Data Visualization

Data visualization turns data into charts, maps, diagrams, and interactive views so people can find patterns, compare values, and communicate evidence. Research develops visual encodings and analysis tools, including dimensionality-reduction methods, scientific visualization workflows, and accessible alternatives such as sonification. A growing line of work uses language and vision models to generate, edit, explain, and interpret charts, while benchmarks test these systems on chart questions, code generation, dashboards, and multi-step scientific tasks. A central challenge is making outputs faithful and useful: models can produce plausible visuals yet misread data, miss design or perceptual rules, and struggle with complex interactions, so evaluation must consider both machine metrics and human interpretation.