ATWL: A Formal Language for Visual Analytics Workflows
Researchers have introduced the Artifact-Transform Workflow Language (ATWL), a domain-agnostic declarative language designed to formally represent visual analytics (VA) workflows. ATWL captures both structure and analytical intent through a modular ontology of eight artifact types—entities, features, arrangements, visualisations, patterns, models, knowledge, and specifications—and transforms characterized by standardized intents such as define-unit, characterise, contextualise, and abstract. The language aims to address the complexity of VA workflows, which are typically described in unstructured prose, hindering systematic comparison, reuse, and training. To facilitate adoption, the authors propose extracting workflows from research papers via supervised interaction with LLM agents, reducing human involvement to review and refinement. The work is detailed in a preprint on arXiv (2605.25489).
Key facts
- ATWL is a domain-agnostic declarative language for representing visual analytics workflows.
- It is built on a modular ontology of eight artifact types.
- Transforms are characterized by standardized intents like define-unit, characterise, contextualise, and abstract.
- The language aims to enable systematic comparison, reuse, and training.
- Workflows are extracted from research papers using supervised LLM agents.
- Human role is reduced to review and refinement.
- The preprint is available on arXiv with ID 2605.25489.
- The work addresses the problem of unstructured prose descriptions in VA.
Entities
Institutions
- arXiv