Refined rules termination analysis through transactions

摘要

We propose two methods for rule termination analysis within active databases. The first method relies on evolution graphs simulating rule processing statically and considering both rule activation and deactivation. The evolution graphs provide a more accurate analysis than traditional graph-based approaches. The second method refines the former one considering the updates forming transactions that trigger active rules. We show that several termination analysis methods are captured with our approaches. The algorithms testing rules termination (with and without transactions) are presented considering both event-preserving and event-consuming policies and immediate rule execution. The proposed approaches turn out to be practical and general with respect to the various rule languages and thus they may be applied to several database systems.