基于语义网络的灾害知识表示研究

发布时间：2018-04-01 15:14

本文选题：语义网络　切入点：灾害知识表示　出处：《华中科技大学》2013年硕士论文

【摘要】：本文基于灾害关联性综合集成研讨项目。该项目拟建立一个“专家体系-知识体系-机器体系”的综合集成研讨系统。其中知识体系的建立所面临的首要问题，便是如何对各种形式的灾害知识进行有效的表示、组织与管理。本研究选取了两种具有代表性的知识——灾害关联性知识与灾害链不确定性知识作为研究对象，利用语义网络表示法进行基本表示，然后结合具体灾害知识背景选择合适的理论对灾害知识展开建模研究。针对灾害关联性知识的表示研究，首先对灾害关联性知识进行了基本语义网络表示并形成了特有的灾害关联性知识语义网络。在明确了建模目的之后，借鉴概念语义相似度的模型方法建立了灾害知识节点关联度计算模型，用于管理和启发潜在未知的关联性知识。最后模拟实验对模型的可用性进行基本检验，并阐述了该模型的实用意义。针对灾害链不确定性知识的表示研究，利用语义网络表示法抽象了一个简单因果型灾害链模型作为基本研究对象。基于贝叶斯网络与语义网络天然的一致性，利用贝叶斯网络构建了灾害链概率传播推理模型，并结合实例演示了该模型对灾害链不确定性知识的表征程度和应用意义；随后分析了贝叶斯网络的静态假定限制并由此引出浸润原理，将静态语义网络推广到动态语义网络，然后通过形式化的定义和描述构建了基于浸润原理的动态推理模型，并将该推理模型应用到灾害链不确定性知识表示中，对模型参数赋予实际知识情景语义内涵，，通过编写算法模拟演示了一个灾害链实例的动态推理过程，验证了该模型的动态表征能力。最后从不同角度比较了两种知识推理模型，说明了它们在灾害链不确定性知识表示方面的重要意义。
[Abstract]:This paper is based on the comprehensive research project of disaster relevance.The project aims to establish an integrated research system of expert system-knowledge system-machine system.The most important problem for the establishment of knowledge system is how to express, organize and manage all kinds of disaster knowledge effectively.In this study, two kinds of representative knowledge-disaster related knowledge and disaster chain uncertainty knowledge are selected as the research objects, and the semantic network representation is used to represent them.Then combining with the specific disaster knowledge background to select the appropriate theory to model disaster knowledge.Based on the research on the representation of disaster related knowledge, the basic semantic network of disaster relevance knowledge is first represented and a unique disaster relevance knowledge semantic network is formed.After defining the purpose of modeling, the model method of conceptual semantic similarity is used to establish the model of disaster knowledge node correlation degree, which can be used to manage and enlighten the potentially unknown correlation knowledge.Finally, the model usability is tested by simulation experiments, and the practical significance of the model is expounded.Aiming at the representation of uncertain knowledge of disaster chain, a simple causal disaster chain model is abstracted by semantic network representation as the basic research object.Based on the natural consistency between Bayesian network and semantic network, the probabilistic propagation inference model of disaster chain is constructed by using Bayesian network, and the representation degree and application significance of the model to uncertain knowledge of disaster chain are demonstrated with examples.Then the static assumption limitation of Bayesian network is analyzed and the infiltration principle is derived. The static semantic network is extended to the dynamic semantic network. Then the dynamic inference model based on the infiltration principle is constructed by formal definition and description.The reasoning model is applied to the uncertain knowledge representation of the disaster chain, and the semantic connotation of the actual knowledge situation is assigned to the model parameters. The dynamic reasoning process of a disaster chain case is simulated and demonstrated by programming the algorithm.The dynamic representation ability of the model is verified.Finally, two kinds of knowledge reasoning models are compared from different angles, and their significance in uncertain knowledge representation of disaster chain is explained.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：X4;TP18

【参考文献】