Query personalization was introduced as an advanced mechanism to allow the reformulation of database queries and adapt them to the user's domain of interest and preferences. Domain of interest and preferences are captured into a profile which is explicitly defined by the user or derived from his or her logged interactions with the information system. A substantial effort was done in recent years to provide reformulation techniques which enrich user queries prior to their execution. These approaches are often limited to a single data source. However, query personalization is much more crucial in a context where data sources are numerous and distributed, i.e. a mediation or P2P systems, as the user is generally overloaded with massive results in response to his or her queries. To deal with such a context of distributed data sources, we have studied two naive approaches and proposed an advanced approach, based on a schemafree profile. A comparison of the three approaches is performe...