摘要
粗糙集理论的概念性框架之一就是利用不可分辨关系和布尔推理作为数据约简和获取决策规则的基础。在分辨矩阵和决策矩阵概念的基础上,提出将约简分为4类,即信息表的对象约简、信息表的全局约简、决策表的对象约简和决策表的全局约简,其中决策表的对象约简对应决策规则。从模式的角度对约简和决策规则进行了分析,利用决策矩阵和决策函数,给出了获取最小决策规则的一种算法,上述结论可以作为设计启发式算法的基础,并用例子对结论进行了说明。
Rough sets theory provides a framework in which indiscemibility relations and boolean reasoning form a foundation for data reduction and decision rule generation. Based on discernibility matrices and decision matrices, four kinds ofreducts are introduced, i.e., object-relative reduct in information table, system-relative reduct in information table, object-relative reduct in decision table, and system-relative reduct in decision table. Object-relative reduct in decision table corresponds to decision rules. Then, reducts and decision rules are considered from the view of patterns. Using decision matrices and decision functions, an algorithm for minimal rule generation is introduced. With the above mentioned results, it is possible to develop efficient heuristics. The results are illustrated with an example.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第7期1773-1776,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(70601013)
关键词
粗糙集
布尔推理
约简
决策规则
模式
rough sets
boolean reasoning
reducts
decision rules
patterns
作者简介
安利平（1971-），男，河北张家口人，博士，副教授，研究方向为智能决策技术；E-mail：anliping2000@sina.com
仝凌云，博士，副教授。
