Automatic Semantic Modeling for Structural Data Source with the Prior Knowledge From Knowledge Graph
Zaiwen Feng, Jiakang Xu, Wolfgang Mayer, Wangyu Huang, Keqing He, Markus Stumptner, Georg Großmann, Hongyu Zhang, Lin Ling · 2021
Mapping structured data to a shared domain ontology is a key step in publishing semantic content on the Web. This problem is known as Relational-To-Ontology Mapping Problem (Rel2Onto). Modeling the semantics of data manually requires huge human cost and expertise, making an automatic method of semantic modeling desired. Most of the related work focuses on semantic annotation of source attributes. However, besides semantically annotating source attributes, it is challenging to explicitly infer the relationships between attributes. In this paper we improve previous work by Taheriyan et al. [4] using Subgraph Matching to take into account frequencies of candidate semantic models occurring in the domain knowledge graph used as background knowledge. Preliminary experiments demonstrate that our method achieves higher precision and recall than the state-of-the-art solutions in the difficult scenarios where only few historical mappings between domain ontology and data sources are available.