Подход к автоматизированному наполнению графов знаний сущностями на основе анализа таблиц

Abstract

The use of Semantic Web technologies including ontologies and knowledge graphs is a widespread practice in the development of modern intelligent systems for information retrieval, recommendation and question-answering. The pro-cess of developing ontologies and knowledge graphs involves the use of various information sources, for example, databases, documents, conceptual models. Tables are one of the most accessible and widely used ways of storing and presenting information, as well as a valuable source of domain knowledge. In this paper, it is proposed to automate the extraction process of specific entities (facts) from tabular data for the subsequent filling of a target knowledge graph. A new approach is proposed for this purpose. A key feature of this approach is the semantic interpretation (annotation) of individual table elements. A description of its main stages is given, the application of the approach is shown in solv-ing practical problems of creating subject knowledge graphs, including in the field of industrial safety expertise of pet-rochemical equipment and technological complexes. An experimental quantitative evaluation of the proposed ap-proach was also obtained on a test set of tabular data. The obtained results showed the feasibility of using the pro-posed approach and the developed software to solve the problem of extracting facts from tabular data for the subsequent filling of the target knowledge graph.