M. Montzieux, S. Nguyen Ba and Madeleine Mutel show how they classed interview fragments into a matrix, and how they then extracted outputs from this (in this case, a typology of actors).