Course outline for Semantic Web and Ontology Engineering

Module 1: Encode shared identities and meanings

• RDF: Encode subject, predicate, and object triples with stable URIs, distinguishing shared identifiers from labels and dataset formatting.
• RDFS: Define a class and property vocabulary, recognizing that domain and range declarations support type inference rather than record rejection.
• OWL: Express bounded ontology axioms and inspect selected entailments without treating inferred types as evidence of closed-world data validity.

Practical outcome: Encode URI-identified facts and a small class/property vocabulary while distinguishing declarations from instance data.

Module 2: Query and inspect a bounded ontology

• SPARQL: Write reproducible relationship queries over supplied triples and compare fixed query results across two ontology versions.
• Ontology modelling: Separate schema declarations from instance facts and choose explicit identity and relationship meanings for the role-and-skills domain.
• Description logic: Express a supplied small class axiom within the selected reasoner's supported fragment and explain its inferential assumptions.

Practical outcome: Run reproducible relationship queries and express a supplied small class axiom within the selected reasoner's supported fragment.

Module 3: Separate inference from validation

• Open-world assumption: Treat absent assertions as unresolved information rather than falsehood, separating missing required data from ontology entailment.
• Reasoning: Compare asserted and entailed facts, document a supported type inference, and inspect the reasoner's response to an inconsistent fixture.
• Knowledge integration: Combine two RDF datasets using shared identities and meanings, recording versions and unresolved information rather than assuming semantic agreement.
• Closed-world validation: Apply supplied SHACL shapes to required values separately from OWL reasoning, reporting validation failures without equating absence with falsehood.

Practical outcome: Integrate two datasets, compare asserted and entailed facts, and run supplied SHACL shapes separately from OWL reasoning.

Practical exercise

• Produce an RDF ontology, SPARQL query set, version manifest, and separate reasoner and validation reports for two synthetic datasets.
• Show one expected type inference and explain why an OWL domain/range declaration can infer a type rather than reject a record.
• Introduce a missing required value and an inconsistent fixture; identify the validation failure and the selected reasoner's reported result without treating absence as false.
• Compare two ontology versions against three fixed queries and record changed entailments, validation results, and unresolved information.