CONICET | Buscador de Institutos y Recursos Humanos

In this work we present our progress in the field of Intelligent User Profiling. Our objective is to build a user profile that captures users? skills rather than classical users? interests. Thus, we propose a novel approach to learn users? skills by observing their behavior during a very common activity: playing games. Specifically, we automatically identify users? skills to manage abstractions by using digital games. Abstraction skills identification is important because it is related to several behavioral tendencies such as career preferences, aptitudes, and learning styles. Traditional skills identification is based on questionnaires whose application implies many complications, including non-intentional influences in the way questions are formulated, difficulty to motivate people to fill them out, and lack of awareness of the consequences or future uses of questionnaires. To address these limitations, we built a user profile that collects users? actions when playing digital games. Then, we built and trained a Hierarchical Naive Bayes network to infer users? skills to manage abstractions. The experiments carried out show that digital games can help us to identify abstraction skills with a promising accuracy.