In logic programming, a program describes relations that hold rather than prescribing one fixed sequence of state-changing commands.

A knowledge base can contain facts such as

parent(alice, bob).
parent(bob, carol).

and rules such as

grandparent(X, Z) :- parent(X, Y), parent(Y, Z).

The rule says that grandparent(X,Z) holds when some Y makes both premises true.

A query asks whether a relation can be satisfied and, when variables are present, which substitutions make it true:

grandparent(alice, Who)?

can produce Who = carol.

Variables in rules are logical variables: they stand for unknown terms constrained by relations, not mutable storage locations that are assigned repeatedly. Matching a query against facts and rules uses unification to find compatible substitutions.

Logic programming therefore changes the programmer's role. The program states relations and implications; an execution strategy searches for a proof of the query. This separation between declarative meaning and search procedure makes logic languages useful for symbolic reasoning, rule systems and relational specifications.