基于语义理解的地理信息分类与融合研究

发布时间：2018-03-06 18:21

本文选题：地理空间语义理解　切入点：语义分类　出处：《武汉大学》2014年博士论文　论文类型：学位论文

【摘要】：随着泛在信息时代的来临,无处不在的地理空间信息正在改变我们的生活以及理解世界的方式。地理空间信息不断跨领域应用,逐渐催生出以人为本,以用户为中心的实时、动态的地图需求。而在此应用过程中,地理信息人机交互与知识共享的需求日益增长,存在以下几个亟需解决的问题。其一,如何增强不同领域专家对相同地理概念的理解,解决基础地理信息分类当中存在的语义交叉与重叠问题；其二,针对不同地理信息标准相互兼容性差等多源空间数据的异质问题,如何开展多源地理信息语义融合研究,并解决在地理概念规模增长背景下,如何削减融合复杂性问题；其三,在地理空间语义分类与融合过程中,如何判定不同地理概念之间的语义关系。针对上述问题,本文以基础地理信息语义分类与多源地理空间信息语义融合研究为切入点,基于形式概念理论、粒计算及描述逻辑等理论,深入开展相关研究,并着重研究了当前国家基础地理信息陆地水系分类问题,多源地理空间信息语义融合的粒度性问题,以及基于描述逻辑的地理空间语义推理问题。总结全文,文章主要研究内容如下： (1)全面梳理了当前地理空间语义的分类、融合以及地理空间语义相关应用领域的应用进展。通过比较分析,指出当前地理空间语义较少关注地理信息分类标准化、语义粒度性及非数值性地理概念匹配问题。在此基础上,提出了本文总体研究结构框架及主要研究内容。 (2)分别从地理本体、描述逻辑及OWL描述语言等方面,对地理空间语义形式化表达基础理论展开剖析。采用BFO的顶层地理本体框架,结合OWL语言对基础地理信息当中的相关地理概念进行语义描述,构建了一个基础地理信息的行政区划本体。 (3)针对基础地理信息标准(GB/T13923—-2006)当中依然存在概念同名与语义交叉重叠等异质问题,提出了一种新的陆地水系地理要素的语义分类方法。通过引入相关领域的语义描述来消除标准分类当中的语义交叉与重叠问题,借助形式概念分析的方法,抽取多源地理空间概念语义,构建融合概念格,经专家判读选择格结点,通过格-树转化方法,消除概念内涵的直接传递关系,最后构建新的分类层次。该方法是对当前形式概念分类方法的一种扩展与完善。 (4)针对多源地理空间数据语义融合复杂性问题,提出了地理空间语义融合的粒化理论。从概念外延角度,提出了地理概念的模糊等价粒度划分方法;从概念内涵角度,提出了一种基于熵权概念格的地理空间语义粒度约简方法。以国家基础地理信息与水利工程领域水系数据为例,开展了系列融合对比实验,并对不同阈值条件下的约减效果进行了分析。该方法符合地理世界的尺度认知,顾及了地理概念不同内涵的重要性,这在地理空间语义融合领域是一种有益尝试。 (5)针对传统基于概念相似度匹配研究,存在计算误差等问题。设计了一种描述逻辑地理空间语义推理方法,对不同地理空间对象的概念与角色等语义关系进行判定。以GB/T13923-2006与GB/T21010-2007分类标准当中的部分农林用地分类部分术语为例,开展了推理实验,在一定程度上解决了概念与角色匹配过程中出现的冲突、覆盖重叠等问题。该方法具有逻辑的精准性,不针对具体的应用领域,一些形式推理规则具有一定的普适性。从语义理解角度,对地理空间语义分类与融合问题开展研究,旨在服务于地理信息共享与互操作。本文的理论构想与技术实践,一方面,拓展了地理空间语义的理论体系,并为地理空间语义进一步应用在下一代空间数据库与地理信息服务网络等方面奠定坚实基础；另一方面,文章中提出的一些理论与方法,对地理空间语义分类与融合问题求解是一种全新视角,可为相关研究提供参考。
[Abstract]:With the advent of ubiquitous information era, geospatial information everywhere is changing our life and the way of understanding the world. Geospatial information continue to cross domain application, has spawned a people-oriented, user centered real-time, dynamic map needs. In this application, the requirements of geographic information interaction and knowledge sharing the growing, there are several problems to be solved. One is how to enhance the experts in different fields of the same geographical concept, to solve the problems of semantic overlap based geographic information classification; second, aiming at the problems of different heterogeneous geographic information standard mutual compatibility of multi-source spatial data, how to carry out the multi-source geographic information semantic fusion research, and solve the geographical concept of scale growth in the background, how to reduce the complexity of the fusion problem; third, in geography In the process of spatial semantic classification and fusion, how to determine the semantic relationship between different geographical concepts.
Aiming at the above problems, this paper based on geographic information semantic classification and semantic multi-source geospatial information fusion research as the starting point, the theory of formal concept based on Granular Computing and description logic theory, carry out the related research, and focuses on the current national basic geographic information system of land classification problem, the problem of information granularity semantic fusion of multi-source the geographical space, and based on the description logic of geospatial semantic reasoning problems.
The main contents of the article are as follows:
(1) a comprehensive classification of the spatial semantic, application integration and geospatial semantic related applications. Through comparative analysis, pointed out that the current pay less attention to the geospatial semantic geographic information classification and standardization, semantic granularity and non numerical geographical concept matching problem. On this basis, put forward the main research contents study on the structure and the overall framework of this paper.
