Guardado en:
Detalles Bibliográficos
Autores principales: Barayan, Abdullah, Camacho-Collados, Jose, Alva-Manchego, Fernando
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2409.20246
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912157779623936
author Barayan, Abdullah
Camacho-Collados, Jose
Alva-Manchego, Fernando
author_facet Barayan, Abdullah
Camacho-Collados, Jose
Alva-Manchego, Fernando
contents Readability-controlled text simplification (RCTS) rewrites texts to lower readability levels while preserving their meaning. RCTS models often depend on parallel corpora with readability annotations on both source and target sides. Such datasets are scarce and difficult to curate, especially at the sentence level. To reduce reliance on parallel data, we explore using instruction-tuned large language models for zero-shot RCTS. Through automatic and manual evaluations, we examine: (1) how different types of contextual information affect a model's ability to generate sentences with the desired readability, and (2) the trade-off between achieving target readability and preserving meaning. Results show that all tested models struggle to simplify sentences (especially to the lowest levels) due to models' limitations and characteristics of the source sentences that impede adequate rewriting. Our experiments also highlight the need for better automatic evaluation metrics tailored to RCTS, as standard ones often misinterpret common simplification operations, and inaccurately assess readability and meaning preservation.
format Preprint
id arxiv_https___arxiv_org_abs_2409_20246
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Analysing Zero-Shot Readability-Controlled Sentence Simplification
Barayan, Abdullah
Camacho-Collados, Jose
Alva-Manchego, Fernando
Computation and Language
Readability-controlled text simplification (RCTS) rewrites texts to lower readability levels while preserving their meaning. RCTS models often depend on parallel corpora with readability annotations on both source and target sides. Such datasets are scarce and difficult to curate, especially at the sentence level. To reduce reliance on parallel data, we explore using instruction-tuned large language models for zero-shot RCTS. Through automatic and manual evaluations, we examine: (1) how different types of contextual information affect a model's ability to generate sentences with the desired readability, and (2) the trade-off between achieving target readability and preserving meaning. Results show that all tested models struggle to simplify sentences (especially to the lowest levels) due to models' limitations and characteristics of the source sentences that impede adequate rewriting. Our experiments also highlight the need for better automatic evaluation metrics tailored to RCTS, as standard ones often misinterpret common simplification operations, and inaccurately assess readability and meaning preservation.
title Analysing Zero-Shot Readability-Controlled Sentence Simplification
topic Computation and Language
url https://arxiv.org/abs/2409.20246