Regresar

Predictive student action model for procedural training in 3D virtual environments

Abstract:

Based on our experience from previous works, a predictive student action model has been created, which uses student logs generated by a virtual environment for procedural training to elaborate summarized information. This predictive model can predict the most common behaviors by considering the sequences of actions more probable, which is useful to anticipate frequent student’ errors. These student logs are read from a student ontology developed in a previous work. This model is represented by an extended automaton, in which each of its states represents the effect of a student’s action (or a failed student’s action). Moreover, the transition function is defined including the frequency of occurrence, which helps in the action prediction. Before creating the automaton, the student logs are clustered based on the number of errors made by each student and the total time that each student spent to complete the entire practice. Next, we create an automaton for each cluster, which allows us to generate predictions more reliable to each student type.