Knowledge Representation and Logical Reasoning
Representing knowledge
From facts to structures
Knowledge representation formalizes entities, relations, and rules so machines can infer new facts. Common formalisms include propositional logic, first-order logic, description logics, semantic networks, frames, and ontologies.
Logic vocabulary
Which formalism is most expressive for quantifying over objects and relations?
First-order logic adds quantification and predicates, making it much more expressive than propositional logic.
Correct answer: First-order logic
What is entailment in logic?
Entailment is semantic consequence, not merely syntactic derivation.
Correct answer: A conclusion is entailed if it must be true whenever the premises are true.
Inference methods
Automated reasoning includes forward chaining, backward chaining, resolution, satisfiability solving, and theorem proving. The choice depends on expressiveness, completeness, and computational tractability.
Reasoning styles
Deductive reasoning
- Derives logically certain conclusions
- Useful for verification and rule-based systems
Defeasible reasoning
- Allows exceptions and uncertainty
- Useful for commonsense and legal reasoning
Expressiveness vs tractability
More expressive logics often become computationally expensive or undecidable, so practical AI systems trade completeness for scalability.
Resolution is most closely associated with which task?
Resolution is a fundamental inference rule in theorem proving for logical formulas in clausal form.
Correct answer: Automated theorem proving
Why are ontologies useful in AI systems?
Ontologies support semantic interoperability, knowledge reuse, and inference.
Correct answer: They provide a shared structured vocabulary and relations for interoperable reasoning.