This paper presents a supporting and consulting infrastructure (methods and tools) assisting educators during the distance learning process. The main focus is on the information sketching a student profile. Having identified all possible behaviors, the proposed infrastructure allows the development of probabilistic distribution models of the emerging educational events. The determination of probabilities is estimated by applying data mining techniques and then using the method of maximum entropy. As an example, the process of learning the semantic structure of motion verbs in Russian and Greek language is taken. Emphasis is given to those cases where verbs structures match entirely, partially overlap, as well as those whose semantic sizes are specific, characterize either the Russian, or solely the Greek language. © Springer-Verlag Berlin Heidelberg 2013.