Map formalism allows specifying processes with a high level of variability. However, this means many variation points, and therefore we need guidance to enact maps by customizing them. Traditional guidance consists in raising decision points to navigate in a map. The limit is that many decision points are raised at the same time, and the user (who enacts the map) does not know which decision to make first. Another kind of guidance, yet to be explored, consists in providing recommendations to the user. Such recommendations can be drawn from collections of profiles collected from map enactment traces using techniques from the data mining domain. This paper proposes a trace management system adapted to maps that was designed to support recommendation-based guidance. The paper shows how data mining algorithms can be used to find profile clusters in a collection of map enactment traces, used then to provide recommendations to the users.