Face-to-face interactions in human gatherings have a significant impact in various contexts, including disease spreading and opinion dynamics. In this thesis, we investigate the temporal properties of group interactions in different settings using data recorded using the SocioPatterns platform. Our analysis focuses on higher-order structures, revealing that the distributions of group durations exhibit large tails, indicating the absence of a typical time scale for higher-order interactions in human gatherings. By examining the accompanying metadata associated with the contact data, we explore the role of homophily, which refers to the tendency of individuals to interact with others with whom share similar attributes, in face-to-face interactions.