RAGE-Vis lets users edit charts via natural language with AI

RAGE-Vis is a relation-aware generative editing interface for natural language-based chart editing. It accepts bitmap chart images as input and converts them into an editable parameterized intermediate representation. Instead of mapping instructions to a single operation, RAGE-Vis parses composite intents, identifies targets and scopes, and generates hierarchical editing panels for underspecified requests. It also identifies potentially affected fields based on visual encoding relations, structural relationships, and expressive consistency relations, organizing them into actionable widgets for cross-component coordinated controls. Two case studies demonstrate its applicability to complex editing tasks like style adjustment, data extension, order rearrangement, legend layout, and color mapping. A user study shows that participants can effectively handle underspecified requests, explore candidate alternatives, and maintain cross-component consistency with RAGE-Vis.