This poster was presented at Computer Graphics & Visual Computing (CGVC24) at City St George's, University of London.
Maeve Hutchinson, Radu Jianu, Aidan Slingsby, Pranava Madhyastha
In this paper, we present a research plan to capture and analyse the reasoning behind visualisation design choices using litvis notebooks. These notebooks provide a detailed narrative of the visualisation process, combining code, visual outputs, and natural language. Our approach aims to use these notebooks to create a dataset that reflects visualisation design in practice, focussing on the reasoning processes behind design choices. We will evaluate state-of-the-art multimodal models on this dataset and will carefully analyse the capacity of the models to generate reasoning traces. We will then build models that are capable of generating valid reasoning traces for visualisations. This research will result in reliable models for data visualisation that are grounded in realistic visualisation design processes.