The repository provides supplementary materials for ACM-SIGCHI 2021 submission: Learning to Automate Chart Layout Configurations Using Crowdsourced Paired Comparison.

Our code is run in the Python 3 environment. In particular, some rely on Jupiter Notebook. You might need to use python3 or pip3 depending on your configurations.

cd LQ2
pip3 install -r requirements.txt

• src: source code for the ranking network and the baseline approaches
• dataset: the dataset used in MTurk studies
  • You may directly download the entire vega-lite json specifications, the corresponding images, and the graphical features from the baseline [Google Drive].
• mturk: code for generating MTurk charts and analyzing results
• user-study: evaluate the application (compared with Human, Default, and Random)