Extraction of Facts from Web-Tables based on Semantic Interpretation Tabular Data
Nikita O. Dorodnykh, Aleksandr Yu. Yurin · 2022
Knowledge graph engineering is a complex and time-consuming task. This process can be automated by using various types of information. Web-tables are worthy sources of relational data and can be used for knowledge graph augmentation with new facts. In this paper, we propose a semi-automated approach for knowledge graph engineering based on extracting data from web-tables. The main feature of our approach is a new method for the automatic recovery of semantics of webtables using an ontological schema included in a target knowledge graph. The proposed method consists of five stages: preprocessing, annotating named entity and literal columns, annotating relationships between columns, and extracting specific entities (facts). The approach is implemented in the form of a prototype of a web-based tool. We also present an experimental evaluation of the proposed approach and a case study in the domain knowledge graph engineering for the TALISMAN framework. The results of our experiments show the prospects of using the proposed solution to support the extraction of new facts from semantically annotated tabular data.