With the considerable amount of data and the diversity of the user’s needs, the personalization of information becomes a real challenge in web-based information systems. This paper presents a personalized access technique, based on a profile modeling. This technique requires two steps: 1) building user’s profile and 2) using profile content for filtering information. The first step consists in modeling a global profile which can be used for different and independent requirements of personalization. The profile modeling takes into account the user’s context when defining user’s profiles. We describe a global profile in three levels: meta-description level, description level for a specific domain, instances (data). The second step consists in selecting the profiles rules which correspond to the user and his current context and to filter information accordingly.