Piece resolution: Towards larger perspectives

Conceptual Graph Rules and Equivalent Rules: A Synthesis

“…The important point for the backward chaining mechanism is that piece-based unification allows it to be guided by the structure of rules and goals. An experimental comparison between piece-based backward chaining and Prolog resolution was led in [CS98].…”

Rules Dependencies in Backward Chaining of Conceptual Graphs Rules

“…Decidability Since FOL formulaes associated with CG rules have the same form as TGDs [7], SR-deduction is semi-decidable (a sound and complete algorithm can compute in finite time whenever the answer is yes, but cannot always halt otherwise). Some decidable subclasses of the problem are proposed in [3]: let us suppose that, after the nth execution of the while loop in Alg.…”

“…In that case, completeness is achieved at the expense of efficiency. Finally, [7] compares BC using piece unification with the standard PROLOG BC that unifies only one predicate at a time. Though piece unification leads to fewer backtracks in the query rewriting mechanism, it does not translate to the overall efficiency of the algorithm, since these backtracks are hidden in the computation of unifications, that relies on a projection.…”

“…Since CG Rules can be translated into FOL formulaes having the form of Tuple Generating Dependencies (TGDs) [7], SR-deduction is semi-decidable.…”

“…Though CG Rules deduction can be computed using a PROLOG-like BC algorithm, successively unifying predicate after predicate in the equivalent FOL formulaes, [11] proposed to rely upon the structure of the graph and unify at the same time whole subgraphs of the rule (called pieces), effectively reducing the number of backtracks [7].…”