US Patent 11960553 Unsupervised user segment discovery system

A computing system generates and a plurality of clusters by clustering the session representations based on similarity between session representations, the plurality of clusters comprising an optimal number of clusters based on key performance indicators corresponding to each cluster of the plurality of clusters. The computing system determines a medoid of each of the plurality of clusters, and then for each session representation generated for each of a second plurality of sessions, the computing system computes a distance between the session representation and each medoid to determine a nearest medoid and assigns the session representation to a cluster associated with the nearest medoid.