Recommendations have long been a means of helping users select services. In a smart city environment, recommendation algorithms should take into account the user’s context to gain in accuracy. What is the context of a smart city user and how can it be captured ? These are the two questions we answer in this paper. After specifying what we understand by context information, we show how the city’s mobility pattern can be used to infer rich contextual information. The main objective of our project will be finally to recommend services according to an estimated trajectory of a user in the smart city.