Ask the Curator: Demonstrating Expert-Driven RDF Data Curation with HERITRACE

πŸ“… 2026-07-24
πŸ“ˆ Citations: 0
✨ Influential: 0
πŸ“„ PDF
πŸ€– AI Summary
This work proposes HERITRACE, a novel system that integrates domain-expert-driven interactive editing, fully automated change snapshots, and temporal rollback mechanisms into RDF data curation to address semantic ambiguities and errorsβ€”such as duplicate records and metadata conflicts. Built upon SHACL validation rules and YAML-based display configurations, HERITRACE automatically generates intuitive form-based interfaces that enable experts to efficiently and traceably edit RDF triples while comprehensively logging provenance and revision history. Evaluated in an OpenCitations Meta bibliographic deduplication scenario, the system successfully assisted experts in identifying DOI conflicts, correcting erroneous merges, and reverting to prior states, thereby demonstrating its effectiveness and practicality in supporting auditable, recoverable semantic data governance.
πŸ“ Abstract
HERITRACE is an open-source Web application for curating RDF data held in triplestores, with provenance and change tracking recorded in RDF as well. It is agnostic to the data model: a technician configures it for a collection by writing SHACL shapes and YAML display rules, after which domain experts edit the data through generated forms, and every change becomes a provenance snapshot that can be inspected and restored. This paper traces the path from configuration to curation on a documented case of bibliographic ambiguity: two unrelated articles that PubMed records under the same DOI. We show how a technician prepares a small OpenCitations Meta subset for curation and how a curator acts on duplicate suggestions to merge two articles, inadvertently confirms a further merge despite conflicting metadata, reverts the mistake through the Time Machine, and corrects the DOI after verification against Crossref.
Problem

Research questions and friction points this paper is trying to address.

RDF data curation
bibliographic ambiguity
DOI disambiguation
data provenance
triplestore
Innovation

Methods, ideas, or system contributions that make the work stand out.

RDF curation
provenance tracking
SHACL
expert-driven editing
Time Machine
πŸ”Ž Similar Papers
No similar papers found.