Back to Study Material Artificial Intelligence

Ontologies in AI

An ontology in AI is a formal, explicit specification of a shared conceptualisation of a domain. The term comes from philosophy, where ontology is the stud...

Ontologies in AI

Formal Vocabularies for Knowledge

An ontology in AI is a formal, explicit specification of a shared conceptualisation of a domain. The term comes from philosophy, where ontology is the study of what exists, but in AI it refers specifically to a structured representation of the concepts, relationships, and rules that govern a particular field of knowledge. Ontologies serve as a common language that allows different software systems, databases, and AI agents to share and interpret information consistently.

An ontology typically defines classes, which are categories of things such as Person, Disease, or Drug. It defines properties, which describe attributes of instances such as hasName, hasAge, or causesSymptom. It defines relationships between classes such as isSubclassOf, which builds a hierarchy, and defines axioms, which are logical constraints that must hold, such as every Disease having at least one Symptom.

The Web Ontology Language, or OWL, is the standard language for publishing ontologies on the web. It is built on description logics and supports automated reasoning: an OWL reasoner can infer implied facts not explicitly stated. For instance, if the ontology states that Paracetamol is an Analgesic and every Analgesic reduces pain, a reasoner automatically infers that Paracetamol reduces pain. The Gene Ontology project is a major real-world example, providing a shared vocabulary for annotating gene products across biological databases worldwide.

Ontologies vs Databases

Databases store instances with a predefined schema but offer limited support for inferring new facts or sharing schema across systems. Ontologies are designed for inference and interoperability. A database might store that aspirin is a drug, but an ontology with appropriate axioms allows a reasoner to derive that aspirin is also an analgesic, an antipyretic, and potentially contraindicated for certain patients.

Exam Tip

Questions on ontologies often ask you to distinguish between a taxonomy, a thesaurus, and an ontology. A taxonomy is a simple hierarchy of classes. A thesaurus adds synonyms and related terms. An ontology adds formal axioms and support for automated reasoning. Placing each on this spectrum of expressiveness and explaining why the additional complexity of an ontology is sometimes worth the engineering cost is a strong exam answer.