datathus

Datathus is a conceptual framework in data science and information design that aims to make data-derived insights more traceable, interpretable, and communicable. It envisions combining data provenance, schema metadata, and narrative explanations into a coherent representation of how conclusions are reached from data.

Origin and use: The term datathus emerged in speculative discussions in the mid-2010s and has appeared in a variety of nonstandard presentations, papers, and blogs. It is not a formal standard but a loose family of approaches that seek to bridge data engineering and storytelling.

Core concepts: Key elements include data provenance graphs mapping sources and transformations; declarative rules or query traces that annotate outputs with justification; structured narratives or storylines that explain results in business terms; and visualizations designed to convey both data lineage and reasoning to diverse audiences.

Applications: Datathus-inspired methods are discussed in data governance, compliance, explainable AI, data journalism, and education, where stakeholders require both technical detail and accessible explanations.

Limitations and reception: As a nonstandard concept, datathus faces issues of interoperability and tool support; maintaining detailed provenance can be costly; there is a risk of confusing users if narratives oversimplify complex processes. Proponents view it as a complement to existing provenance, explainability, and data storytelling practices.