5DV248 · Trustworthy Human-AI Interaction
Explore how knowledge can be represented, structured and queried using different formal languages. Start by building ontologies, then move to Description Logics, First-Order Logic and Semantic Web rules. In each laboratory, tell the knowledge base what is known, ask questions, and inspect the knowledge that can be derived.
Follow the sequence below. Begin by exploring how an ontology organises classes, relationships and individuals. Then compare three ways of expressing knowledge: Description Logics, First-Order Logic and SWRL. In the Ask/Tell laboratories, try to predict the answer before submitting a query, and distinguish facts explicitly stored in the knowledge base from conclusions derived by the supported inference mechanism.
Five activities, progressing from visual modelling to formal knowledge languages.
Build a conceptual model of a fictional district-heating plant. Create classes and subclass relationships, define relationships between classes, and add individuals that belong to those classes. An interactive graph visualises how the ontology changes as you extend it.
Use rdflib.js to construct an RDF/OWL representation of a synthetic plant. Define classes, subclass relationships, object properties with domain and range, named individuals and links between individuals; then export the ontology as Turtle.
Represent a small fictional-university knowledge base using concepts, roles and individuals. Add TBox axioms and ABox assertions, then ask whether statements follow from the knowledge base.
Represent facts and general rules about a fictional university using constants, predicates, variables and quantifiers. Add ground facts and Horn rules, then query the consequences using ground and existential conjunctive queries.
Extend a fictional-university OWL ontology with Semantic Web Rule Language rules. Add class and object-property assertions, write rules using variables, apply the rules, and inspect the additional knowledge that can be inferred.