Rules

Conditional knowledge processing and inference

Rules generalize static graphs: premises must be satisfied before a conclusion is asserted or an action runs. Unlike entailment from a single chain of object properties, rules can combine arbitrary conditions, negation (where allowed), and priorities depending on the engine.

An inference engine can repeatedly evaluate rules until a fixpoint—this is the classical expert-system pattern echoed in [AlHe2020] and in deriver.app’s IF/THEN layer over session state.

Keep rule conditions aligned with the conceptual model: naming predicates consistently with classes and OPs avoids silent drift between “schema” and executable logic.