Visualising category recoding and redistributions

Abstract

This paper proposes graphical representations of data and rationale provenance in workflows that convert data between distinct but semantically related nomenclature. We motivate the graphical representations with a new task abstraction, the Nomenclature Transformation, and associated graph-based information structure, the Crossmap. The structure is illustrated using an example of converting numeric statistics from a country-specific nomenclature to an international classification standard, a common task in ex-post data harmonisation. Crossmaps support the separation of domain expertise and categorical mapping decisions from the specifics of data manipulation. We discuss the opportunities and challenges of using visualisation to audit and communicate nomenclature transformations and present candidate graphical representations.

 

Workshop Details

Workshop on Visualization and Provenance Across Domains at IEEE VIS on Oct 22, 2023

• https://visxprov.github.io/2023/index.html
• https://content.ieeevis.org/year/2023/session_w15.html
• Accepted Papers: https://visxprov.github.io/2023/submission.html

Slides

• HTML
• SOURCE FILES