2018-draco.html

---
layout: publication
authors:
  - Dominik Moritz
  - Chenglong Wang
  - Greg L. Nelson
  - Halden Lin
  - Adam M. Smith
  - Bill Howe
  - Jeffrey Heer
awards:
  - Best Paper Award
doi: 10.1109/TVCG.2018.2865240
blog: https://medium.com/@uwdata/64ce20287e9d
link: https://uwdata.github.io/draco/
osf: 3eg9c
pdf: https://www.domoritz.de/papers/2018-Draco-InfoVis.pdf
recording: https://vimeo.com/308980888
short_doi: 10/cs68
slides: https://www.domoritz.de/talks/Draco-Infovis-2018.pdf
tags:
  - Machine Learning
  - Recommendation
  - Visualization
  - Vega-Lite
title:
  "Formalizing Visualization Design Knowledge as Constraints: Actionable and
  Extensible Models in Draco"
tweet:
  "Draco models visualization guidelines as logical constraints. Shareable design
  guidelines, formal reasoning, vis recommendation & more."
type:
  - Conference
  - Journal
venue: InfoVis
venue_location: Berlin, Germany
venue_tags:
  - VIS
  - InfoVis
venue_url: http://ieeevis.org/year/2018/welcome
year: 2018
---

There exists a gap between visualization design guidelines and their application
in visualization tools. While empirical studies can provide design guidance, we
lack a formal framework for representing design knowledge, integrating results
across studies, and applying this knowledge in automated design tools that
promote effective encodings and facilitate visual exploration. We propose
modeling visualization design knowledge as a collection of constraints, in
conjunction with a method to learn weights for soft constraints from
experimental data. Using constraints, we can take theoretical design knowledge
and express it in a concrete, extensible, and testable form: the resulting
models can recommend visualization designs and can easily be augmented with
additional constraints or updated weights. We implement our approach in Draco, a
constraint-based system based on Answer Set Programming (ASP). We demonstrate
how to construct increasingly sophisticated automated visualization design
systems, including systems based on weights learned directly from the results of
graphical perception experiments.