The modeling of behavior carried out by the construction of machine learning (or brain) on digital information, is today a way on which modern research has most improved. This break in the way of modeling the reality has been mainly built by the provision of varied data in very large quantities.
Indeed, every activity using quasi-free infrastructures of the largest information network (internet) generates data. The digitization of this information firstly made it possible to carry out trend studies by the global processing of information stored in databases. Then the oncoming of machine learning made it possible to treat each information as a single entity, allowing us to understand the behavior of a mass from attributes of the constituent entities.