Abstract: Customization is a key factor for the acceptance and usability of information systems, especially when final users are not familiar with these systems. A major concern for customization is to recognize users' behaviors and expectations to adapt the system's behavior. Trace analysis during previous usage of a system is a promising source to gather information on users' habits and profiles. However, traces have to be interpreted to transform raw data into knowledge and to provide an artificial system with adaptation capacities to different uses. The MIND group of the LITIS lab leads several research projects on customization using trace analysis. In this seminar we will present how this approach has been applied in two different application cases, for information retrieval and for storytelling.