【作者】 陈静；

【导师】 马宗民；

【作者基本信息】 东北大学 ， 计算机软件与理论， 2012， 硕士

【摘要】 语义Web作为一种能够理解人类语言的智能网络,是当前互联网技术研究的热点之一。在语义Web中本体发挥着重要的作用,它能够解决特定应用中的知识共享问题。但是由于各领域都有自己相应的构建本体的标准,本体的建立缺乏统一的标准规范,使本体所描述的内容在语义上往往重叠或关联,在本体表示语言和表示模型上存在差别,造成了本体异构。然而,当前的本体技术并不能很好的解决异构本体里语义识别和资源共享的问题。针对这个问题,需要发现本体潜在的语义来计算出本体之间的语义相似度,而概念相似度计算是本体相似度计算的基础,所以本体之间概念相似度计算是实现大规模异构本体管理的基础问题。大部分现有的概念相似度计算方法都是基于本体概念的语法结构来进行相似度比较,没有充分利用本体内部隐含的语义信息,这使得概念相似度的计算不够精确。OWL DL本体中使用了典型的DL描述构造子,构造子能够表达丰富的语义信息,同时能够进行有效的推理服务,所以考虑描述构造子的语义表达能力来计算概念相似度能够发现本体潜在的信息,进一步提高相似度的精度。然而以DL描述构造子为基础的大部分算法未考虑等价构造子的重要性及忽视了层次结构的影响。本文提出了一种基于语义的OWL本体概念相似性比较算法-CSCSim。该算法以OWL DL本体概念为计算对象,利用描述构造子的语义信息和概念层次的语义信息来进行概念相似度比较。使用OWL DL的有效推理,发现概念内隐含的语义信息和层次关系；在计算RDF三元组描述相似度时对构造子进行分类,突出等价构造子的重要性；在计算RDF图结构相似度的时候,在综合考虑谓词和宾语相似度的基础上,提出了新的计算方法；同时,考虑了本体层次结构对概念相似度的影响。进一步,为了验证CSCSim算法的有效性,本文设计了一个基于语义的本体之间概念相似度计算的原型系统。它能够解析OWL本体文件,并以可视化形式把本体呈现给用户浏览,能够根据用户选择的概念计算出概念对的相似度。实验结果表明本文所提出的算法具有一定的优越性。更多还原

【Abstract】 As an intelligent network to understand the human language, Semantic Web is one of the hotspots of the current Internet technology. Ontologies play an important role in the Semantic Web. It can solve the problem of knowledge sharing in the specific application. However, because each field has its own standards to build ontology, the building of ontology is short of uniform standards, which will cause the content of the ontology are always overlapping or associated. There exist differences between ontology representation language and model, so it causes ontology heterogeneity. The current ontology technology is not good enough solve semantic recognition and resource sharing issues. In order to solve this problem, it needs to find potential semantics in the ontologies so as to calculate the semantic similarity between ontologies. Ontology similarity is based on the concept similarity, so the computing of similarity between two concepts in different ontologies is the basic problem to realize the management of large-scale heterogeneous ontologies.Most of the existing method to compute concept similarity is based on the grammatical structure to compare. It does not take full advantage of implicit semantic information in the body of ontologies which makes the calculation of the concept similatity is not precise enough. OWL DL ontologies use typical DL constructors. Constructors can express rich semantic information and are able to carry out effective reasoning. Considering the the semantic expression abilities of the constructors to compute concept similarity is able to find the potential semantics of ontologies which will further improve the accuracy of the similarity.This paper takes up with a concept similarity comparision algorithm in OWL ontologies-CSCSim. The new algorithm takes OWL.DL ontology concept as calculating object, and compares concept similarity using descriptive constructor and concept hierarchy. It uses the effective reasoning of OWL DL to find the potential semantics in the concepts. It classifies constructors, and highlights the importance of equivalent constructors when calculating the similarity of RDF triple array. This paper also considers the influence of both predicates and object similarity when calculating graph structure similarity of RDF. At the same time, the new algorithm also takes the factor of hierarchy of ontology into consider in calculation. In order to verify the effectiveness of CSCSim algorithm, the paper designs a prototype system to compute concept similarity based on semantics. The system is able to parse OWL ontology file and presents the ontologies to the users in a visual form and compute the concept similarity between concepts the users select. The experiment result proves that the algorithm proposed in this paper is superior.更多还原