ResearchBenchmarksChart Generation on PewResearch and CharXiv Human Study Subset (test)Follow94Type CorrectnessChartDesign89.391.659496.35Apr 6, 2026Evaluation ResultsMethodMethodLinksType CorrectnessOrientation CorrectnessLayout Fidelity (Likert 1-5)Visual Plausibility (Likert 1-5)Preference: OriginalPreference: ChartDesignPreference: NoneChartDesign2026.0494904.44.6256510