Authors: Hrim Mehta, Amira Chalbi, Fanny Chevalier, Christopher Collins
Abstract: Visual storytelling is commonly employed to communicate results of data analyses. Alternatively, automated [1,6] and semi-automated  data narratives or “tours” have been proposed as a means to support exploration of massive multidimensional datasets, substituting the more prevalent static overviews to prompt exploration. While these works demonstrate specific instances of data tours, a concrete model to describe the building blocks of such tours is lacking. We present a hierarchical framework, DataTours, to formalize and guide the design of automated and semi-automated tours for data exploration. We also discuss challenges evoked by the framework in the semi-automated and automated authoring of the data tours.