基于关系数据库的时态RDF建模

Temporal RDF Modeling Based on Relational Database

随着时态数据的不断增加,时态知识图谱的概念得到了普及,如何高效地表示时态知识图谱已成为一个重要的研究方向。RDF(Resource Description Framework)虽然在传统知识图谱建模中被广泛运用,但其只能表示静态语义,缺乏表示时态知识图谱的能力,因此已有几种针对时态知识图谱的时态RDF模型被提出。但这些模型都只是将时态信息简单地附加在谓语或整个三元组上,缺少对时态信息所属对象的准确定位。为了更好地表示时态知识图谱,文中提出了一个新的时态RDF表示模型-tRDF。该模型首先根据宾语的不同类型,选择性地将时态信息附加在宾语或谓语上;其次,结合时态数据库的概念,给出了一种基于关系数据库PostgreSQL的tRDF数据存储方法;最后,从数据存储的时间和空间两个方面对所提出的tRDF数据存储方法进行了验证。实验结果表明,所提方案能有效地表示时态知识图谱。

With the increase of temporal data,the concept of temporal knowledge graph is popularized,and how to represent temporal knowledge graph efficiently has become an important research direction.Although resource description framework(RDF)is widely used in traditional knowledge graph modeling,it can only represent static semantics and lacks the ability to represent temporal knowledge graph.Therefore,several temporal RDF models have been proposed for temporal knowledge graph,but all these models simply attach temporal information to the predicate of RDF or the whole triple,and lack the accurate positioning of the object to which the temporal information belongs.In order to better represent temporal knowledge graph,firstly,this paper proposes a new temporal RDF representation model called tRDF,which attaches temporal information to the object or predicate according to the type of object.Secondly,by combining the concept of temporal database,this paper presents a tRDF data storage method based on the relational database,PostgreSQL.Finally,the proposed tRDF data storage method is verified from two aspects,the time of storing and the size of space.Experimental results show that the proposed scheme can effectively represent temporal knowledge graph.