Knowledge Representation, Ontology, Reasoning, and Uncertainty

Knowledge representation

Storing knowledge in a form a computer can use for reasoning.

Automated reasoning

Using a computer to answer questions and draw new conclusions.

Ontology

An organization of things into categories and relationships; often arranged in a hierarchy.

General ontology

A broad category hierarchy for representing general things in the world.

Hierarchy of categories

A structure where broad categories contain narrower subcategories.

Semantic networks

Graph-style knowledge representations using nodes and links to show concepts and relationships.

Description logics

Logic-based knowledge representation designed to support efficient/tractable reasoning.

Existential graphs

Diagrammatic logical representations; important historically but not the tractability-focused answer here.

Overriding/default reasoning

Handling exceptions to general rules, such as “birds fly” but “penguins do not.”

Tractability

Whether a problem can be solved efficiently enough in practice.

Belief state

The agent’s representation of which states might be true when there is uncertainty.

Contingency plan

A plan that includes different actions depending on what is observed or what happens.

Unknowns

Information the agent does not have; realistic agents track or reduce uncertainty rather than ignore it.

Deterministic outcome

An outcome that is fixed with certainty from a given state/action.

Risk

Possibility of loss or undesirable outcome; rational systems often manage risk rather than avoid all action.