Track analysis in hypermedia environments has become a hot research topic over the last decades. The main goal of this research is to help teachers to perceive and interpret the learner's activities in elearning situations, by exploiting and analyzing the tracks and providing knowledge on the activities, that we call learning indicators. In this paper, we present our initial proposal to automatically determine the learners' learning style in web based learning from learning indicators. The advantage of this approach is that it allows having a dynamically updated value of the learning style provided by the interpretation of tracks on the learner's activities. Furthermore, we propose the modeling of the learning style as a classification of the styles models proposed in the literature that can be detected in such a context. Thus the approach is independent of a particular learning style model. It is implemented in IDLS, a track based system. Through this system, we describe the main detections steps.