We present an optimized recursive query evaluation for deductive databases with the presence of negative literals, based on the Deductive High-Level Petri Net (DHLPN) model. The syntax and semantics of this model are derived frown High-Level Petri Nets. Such a model benefits from the control structure inherent to Petri nets, allowing the capture of rules' stratification, in order to interpret, without ambiguities, negative literals in rules, particularly during a recursive query evaluation. Voir le lien ici