Rule randomization for propositional logic-based workflow verification

Abstract

Workflow verification has been a well studied research topic during the past few years. Theorem proof based approaches to workflow verification become popular due to several advantages including being based on formal characterization with rigorous and non-ambiguous inference mechanisms. However, a common problem to these inference mechanisms is combinatorial explosions, which forms a major performance hurdle to workflow verification systems based on inference. In this paper, we study how randomization enables reuse and reduces processing time in logic based workflow verification approaches. We, in particular, look at a propositional logic based workflow verification technique. For the logic inference rules, which are used to infer new truthful propositions from existing truthful propositions in this logic, we apply randomization to the inference rules after each verification task such that new inference rules reflecting the componentized verification are added to the inference rule sets. We reviewed the savings incurred in verifying a workflow pattern and provide a theoretical analysis.