水环境本体构建和组织方法研究及应用

发布时间：2018-12-31 10:35

【摘要】：随着水利信息化的发展,由于没有规范的信息表达形式,造成了许多信息无法共享和融合,制约了水利科技的进步。而本体作为一种规范化的表达形式可以得到很好的应用。为了实现数字流域框架下的信息共享,需要以流域水环境涉及的领域知识为研究对象构建水环境领域本体。本文提出了水环境本体构建和组织方法,构建的水环境领域本体包含了一系列水环境领域的概念,同时也包含了对这些概念分类关系的描述。本文针对现有本体构建的不足,采用种子概念和TFIDF混合法提取领域概念,考虑到词语在文本集中分布不均匀的情况,改进了原始的TFIDF公式,提高了本体概念提取的精度。通过引入种子概念解决了TFIDF法存在的低频词提取较少的缺陷,其中选取的种子概念来源于地球与环境术语语义网(SWEET)和CUAHSI本体里面的相关概念,这些本体都提供了水环境系统学科概念的顶层本体。然后通过计算候选概念的领域相关度(DR)和领域一致度(DC)得到领域概念集合。其次,在传统的K-means聚类算法中加入遗传算法进行参数优化,从而提取水环境领域概念的分类关系。该优化措施有效地消除了K-means算法人工确定分类数K的随机性和无法自动生成本体层次结构的局限性,提高了聚类分析的精确度。再次,本文提出了多策略本体映射,主要从名称、结构、属性以及实例策略得到本体概念之间的相似度,并通过引入Sigmoid函数来确定各个策略的权值得到综合相似度,提高了本体映射的精度和效率,完成水环境本体组织。最后,本文设计并开发了水环境本体构建和组织系统,根据本体构建的方法和流程,该系统分为5个模块:预处理模块、领域概念提取模块、概念关系提取模块、本体映射与整合模块、本体可视化模块,完成了文本数据源的本体构建并进行了可视化展示。通过实验验证了本体构建方法的有效性,为本体构建和应用提供了技术支持。
[Abstract]:With the development of water conservancy information, due to the lack of standardized information expression, many information can not be shared and integrated, which restricts the progress of water science and technology. As a kind of standardized expression, ontology can be applied well. In order to realize information sharing in the framework of digital watershed, it is necessary to construct the domain ontology of water environment by taking the domain knowledge of watershed water environment as the research object. In this paper, a method of constructing and organizing water environment ontology is proposed. The water environment domain ontology contains a series of concepts of water environment domain, and also includes a description of the classification of these concepts. In this paper, the concept of domain is extracted by using seed concept and TFIDF mixed method. Considering the uneven distribution of words in text set, the original TFIDF formula is improved and the precision of ontology concept extraction is improved. By introducing the concept of seed, the defects of low frequency word extraction in TFIDF method are solved. The selected seed concept is derived from the related concepts in the semantic web of the earth and environment, (SWEET) and CUAHSI ontology. These ontologies all provide the top ontology of the subject concept of water environment system. Then the domain concept set is obtained by calculating the domain correlation degree (DR) and domain consistency degree (DC) of candidate concepts. Secondly, genetic algorithm is added to the traditional K-means clustering algorithm to optimize the parameters, so as to extract the classification relationship of the concept in the field of water environment. The optimization method effectively eliminates the randomness of the K-means algorithm to determine the classification number K manually and the limitation that the hierarchical structure of ontology can not be generated automatically and improves the accuracy of clustering analysis. Thirdly, this paper proposes a multi-policy ontology mapping, which obtains the similarity between ontology concepts from name, structure, attribute and instance strategy, and obtains the comprehensive similarity by introducing Sigmoid function to determine the weights of each strategy. It improves the precision and efficiency of ontology mapping and completes the organization of water environment ontology. Finally, the water environment ontology construction and organization system is designed and developed. According to the method and process of ontology construction, the system is divided into five modules: pretreatment module, domain concept extraction module, concept relation extraction module. Ontology mapping and integration module, ontology visualization module, the text data source ontology construction and visualization display. The validity of ontology construction method is verified by experiments, which provides technical support for ontology construction and application.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.1

【参考文献】