From Visual Exploration to Storytelling and Back Again

Abstract

The primary goal of visual data exploration tools is to enable the discovery of new insights. To justify and reproduce insights, the discovery process needs to be documented and communicated. A common approach to documenting and presenting findings is to capture visualizations as images or videos. Images, however, are insufficient for telling the story of a visual discovery, as they lack full provenance information and context. Videos are difficult to produce and edit, particularly due to the non-linear nature of the exploratory process. Most importantly, however, neither approach provides the opportunity to return to any point in the exploration in order to review the state of the visualization in detail or to conduct additional analyses. In this paper we present CLUE (Capture, Label, Understand, Explain), a model that tightly integrates data exploration and presentation of discoveries. Based on provenance data captured during the exploration process, users can extract key steps, add annotations, and author “Vistories”, visual stories based on the history of the exploration. These Vistories can be shared for others to view, but also to retrace and extend the original analysis. We discuss how the CLUE approach can be integrated into visualization tools and provide a prototype implementation. Finally, we demonstrate the general applicability of the model in two usage scenarios: a Gapminder-inspired visualization to explore public health data and an example from molecular biology that illustrates how Vistories could be used in scientific journals. (see Figure 1 for visual abstract)

HiPiler: Visual Exploration of Large Genome Interaction Matrices with Interactive Small Multiples

“…2.3), which requires first investigating appropriate methods for aggregating snippets of different sizes without destroying patterns in the data. Also, we want to provide ways to efficiently show a user’s exploration history and to support collaborative scenarios [21] as mentioned by P1, P3, and P4 during the interviews. Eventually, we want to visually summarize and aggregate data attributes of piles that will enable experts to more seamlessly transition from context-driven (interaction matrix) to knowledge-driven (data attributes) pattern exploration.…”

HiGlass: Web-based Visual Exploration and Analysis of Genome Interaction Maps

“…This JSON state representation can either be saved locally or stored in HiGlass's database and shared as a link to an interactive figure (Figures 2,3,4,5). By capturing the current composition and storing its complete state on the server, we create the opportunity to integrate HiGlass with tools for documenting and exploring the provenance of the composition to better understand the steps that the analyst took to reach their conclusions [45].…”

Communicating Genome Architecture: Biovisualization of the Genome, from Data Analysis and Hypothesis Generation to Communication and Learning

“…Saving session states and data collections is found in most tools, but a more detailed recording of the choices made in encoding and interacting with the data is not. Capturing this visualization process and retelling the intricate histories of discovery that result from it remains a significant challenge which is incompletely addressed in current genomic tools [122].…”