Authors: Ali Sarvghad, Melanie Tory
Abstract: When analysts work in a distributed fashion, they need to understand what their collaborators have done and what avenues of analysis remain uninvestigated. Although visualization history has the potential to communicate such information, the common representations are often limited to sequential lists of past work. Such representations do not make it easy to understand the analytic coverage of the dimension space (i.e. which dimensions have been investigated and which have not). This makes it difficult for an analyst to plan their next steps, particularly when the number of dimensions is large. In this paper, we propose representing the prior analysis from a dimension coverage perspective. Dimension view provides a unique perspective that can facilitate exploratory analysis by enabling analysts to easily identify what dimensions have been examined and in what combinations. We hypothesize that addition of this view to common representations of visualization history will reduce cognitive and interaction costs by helping the analyst to discover data subsets to explore. We studied the effects of this view on a distributed collaborative visualization process. Our findings show that providing views of the dimension and data space reduces time required for identifying and investigating unexplored regions and increases the accuracy of this understanding. In addition, providing these views results in a larger coverage of entire dimension space.