Keywords: Knowledge Graph Construction (KGC), RDF Mapping Language (RML), Error reporting
Abstract: The formalization of the RDF Mapping Language (RML) has led the KGC community to propose challenge tracks focused primarily on compliance coverage and performance, where compliance coverage was limited to "pass/fail", with test batteries.
The formalization of RML benefited from community building and gathered insights from these tracks.
While this drove the development of RML and its implementation, the focus on compliance and performance led to Developer Experience lagging behind.
Current RML processors function as opaque black boxes, where the development lifecycle is reduced to a "Write-Run-Fail" loop, which is known to be frustrating and creates a barrier for uptake.
Such barriers hinder the adoption of RML by non-experts.
The issue is that there is no meaningful representation of the various errors for authoring and executing RML, which could aid agents (both human and computer-based, such as RML authors and RML IDEs), in engaging in meaningful ways with RML (processors).
To bridge this gap, we propose the RML Execution Report (RER), a formalised, machine-readable artefact designed to facilitate such exchange by making issues transparent, structured, and meaningful.
By interlacing execution statistics, plans, and a structured error taxonomy, we aim to transform the opaque `exit code 1` used in many processors into actionable feedback.
We present RER to the community not as a finalised artefact, but as a foundation for discussion on improving the RML ecosystem.
Submission Number: 2
Loading