Methods and tools for identifying names of entities and relationships based on lexical and syntactic templates are considered in the framework of the task of semantic indexing of document texts. The content of the text is considered as a set of elementary facts represented by triplets, including the names of entities and relationships. Relations are divided into immanent, situational, and structural–linguistic. A taxonomy of relations is used to typify situational relations, whose classes include linguistic constructions. Immanent relations are formed on the basis of a network of concepts (thesaurus). A taxonomy of properties and units of measurement is used to identify the properties of entities. The proposed approach allows one to use the names of entities and the names of relationships, as well as elementary facts and complete semantic constructions made up of them, as a search query.
Automatic Documentation and Mathematical Linguistics – Springer Journals
Published: Jul 12, 2021
Keywords: semantic information retrieval; text processing; ontologies; semantic relations; conceptual framework; method of identification of semantic relations