Named Entity Recognition and Knowledge Linkage

Abstract

Named entity recognition (NER) is an important method for Natural Language Processing. In this work, the Named Entity Recognition model FLERT XLM-R by Schweter and Akbik (2021) is investigated with respect to Bertuch (1790) Bilderbuch für Kinder (1790- 1830). The objective is to successfully apply the model for recognizing named entities of the book Bilderbuch für Kinder and to link the recognized entities to corresponding Wiki- data items. Achieving this undertaking leads to linking the entities to the current knowledge of the Wikidata database and gives researchers the opportunity to gain further knowledge in the historical education field. To achieve this goal, the author of this paper developed a data curation pipeline, starting with the extraction of named entities, followed by the creation of a gold-standard validated with the Inter- Annotator Agreement (IAA). The result of the extracted entities has been approved using F-measure and was processed through OpenRefine. The delineated procedure can be used to create similar knowledge linkages.

More Information

Jonathan Baum, Information Scientist

Website: jobaum.page
Mail: jonathan (dot) baum (at) stud.h-da.de

Information Science in Darmstadt, Germany:  https://sis.h-da.de/
Information Science Contents: https://h-da.de/information-science/