基于特征网络的突发事件关键属性识别方法

发布时间：2018-05-30 20:30

本文选题：突发事件 + 知识元　；参考：《大连理工大学》2015年硕士论文

【摘要】：进入21世纪之后,发生了一些诸如中国SARS事件、四川汶川大地震、日本福岛核泄露事故、上海外滩踩踏事件等具有较大社会危害的突发事件,引起国内国外民众、政府、学者的高度关注。从系统论的角度看,突发事件与自然系统、社会系统、人文系统等有错综复杂的联系,具有极强的系统复杂性和不确定性。同时,事件系统的复杂性还体现在事件的突发性和信息爆炸的特性,突发事件爆发后的大量的事物信息增加了应急管理的难度。使得较为严谨的逻辑推理式应急策略不能在事件时空限制下很好地满足需求,相反,基于少量事件关键事物信息的事件应急策略,解决了时空条件对事件应急的限制。因此,如何从大量事物属性中识别出事件的关键属性成为影响事件应对效率的关键问题之一。针对现有研究中,在描述突发事件演化态势和事件关键属性识别时,忽略推理规则作为事件演化内在逻辑对事件发展态势的影响。本文以系统科学、知识科学、管理学、数学等基本理论为基础,提出基于特征网络的突发事件关键属性识别方法,主要进行以下相关内容的研究。首先,基于知识元模型的事物知识表示方法,提出突发事件问题域的对象-属性-关系模型,从三个层次分析问题域内事物属性和推理规则之间的逻辑关系,并从知识元实例化的角度,给出对象-属性-关系模型中三层知识的使用方法。其次,结合属性知识和规则知识,提出用于表示事件发展态势的特征网络模型的构建方法,该特征网络以客观事物的属性为节点,以事物属性在推理规则的最简结构中关系为有向边。然后,从可推理角度,分析突发事件关键属性的识别流程,并结合网络中心性分析的方法,提出基于特征网络的突发事件关键属性分析方法。最后,以舟曲泥石流事件中的一个问题域实例验证以上方法的有效性。本文从事物知识组织管理的角度对突发事件进行研究,对突发事件的演化推演和应急决策都有实践意义。基于知识元模型的对象-属性-关系层次模型,实现了对突发事件事物知识的立体分层组织,降低了应急决策者对专业性较强的突发事件领域知识的使用难度。使用特征网络的事件态势描述方法,更符合突发事件复杂性的特点,可增强对突发事件发展形势的判断准确性。基于特征网络的事件关键属性识别方法,为应急决策者提供了一种基于少数关键事件信息的应急管理策略,可提高公共部门在面对突发事件时的应对水平。
[Abstract]:After entering the 21st century, some unexpected events, such as the SARS incident in China, the Wenchuan earthquake in Sichuan, the Fukushima nuclear leakage accident in Japan and the Shanghai Bund stampede, have caused great social harm to the people and the government at home and abroad. Scholars are highly concerned. From the point of view of system theory, unexpected events have intricate relations with natural system, social system, humanistic system and so on, and have strong system complexity and uncertainty. At the same time, the complexity of the event system is also reflected in the unexpected events and the characteristics of information explosion, a large number of things information after the outbreak of emergencies increased the difficulty of emergency management. The more rigorous logical reasoning emergency strategy can not meet the demand well under the event space-time limit. On the contrary, the event emergency strategy based on a small amount of information of the key thing solves the limitation of the space-time condition to the event emergency. Therefore, how to identify the key attributes of events from a large number of object attributes has become one of the key issues affecting the response efficiency of events. In view of the existing researches, the influence of inference rules as the intrinsic logic of event evolution on event development situation is ignored in describing the evolution situation of unexpected events and identifying the key attributes of events. Based on the basic theories of system science, knowledge science, management science, mathematics and so on, this paper proposes a method of identifying the key attributes of unexpected events based on feature network. Firstly, based on the knowledge metamodel, the object-attribute-relation model of the emergency problem domain is proposed, and the logical relationship between the object attribute and the reasoning rule in the problem domain is analyzed from three levels. From the point of view of knowledge element instantiation, the method of using three-layer knowledge in object-attribute relation model is given. Secondly, combining attribute knowledge and rule knowledge, a method of constructing feature network model is proposed, which takes the attribute of objective thing as the node, which is used to express the situation of event development. The relation of object attribute in the simplest structure of inference rule is directed edge. Then, from the point of view of reasoning, the identification process of critical attributes of emergencies is analyzed, and combining the method of network centrality analysis, a method of critical attribute analysis of emergencies based on feature network is proposed. Finally, an example of the problem domain in Zhouqu debris flow event is given to verify the effectiveness of the above method. In this paper, from the perspective of knowledge organization and management of things, the research on unexpected events is of practical significance to the evolution of emergencies and emergency decision-making. The object-attribute-relational hierarchical model based on the knowledge metamodel realizes the hierarchical organization of the knowledge of unexpected events and reduces the difficulty for emergency decision makers to use the specialized knowledge in the field of emergencies. The method of event situation description based on feature network is more in line with the complexity of unexpected events and can enhance the accuracy of judging the developing situation of emergencies. Event critical attribute recognition based on feature network provides an emergency management strategy based on a small number of critical event information for emergency decision makers, which can improve the level of public sector response to emergencies.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：D63;TP18

【参考文献】