【作者】 王庆全；

【导师】 荣莉莉；

【作者基本信息】 大连理工大学 ， 管理科学与工程， 2009， 博士

【摘要】 突发公共事件应急反应过程中,决策者在面对复杂多变的应急决策问题时,需要迅速获得相关决策知识,以辅助进行快速而有效的决策。这些决策知识的获取通常需要所有应急反应部门参与,并依照各类法律、法规、预案和经验总结,将问题情境相关的片断性决策知识进行快速的检索、组织和整合。但由于这些分散的决策知识片断结构化程度低、同应急问题关联无序、自动获取难度大的特点,面向语义的决策知识获取就成为辅助应急决策的难点。利用现有知识管理和计算机技术,为实现有效的面向应急问题的决策知识供给,研究应急决策知识供需匹配方法具有重要理论和实践意义。本文在分析应急决策知识需求与供给机制的基础上,以实现应急问题情境下有效的决策知识语义关联获取为目标,以范畴论为主要工具,运用管理科学和计算机科学的方法,从辅助决策角度出发,重点研究在多变应急决策环境和紧迫知识供给条件下决策知识的供需匹配方法,提出基于范畴论的应急决策知识供需匹配的实现框架,研究面向范畴结构语义关联的决策知识表示、完备化处理、相似度计量、融合度计量和重组方法,为辅助应急决策提供必要的理论和方法上的支持。论文的主要研究工作如下:(1)应急决策知识需求与供给机制研究。从应急决策知识的来源、表现形式、敏捷特性和产品特性角度,分析应急决策支持过程中实现快速知识获取的需求,借鉴敏捷供应链的思想,将应急决策知识看作一种具有较强时效性和特殊结构的产品,提出基于敏捷供应链的应急决策知识供给机制,并以实现决策知识供需匹配中有效的语义关联为目标,引入范畴论和定型范畴论,提出一种面向语义的决策知识供需匹配实现框架。(2)应急决策知识的范畴化表示方法研究。分析应急决策知识的语义特性,将范畴论和定型范畴论引入应急决策知识的结构化表示。提出面向非约简语义关系处理的范畴化概念建模和知识表示方法,赋予应急决策中知识片断更为丰富的语义结构信息,增强语义关联的可靠性,实现应急问题情境和决策知识的形式化。从概念性和结构性知识相结合的角度,研究了应急决策领域建模和知识表示问题。(3)面向范畴知识结构的语义管理计量方法研究。在范畴化知识表示的基础上,为提高范畴化知识片断的完备性,提出运用范畴的类推函子和推出来实现知识扩展推理方法,通过联想和推演过程,实现范畴化知识结构的对象和态射补充,并从对象、态射、定型和定型范畴四个层面量化知识片断相似性,提出基于最大公约子范畴的决策知识相似度计量方法,以及知识片断的语义融合度计量方法。(4)语义关联的范畴化知识重组方法研究。针对应急问题情境同背景知识关联不规律的特点,提出基于遗传算法的语义关联重组方法,它是在范畴化知识结构基础上,以定型作为遗传进化的基因代码来进行遗传编码,以知识片断的相似度计量作为个体适应度的测定,以知识片断间的语义融合度作为启发式的遗传进化策略,来解决应急决策知识供需匹配的组合优化问题。通过对突发公共卫生事件样本的仿真实验分析,该重组方法能够建立起问题情境和背景知识之间较为可靠的语义关联。本文从理论和方法的角度,研究快速应急决策知识供给问题,提出基于范畴论的应急决策知识供需匹配方法,从宏观和微观的角度,为有效的辅助应急决策知识获取提供一条可行的新途径,并拓展范畴论在知识管理中的应用。更多还原

【Abstract】 When a public emergency emerges,a decision maker will be confronted with intricate problems in the emergency responses.It is of great necessity for decision makers to call for relevant decision-making knowledge so as to facilitate effective decision-making.Such decision-making knowledge acquisition usually depends on the participation of all the emergency departments.They need search,organize and integrate the fragmented decision-making knowledge that is related with specific problem scenarios,according various laws,regulations,plans and summaries.However because of the lower structuralized degree, the disordering related with emergency problems,the great difficulties of automated acquisition,the semantic oriented decision-making knowledge piece acquisition becomes the major problems of supporting emergency decision-making.So the studies on supply and demand matching method to fulfill effective problem-oriented knowledge supply will make an important impact both in theory and practice with the help of current knowledge management and computer technology.This dissertation aims to provide the supply and demand matching method of emergency decision-making knowledge based on the analysis of emergency decision-making knowledge demand and supply mechanism.Taking the category theory as the major tool and the administrative and computer sciences as the proper approach,from the angle of supporting decision-making,this dissertation lays an emphasis on the method of matching supply and demand under the complex decision situations and urgent knowledge supply requirements,as well as the implementation framework of supply and demand matching,knowledge representation,completion,similarity measure,merging and reorganization,in order to provide the theoretical and methodological support for the decision-making in emergency responses.This dissertation elaborates on the following items of research:(1) The demand and supply mechanism of emergency decision-making knowledge.This dissertation analyzes the requirements of quickly achieving knowledge acquisition from the angle of knowledge sources,forms of expression,agility and product properties.By means of the agile supply chain,we regard emergency decision-making knowledge as a distinctive product of considerable time-sensitiveness and structure,and present a supply mechanism of emergency decision-making knowledge based on it.For the Objectives of acquiring the effective semantic association of the supply and demand matching of decision making knowledge,this dissertation also proposes a semantic oriented implementation framework, through the introduction of category theory,to the supply and demand matching. (2) The categorical representation methods of emergency decision-making knowledge. This dissertation analyzes the semantic characteristics of decision-making knowledge and employs category theory as well as type category theory for the structural representation of decision-making knowledge.We propose the methods of conceptual modeling and knowledge representation,which are oriented to non-deduce semantic relationships,based upon category theory.These methods give much better semantic structure information to the knowledge pieces in emergency decision-making,and so to enhance the reliability of knowledge acquisition and provide an effective way for the two types of knowledge:problem scenario and decision-making knowledge,to be formalized in the process of knowledge supply.It conducts a research into the modeling and representing of emergency decision-making knowledge from a perspective of conceptual information integrated with structural information.(3) The semantic measure methods on categorical knowledge structures.Based on the categorical knowledge representation and improving the completeness of categorical knowledge pieces,we present a knowledge reasoning method using the categorical construction of analogical functors and pushouts,through knowledge association and deduction,to achieve the supplements of objects and morphisms.This dissertation proposes a similarity measure method for decision-making knowledge acquisition based on greatest common subcategory,and this method calculates the similarity between knowledge pieces from the perspectives of objects,morphisms,types and type categories.A knowledge merging method and the measure of multiple knowledge pieces merging are also presented.(4) The knowledge reorganization method of categorical knowledge structures.According the erratic property of semantic association,this paper present a reorganization methods based on genetic algorithms.This method conducts the genetic encoding by means of categorical type,estimates the individual adaptability with the similarity measure of knowledge pieces, and employs the semantic merging degree between knowledge pieces to be the heuristic evolution strategy,so as to achieve the reorganization of emergency decision-making knowledge.Through the simulated experiments on public health emergencies,the reorganization method can effectively establish reliable semantic association between problem scenarios and decision-making knowledge,and have better performance.From the perspective of theory and method,this paper studies the issues of emergency decision-making knowledge supply,proposes the supply and demand matching method based on the category theory.This research explores a new feasible way for effective emergency decision-making knowledge acquisition with the macroscopical and microcosmic views,and expands the application of the category theory in the fields of knowledge management.更多还原