Publication: Identificación automática de la personalidad de un usuario en videojuegos
Loading...
Official URL
Full text at PDC
Publication Date
2020
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Hasta la fecha, no existe ningún medio que permita determinar la personalidad de un usuario con suficiente precisión de forma no invasiva, sin embargo, sí existen estudios previos a este que demuestran una relación entre la personalidad y el comportamiento en videojuegos de un jugador.
Con esta investigación se busca definir una metodología eficiente y estable que permita obtener el perfil de personalidad de un usuario de manera no intrusiva a través de una experiencia interactiva, estudiando la personalidad mediante el videojuego.
Para ello, se ha escogido como referencia en la valoración de la personalidad un cuestionario validado que la determina en función de cinco cualidades, recogidas según el modelo Big Five, y se ha diseñado una experiencia interactiva centrada en la narrativa y los diálogos en el entorno de Minecraft, asignando unos valores para cada una de estas cualidades a cada interacción del jugador, considerando el contexto en el que aparecen.
Finalmente, los datos recogidos han sido analizados y comparados con los resultados del cuestionario para determinar su precisión y equilibrado. Este equilibrado, que busca acercar los valores de las opciones al óptimo para minimizar el error en el conjunto de datos, se ha realizado de forma manual y se ha planteado el modo de realizarlo a través de aprendizaje máquina, sin llevarlo a cabo debido a la reducida cantidad de datos.
Como resultado se ha conseguido una metodología viable para determinar la personalidad de un usuario de una manera no invasiva y que, con la validación de un experto y una mayor cantidad de muestras, se podría obtener una experiencia interactiva capaz de determinar la personalidad de un sujeto.
Until today, there is no existing means to determine a user’s personality with enough accuracy in a non-intrusive way, although there are previous studies that prove an existing relationship between personality and player behavior within a videogame. This study is targeted to define a stable and efficient methodology, allowing to identify user personality profile non-intrusively, through an interactive experience, studying personality using videogames. As reference, it was used a validated questionnaire that determines personality split into five qualities, according to the Big Five model, and an interactive experience was built around these categories in Minecraft, focused on narrative experience and dialogs, assigning values to each of these factors in every player interaction, taking into account the context in which they are. Finally, obtained data has been analyzed, comparing it to the results derived from the questionnaire to determine accuracy and balance. This balancing was meant to get the option’s values closer to the optimal output in order to minimize errors in the data group and was done manually, proposing an automatization through machine learning that could not be developed due to the reduced data amount. As result, a viable methodology to determine user profiling was achieved in a non-intrusive way and, with an expert validation and greater data pool, user personality could be obtained via an interactive experience.
Until today, there is no existing means to determine a user’s personality with enough accuracy in a non-intrusive way, although there are previous studies that prove an existing relationship between personality and player behavior within a videogame. This study is targeted to define a stable and efficient methodology, allowing to identify user personality profile non-intrusively, through an interactive experience, studying personality using videogames. As reference, it was used a validated questionnaire that determines personality split into five qualities, according to the Big Five model, and an interactive experience was built around these categories in Minecraft, focused on narrative experience and dialogs, assigning values to each of these factors in every player interaction, taking into account the context in which they are. Finally, obtained data has been analyzed, comparing it to the results derived from the questionnaire to determine accuracy and balance. This balancing was meant to get the option’s values closer to the optimal output in order to minimize errors in the data group and was done manually, proposing an automatization through machine learning that could not be developed due to the reduced data amount. As result, a viable methodology to determine user profiling was achieved in a non-intrusive way and, with an expert validation and greater data pool, user personality could be obtained via an interactive experience.
Description
Trabajo de Fin de Grado en Desarrollo de Videojuegos, Facultad de Informática UCM, Departamento de Ingeniería de Software e Inteligencia Artificial, Curso 2019/2020