Ramón López-Cózar Delgado

Technologies for Inclusive Education: Beyond Traditional Integration Approaches

Students’ performance and motivation are influenced by their emotions. Game-based learning (GBL) environments comprise elements that facilitate learning and the creation of an emotional connection with students. GBL environments include Intelligent Tutoring Systems (ITSs) to ensure personalized learning. ITSs reason about students’ needs and characteristics (student modeling) to provide suitable instruction (tutor modeling). The authors’ research is focused on the design and implementation of an emotional student model for GBL environments based on the Control-Value Theory of achievement emotions by Pekrun et al. (2007). The model reasons about answers to questions in game dialogues and contextual variables related to student behavior acquired through students’ interaction with PlayPhysics. The authors’ model is implemented using Dynamic Bayesian Networks (DBNs), which are derived using Probabilistic Relational Models (PRMs), machine learning techniques, and statistical methods. This work compares an earlier approach that uses Multinomial Logistic Regression (MLR) and cross-tabulation for learning the structure and conditional probability tables with an approach that employs Necessary Path Condition and Expectation Maximization algorithms. Results showed that the latter approach is more effective at classifying the control of outcome-prospective emotions. Future work will focus on applying this approach to classification of activity and outcome-retrospective emotions. Karla Muñoz University of Ulster, UK Paul Mc Kevitt University of Ulster, UK Tom Lunney University of Ulster, UK Julieta Noguez Tecnológico de Monterrey, México Luis Neri Tecnológico de Monterrey, México

Merging Intention and Emotion to Develop Adaptive Dialogue Systems

In this paper we propose a method for merging intentional and emotional information in spoken dialogue systems in order to make dialogue managers more efficient and adaptive. The prediction of the user intention and emotion is carried out for each user turn in the dialogue by means of a module conceived as an intermediate phase between natural language understanding and dialogue management in the architecture of these systems. We have applied and evaluated our method in the UAH system, for which the evaluation results show that merging both sources of information improves system performance as well as its perceived quality.