The objective of the thesis is to explore the combination of contextual information as well as sequential aspect of user's behavior in order to provide him/her with a personalised itinerary of spatio-temporal activities. The study aims at maximising user's satisfaction with his/her attendance of big distributed event (such as Festival Lumière or Fête de la Musique, cruise or conference with multiple parallel sessions). This work requires the modelling of user's interests in activities with respect to different constraints, i.e. temporal (simultaneity of activities, etc.), geographic (closeness of an activitiy to the user's current position), and users profiles. Another step includes the contraction of an optimal itinerary taking into account the set of constraints. Finally, an ecaluation protocol will be proposed and applied for validation of proposed approach.