Mobile applications changed unexpectedly people life and business models around the world. Nevertheless, there are old applications, called legacies, without adaptation to mobile devices, because this adaptation or migration have a considerable cost in dependence of software scope. Currently, most users bring constantly their smartphones and other devices with them, especially millennials. For that reason, some approaches try to solve this portabilization, generating certain improvements. However, in the majority of these solutions there is not a direct participation of users; do not consider their visual identity, analysis of feeling or mining of opinion. This paper proposes getting the behavior web application model with Markov heuristics from the widgets closeness matrix, prior to adaptation in order to include the user logic.