Backward chaining (source code)

= Backward chaining
{wiki}

Goal-directed rule evaluation that replaces a goal by premises of a matching rule. Shared variables require consistent answer <first-order substitutions> across the premises.