Feature-Action Design Patterns for Storytelling Visualizations with Time Series Data

TL;DR

The paper addresses delivering data-driven storytelling for dynamic time-series data tailored to individual audiences. It introduces meta-authoring with feature-action design patterns to map anticipated data features to visualization actions, enabling a common meta-storyboard adaptable to unseen data streams. The contributions include a meta-authoring framework, an algorithmic pipeline for feature detection, ranking, and segmentation, and six COVID-19 and three ML-workflow storyboards demonstrating generality across domains. This approach enables scalable, data-dependent storytelling with reduced manual authoring, improving accessibility for public engagement, dashboards, and AI-for-science contexts. The work highlights practical pathways to automate and customize temporal narratives without sacrificing narrative control.

Abstract

We present a method to create storytelling visualization with time series data. Many personal decisions nowadays rely on access to dynamic data regularly, as we have seen during the COVID-19 pandemic. It is thus desirable to construct storytelling visualization for dynamic data that is selected by an individual for a specific context. Because of the need to tell data-dependent stories, predefined storyboards based on known data cannot accommodate dynamic data easily nor scale up to many different individuals and contexts. Motivated initially by the need to communicate time series data during the COVID-19 pandemic, we developed a novel computer-assisted method for meta-authoring of stories, which enables the design of storyboards that include feature-action patterns in anticipation of potential features that may appear in dynamically arrived or selected data. In addition to meta-storyboards involving COVID-19 data, we also present storyboards for telling stories about progress in a machine learning workflow. Our approach is complementary to traditional methods for authoring storytelling visualization, and provides an efficient means to construct data-dependent storyboards for different data-streams of similar contexts.

Figures (9)

• Figure 1: (a) In a typical workflow for creating storytelling visualization, an author defines a storyboard for a known dataset, which is then developed as a web-based visualization, usually for a specific target audience. (b) With our approach, the author creates a meta-storyboard that works with multiple, dynamic, and often not-yet-inspected datasets. The storyboard is converted by a developer, following rules that facilitate the automatic or semi-automatic depiction of user-driven stories, for different target audiences.
• Figure 2: Overview of the proposed meta-authoring process, involving a story meta-author and a developer. The resulting storytelling visualization software can be applied to many similar, dynamic, and often not-yet-inspected data streams.
• Figure 3: Actions in our implementation with their parameters: (a) change the colour of a section of the time series line for highlighting. (b) draw a circle at a datapoint. (c) annotate the graph with a line attached at a particular point. (d) While animating a time series segment, place a text description on the opposite half of the graph. It is possible to only animate a time segment or animating a segment and annotating a point.
• Figure 4: Demonstration of the single-location story. Programmed features are located. When the Play button is pressed the delivery system actions to progresses to the next feature, the appropriate text is concatenated with real data, and placed in a suitable position (if there is space it will be shown to the right of the vertical line, otherwise to the left). (a) The full interface; (b) story start, (c) highest deaths per day, (d) booster vaccination program starts.
• Figure 5: Comparison story demonstration, where (a) depicts the final frame. The story is shown in stages, moving key 'features', and alternating 'actions' between region 1 and 2. The insets (b-f) depict several key event features, which are incrementally shown as the story progresses; (b) a single feature and action about Bradford (region 2); (c) story action focusing on peaks, with data specific to the local site; (d and f) comparison feature showing differences in terms of days; (e) feature comparison based on calculated data.