Frank Puhlmann

Using the π-calculus for formalizing workflow patterns

This paper discusses the application of a general process theory – the π-calculus – for describing the behavioral perspective of workflow. The π-calculus is a process algebra that describes mobile systems. Mobile systems are made up of components that communicate and change their structure as a result of communication. The ideas behind mobility, communication and change can also enrich the workflow domain, where flexibility and reaction to change are main drivers. However, it has not yet been evaluated whether the π-calculus is actually appropriate to represent the behavioral patterns of workflow. This paper investigates the issue and introduces a collection of workflow patterns formalizations, each with a sound formal definition and execution semantics. The formalizations can be used as a foundation for pattern-based workflow execution, reasoning, and simulation as well as a basis for future research on theoretical aspects of workflow.

Investigations on soundness regarding lazy activities

Current approaches for proving the correctness of business processes focus on either soundness, weak soundness, or relaxed soundness. Soundness states that each activity should be on a path from the initial to the final activity, that after the final activity has been reached no other activities should become active, and that there are no unreachable activities. Relaxed soundness softens soundness by stating that each activity should be able to participate in the business process, whereas weak soundness allows unreachable activities. However, all these kinds of soundness are not satisfactory for processes containing discriminator, n-out-of-m-join or multiple instances without synchronization patterns that can leave running (lazy) activities behind. As these patterns occur in interacting business processes, we propose a solution based on lazy soundness. We utilize the π-calculus to discuss and implement reasoning on lazy soundness.

Formalizing service interactions

Cross-organizational business processes are gaining increased attention these days, especially with the service oriented architecture (SOA) as a realization for business process management (BPM). In SOA, interaction agreements between business partners are defined as choreographies containing common interaction patterns. However, complex interactions are difficult to specify, basically because a formal, common standard supporting all interaction patterns is missing. This paper motivates the use of the π-calculus for formally representing service interaction patterns.

Extending BPMN for modeling complex choreographies

Capturing the interaction behavior between two or more business parties has major importance in the context of business-to-business (B2B) process integration. The Business Process Modeling Notation (BPMN), being the de-facto standard for modeling intraorganizational processes, also includes capabilities for describing cross-organizational collaboration. However, as this paper will show, BPMN fails to capture advanced choreography scenarios. Therefore, this paper proposes extensions to broaden the applicability of BPMN. The proposal is validated using the Service Interaction Patterns.

Soundness verification of business processes specified in the Pi-calculus

Recent research in the area of business process management (BPM) introduced the application of a process algebra--the π-calculus-- for the formal description of business processes and interactions among them. Especially in the area of service-oriented architectures, the key architecture for todays BPM systems, the π-calculus--as well as other process algebras--have shown their benefits in representing dynamic topologies. What is missing, however, are investigations regarding the correctness, i.e. soundness, of process algebraic formalizations of business processes. Due to the fact that most existing soundness properties are given for Petri nets, these cannot be applied. This paper closes the gap by giving characterizations of invariants on the behavior of business processes in terms of bisimulation equivalence. Since bisimulation equivalence is a well known concept in the world of process algebras, the characterizations can directly be applied to π-calculus formalizations of business processes. In particular, we investigate the characterization of five major soundness properties, i.e. easy, lazy, weak, relaxed, and classical soundness.

Interaction soundness for service orchestrations

Traditionally, service orchestrations utilize services according to a choreography where they are a part of. The orchestrations as well as the choreographies describe pre-defined sequences of behavior. This paper investigates if a given orchestration can be enacted without deadlocks, i.e. is interaction sound, inside an environment made up of different services. In contrast to existing approaches, we utilize link passing mobility to directly represent dynamic binding as found in service oriented architectures. Thus, the sequences of interaction behavior are not statically pre-defined but rather depend on the possible behavior of the services in the environment.

Structural Detection of Deadlocks in Business Process Models

A common task in business process modelling is the verification of process models regarding syntactical and structural errors. While the former might be checked with low efforts, the latter usually requires a complex state-space analysis to prove properties like deadlock-freedom of the models. In this paper, we address the issue of deadlock detection with a novel approach based on business process querying. Using queries, we are able to detect a broad range of common structural errors that lead to deadlocks, such as misaligned splits and joins. While not being complete, the proposed approach has the advantages of low computational efforts as well as providing graphical outputs that directly lead to the errors.

Variability Modeling and Product Derivation in E-Business Process Families

In this paper we present our findings in the area of process family architectures for e-business systems, described as variant-rich process models in the Business Process Modeling Notation (BPMN). We moreover address product derivation issues specific to process family engineering.

A Look Around the Corner: The Pi-Calculus

While Petri nets play a leading role as a formal foundation for business process management (BPM), other formalizations have been explored as well. This chapter introduces the *** -calculus as a formal foundation for BPM. The approach presented is pattern-centric, thus allowing for direct comparisons between the *** -calculus and different formalizations. In particular, selected basic and advanced control flow patterns as well as service interaction patterns are discussed. The chapter furthermore introduces the application of bisimulation techniques for proving soundness properties of business processes.

Implementation Framework for Production Case Management: Modeling and Execution

Nowadays, business process modeling and system-supported executions have become a commodity in many companies. Most systems, however, focus on modeling and execution of static, pre-defined processes with standards like the Business Process Model and Notation (BPMN). While these static process executions are applicable to a number of traditional processes like purchase orderings or back orderings, they fail at representing variant-rich, flexible processes. One solution for supporting flexible processes is Adaptive Case Management (ACM), where a case manager creates an individual execution path for each process instance, such as a doctor defining a clinical pathway for a specific patient. We found out, however, that both approaches are too strict, either supporting static process definitions with only a limited set of pre-defined flexibility or allowing maximum flexibility but requiring a highly skilled knowledge worker. To overcome this problem, we propose an implementation framework for Production Case Management (PCM) that combines concepts from traditional process management and adaptive case management. PCM combines the modeling of small, static process fragments with the execution flexibility of ACM.