The Gaia DR3 all-sky classifications of many variability types multiplied the opportunities of data exploitation. The large increase in the number of classes and classified sources with respect to Gaia DR2 (by factors of 6 and more than 30, respectively) was achieved by supervised classification techniques applied to general multi-class and specialized binary classifiers, making it possible to include both rare and common classes, which returned from a few hundred up to millions of candidates, respectively.
These classifications represent candidates, which aimed more at completeness than purity. A subset of these classes was further processed by class-specific SOS work packages (described in the following sections, except for extractor- and SVD-dependent work packages that are reported in Sections 10.9, 10.10, 10.13, and 10.12). This Section highlights the main aspects for a general understanding of the classification results and how they were achieved. Full details are presented in Rimoldini et al. (2023).