基于语义标注的数据资源库元数据质量自动评估方法研究  

RESEARCH ON AUTOMATIC EVALUATION METHOD OF METADATA QUALITY OF DATA REPOSITORIES BASED ON SEMANTIC ANNOTATION

作　　者：郭晓明[1] 马良荔[1] 苏凯[1] 孙煜飞[1] Guo Xiaoming;Ma Liangli;Su Kai;Sun Yufei(Department of Computer Engineering, Naval University of Engineering, Wuhan 430033, Hubei, China)

机构地区：[1]海军工程大学计算机工程系,湖北武汉430033

出　　处：《计算机应用与软件》2018年第6期23-27,82,共6页Computer Applications and Software

基　　金：总装预研基金项目(9140A04020215JB11050)

摘　　要：针对海量信息资源库元数据缺乏对信息内容描述、语义表现力差的问题,在基于元数据信息组织的基础上,研究关系型数据库元数据质量自动评估方法,提出基于相似度计算的数据库元数据语义标注算法进行元数据自动语义标注。该算法综合考虑关系表与本体类名称的相似度计算以及结构相似度计算。其中结构相似度细分为表所含列和类所含属性之间的相似度,以及同它们各自连接的表和类之间的相似度。经综合计算后相似度值大于阈值的本体类用于语义标注。因必须同时满足名称和结构的相似度的本体概念和属性才能用于语义标注,标注准确性较高。另外该算法无需迭代计算,标注效率高。Against short of information content description and badness of semantic represent ability in massive data repository metadata. On the basis of metadata information organization,this paper studied the method of metadata automatic evaluation of relational database,and proposed database metadata semantic annotation algorithm based on similarity to realize auto semantic annotation of metadata. This algorithm considered the similarity calculation between the relational table and the ontology class name,and also considered the relational table structure similarity calculation. The structural similarity was subdivided into the similarities between the columns contained in the relational table and the attributes contained in the ontology class,and the similarity between the tables and classes that were connected. After comprehensive calculation,the ontology class with the similarity value greater than the threshold value was used for semantic annotation. Because the ontologies and attributes that satisfied the similarity between the name and the structure at the same time was used for semantic annotation,the annotation accuracy was high. In addition,the algorithm didn＇t require iterative calculations and had high labelling efficiency.

关 键 词：元数据质量 语义元数据 语义标注 本体映射 相似度计算 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]