Semantic Snapping for Guided Multi-View Visualization Design

Abstract

Fig. 1. Semantic snapping allows the user to perform iterative operations to improve the compactness and consistency of a multi-view visualization. Underlying algebraic rules called relations define the available operations for each iteration. In this example showing the 2016 US Election poll percentages and pollsters, from our initial composition we (1) identify a confuser relation between the bottom left and rightmost views showing the same color (red). We differentiate these views by selecting green as the fill color for the scatter plot. We next identify a multiples relation in the two left views. We resolve this through one of two integration operations. (2*) Overlay produces an unsatisfactory result, so we revert and (2) perform a mirroring operation to arrive at our resulting composition. The semantic map to the right illustrates our path through semantic space.