Please use this identifier to cite or link to this item:
Title: Entrenamiento y evaluación automática y humana de un motor de traducción automática neuronal para la traducción de textos literarios del francés al español en el marco del proyecto MTUOC
Author: Peñagarikano Calvo, Jon Mikel
Tutor: Rodríguez Vázquez, Silvia  
Abstract: The use of machine translation systems to translate literary texts has recently aroused great interest. This has been motivated by the increasing potential of these systems and because, theoretically, literary texts do not seem suitable for machine translation. In this paper, a parallel corpus aligned by means of the SBERT algorithm is created to train a neural machine translation engine specialized in the translation of literary texts from French into Spanish. This training is performed using tools from the MTUOC project of the Universitat Oberta de Catalunya and a sentence-weighting technique so that segments from the specialized corpus are prioritized. This represents one of the main changes when compared to previous research. Afterwards, both automatic and human evaluation of the engine are carried out and the results are compared with those of a previously trained engine with a smaller specialized corpus. Although the translations provided by the trained engine are far from reaching the quality of a human translation, the results obtained in this study seem to indicate that the use of a larger specialized corpus and a sentence-weighting technique in the training process enhances the quality of the engine's output.
Keywords: neural machine translation
literary translation
parallel corpus
neural machine translation system training
automatic evaluation
human evaluation
Document type: info:eu-repo/semantics/masterThesis
Issue Date: Jun-2023
Publication license:  
Appears in Collections:Trabajos finales de carrera, trabajos de investigación, etc.

Files in This Item:
File Description SizeFormat 
jpenagarikanoTFM0623.pdfMemoria del TFM1,32 MBAdobe PDFThumbnail
View statistics

This item is licensed under aCreative Commons License Creative Commons