Anomalies in Rule-Adapted Workflows - A Taxonomy and Solutions for vBPMN

Modern approaches for flexible workflow modeling and execution foster the convenient combination of graph- and rule-based business logic. Specifically, the pattern-based adaptation of reference workflows to changing data contexts allows for a better modeling of variance and its evolution over time than traditional approaches. We recently presented variant BPMN (vBPMN) as an extended business process metamodel for realizing suchlike flexible and event-aware workflows by the rule-based application of reusable BPMN adaptation patterns to a reference workflow at runtime. However, as for vBPMN and in general, increased modeling flexibility usually comes at the cost of an increased danger of defining inconsistent business logic. Therefore we reviewed literature in order to create a generic taxonomy of potential anomalies which can occur when mixing graph- and rule-based business logic, comprising control-flow, rule-base and data-flow issues. To tackle these issues concretely for control-flow and rule-base issues in vBPMN, we developed a checking pipeline using the vBPMN metamodel and OCL to identify special structural error spots in a vBPMN model and a subsequent selective rule condition checking. In this paper we discuss by which means the most challenging anomalies can be prevented throughout the checking pipeline. The pipeline itself has been implemented by integrating eCore/oclInEcore facilities with the jBoss Drools Rule Verifier.