=Paper=
{{Paper
|id=Vol-2097/paper2
|storemode=property
|title=A Flexible Event Handling Model for Using Events in Business Processes
|pdfUrl=https://ceur-ws.org/Vol-2097/paper2.pdf
|volume=Vol-2097
|authors=Sankalita Mandal
|dblpUrl=https://dblp.org/rec/conf/emisa/Mandal18
}}
==A Flexible Event Handling Model for Using Events in Business Processes==
https://ceur-ws.org/Vol-2097/paper2.pdf
A Flexible Event Handling Model
for Using Events in Business Processes
Sankalita Mandal 1
Abstract: Business process management (BPM) enables modeling, implementing and monitoring
organizational processes to achieve certain business goals. As organizations continue to strive for
agility, they have started taking advantage of the digitization and bring flexibility in their processes by
several means. One of these is to integrate complex event processing (CEP) with business processes.
Event handling specifies how a process communicates with its environment and how this environment
influences the execution of the process. Though highly expressive and feature-rich languages like
BPMN exist for process specification, they still lack an unambiguous semantics for event handling
in different situations. In this work, an event handling model is proposed that take into account the
possibilities of subscribing to an event at different point in time with respect to process execution.
The model is grounded with formal semantics using Petri Nets and trace analysis to ensure correct
execution of process behavior as well as the temporal dependencies among event subscription, event
occurrence, event consumption and event unsubscription.
Keywords: Process execution; Event handling; Event subscription; BPMN
1 Introduction
Process model defines a set of activities to achieve certain business goals [We12] in an
organizational context. These activities are connected with causal and temporal dependencies
using control-flow. A process model also specifies how a process is supposed to interact with
the environmental occurrences, represented as events [EN10]. Process engines subscribe
to event sources and react to events emitted by these sources according to the process
specification. Often, a complex event processing engine is used to abstract from the
complexities of connecting to different event sources emitting events in different formats
and aggregating them to receive the business level event [HMW13].
Business Process Model and Notation (BPMN) [OM11] is the industry standard for modeling
and executing processes. BPMN includes explicit event constructs to depict the production
and consumption of events, message flows to link external process participants, and control-
flow routing based on events such as event-based gateways or boundary events. However,
the event handling semantics in BPMN is pretty limited while it comes to specify when
1 Business Process Technology Group, Hasso Plattner Institute, University of Potsdam, Germany
sankalita.mandal@hpi.de
12 Sankalita Mandal
to subscribe to an event source or when to stop listening to an event stream [MWW17].
According to BPMN, when the control-flow reaches the event construct, the node is enabled
and the process instance waits for the event occurrence. Once the event occurs, the control-
flow is passed to downstream activities. The above semantics is a severe limitation as in a
distributed setup, the event sources are mostly unaware of the process execution status and
therefore the events can occur anytime irrespective of the process being ready to consume it.
Fig. 1: Motivating example showing the need of flexible event subscription.
In Fig. 1, a process from logistics domain has been shown. The Logistic Company manages
a process engine to control several transportation where trucks carry goods via Euro Tunnel
following the transport plan sent by the company. However, when the truck is still driving
(before Arrival Time), the engine gets notification if there is a Significant Delay at
Euro Tunnel, e.g. caused by technical difficulty or traffic congestion. In this situation,
the alternative route ferry is taken to cross the Strait of Dover. The ferry status update is
considered periodically to adapt the local delivery plan at the destination and the orders
are distributed once the ferry is taken. Now, Euro Tunnel or ferry do not know about the
processes run by the logistic companies and publish the status update events whenever
there is a change of situation which might not coincide with the time when the process is
ready to receive the update. As a result, a process instance may not react to an event which
occurred before its control-flow reached the respective event construct, even if the event
is still relevant for process execution. The process execution might even get stuck waiting
for an event which has already occurred. On the other hand, the process engine does not
need to listen to an event stream if at certain point of process execution the event becomes
irrelevant, e.g. Euro Tunnel updates become irrelevant after ferry is chosen.
So far, there is no clear semantics for all the possible scenarios supporting flexible event
handling from a business process perspective. The work in this thesis proposal addresses
these shortcomings of event handling semantics and aims to offer a flexible event handling
model accommodating the possibilities of event-process interactions starting from the
initiation of process engine, via process execution, to the engine termination. The next
section discusses the steps followed so far as well as the future plans to achieve this.
A Flexible Event Handling Model 13
2 Research Plan: Past, Current & Future Steps
The goal of the research is to build a flexible event handling model from the perspective of
business process execution. The event handling semantics should be grounded formally and
provide conceptual as well as technical analysis of using events in business processes. The
rest of the section describes the steps required to reach the goal.
Integration of Real-World Events and Business Processes. The whole research is pre-
dominantly motivated considering the importance and advantages of using event information
flexibly during process execution. There can be several applications (such as [Pu17]) of using
event processing techniques that can improve process flexibility and efficiency. The first
step was, therefore, to build an integrated architecture enabling event-process interaction.
Though the current process engines have provisions for receiving events from external
sources, there was no end-to-end framework considering the conceptual and technical
challenges for integrating real-life events in processes. In our work associated with a project
with industry partners [Be16], we explored the basic aspects to consider while integrating
the two worlds of BPM and CEP [MHW17]. This framework identifies three requirements:
1) Separation of concerns between the logic of process execution and event processing, 2)
Representation of event hierarchies to show the connection among event sources, low-level
events generated from the sources and higher-level business events aggregated from them,
and finally 3) Execution of event integration that deals with the technical challenges such
as binding events, receiving events and reacting on events. Based on these requirements,
a conceptual framework is provided. Also, an integrated architecture consisting of the
process engine Chimera2, the event processing platform Unicorn3 and an extended version
of Camunda process modeler4 is implemented to realize the concepts.
