PaperlayerView exact revision on arXiv

Source-linked AI summary

X-DigCheck: Co-Evolving Application Profiles and Knowledge Graphs, Demonstrated on the RTI Documentation of Rupe Magna

Celian Ringwald, Huseyin Erdogan, Valentina Presutti, Cristiano Putzolu

arXiv:2609.07694v2cs.DBcs.DL

TL;DR

Application profiles must evolve with domain knowledge and produced data to avoid misalignment between modelling assumptions, workflows, and validation requirements. X-DigCheck integrates competency questions, SHACL, and coverage analysis into an iterative profile–graph co-evolution workflow, demonstrated through RupeMagna-RTI. The loop distinguishes transformation artefacts from field-data gaps, and conformance recovered after graph regeneration following a temporary coverage regression.

  • Problem

    Application profiles evolve with domain knowledge and produced data, so treating them as static extensions risks misalignment between modelling assumptions, workflows, and validation requirements.

    Source: §1 · p. 2

  • Method

    X-DigCheck combines competency questions, SHACL validation, and profile–graph coverage analysis to guide revisions as RDF is generated against an evolving profile.

    Source: §1 · p. 2, §2 · p. 4

  • Results

    Iterative validation distinguished transformation artefacts from field-data gaps, and conformance recovered after the graph was regenerated following a temporary coverage regression.

    Source: §5 · p. 6

  • Takeaways & Limitations

    The domain-agnostic environment supports profile–graph co-evolution, and its validation produces a reusable RTI semantic profile.

    Source: §5 · p. 7

    The work is demonstrated on a single RTI use case; extending the profile to acquisition quality assessment, archaeological interpretation, and literature links remains future work.

    Source: §5 · p. 7

Abstract

from arXiv · show

We demonstrate X-DigCheck, a domain-independent environment for building and maintaining application profiles as they co-evolve with the data they describe. Profiles developed against a fixed ontology quickly drift from the schema they were meant to capture. X-DigCheck treats profile construction as a continuous ontology-data co-evolution loop: data are lifted into RDF against the profile, checked through competency questions and SHACL, and the resulting reports jointly drive revisions of the ontology, mappings, constraints, and graph. The loop is agnostic to the domain and to the pipeline that produces the graph. We validate and demonstrate the tool in the cultural heritage domain, on the construction of RupeMagna-RTI, the first Reflectance Transformation Imaging (RTI) specialisation of the Cultural Heritage Survey ODP (CHS-ODP), aligned with CIDOC-CRM/CRMdig, ArCo, CHAD-KG, and Getty AAT, with semRTI as the lifting pipeline of this use case. The demonstration lets visitors run one full turn of the loop -on the shipped Rupe Magna (Grosio, Italy) RTI survey, or on a profile and graph of their own -executing the competency-question and SHACL checks live and reading the bidirectional coverage report that flags modelling gaps and stale assumptions. The result is a portable co-evolution environment for profile engineering, together with a reusable RTI application profile produced through it. A screencast of the demonstration is available at https://zenodo.org/records/22210609.

1 Introduction

Application profiles specialise shared ontologies for particular communities, but they must evolve with domain knowledge and produced data to avoid misalignment. X-DigCheck makes that co-evolution observable and actionable across profile, shapes, and data layers.

1 Introduction

  • Application profiles select community-relevant terms, cardinalities, and value spaces from a domain ontology, with shapes validating instance data against the profile.The paper specialises the Cultural Heritage Survey Ontology Design Pattern (CHS-ODP) as a shared conceptual foundation.Source: §1 · p. 2
  • Profiles evolve with domain knowledge and produced data; treating them as static extensions risks misalignment between modelling assumptions, workflows, and validation requirements.The proposed response is iterative modelling, generation, validation, and revision of profiles, mappings, constraints, and graph instances.Source: §1 · p. 2
  • X-DigCheck assesses a profile against its generated graph through competency questions, SHACL, and bidirectional coverage analysis, turning reports into revisions of the ontology, mappings, constraints, or data.The loop is domain- and graph-generation-pipeline-independent.Source: §1 · p. 2
  • The paper demonstrates the approach by engineering RupeMagna-RTI, the first RTI specialisation of CHS-ODP, with semRTI instantiating the graph-generation side of the loop.The workflow lifts acquisition outputs and field annotations into RDF, then uses alignment, competency-question, and SHACL reports to guide revisions.Source: §1 · p. 2, Figure 1 · p. 3
  • Unlike tools focused on deriving shapes or assessing a single ontology or dataset layer, X-DigCheck makes alignment across profile, shapes, and generated data observable over successive regenerations.Astrea and OWL2SHACL derive SHACL constraints; ABSTAT and OOPS! focus on ontology quality and dataset profiling.Source: §1 · p. 2
  • The user-oriented interface extends TestALod [5], previously focused exclusively on ontology validation through competency questions, to support validation and evolution of profiles, mappings, constraints, and generated data.It could also support CLI automation.Source: §1 · p. 3

2 The Co-Evolution Workflow

The workflow couples RDF generation against an evolving profile with graph validation and revision. semRTI supplies the demonstrated RTI lifting pipeline, while X-DigCheck provides reusable checks and reports for iterating on the profile, mappings, constraints, and source data.

