We propose a context-based filtering process which aims at adapting the awareness information delivered to mobile users by collaborative web systems. This filtering process relies on a model of context which integrates both a physical and an organizational dimensions and allows to represent the user's current context as well as general profiles. These profiles are descriptions of user's potential contexts and express the awareness information filtering rules to apply when the user's current context matches one of them. These rules reflect the user's preferences given the context described. We describe how the filtering process performs in two steps, one for identifying the general profiles that apply, and a second for selecting the awareness information. We also discuss the patterns matching algorithms used in the filtering process to compare the contexts descriptions.