可能性证据——非统计不确定性信息的一种知识表示方法

POSSIBILITY-EVIDENCE:A KNOWLEDGE REPRESENTATION FOR NON-STATISTIC UNCERTAINTY INFORMATION

在经济管理科学中,单纯依靠数据支持,通过建模进行仿真和优化计算,以寻求合理的决策方案的数量化研究方法,面对许多非结构化或半结构化的,并含有大量复杂的不确定性因素的社会经济问题,暴露出严重的不足.需要依靠有关领域专家的知识、经验和智慧来解决问题.发展知识工程和带知识库的决策支持系统,成为信息——决策科学的前沿领域.

In this paper,a concept of possibility-evidence is proposed and a knowledge representa-tion method for non-statistic information provided by experts,which is applied to estimatinguncertainties,is given.The method combines organically both the possibility theory proposedby L.A.Zadeh and the theory of evidence proposed and developed by A.T.Dempster andG.Shafer.The method is conveniently applied to introducing representation of expert know-ledge from a possibility proposition which is shown in natural language.It is proved thatthe method is reasonable and human's knowledge estimating uncertainty has its sturcturalstability in statistics.A theoretical foundation for developing an uncertainty analysis knowledge-base ofEXDASS is established.