Aplicación de los videojuegos a la medida de la inteligencia
Loading...
Official URL
Full text at PDC
Publication date
2020
Defense date
06/2020
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
Diversos proyectos de investigación recientes han demostrado un marcado grado de correlación entre el factor g (o factor de inteligencia general) y el rendimiento global en los videojuegos. Los experimentos llevados a cabo en ellos incluyeron no sólo juegos “mentales” clásicos (o brain games), sino también diferentes títulos comerciales ya existentes. Esos juegos “cerrados” no permiten la extracción automática de datos de comportamiento en el juego. Como resultado, los investigadores a menudo recurrieron a métodos manuales externos para registrar las medidas de rendimiento e interacción con el juego. El objetivo de este trabajo es ayudar a mejorar el proceso de recopilación de datos utilizado en dichos estudios mediante (1) la reimplementación de un pequeño subconjunto que consta de tres de los juegos empleados en estos estudios anteriores (2) desarrollar un sistema de telemetría para automatizar y mejorar la recopilación de eventos y variables de usuario en el juego y (3) implementar una plataforma web para realizar un experimento en línea con el fin de reunir dichos datos. En tres semanas, un total de 53 participantes completaron los tres juegos disponibles en el sistema. A pesar del tamaño modesto del conjunto de datos, pudimos mostrar un factor de inteligencia latente subyacente a través de un análisis factorial exploratorio. Además, intentamos predecir el resultado final de un jugador en un juego dado a partir de registros de partidas truncadas (hasta cierto punto en el tiempo) a través de redes neuronales y bosques aleatorios. Este análisis posterior podría permitir a futuros estudios acortar sus tiempos de experimentación, aumentando así la viabilidad de la evaluación de inteligencia basada en juegos.
Several recent research projects have shown a high correlation between the g factor (or general intelligence factor) and overall performance in video games. The experiments carried out in them included not only classical brain games, but also different existing comercial titles. These closed games do not allow the automatic extraction of in-game behaviour data. As a result, researchers often resorted to manual methods of registering measurements externally. The aim of our work is to help to improve the data collection process used in those studies by (1) reimplementing a small subset consisting of three of the games administered in these former studies (2) developing a telemetry system to automate and enhance the recording of in-game user events and variables and (3) deploying a web platform to conduct an online experiment to collect such data. Within three weeks, a total of 53 participants completed all three games available on the system. Despite the dataset’s modest size, we were able to show an underlying latent intelligence factor through an exploratory factor analysis. Moreover, we attempted to predict a player’s final score in a given game from truncated play logs (up to a certain point in time) using neural networks and random forests. This later analysis could potentially allow future studies to shorten experiment times, thus increasing the viability of game-based intelligence assessment.
Several recent research projects have shown a high correlation between the g factor (or general intelligence factor) and overall performance in video games. The experiments carried out in them included not only classical brain games, but also different existing comercial titles. These closed games do not allow the automatic extraction of in-game behaviour data. As a result, researchers often resorted to manual methods of registering measurements externally. The aim of our work is to help to improve the data collection process used in those studies by (1) reimplementing a small subset consisting of three of the games administered in these former studies (2) developing a telemetry system to automate and enhance the recording of in-game user events and variables and (3) deploying a web platform to conduct an online experiment to collect such data. Within three weeks, a total of 53 participants completed all three games available on the system. Despite the dataset’s modest size, we were able to show an underlying latent intelligence factor through an exploratory factor analysis. Moreover, we attempted to predict a player’s final score in a given game from truncated play logs (up to a certain point in time) using neural networks and random forests. This later analysis could potentially allow future studies to shorten experiment times, thus increasing the viability of game-based intelligence assessment.
Description
Trabajo de fin de Grado en Grado en Desarrollo de Videojuegos - Doble Grado en Ingeniería Informática y Matemáticas, Facultad de Informática UCM, Departamento de Ingeniería del Software e Inteligencia Artificial, Curso 2019/2020