摘要
如何根据规则进行决策是决策规则应用必须解决的核心问题之一。针对该问题,提出了一种决策规则的合成方法,为根据多条规则进行决策提供了崭新的思路。首先,获取规则,计算支持度、信度等信息;然后,将规则信度转换为可信度分配,综合属性重要性和规则支持度两方面因素,计算可信度分配的权重;最后,运用证据理论对可信度分配进行合成。实验表明,规则合成方法能够对决策规则,尤其是冲突规则进行有效合成,得出可靠的决策。
A combination method to make decision using decision rules was proposed. The method which took advantage of the information derived from the process of rule extraction gave a new idea to make decision according to decision rules. First, rule extraction methods were employed to obtain decision rules and their support and confidence degrees. Second, the basic probability assignments were calculated according to the confidence degrees of rules. Then, the attribute significances and support degrees of rules were considered to compute the evidence weights. Last, the combination method of evidence theory was used to obtain the aggregated decision. The experiment shows that the combination method performs well to combine decision rules, especially conflicting rules.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2008年第4期951-955,共5页
Journal of System Simulation
基金
国家自然科学基金项目(70672097)
国家自然科学基金重点项目(70631003)
合肥工业大学研究生科技创新基金(038902)
关键词
决策规则
粗糙集
证据理论
合成方法
decision rule
rough sets
evidence theory
combination method
