Mobile applications are increasingly taking advantages of the diverse geospatial web services to meet the information needs of their users. However, matching available web services to user's information needs is not a trivial task, as there are many contextual factors that may influence the fitness of use. In addition, mobile activities can be highly dynamic and interleaving, which demand certain level of context-adaption for web service matching policies. Previous work on context-based service matching and composition tends to focus on environmental and functional contexts that can be either sensed directly or defined a prior, and they assume relatively stable user activities. In this paper, we describe a method for representing and reasoning the intentional structures of mobile activities and use it to contextualize mobile map services. Our context model treats a user's mobile activity as an evolving collaborative plan situated in a set of physical and mental factors. The model explicitly reasons on the intentional structure of the mobile activity to determine the appropriate service matching policy on the fly. The feasibility and benefit of using this model is demonstrated through the implementation of a prototype system, MyTour - a mobile city tour guide application.