DVrel

DVrel, short for Data Visualization Relationship Language, is a proposed declarative framework for encoding relationships among data sources, visual encodings, and user interactions within data visualization systems. It aims to capture how components of a visualization relate to one another—such as data provenance, mappings from data fields to visual channels, and the sequence of user-driven interactions—in a structured, machine-readable form. The goal is to improve interoperability, reproducibility, and explainability of visual analytics across tools and platforms.

DVrel originated from conversations in the visualization and human–computer interaction communities in the early 2020s, as researchers and practitioners sought to reduce ad hoc wiring between data sources and visual components. It is designed as an open, language-agnostic specification with a modular schema, often envisioned to be expressed in JSON or a similar data format so that it can be consumed by multiple visualization engines.

A typical DVrel model defines core entities such as DataSource, Visualization, and Interaction, and core relationships such as uses, mapsTo, triggers, and derives. A DVrel document describes a visualization pipeline as a graph of components with typed relations and optional provenance metadata that record data origins, transformations, and versioning. The specification emphasizes explicit semantics to support validation, recomputation, and reassembly of visualizations.

Adoption of DVrel has been limited to exploratory prototypes and research tools. Advocates argue that DVrel could enable reusable visualization recipes, easier sharing of dashboards, and better auditing of analytic workflows. Critics point to the added complexity and the challenge of aligning diverse tools to a common schema. As of the mid-2020s, a formal standard had not been universally adopted.

See also: Data visualization, Visualization grammar, JSON Schema, Reproducible research.