In this thesis we focus on improving current modeling and verification techniques for complex business processes. The objective of the thesis is to consider several aspects of real-life business processes and give specific solutions to cope with their complexity. In particular, we address verification of a proper termination property for workflows, called generalized soundness. We give a new decision procedure for generalized soundness that improves the original decision procedure. The new decision procedure reports on the decidability status of generalized soundness and returns a counterexample in case the workflow net is not generalized sound. We report on experimental results obtained with the prototype implementation we made and describe how to verify large workflows compositionally, using reduction rules. Next, we concentrate on modeling and verification of adaptive workflows — workflows that are able to change their structure at runtime, for instance when some exceptional events occur. In order to model the exception handling properly and allow structural changes of the system in a modular way, we introduce a new class of nets, called adaptive workflow nets. Adaptive workflow nets are a special type of Nets in Nets and they allow for creation, deletion and transformation of net tokens at runtime and for two types of synchronizations: synchronization on proper termination and synchronization on exception. We define some behavioral properties of adaptive workflow nets: soundness and circumspectness and employ an abstraction to reduce the verification of these properties to the verification of behavioral properties of a finite state abstraction. Further, we study how formal methods can help in understanding and designing business processes. We investigate this for the extended event-driven process chains (eEPCs), a popular industrial business process language used in the ARIS Toolset. Several semantics have been proposed for EPCs. However, most of them concentrated solely on the control flow. We argue that other aspects of business processes must also be taken into account in order to analyze eEPCs and propose a semantics that takes data and time information from eEPCs into account. Moreover, we provide a translation of eEPCs to Timed Colored Petri nets in order to facilitate verification of eEPCs. Finally, we discuss modeling issues for business processes whose behavior may depend on the previous behavior of the process, history which is recorded by workflow management systems as a log. To increase the precision of models with respect to modeling choices depending on the process history, we introduce history-dependent guards. The obtained business processes are called historydependent processes.We introduce a logic, called LogLogics for the specification of guards based on a log of a current running process and give an evaluation algorithm for such guards. Moreover, we show how these guards can be used in practice and define LogLogics patterns for properties that occur most commonly in practice.
|Qualification||Doctor of Philosophy|
|Award date||11 Dec 2007|
|Place of Publication||Eindhoven|
|Publication status||Published - 2007|