Decoupling Data and Tooling in Interactive Visualization

TL;DR

This paper proposes a modular architecture for interactive visualization tools that separates data handling from visualization, reducing redundancy and improving flexibility in data analysis workflows.

Contribution

It introduces a decoupled, modular approach to data and visualization integration, enabling interoperability and reducing development overhead.

Findings

Prototype demonstrates feasibility of modular data-visualization architecture.

Separation of data handling improves flexibility and reduces redundancy.

Future directions include integration with IDEs and notebooks.

Abstract

Interactive data visualization is a major part of modern exploratory data analysis, with web-based technologies enabling a rich ecosystem of both specialized and general tools. However, current visualization tools often lack support for transformation or wrangling of data and are forced to re-implement their own solutions to load and ingest data. This redundancy creates substantial development overhead for tool creators, steeper learning curves for users who must master different data handling interfaces across tools and a degraded user experience as data handling is usually seen as an after-thought. We propose a modular approach that separates data wrangling and loading capabilities from visualization components. This architecture allows visualization tools to concentrate on their core strengths while providing the opportunity to develop a unified, powerful interface for data handling. An additional benefit of this approach is that it allows for multiple tools to exist and be used side by side. We demonstrate the feasibility of this approach by building an early prototype using web technologies to encapsulate visualization tools and manage data flow between them. We discuss future research directions, including downstream integrations with other tooling, such as IDEs, literate programming notebooks and applications, as well as incorporation of new technologies for efficient data transformations. We seek input from the community to better understand the requirements towards this approach.