Improving Estonian Text Simplification through Pretrained Language Models and Custom Datasets

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Barbu, Eduard, Muru, Meeri-Ly, Malva, Sten Marcus
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908782849687552
author Barbu, Eduard
Muru, Meeri-Ly
Malva, Sten Marcus
author_facet Barbu, Eduard
Muru, Meeri-Ly
Malva, Sten Marcus
contents This paper presents a method for text simplification based on two neural architectures: a neural machine translation (NMT) model and a fine-tuned large language model (LLaMA). Given the scarcity of existing resources for Estonian, a new dataset was created by combining manually translated corpora with GPT-4.0-generated simplifications. OpenNMT was selected as a representative NMT-based system, while LLaMA was fine-tuned on the constructed dataset. Evaluation shows LLaMA outperforms OpenNMT in grammaticality, readability, and meaning preservation. These results underscore the effectiveness of large language models for text simplification in low-resource language settings. The complete dataset, fine-tuning scripts, and evaluation pipeline are provided in a publicly accessible supplementary package to support reproducibility and adaptation to other languages.
format Preprint
id arxiv_https___arxiv_org_abs_2501_15624
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Improving Estonian Text Simplification through Pretrained Language Models and Custom Datasets
Barbu, Eduard
Muru, Meeri-Ly
Malva, Sten Marcus
Computation and Language
This paper presents a method for text simplification based on two neural architectures: a neural machine translation (NMT) model and a fine-tuned large language model (LLaMA). Given the scarcity of existing resources for Estonian, a new dataset was created by combining manually translated corpora with GPT-4.0-generated simplifications. OpenNMT was selected as a representative NMT-based system, while LLaMA was fine-tuned on the constructed dataset. Evaluation shows LLaMA outperforms OpenNMT in grammaticality, readability, and meaning preservation. These results underscore the effectiveness of large language models for text simplification in low-resource language settings. The complete dataset, fine-tuning scripts, and evaluation pipeline are provided in a publicly accessible supplementary package to support reproducibility and adaptation to other languages.
title Improving Estonian Text Simplification through Pretrained Language Models and Custom Datasets
topic Computation and Language
url https://arxiv.org/abs/2501.15624