Early Event Subscription and Event Buffering. The basic framework defines the concept
and technicalities required to subscribe to an event, get notified when a matching event
occurs and to react on the event according to process specification. However, in a distributed
IT setup, the event sources might be unaware of the process execution status. Therefore,
the BPMN assumption of event occurrence only when the process is ready to consume it
seemed unrealistic and restricting. Hence, the notion of early event subscription using an
explicit subscription task is introduced in later work [MWW17]. To accommodate early
event occurrences, an event buffer is discussed in this context. The buffer comes with
specific policies to store, retrieve and reuse the events.
Flexible Event Handling Model with Petri Net Mapping. The next step prescribes the
formal semantics for handling events considering the fact events can occur anytime – before,
during or after the process is ready to consume it. Based on the grounding semantics, we first
2 https://bpt.hpi.uni-potsdam.de/Chimera
3 https://bpt.hpi.uni-potsdam.de/UNICORN/WebHome
4 http://bpmn.io/
14 Sankalita Mandal
analyze different possibilities of modeling an external event in a process flow, represented
as intermediate catching message events. We assume that the subscription does not have
any unresolved data dependency, i.e. the subscription is not dependent on any data that
must be generated during process execution. The subscription, occurrence and consumption
possibilities of events are considered based on a timeline starting from the initiation of
process engine and ending at the engine shut down, as shown in Fig. 2.
Fig. 2: Process execution timeline.
The CASU Framework proposed in [DM09] talks about (un)-subscription patterns with
respect to process instantiation semantics whereas we focus on the environmental occurrences
modeled as intermediate catching events. Fig. 3 is an extended version of the causality
proposed by [BDG07] that said an event
can only be consumed if there is a subscrip-
tion for this event and the event has actually
occurred. We proposed in [MWW17] that
to make an event relevant for a process,
the subscription should be done before the
event occurrence. An unsubscription can Fig. 3: Dependencies among event subscription, event
be done only when a subscription exists, occurrence, consumption, and unsubscription.
but it is independent of the event consumption, or even event occurrence, e.g. if at certain
point of process execution, the event becomes irrelevant, unsubscription can be done. The
dependencies show that along with maintaining the causality of process execution, the
event subscription, occurrence, consumption and unsubscription should also follow certain
temporal ordering. Currently we are working on mapping the concepts required for flexible
event handling model to Petri Net modules in order to specify the implementation semantics
and analyzing the traces to ensure correct execution.
3 Conclusion
Event processing informs business processes about the environmental occurrences, such
that the process can adapt to the changed environment as required. This improves process
execution with respect to flexibility and efficiency. Hence, we presented a basic framework
to realize the integration of event processing in business processes. Further, we developed
the notion of early event subscription and event buffering to accommodate the need of
flexible event-process communication in a distributed IT system. To this end, a formal event
A Flexible Event Handling Model 15
Extensive literature study Dagstuhl Seminar on “Integrating Process‐ EPBPM Workshop, DEBS 2017 Focused and detailed research
GET Service Project (Jul – Sep ‘15) Oriented and Event‐Based Systems” (Aug ‘16) Bosch Project (Sep ‘16 – Sep ‘17) Event handling model development, formalization & evaluation
Role of events in Integration of Flexible event handling
Early event subscription
business processes and real‐world events and model with Petri Net
and event buffering
several relevant aspects business processes mapping
LEGEND
Highlighted
Phase activities
Publication
Fig. 4: Research progress and planned steps.
handling model is proposed that specifies the semantics for event subscription, occurrence,
consumption and unsubscription taking into account the complete runtime of a process
engine and several points in time when subscription for an intermediate event can be made.
Next steps include correctness check of process specification in the light of the interactions
with environmental events. Fig. 4 summarizes the research progress reflecting on the phases
discussed, the associated publications for each phase and the activities that inspired further
research directions.
References
[BDG07] Barros, A.; Decker, G.; Grosskopf, A.: Complex Events in Business Processes. In:
Business Information Systems. Springer, 2007.
[Be16] Beyer, J.; Kuhn, P.; Hewelt, M.; Mandal, S.; Weske, M.: Unicorn meets Chimera:
Integrating External Events into Case Management. In: BPM Demo Session. CEUR-
WS.org, 2016.
[DM09] Decker, Gero; Mendling, Jan: Process instantiation. Data Knowledge Engineering,
68(9):777–792, 2009.
[EN10] Etzion, O.; Niblett, P.: Event Processing in Action. Manning Publications, 2010.
[HMW13] Herzberg, N.; Meyer, A.; Weske, M.: An Event Processing Platform for Business Process
Management. In: EDOC. IEEE, 2013.
[MHW17] Mandal, S.; Hewelt, M.; Weske, M.: A Framework for Integrating Real-World Events and
Processes in an IoT Environment. In: CoopIS. Springer, 2017.
[MWW17] Mandal, S.; Weidlich, M.; Weske, M.: Events in Business Process Implementation: Early
Subscription and Event Buffering. In: BPM Forum. Springer, 2017.
[OM11] OMG: , Business Process Model and Notation (BPMN), Version 2.0. http://www.omg.
org/spec/BPMN/2.0/, January 2011.
[Pu17] Pufahl, L.; Mandal, S.; Batoulis, K.; Weske, M.: Re-evaluation of Decisions based on
Events. In: BPMDS. Springer, 2017.
[We12] Weske, M.: Business Process Management - Concepts, Languages, Architectures, 2nd
Edition. Springer, 2012.