backward chaining


backward chaining

[¦bak·wərd ′chān·iŋ] (computer science) In artificial intelligence, a method of reasoning which starts with the problem to be solved and repeatedly breaks this goal into subgoals that are more readily solvable with the relevant data and the system's rules of inference.

backward chaining

(algorithm)An algorithm for proving a goal by recursivelybreaking it down into sub-goals and trying to prove theseuntil facts are reached. Facts are goals with no sub-goalswhich are therefore always true. Backward training is theprogram execution mechanism used by most logic programminglanguage like Prolog.

Opposite: forward chaining.

backward chaining

In AI, a form of reasoning that starts with the conclusion and works backward. The goal is broken into many subgoals or sub-subgoals which can be solved more easily. Known as top-down approach. Contrast with forward chaining.