Skip to main content

Semi-Automated Visual Analytics

Semi-Automated Visual Analytics

Visualization should lead to insight in data, insight leads to new knowledge, supports reasoning, and aids in decision making. These are fundamental goals of information visualization and visual analytics. Effective visual and immersive analytics design depends on a range of factors including data domain and attribute semantics, visual representation, interaction, user characteristics, tasks and insight type, and device characteristics. While studying these factors in isolation provides valuable insight, defining a holistic visual analytics design space would allow for an overall process optimization. Previous work and especially the seminal grammar of graphics by Wilkinson (2006) provide invaluable guidance towards this goal but still lack a holistic integration and a language describing all elements of modern visual analytics systems.

In this project, we aim at defining a holistic visual analytics design process and relate it to a visual analytics algebra. The purpose of the algebra is to define a formal system with operands as declarative specifications of a visual interactive system and operators for manipulating specifications to achieve a desired goal. With such an algebra we aim to define visual analytics design as an optimization problem to improve task-solving performance and user experience for visual analytics systems.

Investigators

Dr. Ulrich Engelke, Decision Sciences, CSIRO Data61
Dr. Eser Kandogan, Accelerated Discovery Lab, IBM Research

Lifetime

2016 – ongoing