Multilingual Fine-Grained Entity Typing

M.G.J. van Erp, P.T.J.M. Vossen

Research output: Chapter in Book / Report / Conference proceedingChapterAcademicpeer-review

22 Downloads (Pure)

Abstract

Many entity recognition approaches classify recognised entities into a limited set of coarse-grained entity types. However, for deeper natural language analysis and end-user tasks, fine-grained entity types are more useful. For example, while standard named entity recognition may determine that an entity is a person knowing whether that entity is a politician or an actor is important for determining whether, in a subsequent relation extraction task, a relation should be acts or governs. Currently, fine-grained entity typing has only been investigated for English. In this paper, we present a fine-grained entity typing system for Dutch and Spanish using training data extracted from Wikipedia and DBpedia. Our system achieves comparable performance to English with an F1 measure of .90 on over 40 types for both Dutch and Spanish.
Original languageEnglish
Title of host publicationLanguage, Data, and Knowledge
Subtitle of host publicationFirst International Conference, LDK 2017, Galway, Ireland, June 19-20, 2017, Proceedings
EditorsJorge Gracia, Francis Bond, John P. McCrae, Paul Buitelaar, Christian Chiarcos, Sebastian Hellmann
PublisherSpringer
Pages262-275
Number of pages14
ISBN (Electronic)9783319598888
ISBN (Print)9783319598871
DOIs
Publication statusPublished - 2017

Publication series

NameLecture Notes in Computer Science (subseries Lecture Notes in Artificial Intelligence)
PublisherSpringer Verlag
Volume10318 (LNAI)
ISSN (Print)0302-9743

Funding

The research for this paper was made possible by the CLARIAH-CORE project financed by NWO.

FundersFunder number
Nederlandse Organisatie voor Wetenschappelijk Onderzoek

    Fingerprint

    Dive into the research topics of 'Multilingual Fine-Grained Entity Typing'. Together they form a unique fingerprint.

    Cite this