Rule Termination Analysis based on a Rule

Rule processing can be subject to innnite loops-that is, rules may trigger one another indee-nitely. In database systems, this behaviour can be catastrophic; e.g., looping rules would certainly inhibit concurrency and saturate memory thus slowing down system throughput. In general it is unde-cidable to determine in advance if rules are guaranteed to terminate. If analysis algorithms are applied they work on a very rough level, i.e. although their conditions are suucient they are usually far away from being necessary. Existing rule analysis algorithms are designed for special rule models and only consider parts of the semantics of rules (e.g., do not consider priorities of rules). Therefore they are either too inaccurate and/or can only be used in a special environment (they are based on a feature of the underlying rule model, not/seldomly being used in other rule models). In this paper we present a static rule termination analysis based on the rule meta model Vampire. Vampire uses Petri nets as a formal method for a precise and exhausting specii-cation of many of the diierent elementary semantic aspects of rule models. Our rule termination anal-CADLAB is a Joint Venture of the University of Paderborn and Siemens Nixdorf Informationssys-teme AG ysis method is based on an analysis algorithm from the Petri net theory which exploits all the information represented in the Petri net. Thus, our method works on a much more precise level than other approaches known from literature. Additionally our approach can be applied to a broad spectrum of rule models.