Publication: EmoTraductor 2.0. Análisis emocional de textos en español
Loading...
Official URL
Full text at PDC
Publication Date
2020
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Existen personas para las que resulta difícil interpretar y reconocer el contenido afectivo de un texto. En personas con Trastornos de Espectro Autista (TEA) estas dificultades se incrementan, surgiendo la necesidad de crear herramientas para conseguir la integración digital de personas con este tipo de dificultad.
El EmoTraductor 1.0 es una primera aproximación a este tipo de herramientas para el castellano. Este trabajo tiene como objetivo evaluar dicha herramienta e intentar mejorar su marcado emocional añadiendo distintas mejoras como el tratamiento de la negación o los modificadores. Este TFM realiza un análisis de los resultados obtenidos con el EmoTraductor 1.0 para identificar sus deficiencias y mejorarlo. Entre las mejoras propuestas caben destacar: el tratamiento de los modificadores y el tratamiento de la negación. Cada una de las mejoras propuestas ha sido implementada y posteriormente evaluada para comprobar su validez y la mejora en el marcado conseguida. Todas aquellas mejoras que se tradujeron en un avance en la detección de la carga emocional de un texto han sido incorporadas a una nueva versión de la herramienta, el EmoTraductor 2.0.
Finalmente, tras una evaluación final de los resultados, la nueva versión del EmoTraductor obtiene mejores resultados que la primera versión. No obstante, también se identifica aún trabajo por realizar para continuar mejorando el marcado emocional como mejorar el tratamiento de la negación o la ampliación del diccionario emocional que usa el sistema como base.
Sometimes, it is difficult to interpret the information we find on the internet and recognize the affective content of a text, and the diculty increases as a result of the growth of the Internet. In people with Autism Spectrum Disorders (ASD) these difficulties increase even more. As a consequence, the need to create tools to achieve digital integration for people with different abilities appears. Thanks to affective computing, great advances are being made for its digital integration by creating tools capable of emotionally labeling texts using emotional marking methods. Although most of the work and advances in affective computing have been done in English, we know some tools in Spanish that are doing the same task or we expect them to do it in the future. One of these tools is the EmoTraductor 1.0 developed in 2018 at the Faculty of Computing at the Complutense University of Madrid. This project aims to evaluate this tool and try to improve its emotional marking of texts by adding different improvements such as nagations or modifiers. A study of EmoTraductor 1.0 is carried out to improve its deficiencies, the treatment of modiers is studied and a treatment method is applied. The treatment of negation is also studied, presenting different methods. Each of the proposed improvements is associated with a serie of evaluations to check their validity. Finally, it is possible to improve the emotional marking of texts by correcting the malfunctions of the EmoTraductor 1.0 application, adding the treatment of modifiers and the treatment of negation.
Sometimes, it is difficult to interpret the information we find on the internet and recognize the affective content of a text, and the diculty increases as a result of the growth of the Internet. In people with Autism Spectrum Disorders (ASD) these difficulties increase even more. As a consequence, the need to create tools to achieve digital integration for people with different abilities appears. Thanks to affective computing, great advances are being made for its digital integration by creating tools capable of emotionally labeling texts using emotional marking methods. Although most of the work and advances in affective computing have been done in English, we know some tools in Spanish that are doing the same task or we expect them to do it in the future. One of these tools is the EmoTraductor 1.0 developed in 2018 at the Faculty of Computing at the Complutense University of Madrid. This project aims to evaluate this tool and try to improve its emotional marking of texts by adding different improvements such as nagations or modifiers. A study of EmoTraductor 1.0 is carried out to improve its deficiencies, the treatment of modiers is studied and a treatment method is applied. The treatment of negation is also studied, presenting different methods. Each of the proposed improvements is associated with a serie of evaluations to check their validity. Finally, it is possible to improve the emotional marking of texts by correcting the malfunctions of the EmoTraductor 1.0 application, adding the treatment of modifiers and the treatment of negation.
Description
Trabajo de Fin de Máster en Ingeniería Informática, Facultad de Informática UCM, Departamento de Ingeniería de Software e Inteligencia Artificial, Curso 2019/2020