2026-08-17T09:28:42Z https://drops.dagstuhl.de/oai

oai:drops-oai.dagstuhl.de:18517 2024-03-06T10:32:08Z ddc:004 open_access

A Pseudonymization Prototype for Hungarian Novák, Attila Novák, Borbála named entity recognition morphological reinflection pseudonymization entity embedding model In this paper, we present a pseudonymization prototype for Hungarian, an agglutinating language with complex morphology, implemented as a web service. The service provides the following functions: entity identification and extraction; automatic generation and selection of replacement candidates; automatic and consistent replacement and reinflection of entities in the final pseudonymized document. The named entity recognition model applied handles names of persons well, and it has decent performance on other entity types as well. However ID-like entities need to be handled separately to achieve proper performance (not handled in the current prototype version). For automatic replacement candidate generation, a simple entity embedding model is used. We discuss the performance and limitations of the prototype in detail. Schloss Dagstuhl – Leibniz-Zentrum für Informatik Attila Novák and Borbála Novák 2023 Is Part Of OASIcs, Volume 113, 12th Symposium on Languages, Applications and Technologies (SLATE 2023) InProceedings Text doc-type:ResearchArticle publishedVersion application/pdf doi:10.4230/OASIcs.SLATE.2023.3 urn:nbn:de:0030-drops-185177 https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.SLATE.2023.3 eng https://creativecommons.org/licenses/by/4.0/legalcode