Combining Process Modelling and Case Modelling

Adaptive Case Management deals with processes that are not predefined or repeatable, but depend on evolving circumstances and decisions regarding a particular situation. While case management is often considered as different from conventional business process management, in reality they cannot be strictly separated. A structured business process can contain parts which deal with non-routine cases. The Object Management Group (OMG) published the Business Process Model & Notation (BPMN) as well as the Case Management Model & Notation (CMMN). There is an ongoing debate whether these two languages should be combined are kept independent. After a short introduction into CMMN and BPMN we analyse an application process as it is typical for public administration in order to identify strengths and weaknesses of both BPMN and CMMN. We show that typical processes contain both structured and non-structured parts and neither BPMN nor CMMN alone is adequate to model the process. Finally, we propose recommendations for a metamodel, which combines elements of BPMN and CMMN. 1. – Introduction The benefits of Business Process Management (BPM) --reduction of cycle time, automation of routine processes, standardization and compliance, business integration and end-to-end performance visibility-are well established in many companies and public administrations. However, an important class of business processes has been unable to enjoy them: case management [1]. Case management processes are not predefined or repeatable, but instead, depend on evolving circumstances and decisions regarding a particular situation, a case [2]. Human judgment is required in determining how to proceed [3]. Case management processes are a common occurrence in public administrations and many industry segments. In a case management survey it was observed that two third of a knowledge worker's day is spent in unstructured and often unpredictable work patterns [4]. In government we can find case management in particular in decision processes, where human judgment is required. It is well established in many segments of benefits administration, like disability, welfare assistance, student financial aid and grants programs [1]. Adaptive Case Management (ACM) has been proposed as alternative data-centred approach for the management of such case processes [2], in contrast to the activity-centred approaches for structured business processes [4]. However, while case management is often considered as different from conventional business process management (see [4]), in reality they cannot be strictly separated. A structured business process can contain parts which deal with non-routine cases requiring additional investigation by different people, for example the underwriting in 1 Also called Adaptive Case Management (ACM) Page 2 Combining Process Modelling and Case Modelling financial services [1]. Nevertheless, in 2013 the OMG published the new Case Management Model and Notation (CMMN) [5], a modelling language specific for case management which is separate from Business Process Model and Notation (BPMN) [6], which is a standard for business process modelling. In this research we investigate, whether and how business process modelling and case modelling should be integrated. We analyse a concrete case and show that neither BPMN nor CMMN alone are appropriate to model all aspects. We analyse the commonalities and differences of BPMN and CMMN and suggest an integrated meta-model that allows to model structured and non-structured aspects of a business process in a single environment. 2. – Related Work Adaptive Case Management (ACM) has recently received growing attention [7]–[9]. In this section the differences between Adaptive Case Management (ACM) and Business Process Management (BPM) are briefly described. BPM focuses on business processes consisting of activities in a flow. According to The business process defines the flow of control, all the possible paths and permutations are defined in advance [4]. As an activity occurs, the state of the process changes (see Figure 1). As the state changes, it is transferred from one activity to the next, control flows similarly from one activity to the next. Figure 1: States in BPM [10] ACM, on the other hand, focuses on cases consisting on events and outcomes. More precisely a case is a collection of processes and isolated tasks, the number and identity of which cannot be fixed by a predefined template or rules. While the circumstances , which define how and when the case is completed are known, it is not possible to fully predetermine all the tasks needed to complete the case, as stated by Palmer [4]. A case evolves over time in the direction of achieving a goal and it can take unpredictable directions. The activities to reach the completion of the case are determined by the content and context and each of them creates information to add to the case. This defines the state of a case as shown in Figure 2. In particular where human judgment is predominant, ACM is a more appropriate approach. Figure 2: States in ACM [10] K. Hinkelmann, A. Pierfranceschi Page 3 Modelling cases has specific requirements compared to process modelling, because with ACM sequence and pathways of execution are not predetermined [4]. In the standard process modelling language BPMN there is the possibility to indicate subprocesses [6] as adhoc, but there is no further support to express specific aspects of cases. As a consequence, the OMG developed the Case Management Model and Notation CMMN [5] as a separate modelling language. There is an ongoing debate whether a new modelling language is necessary for case management or whether case modelling can be an extension of business process modelling. While Swenson claims that "BPMN is incompatible with ACM" [11], Silver argues that BPMN and CMMN should be merged [12]. He proposes to extend BPMN with case-specific element of CMMN. To illustrate his argument, he shows an example with case elements added to BPMN. Also IBM works on a case extension of their business process modelling in tool [13] offering the possibility to add discretionary tasks in business processes, instantiating them at runtime either manually by a user or by a few defined preconditions. In other words IBM's BPM 8.5.5 seamlessly blends runtime and non runtime processes, and all combinations thereof, in a single product. Other approaches combine CMMN and BPMN in the same tool by reusing model elements (see for example [14]). In contrast, we focus on developing an integrated language with elements from both CMMN and BPMN. 3. – Overview of BPMN and CMMN In this section we give a short overview of BPMN [6] and CMMN [5]. This section is not intended as an introduction into the languages but serves as a reference. For convenience we show in Table 1 the main elements of BPMN, which can be divided into 4 categories: flow objects, connectors, artefacts and swimlanes. The focus of BPMN is to model the control flow of business processes (with flow objects and sequence flow connector), the assignment of activities to participants (using swimlanes) and information flow (using data objects and message flow). Element Name Element Icon Element Description