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

Joohee Kim, Hyunwook Lee, Duc M. Nguyen, Minjeong Shin, Bum Chul Kwon, Sungahn Ko, Niklas Elmqvist

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.

Introduction: The Challenge of Dynamic Graph Storytelling

Dynamic graphs—networks that evolve over time—are ubiquitous, ranging from geopolitical alliances to academic collaboration networks. While these graphs are rich with information, communicating their "story" is notoriously difficult. Static visualizations often result in cluttered "hairballs," and animations can be cognitively taxing to follow.

An example of a graph comic narrating the events of World War II, generated using DG Comics.

Graph comics offer a compelling alternative, using sequential panels to narrate network evolution. However, the manual creation of these comics presents a significant "authoring bottleneck." Storytellers must decide how to segment time, identify "main characters" for specific eras, and write explanatory captions—tasks that are time-consuming and error-prone for large datasets. DG Comics addresses this by providing a semi-automatic authoring tool that bridges the gap between automated data analysis and human narrative agency.

Methodology: From Data to Narrative

The core of DG Comics is a workflow that simplifies the transition from raw temporal snapshots to a structured comic layout.

Causality-Preserving Temporal Segmentation

Unlike standard clustering algorithms that might group any two similar snapshots regardless of their timing, DG Comics utilizes a causality-preserving hierarchical clustering algorithm. This algorithm only merges snapshots that are adjacent in time, ensuring that the resulting dendrogram respects the chronological flow of the data. This allows users to use a "depth slider" to automatically generate a set of comic panels based on significant temporal shifts.

The causality-preserving hierarchical clustering algorithm allows users to segment temporal data and visualize changes (added, deleted, or maintained) between snapshots.

Enhanced Distance Metric

To accurately detect changes, the tool employs a metric combining Graph Edit Distance (GED) with Weighted Jaccard Similarity. This approach captures not only topological changes (node/edge additions or deletions) but also shifts in node or edge attributes, such as changes in trade volume or centrality.

Semi-Automatic Authoring Interface

The tool provides several features to assist the storyteller:

  • Character Identification: Automatically identifies "Main" and "Supporting" characters based on centrality metrics.
  • Automated Captioning: Generates template-based captions that can be further refined using an integrated Large Language Model (Mixtral 8x7B).
  • Mental Map Preservation: Fixes node positions across panels to help readers track entities without losing their spatial context.

Evaluation and Results

The authors evaluated DG Comics using three diverse datasets: World War II history, IEEE Visualization publication data (Vispubdata), and international trade data (COW).

A user study with 13 participants yielded a System Usability Scale (SUS) score of 67.88, indicating "acceptable" to "good" usability for a complex research prototype. Participants successfully created cohesive historical narratives from decades of trade data in approximately one hour—a task that would typically take days using traditional design software.

Expert reviews from specialists in history, communication, and data science further validated the tool. Experts praised its ability to "extract outstanding stories" and noted its potential applications in data journalism and supply chain management.

Conclusion and Future Work

DG Comics demonstrates that semi-automation can significantly lower the barrier to creating data-driven graph comics. By providing a "best guess" template that users can refine, the tool preserves storytelling agency while automating tedious data processing tasks. Future work will focus on improving the learning curve for the feature-rich interface and exploring non-linear temporal granularities within a single comic layout.

Summary updated

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.
BibTeX
@article{kim2025dg,
  author = {Joohee Kim and Hyunwook Lee and Duc M. Nguyen and Minjeong Shin and Bum Chul Kwon and Sungahn Ko and Niklas Elmqvist},
  title = {{DG Comics: Semi-Automatically Authoring Graph Comics for Dynamic Graphs}},
  journal = {IEEE Trans. Vis. Comput. Graph.},
  year = {2025},
  doi = {10.1109/tvcg.2024.3456340},
}

Loading related content...

Figure