概要
ChinaVis 2026 paper introducing a Generative UI for natural-language chart editing. RAGE-Vis turns bitmap charts into editable parameterized representations, parses composite requests, and generates hierarchical panels with coordinated global and local controls for underspecified editing intents.
要約
Why RAGE-Vis Adds New Signal:
1. Generates the control surface: Creates task-specific editing panels instead of mapping each request to one fixed widget
2. Handles ambiguous intent: Lets users inspect and adjust alternatives when a natural-language request is underspecified
3. Relation-aware coordination: Links controls through visual encoding, structure, and expressive-consistency relationships
4. Works from bitmap input: Reconstructs an editable parameterized chart before applying coordinated changes
5. Evidence-backed interaction: Case studies and a user study cover style, data, order, legend, and color-mapping edits
1. Generates the control surface: Creates task-specific editing panels instead of mapping each request to one fixed widget
2. Handles ambiguous intent: Lets users inspect and adjust alternatives when a natural-language request is underspecified
3. Relation-aware coordination: Links controls through visual encoding, structure, and expressive-consistency relationships
4. Works from bitmap input: Reconstructs an editable parameterized chart before applying coordinated changes
5. Evidence-backed interaction: Case studies and a user study cover style, data, order, legend, and color-mapping edits
タグ
generative-uidata-visualizationnatural-language-editingadaptive-controlshuman-in-the-loop