Does naming an object imply domain closure?

No. Naming some objects does not say that no others exist.

Does closure make every unknown property false?

No. Object completeness and fact completeness are different assumptions.

Can domain closure be revised?

Yes. A changing inventory may require a revised theory, with previously derived universal conclusions reassessed under the new scope.

AI Glossary / Graphs and symbolic AI / Intermediate

Domain closure

Definition

Domain closure is an assumption that every object in a modeled domain is accounted for by the specified names or ground terms. In a finite setting with named objects a, b, and c, it can be expressed as: every object equals a, b, or c. It rules out additional unnamed objects within that scope.

What it means

A logical knowledge base can name some objects without saying that those are all the objects that exist in its domain. Domain closure supplies that missing completeness claim. It allows reasoning that depends on an exhaustive inventory, such as checking a property for every member by examining a complete finite list.
The scope matters. Closing the modeled list of machines in one test does not claim that no other machines exist anywhere. A clearly bounded modeling assumption is useful; an undocumented assumption that a partial list is complete can lead to incorrect universal conclusions.

How it works

For a finite domain named by constants a and b, a closure axiom is: for every x, x = a or x = b. Any admissible model must interpret each domain element as one of those named objects. If the model includes other kinds of entities, closure may instead be restricted to a predicate, such as every Machine being one of the listed machines.
Closure is separate from distinctness. The axiom alone permits a and b to name the same object. An additional statement a != b is needed when they must be different. Closure is also separate from deciding which properties those objects have; an exhaustive object list need not be an exhaustive fact list.

Three practical examples

A scheduling model explicitly closes the set of available rooms to the inventory supplied for a particular event.
A configuration test restricts possible component objects to a fixed set of test fixtures so universal checks have a defined scope.
A logic exercise compares conclusions before and after declaring that the named objects exhaust the domain.

Think of it this way

Think of a guest list labeled these are all the guests, rather than some guests we know about. The completeness label changes what checking every name establishes. The analogy is limited because logical names can refer to the same object unless distinctness is specified, and an exhaustive guest list says nothing by itself about every guest's preferences or actions.

Worked example

Suppose a knowledge base contains Ready(a) and Ready(b). Without domain closure, these facts do not imply that every object is ready. A countermodel can contain a third, unnamed object u for which Ready(u) is false, while both stated facts remain true.
Now add the closure axiom: every x equals a or b. For any object x, it must be one of the named objects, and each is ready. The conclusion that every object is ready now follows.
Do we also know that exactly two objects exist? Not yet. The names a and b could denote one object. Adding a != b, together with closure, establishes exactly two domain elements. Conversely, introducing a new constant c does not automatically add a third object: under the original closure, c must equal a or b. To represent a genuinely third object, revise the closure and add the required distinctness statements. These consequences show why object completeness and unique naming must be tracked independently.

Illustrative case study

At fictional Alder Inspection, a prototype declares all modeled devices checked after processing the two named devices. Reviewers discover that the imported list was only a partial snapshot. The logical conclusion relied on a closure axiom that the available evidence did not justify.
The team restricts the claim to the documented test inventory and makes completeness a required input condition. This fictional case illustrates the importance of modeling scope, without claiming that formal logic independently verifies whether a real inventory is complete.

Common misconceptions and mistakes

Domain closure is not the same as the closed-world assumption. The former restricts which objects exist in the modeled domain; the latter commonly treats unestablished facts as false under a specified reasoning convention. A system can close its object list while leaving many properties unknown.
Nor does closure imply unique names. Several constants can denote one object unless the theory excludes that possibility. In languages with function symbols, ground terms can name additional objects and may form an infinite set, so a simple finite-constants explanation should not be silently generalized to every formalism. State the exact domain and language assumptions used by the application.

Business use

Businesses can use domain closure when a decision depends on a controlled, exhaustive inventory. Record the inventory's scope, version, owner, and completeness conditions alongside the logical model. Revisit the axiom when new objects arrive. A result that holds for all listed assets is useful, but it should not be presented as covering unlisted assets unless completeness is established. Where records combine people, locations, and equipment, prefer clearly typed or predicate-restricted claims over an accidental closure of everything.

Check your understanding

A theory says every object equals a or b, but contains no inequality between them. Must the domain contain exactly two objects?

Answer: No. It may contain one object with two names. To require two distinct objects, add a != b as well as the closure axiom.

Related terms

Learn next

Write a predicate-restricted closure statement saying every Room is RoomA or RoomB, while allowing people and equipment in the broader domain. Explain why it does not imply that every object in the system is one of the two rooms.

Sources

Stanford knowledge-representation course, original sample axioms and domain-closure discussion: https://www-formal.stanford.edu/leora/krcourse/homeworks/hw2sampleaxioms.html Russell and Norvig, First-Order Logic, author-hosted chapter for domain and naming semantics: https://aima.cs.berkeley.edu/4th-ed/pdfs/newchap08.pdf

Previous: Domain adaptation Next: Domain Constraint