval valued decision tables.The effectiveness of the obtained attribute reduction results and algebraic reduction was verified by specific cases.These results enriched and perfected the attribute reduction theory of information system.
作者
张晓燕
李璐
ZHANG Xiaoyan;LI Lu(College of Artificial Intelligent,Southwest University,Chongqing 400715,China)
出处
《西南大学学报(自然科学版)》
CAS
CSCD
北大核心
2024年第5期67-76,共10页
Journal of Southwest University(Natural Science Edition)
基金
国家自然科学基金项目(12371465)
重庆市自然科学基金项目(CSTB2023NSCQ-MSX1063).
关键词
粗糙集
区间值决策表
相邻关系
相对优势领域粒度
属性约简
rough set
interval-valued decision table
adjacent relation
relative dominant neighborhood granularity
attribute reduction