Most frequent Spatio-Temporal pattern mining and visualization in a large graph – MoS-T

Recent technological advances are leading to a massive production of spatio-temporal data that can be modeled by a graph. This is crucial to analyse, however, these data are very often characterised by a volume and a complexity never previously seen. Despite the dynamism of data mining research, few unsupervised methods are usable for extracting both original and relevant information due to their runtime. It is therefore essential to design and develop innovative data mining approaches that are capable of processing these large volumes of data while taking into account the spatial and temporal aspects. The objective of the MoS-T project is to develop innovative methods to extract the most frequent approximate patterns in a spatiotemporal graph, by exploiting deep neural networks, in order to provide a synthetic visualization of the spatiotemporal phenomenon being studied.