Publication: Uso de modelos de lenguaje natural para la interacción humano-robot
Loading...
Identifiers
Publication date
2024-01
Defense date
2023-09-27
Authors
Advisors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Este proyecto fin de Máster se ha realizado en el marco del proyecto "Minería de
Datos de Redes Sociales para la Interacción Humano Robot", acción financiada por la
Comunidad de Madrid a través de la línea "Estímulo a la Investigación de Jóvenes
Doctores/as" del Convenio Plurianual con la UC3M (SMM4HRI-CM-UC3M), en el
marco del V PRICIT.
El principal propósito del proyecto es el desarrollo de una nueva funcionalidad
para el robot social Mini, que utilice las redes sociales para mantener a las personas
informadas y conectadas con la actualidad. Para ello, se obtiene información actualizada
de las noticias más recientes publicadas en Twitter, se procesa el contenido de estas, se
genera un resumen mediante modelos de Inteligencia Artificial conocidos como grandes
modelos de lenguaje y se responden las preguntas de los usuarios relacionadas con dichos
artículos mediante otros modelos similares. Con todo ello, se ha implementado una
solución adaptable a las gustos individuales de los usuarios que permite la personalización
del contenido, estimulando una mayor adopción e interacción con el sistema.
Los resultados obtenidos en la experimentación avalan un buen resultado en las
respuestas generadas por los diferentes modelos de procesamiento de lenguaje natural que
se han empleado, así como la capacidad para proporcionar información útil y responder
preguntas de manera efectiva. Las conclusiones obtenidas respaldan una experiencia de
usuario totalmente satisfactoria y una percepción del robot con altas capacidades para la
tarea desarrollada. Por tanto, el proyecto demuestra ser una solución prometedora,
ofreciéndose como una herramienta útil para mejorar la calidad de vida, especialmente
para las personas mayores. La aplicación y uso están destinados en última instancia a la
asistencia de personas mayores, con el objetivo de reducir su aislamiento social causado
por las nuevas tecnologías, promoviendo así una sociedad más inclusiva. El proyecto se
ha llevado a cabo de manera interdisciplinaria en conjunto con el Laboratorio de Robótica
Social y el Grupo de Sistemas Interactivos, ambos de la Universidad Carlos III de Madrid.
This master's thesis is part of the project "Social Media data mining for HumanRobot Interaction", funded by the Madrid Regional Government through "Estímulo a la Investigación de Jóvenes Doctores/as" line under the agreement with UC3M (SMM4HRICM-UC3M), in the V PRICIT framework. The main purpose of the project is to develop a new functionality for the Mini social robot, which uses social networks to keep people informed and connected with current events. To do this, updated information is obtained from the most recent news published on Twitter, the content of these is processed, a summary is generated using Artificial Intelligence models known as large language models and users' questions related to these articles are answered using other similar models. With all this, a solution adaptable to the individual preferences of the users has been implemented that allows the personalisation of the content, stimulating a greater adoption and interaction with the system. The results obtained in the experimentation support a good result in the answers generated by the different natural language processing models that have been used, as well as the capacity to provide useful information and answer questions effectively. The conclusions obtained support a totally satisfactory user experience and a perception of the robot with high capabilities for the task developed. Therefore, the project proves to be a promising solution, offering itself as a useful tool to improve the quality of life, especially for the elderly. The application and use are ultimately aimed at assisting elderly people, with the objective of reducing their social isolation caused by new technologies, thus promoting a more inclusive society. The project has been reached in an interdisciplinary manner in conjunction with the Social Robotics Laboratory and the Interactive Systems Group, both at the Carlos III University of Madrid.
This master's thesis is part of the project "Social Media data mining for HumanRobot Interaction", funded by the Madrid Regional Government through "Estímulo a la Investigación de Jóvenes Doctores/as" line under the agreement with UC3M (SMM4HRICM-UC3M), in the V PRICIT framework. The main purpose of the project is to develop a new functionality for the Mini social robot, which uses social networks to keep people informed and connected with current events. To do this, updated information is obtained from the most recent news published on Twitter, the content of these is processed, a summary is generated using Artificial Intelligence models known as large language models and users' questions related to these articles are answered using other similar models. With all this, a solution adaptable to the individual preferences of the users has been implemented that allows the personalisation of the content, stimulating a greater adoption and interaction with the system. The results obtained in the experimentation support a good result in the answers generated by the different natural language processing models that have been used, as well as the capacity to provide useful information and answer questions effectively. The conclusions obtained support a totally satisfactory user experience and a perception of the robot with high capabilities for the task developed. Therefore, the project proves to be a promising solution, offering itself as a useful tool to improve the quality of life, especially for the elderly. The application and use are ultimately aimed at assisting elderly people, with the objective of reducing their social isolation caused by new technologies, thus promoting a more inclusive society. The project has been reached in an interdisciplinary manner in conjunction with the Social Robotics Laboratory and the Interactive Systems Group, both at the Carlos III University of Madrid.
Description
Keywords
Software, Inteligencia artificial, Robótica social, Minería de redes sociales, Diálogos, Artificial intelligence, Social robotics, Social media mining, Dialogues