GraphDance: A Semantics-Driven Framework for Authoring Narrative Data Video in Graph Structure

doi:10.1007/978-981-95-5764-6_25

Return

GraphDance: A Semantics-Driven Framework for Authoring Narrative Data Video in Graph Structure

2026-01-01

PRE

DOI：10.1007/978-981-95-5764-6_25

Original

Original request for help

Save

Abstract

En 中文

Narrative Data Videos (NDVs) are gaining traction as powerful information mediums by combining storytelling and visualization. Among the core elements of NDVs, graphs visualize abstract relational data, rendering complex relational narratives perceptible and comprehensible. However, existing NDV tools rarely focus on graphs. Moreover, their complex, manual authoring process poses a significant challenge for non-expert users, especially when dealing with unstructured narrative text. To address this gap, we propose GraphDance, a human-AI collaborative framework designed for the semi-automated authoring of graph NDVs from raw narrative text. We further develop an interactive system that, via a GUI, supports users in narrative text editing, animation intent annotation, and the fine-tuning of animation parameters. To evaluate GraphDance's effectiveness and practicality, we detail the graph NDV authoring process and conduct a mixed-methods user study. Results indicate that GraphDance enables users to easily and efficiently author graph NDVs, delivering a positive user experience. This work pioneers exploration into narration text-driven graph NDV authoring, providing a human-AI collaborative framework and a user-friendly interactive system. It also offers practical insights for this domain while pointing to future research on graph animation intent understanding, mechanisms, and user experience optimization.

Keywords:

Narrative Visualization

Data Video

Knowledge Graph