An evidential neural network (ENN) for predicting individual travel mode is presented. This model can be used to support management decision making and to build predictions under uncertainty related to changes in people's behavior, the economic context, or the environment and policy. The presented model uses individuals’ characteristics, transportation mode specifications, and data related to places of work and residence. The data set analyzed was taken from a survey conducted in 2007 and contains information on the daily mobility (e.g., from home to work) of individuals who either lived or worked in Luxembourg. Individual characteristics were extracted to relate daily mobility (journeys between home and work, in particular) to the characteristics of working individuals. Information about public transportation specification and some geographical particularities of residential areas and workplaces were used. Rates of successful prediction obtained by the ENN and several alternative approaches were compared by cross-validation. The results showed that the ENN was superior to the studied alternatives.

We submit a method (EMPI: Evaluation of Multimedia, Pedagogical and Interactive software) to evaluate multimedia software used in educational context. Our purpose is to help users (teachers or students) to decide in front of the large choice of software actually proposed. We structured a list of evaluation criteria, grouped through six approaches: the general feeling, the technical quality, the usability, the scenario, the multimedia documents, and the didactical aspects. A global questionnaire joins all this modules. We are also designing software that could make the method easier to use and more powerful. We present in this paper the list of the criteria we selected and organised, along with some examples of questions, and a brief description of the method and the linked software.

No abstract

Emotions have a major role in the player-game interaction. In serious games playing contexts, real-time assessment of the player's emotional state is crucially important to enable an emotion-driven adaptation during gameplay. In addition, a personalized assessment and adaptation based on the player's characteristics remains a challenge for serious games designers.This paper presents a generic and efficient emotion-driven approach for personalized assessment and adaptation in serious games, in which two main methods and their algorithms are proposed. The first one is a method for assessing, in real time, the player's emotion taking into account the personality type and the playing style of the player.The second one is an emotion-driven personalized adaptation method based on Markov modeling of dependency between the serious game events and the change in the player's emotional state. Therefore, the proposed approach has been evaluated by playing an affective vs. non-affective version of a serious game that we have developed to illustrate the applicability of the above-mentioned methods. The overall results showed that owing to our approach, a serious game become able to enhance its adaptivity toward playing outcomes and improve its overall playability.