originswhere

Originswhere is a conceptual framework for information provenance that models the origins and context of data items within a knowledge graph or dataset. It extends traditional provenance by explicitly encoding where a datum originated, when, and how it propagates through transformations and use. The framework emphasizes the spatial and situational context of data as a core element of trust and interpretability.

The term originswhere blends origin and where and was introduced in scholarly discourse in the early 2020s to address the need for location-aware provenance across distributed systems. As a concept, it is not tied to a single product but describes a family of representations and methods for capturing origin-context alongside existing provenance attributes.

Key components include origins (the source events or entities that generate data points), locations (geographic, systems, or operational contexts), timestamps, agents (persons or processes), and processes (transformations affecting data). A provenance trail records the sequence of steps and data flows, supporting queries about the origin path, lineage, and context. The model is designed to be compatible with established standards such as the W3C PROV data model and RDF/JSON-LD representations, facilitating interoperability across tools.

Applications span scientific data management, supply chain traceability, and digital humanities. In research, originswhere supports reproducibility by enabling reconstruction of data from its original sources. In industry and governance, it aids verification and accountability by tracing information through transformations and recipients.

Critics note that adding provenance context can incur storage, processing, and privacy costs, requiring careful consideration of granularity and access controls.

See also: data provenance, provenance graph, W3C PROV, knowledge graph, lineage.