Ontology and Thesaurus Integration Best Practices with References

Here are the key integration best practices with references:
1. Separate the domain model from the vocabulary Keep ontology classes distinct from thesaurus concepts. The ontology defines the schema; the thesaurus populates it with terms. This is the foundational principle behind using SKOS alongside OWL.

• Allemang & Hendler, Semantic Web for the Working Ontologist (3rd ed.)
• Miles & Bechhofer, SKOS Simple Knowledge Organization System Reference, W3C 2009

2. Use dual-typing judiciously Typing a resource as both skos:Concept and an OWL class instance (as you do with the:maskingFluid a skos:Concept, vad:WatercolorAdditive) is a recognized bridging pattern but should be applied deliberately. It conflates the concept-as-term with the concept-as-class-instance.

• Isaac & Summers, SKOS Simple Knowledge Organization System Primer, W3C 2009
• van Assem et al., Converting and Integrating Vocabularies for the Semantic Web, ISWC 2006

3. Use SKOS mapping properties for cross-vocabulary alignment When bridging independently evolved vocabularies, prefer skos:exactMatch, skos:closeMatch, skos:broadMatch, skos:narrowMatch, and skos:relatedMatch over OWL axioms like owl:equivalentClass. The SKOS mapping properties are lighter weight and don't trigger unintended OWL inferences.

• Miles & Bechhofer, SKOS Reference (above)
• Baker et al., Library Linked Data Incubator Group Final Report, W3C 2011

4. Declare ConceptSchemes explicitly Every skos:inScheme reference should point to a declared skos:ConceptScheme resource. Undeclared schemes break SKOS validation and make navigation unreliable.

• Miles & Bechhofer, SKOS Reference (above)
• Bechhofer & Miles, SKOS Core Guide, W3C 2005

5. Maintain namespace discipline Separate namespaces for separate concerns ? as you do with vad: for the ontology and the: for the thesaurus ? is a best practice. It preserves modularity and makes provenance clear.

• Heath & Bizer, Linked Data: Evolving the Web into a Global Data Space, 2011
• Noy & McGuinness, Ontology Development 101, Stanford 2001

6. Use a separate alignment module for cross-ontology mappings Rather than burdening either ontology with knowledge of the other, declare mappings in a dedicated alignment file that imports both. This preserves the universality of each model.

• Euzenat & Shvaiko, Ontology Matching (2nd ed.), Springer 2013
• The Linked Open Vocabularies project demonstrates this pattern at scale

7. Align with established upper ontologies where possible Grounding your ontology in well-known upper models (SKOS, Dublin Core, FOAF, Schema.org ? all of which you already use) improves interoperability and reduces the need for custom alignment work later.

• Niles & Pease, Towards a Standard Upper Ontology, FOIS 2001
• Your current use of dc:, foaf:, schema:, and skos: already reflects this practice well

8. Document design decisions in the ontology itself Use rdfs:comment, skos:scopeNote, and skos:editorialNote to record rationale inline. Your use of # MERGE NOTE: comments in CWVA is good practice ? formalizing some of those as rdfs:comment annotations would make them machine-readable.

• Rector et al., OWL Pizzas: Practical Experience of Teaching OWL-DL, EKAW 2004
• Allemang & Hendler (above)

9. Validate regularly against SKOS and OWL profiles Tools like the SKOS Play validator, OOPS! ontology pitfall scanner, and Apache Jena's riot can catch issues like undeclared schemes, dangling references, and property typos before they propagate.
Your system already reflects most of these practices ? the gaps noted in the review (undeclared pigment scheme, the skos:altfLabel typo, the DrawingForm naming inconsistency) are precisely the kind of thing routine validation catches early.