Enhance sequential pattern mining with time and reasoning

Sequential pattern mining (SPM) aims at identifying interesting patterns in large database of sequences. For instance, in pharmaco-epidemiology, SPM enables to identify care sequences in a collection of patient care pathways. Thus, it supports epidemiologists to reveal adverse medical events. But, SPM suffers from limitations:(i) the temporal dimension is poorly capture by sequential patterns, (ii) the analysts require less but much meaningful patterns. During this defense, I will present some contributions to tackle these limitations.