14 de junio de 2024
Resumen:
This thesis explores the transformative potential of integrating Serious Games (SG) and causal Artificial Intelligence (AI) into social science research. This integration holds the key to addressing persistent challenges in social science research, such as limited sample sizes or the difficulty of reaching certain populations, like minors. Designed with a purpose beyond entertainment, SGs are emerging as a promising tool to engage diverse audiences and challenge traditional social research paradigms. The interactive and immersive nature of video games opens up opportunities to increase participant involvement, simulate real-life scenarios with fewer ethical constraints, and reach populations with greater demographic variability.
The main contribution of this thesis is to advocate the use of Probabilistic Graphical Causal Models (PGCM) to analyze data derived from the use of SGs in research contexts. These models facilitate the understanding of complex causal relationships and provide an analytical framework that overcomes some of the limitations of traditional statistical and machine learning approaches. In particular, we propose using these models to analyze sensitive social issues, such as cyberbullying (CB).
Firstly, we present a methodology for constructing Directed Acyclic Graphs (DAG) that harmoniously combine expert knowledge with data-driven algorithms. This hybrid approach aims to produce robust and realistic causal models that accurately reflect the complexity of the phenomena under study. Secondly, we introduce a novel synthetic data generation technique for SGs, based on a psychometric theoretical framework. This technique allows us to simulate player behavior in a parameterized and realistic manner. The incorporation of synthetic data into this area of research enhances the efficiency of development processes by overcoming obstacles such as data scarcity, and facilitates the design, analysis and management of data.
The empirical validation of these methods was carried out within the framework of the European H2020 project RAYUELA. This project is a pioneering multidisciplinary initiative that studies cybercrime involving minors and seeks to propose methodologies to better understand the factors that influence it. The results demonstrate the potential of SGs to observe and understand the complex dynamics of CB. This thesis posits that the integration of SGs and causal AI in social science research can significantly advance our capacity to address complex social issues by enabling large-scale ethical experiments to explore into the causal relationships that define such issues.
In conclusion, this thesis makes a significant contribution to demonstrating the potential of SGs as a valuable research tool in the social sciences and deepens existing methodologies in this field by integrating causal AI techniques. In doing so, it contributes to the multidisciplinary scientific community by offering new ways to explore and address relevant social challenges. The implications of this research extend beyond academia, providing practical tools and knowledge to develop effective prevention and intervention strategies against CB, thus improving the online well-being of minors.
Resumen divulgativo:
Esta tesis demuestra cómo la integración de los Juegos Serios y la IA causal en la investigación en ciencias sociales puede mejorar la comprensión de problemas complejos como el ciberacoso. Aboga por los modelos causales gráficos probabilísticos para analizar los datos de la investigación en juegos.
Descriptores: Inteligencia Artificial, Modelos Causales
Palabras clave: Serious Games, Artificial Intelligence, Causality, Computational Social Science, Cyberbullying, Adolescents.
Cita:
J. Pérez (2024), Impact of Serious Games and causal Artificial Intelligence on social science research: a case study on cyberbullying. Madrid (España).
