DG Comics: Semi-Automatically Authoring Graph Comics for Dynamic Graphs

TL;DR

DG Comics introduces a semi-automatic authoring tool that converts dynamic graphs into narrative graph comics via a causality-preserving hierarchical clustering. The system provides multiple views (Summary, Graph Comic, Timeline, Community) and rich editing tools, enabling analysts to balance automation with editorial control. A causality-aware clustering backbone together with a GED-based similarity measure underpins automatic panel generation, while user studies and expert feedback validate usefulness and reveal paths for scalability and data-preprocessing enhancements. The work demonstrates the practicality of data-driven storytelling for dynamic networks and outlines concrete future work to broaden applicability across domains and improve usability.

Abstract

Comics are an effective method for sequential data-driven storytelling, especially for dynamic graphs -- graphs whose vertices and edges change over time. However, manually creating such comics is currently time-consuming, complex, and error-prone. In this paper, we propose DG Comics, a novel comic authoring tool for dynamic graphs that allows users to semi-automatically build and annotate comics. The tool uses a newly developed hierarchical clustering algorithm to segment consecutive snapshots of dynamic graphs while preserving their chronological order. It also presents rich information on both individuals and communities extracted from dynamic graphs in multiple views, where users can explore dynamic graphs and choose what to tell in comics. For evaluation, we provide an example and report the results of a user study and an expert review.

Figures (7)

• Figure 1: Comparison of graph distance metrics for two graphs.$\mathcal{G}_t$ and $\mathcal{G}_{t+1}$ with different labels and attributes with (a) Weisfeiler-Lehman kernel-based distance, (b) deep learning-based similarity (or distance), (c) graph edit distance with MCS or unweighted set similarity, and (d) graph edit distance with weighted Jaccard similarity (our choice). The thickness of the line is proportional to the node or link attributes.
• Figure 2: DG Comics overview. DG Comics offers (A) a Summary View, (B) sliders for filtering and highlighting nodes, (C) a Graph Comic View, (D) Main Character and (E) Supporting Character tables, and (F) a Timeline View. Users can switch to (G) the Node Attribute Table or (H) Community View using the tab. It supports (M) mental map preservation by fixing nodes across displays, and (O) community changes using bubble sets.
• Figure 3: Data abstraction and presentation. Each cluster below the depth slider (left) represents a panel (right) through a set operation of two subcluster snapshots. The aggregate graph snapshot is computed by the union of graphs for timespans T1 and T2.
• Figure 4: Node Attribute Table. This table shows a list of all the nodes and their values over time.
• Figure 5: Community View changes when communities are selected, as shown in the left inset, where red and blue colors mean Keim and Schreck, respectively, whose communities diverged since 2014.