(2) from the geographic ontology, description logic and OWL description language and other aspects of the geospatial semantic representation theory are explored. The use of the top-level geographic ontology framework BFO, combined with the geographical concept of the basic geographic information of the OWL language to describe the semantic construction of administrative divisions, a body of Fundamental Geographic information.
(3) the basic geographic information standard (GB/T13923 - -2006) which still exists with the same name as the concept of semantic overlapping heterogeneous problems, put forward a new method of semantic classification of land system of geographical factors. Through the introduction of semantic description of related fields to eliminate the semantic overlap problem of standard class, by means of the form concept analysis, concept extraction of multi-source geospatial semantic fusion, constructing concept lattice, the lattice node selection expert interpretation, through the lattice tree transformation method to eliminate the direct transfer relationship concept, finally built a new classification. This method is perfect and an extension of the current form of the concept of classification.
(4) to solve the problem of multi-source spatial data fusion complexity, propose geospatial semantic fusion granulation theory. From the concept angle, puts forward the fuzzy equivalent granularity of geographical concepts; from the concept angle, proposes a reduction method of semantic granularity entropy concept lattice based on national geographic space. The basic geographic information and data stream field of water conservancy project as an example, to carry out a series of experiments and the fusion, different threshold reduction effect is analyzed. The method accords with the cognitive scale geography of the world, and the importance of Gu geographical concept of different connotation, it is a beneficial attempt in the field of geospatial semantic fusion.
(5) according to the research, based on the traditional concept similarity, the calculation error problem. Design a description logic semantic geospatial reasoning method to judge the different geographical object concept and role semantic relations. Land use classification terminology case study on GB/T13923-2006 and GB/T21010-2007 and part of the standard, carry out the experiments in a certain extent to solve the conflicts, the concept and role in the process of overlapping problems. This method has accurate logic, not for specific applications, some form of reasoning rules are universal.
From the perspective of semantic understanding, research on the problem of geospatial semantic classification and fusion, to serve the geographic information sharing and interoperability. On the one hand, the theory and practice of technology, and expand the theoretical system of geospatial semantic, and geospatial semantic further lay a solid foundation in the next generation of spatial database with the geographic information service network etc; on the other hand, some theories and methods proposed in this paper, the spatial semantic classification and fusion problem solving is a new perspective, which can provide reference for the related research.

【学位授予单位】：武汉大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：P208

【相似文献】