The whole is other than the sum of its parts: a framework for classifying and deconstructing composite visualization
Muxing Chen · 2024
As we are able to produce, collect, and store greater quantities and varieties of information, complex data has become increasingly commonplace. This accessibility poses a significant challenge: a single visualization is often insufficient to represent such intricate information. To address this, researchers and designers have devised techniques for visualizing various facets of a complex dataset by combining multiple visual representations in the same visual space, known as composite visualization. Despite a plethora of composite visualizations has emerged within the visualization community, mainstream theories have yet to fully characterize them. Established theories, exemplified by Bertin's visual variables, elucidate how data attributes can be mapped to graphical features, such as position, color, etc., but fail to explain the interplay between visualizations. To overcome this limitation, this thesis formulates a framework that identifies ten composition patterns for combining multiple visualizations to form a cohesive whole, based on spatial and data relations between the involved visualizations. This framework characterizes the interrelationships between visualizations and therefore augments the established theories. Meanwhile, it offers tools for designers and researchers to communicate, classify, and deconstruct composite visualization.--Author's abstract