Network science is a fundamental tool for representing, understanding and analyzing an ever increasing variety of systems. However, the majority of works so far consider and represent the interactions between network components as pairwise. It is known that this is not always the real case as interactions themselves can involve more than two nodes, having substantial consequences on the dynamical processes affecting the different systems. The aforementioned raises the need of a new language able to describe and take into account these so-called higher order interactions, as well as developing efficient methods for their detection since the vast majority of datasets, in particular those displaying social interactions, are formulated with dyadic interactions. During this talk we will first introduce some of the recent methods implemented to recover higher order interactions from empirical data. Then, we will describe characteristics of higher order systems in social dynamics and focus on the implications for epidemic processes.
