【作者】 李勤超；

【导师】 王家耀；

【作者基本信息】 解放军信息工程大学 ， 地图制图学与地理信息工程， 2011， 博士

【摘要】 论文将本体论思想与方法论和语言语义学观点引入到地理信息科学领域,阐述了地理本体的特征与含义以及本体在地理信息科学领域的作用和意义;分析了语言表述语义的机制,以及语言与本体的关系。重点讨论了基于本体的地理信息语义转换理论问题,提出语言分析与本体构建相结合实现地理信息不同描述语言形式之间的语义转换,并给出了语义转换模型;对地理领域本体层次进行了分析,给出了领域本体构建的原则与方法,并对领域本体操作包括本体裁剪、语义相似度计算等进行了全面分析,为语义转换提供了基础;应用MDA的元建模框架与机制,进行了GIS数据本体元建模和元建模下的GIS数据转换实验,探索了结构性语言描述下的地理信息语义转换;使用本体标注与自然语言处理技术实现从自然语言描述中提取地理知识,完成了非结构性空间信息描述向结构化地理本体的转换实验。论文的具体贡献包括:1.深入探讨了语言的语义表达机制;比较了语言与本体的相互关系,指出了语言语义与本体论方法的结合点;分析了地理本体的不同描述语言之间的关系,为使用本体论技术实现地理信息转换提供了理论基础。2.提出了基于本体论的地理信息语义转换模型。对地理信息的特点进行了分析,并比较了地理信息描述的典型语言,探索基于本体的地理信息描述方法,给出了构建地理信息领域本体的原则与步骤;利用叙词表、地形图、地理信息分类词表等资源和本体裁剪等方法进行了地理信息领域本体的创建;提出从分析地理信息描述语言着手,分解语言的形式与语义,并利用高阶-低阶模型理论,建立了基于本体的地理信息语义转换模型;讨论了基于此模型的转换流程和转换模式。3.进行了基于本体元建模的GIS数据转换实验。论文对软件工程中的建模和元建模技术与本体技术进行了比较,将本体抽象层次与元模型层次对应起来,提出利用本体元建模的方法进行GIS数据的转换。利用元建模技术架构MDA和UML扩展机制,分析、创建了GIS数据元模型,并利用元模型进行了GIS数据建模。依据提出的基于本体的地理信息语义转换模型,制订了本体元建模技术下的GIS数据转换流程,进行了GIS数据转换的相关实验。4.建立了本体相似度计算实验平台。通过对当前的语义相似度算法以及本体映射原型系统的分析,指出对不同的任务应使用不同的算法或算法组合;设计并实现了地理信息语义相似度计算的实验系统,作为算法分析和评估的平台;实现了MDSM算法、基于HowNet词典语义距离的算法以及改动的SF算法等语义相似度算法;在实验平台中,选用案例测试了几种算法的性能,并评估了本体裁剪对相似度计算的影响。5.研究了非结构化空间信息的处理问题。从非结构化的空间描述中获取地理认知,和将形式化的地理知识转换为习惯的自然语言形式,这是地理信息转换的重要模式。论文分析了现有的自然语言处理和知识抽取方法,以及这些方法在地理信息领域中的应用现状,提出采用本体标注与自然语言处理相结合的技术途径,来实现从自然语言形式的地理信息描述中提取地理知识,从而完成非结构化的地理信息描述向结构化的地理本体的转换。论文设计了支持多本体语义标注的系统,对从HowNet中提取形式化地理本体进行了阐述;对利用领域本体改进分词结果,应用概念复合生成短语结构,以及领域句法模式规则描述等关键技术进行了分析,并在实验系统中对相关算法与规则进行了检验。更多还原

【Abstract】 The paper leads the ontology thoughts, the methodology and language semantics viewpoint into the geographic information science domain, elucidates the exact character and meaning of geographic ontology and the function and the sense of ontology in the geographic information science domain, analyses the mechanism that language expresses semantics and the relation between the language and the ontology. The paper deeply discusses the geographic information semantic conversion theory problem based on the ontology, combines language analysis with ontology building to realize the semantic conversion in different language forms, and designs the semantic conversion model. The paper analysis geographic domain ontology, suggests the principle and method for building domain ontology, thoroughly analysis domain ontology manipulation which contains ontology cutting and semantic similarity degree computing, and establishes the foundation for the semantic conversion. The paper does an experiment on GIS data transformation between GIS data ontology metamodeling and metamodeling based on MDA metamodeling framework and mechanism, explores geographic information semantic conversion under structural linguistics. The paper distills geographic knowledge with the ontology annotation and natural language management technology in natural language, realizes the experiment on the conversion from non-structural spatial information description to structural geographic ontology.The main works of the paper is as follow:1. This paper thoroughly discusses the mechanism semantic expression of language, compares the relations between language and ontology, indicates the relation between language semantic and ontology theory method, analysis the connection in different description language, this offer theory base for using ontology technology to realize the conversion of geographic information.2. This paper puts forward geographic information semantic conversion model based on ontology. It analysis the characteristics of geographic information, compares typical languages of describing geographic information, explores the methods of geographic information based on ontology, indicates the principles and steps for establishing geographic information domain ontology. This paper uses some sources such as thesauri, topographic map, geographic information classify vocabulary, some methods such as ontology cutting, to establish geographic information domain ontology. This paper launches on languages of describing geographic information, breaks up forms and semantics of language, establishes geographic information semantic conversion model based on ontology by high-low order model theory. The paper discusses conversion flow and pattern based on this model.3. This paper does an experiment on GIS data transformation based on ontology metamodeling. The paper compares modeling in software engineering with metamodeling technology and ontology, abstract ontology level and element model are corresponding, puts forward GIS data transformation methods based on ontology metamodeling. The paper analysis and establishes GIS data element model by MDA and UML extended mechanism, modeling GIS data with element model. The paper constitutes GIS data conversion flow by using geographic information semantic conversion model based on ontology, does related experiment on GIS data conversion.4. This paper establishes experimental platform for ontology similarity calculation. Through analyzing existing semantic similarity algorithm and ontology mapping prototype system, indicates that it should use different algorithm or algorithm combinatorial according as different tasks. The paper designs and realizes the experiment based on geographic information semantic similarity algorithm, makes it as the platform for algorithm analysis and evaluation. The paper realizes some semantic similarity algorithm such as MDSM algorithm, semantic distance algorithm based on HowNet dictionary and altered SF algorithm. In experiment platform, the paper tests capability of these algorithm with related cases, evaluates the influence to similarity algorithm of ontology cutting.5. This paper researches the problem for dealing with non-structural spatial information. Getting geographic cognition from non-structural spatial description, transforming formal geography knowledge into customary natural language form, all of this is important pattern for conversion geographic information. The paper analysis existing methods of natural language processing and knowledge extraction, discusses its application situation in geographic information domain. The paper puts forward the technology approach combining ontology annotation and natural language processing, extracts geography knowledge from geographic information description of natural language form, then accomplishes the conversion from non-structural geographic information to structural geographic ontology. The paper designs a system which supports multi-ontology semantic annotation, describes the formal geographic ontology which is extracted from HowNet. The paper analysis some key technologies such as improving the word segmentation results with domain ontology, creating phrase structure by compounding conception, describing rules with domain syntactical patter. The paper tests related algorithms and rules in experimental system.更多还原