Automatic Simplification of Lithuanian Administrative Texts

Text simplification reduces the complexity of text while preserving essential information, thus making it more accessible to a broad range of readers, including individuals with cognitive disorders, non-native speakers, children, and the general public. In this paper, we present experiments on text...

Full description

Saved in:
Bibliographic Details
Main Authors: Justina Mandravickaitė, Eglė Rimkienė, Danguolė Kotryna Kapkan, Danguolė Kalinauskaitė, Tomas Krilavičius
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Algorithms
Subjects:
Online Access:https://www.mdpi.com/1999-4893/17/11/533
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850217892169121792
author Justina Mandravickaitė
Eglė Rimkienė
Danguolė Kotryna Kapkan
Danguolė Kalinauskaitė
Tomas Krilavičius
author_facet Justina Mandravickaitė
Eglė Rimkienė
Danguolė Kotryna Kapkan
Danguolė Kalinauskaitė
Tomas Krilavičius
author_sort Justina Mandravickaitė
collection DOAJ
description Text simplification reduces the complexity of text while preserving essential information, thus making it more accessible to a broad range of readers, including individuals with cognitive disorders, non-native speakers, children, and the general public. In this paper, we present experiments on text simplification for the Lithuanian language, aiming to simplify administrative texts to a Plain Language level. We fine-tuned mT5 and mBART models for this task and evaluated the effectiveness of ChatGPT as well. We assessed simplification results via both quantitative metrics and qualitative evaluation. Our findings indicated that mBART performed the best as it achieved the best scores across all evaluation metrics. The qualitative analysis further supported these findings. ChatGPT experiments showed that it responded quite well to a short and simple prompt to simplify the given text; however, it ignored most of the rules given in a more elaborate prompt. Finally, our analysis revealed that BERTScore and ROUGE aligned moderately well with human evaluations, while BLEU and readability scores indicated lower or even negative correlations
format Article
id doaj-art-3e5537f4e7fe463881646da56858c217
institution OA Journals
issn 1999-4893
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Algorithms
spelling doaj-art-3e5537f4e7fe463881646da56858c2172025-08-20T02:07:57ZengMDPI AGAlgorithms1999-48932024-11-01171153310.3390/a17110533Automatic Simplification of Lithuanian Administrative TextsJustina Mandravickaitė0Eglė Rimkienė1Danguolė Kotryna Kapkan2Danguolė Kalinauskaitė3Tomas Krilavičius4Faculty of Informatics, Vytautas Magnus University, 53361 Akademija, Kaunas District, LithuaniaFaculty of Informatics, Vytautas Magnus University, 53361 Akademija, Kaunas District, LithuaniaFaculty of Informatics, Vytautas Magnus University, 53361 Akademija, Kaunas District, LithuaniaFaculty of Informatics, Vytautas Magnus University, 53361 Akademija, Kaunas District, LithuaniaFaculty of Informatics, Vytautas Magnus University, 53361 Akademija, Kaunas District, LithuaniaText simplification reduces the complexity of text while preserving essential information, thus making it more accessible to a broad range of readers, including individuals with cognitive disorders, non-native speakers, children, and the general public. In this paper, we present experiments on text simplification for the Lithuanian language, aiming to simplify administrative texts to a Plain Language level. We fine-tuned mT5 and mBART models for this task and evaluated the effectiveness of ChatGPT as well. We assessed simplification results via both quantitative metrics and qualitative evaluation. Our findings indicated that mBART performed the best as it achieved the best scores across all evaluation metrics. The qualitative analysis further supported these findings. ChatGPT experiments showed that it responded quite well to a short and simple prompt to simplify the given text; however, it ignored most of the rules given in a more elaborate prompt. Finally, our analysis revealed that BERTScore and ROUGE aligned moderately well with human evaluations, while BLEU and readability scores indicated lower or even negative correlationshttps://www.mdpi.com/1999-4893/17/11/533text simplificationLithuaniantransformersfine-tuningmT5mBART
spellingShingle Justina Mandravickaitė
Eglė Rimkienė
Danguolė Kotryna Kapkan
Danguolė Kalinauskaitė
Tomas Krilavičius
Automatic Simplification of Lithuanian Administrative Texts
Algorithms
text simplification
Lithuanian
transformers
fine-tuning
mT5
mBART
title Automatic Simplification of Lithuanian Administrative Texts
title_full Automatic Simplification of Lithuanian Administrative Texts
title_fullStr Automatic Simplification of Lithuanian Administrative Texts
title_full_unstemmed Automatic Simplification of Lithuanian Administrative Texts
title_short Automatic Simplification of Lithuanian Administrative Texts
title_sort automatic simplification of lithuanian administrative texts
topic text simplification
Lithuanian
transformers
fine-tuning
mT5
mBART
url https://www.mdpi.com/1999-4893/17/11/533
work_keys_str_mv AT justinamandravickaite automaticsimplificationoflithuanianadministrativetexts
AT eglerimkiene automaticsimplificationoflithuanianadministrativetexts
AT danguolekotrynakapkan automaticsimplificationoflithuanianadministrativetexts
AT danguolekalinauskaite automaticsimplificationoflithuanianadministrativetexts
AT tomaskrilavicius automaticsimplificationoflithuanianadministrativetexts