14.12 Thematic Takeaways

• In declarative programming, the programmer specifies what they want to compute, not how to compute it.

• In logic programming, the programmer specifies a knowledge base of known propositions—axioms declared to be true—from which the system infers new propositions using a deductive apparatus:

  

• Propositions in a logic program are purely syntactic, so they have no intrinsic semantics—they can mean whatever the programmer wants them to mean.

• In Prolog, the programmer specifies a knowledge base of facts and rules as a set of Horn clauses—a canonical representation for propositions—and the system uses resolution to determine the validity of goal propositions issued as queries, which are also represented as Horn clauses.

• Unlike Prolog, which uses backward chaining, CLIPS uses forward chaining— there is no concept of a goal in CLIPS.

• There is a mismatch between predicate calculus and Prolog. Some things can be modeled in one but not the other, and vice versa.

• While Prolog primarily supports a logic/declarative style of programming, it also supports functional and imperative language concepts.

• The ultimate goal of logic/declarative programming is to make programming entirely an activity of specification—programmers should not have to impart control upon the program. Prolog falls short of the ideal.