The Co-Evolution Workflow

  • The workflow treats changes to the profile, lifting rules, validation requirements, and source annotations as progressive inputs to generation, evaluation, and revision.X-DigCheck is domain-independent; any pipeline that generates RDF against an evolving profile can replace semRTI.Source: §2 · p. 3

semRTI: acquisition and semantic lifting in the use case

  • semRTI lifts RTI products, metadata, and manual annotations into RDF through profile-based CONSTRUCT mappings, retaining traceability to source fields and enabling automatic graph regeneration after changes.The pipeline also uses RelightLab Turtle sidecars and OWL reasoning to materialise profile-prescribed entailments.Source: §2 · pp. 3–4

X-DigCheck: validation and co-evolution tracking

  • X-DigCheck combines profile–graph alignment analysis, competency-question regression tests, and SHACL validation to identify gaps and guide revisions across profile, mappings, constraints, and source data.It runs against local or external SPARQL 1.1 endpoints; SHACL shapes can be OWL2SHACL-derived or manually authored, and reports can be exported as Markdown.Source: §2 · p. 4

3 The Validation Use Case: RupeMagna-RTI and Its Knowledge Graph

The Rupe Magna RTI survey provides the use case for a modular profile and the co-evolution loop. Its stable iteration reports competency-question and SHACL outcomes alongside remaining unused profile terms, while the survey and profile artefacts are shipped for reuse.

Demo video

  • A three-minute screencast shows live competency-question and SHACL runs and the coverage report on the shipped Rupe Magna graph.It is available at https://zenodo.org/records/22210609.Source: §3 · p. 4

Use case and dataset

  • The demonstration artefacts come from ten loop turns on the Rupe Magna RTI survey, whose September 2025 campaign produced 40 sessions covering 11 engraved figures.Each session used 48 images under a custom 48-LED dome; 24 datasets are published, and full graph regeneration yields roughly 148k triples.Source: §3 · p. 5

Application Profile structure

  • The profile evolved from a monolithic model into six modules built on a shared core, with automated checks synchronising independently complete modules.Its digitisation process is modelled as a provenance chain and aligned with CIDOC CRM/CRMdig, ArCo, CHAD-KG, DCAT, OntoPiA, and Getty AAT.Source: §3 · p. 5

What the loop produced

  • The eight inherited CHS-ODP competency questions expanded into 29 profile-specific questions across seven themes, while cardinality constraints moved from OWL restrictions into SHACL.Two shape sets are maintained: one derived from profile axioms through OWL2SHACL and one authored for Rupe Magna requirements.Source: §3 · p. 5
  • 26 of 28 executable competency questions pass (93%), and 32 of 42 shapes conform (76%) at the current stable iteration.Additionally, 8% of profile classes and 3% of properties are never instantiated, with no undeclared terms remaining.Source: Figure 2 · p. 5, §3 · p. 6

4 Demonstration

The demonstration walks visitors through a project’s validation history, live checks, and resulting reports, then shows the same checks on an independently supplied profile and graph. It also demonstrates headless execution and provides reusable software and data artefacts.

Setup and walkthrough

  • The walkthrough shows semRTI and the RupeMagnaViewer transforming survey data into RDF alongside X-DigCheck’s role in developing the specialised RTI profile.Open-source software, a project template, competency questions, the profile, shapes, and RDF dataset are available for reuse.Source: §4 · p. 6

Demonstration

  • Attendees inspect a selected project iteration, run competency questions and SHACL validation, compare profile-derived constraints with contextual shapes, and export a report.They first load the git-tracked RupeMagnaOnto project and inspect statistics from its agile development cycle.Source: §4 · p. 6
  • A sandbox profile and graph from another use case or an attendee’s own data demonstrate that the checks generalise beyond RupeMagna-RTI.The demonstration also runs every UI validation headlessly through the xdigcheck CLI, including ontology coverage and model-drift inspection.Source: §4 · p. 6

5 Discussion and Conclusion

The RupeMagna-RTI validation shows ontology engineering and graph construction evolving together: iterations exposed transformation artefacts and field-data gaps, while graph regeneration restored conformance. X-DigCheck provides a reusable co-evolution tool and RTI profile, demonstrated in one use case, with broader profiles and capabilities left for future work.

5 Discussion and Conclusion

  • Iterative validation exposed incorrect mappings, missing properties, absent inverses, and field-data gaps, distinguishing transformation artefacts from gaps in the source data.A coverage regression after the digitisation-module rewrite reflected lag between an enriched TBox and its ABox, not a modelling failure; conformance recovered after graph regeneration.Source: §5 · p. 6
  • X-DigCheck integrates competency questions, SHACL validation, and ontology–knowledge graph coverage analysis throughout profile development.The work demonstrates it through RupeMagna-RTI, the first RTI application profile derived from CHS-ODP, using semRTI as the lifting pipeline.Source: §5 · pp. 6–7
  • The work establishes a foundation for deriving reusable domain-specific profiles from shared conceptual cores, but is demonstrated on a single RTI use case.Future use-case work would extend the RTI profile to acquisition quality assessment, archaeological interpretation, and links to archaeological literature.Source: §5 · p. 7
  • Future X-DigCheck work targets Git-based tracking of coverage, SHACL, and competency-question results, possible reporting of mapping-file errors, and agentic LLM assistance.The proposed LLM assistance would address mapping repair, shape generation, and query maintenance while preserving executable validation criteria.Source: §5 · p. 7

Declaration of use of Generative AI

Generative AI supported language editing and parts of the engineering process. The authors reviewed its output and retained it only after validation and competency-question testing.

Declaration of use of Generative AI

Loading 2609.07694v2…