The BPMN standard is largely adopted by industry and academia due to its intuitive graphical notation. Nevertheless, fully understanding the behaviour of BPMN collaboration models may be difficult when dealing at the same time with multiple instances, exchange of messages, and data manipulation. Figuring out the interplay between such concepts by statically looking at models is in general error-prone and time-consuming. To overcome this issue we provide a novel model animator tool, called MIDA. It turns out to be an effective supporting tool for enhancing the understanding of BPMN collaborations and debugging errors that can easily arise in modelling them.
The BPMN standard [
To clarify this issue, we resort to a multi-instance collaboration model that will be used as a case study throughout this demo paper. The collaboration diagram depicted in Fig. 1 concerns with the management of a hotel booking involving three participants: the Customer, the Booking System, and the Hotel. The collaboration represents a scenario in which the Customer interacts with the Booking System in order to get a list of hotel quotes for a desired travel. To serve the customer request, the Booking System creates a process instance. This, in its turn, sends multiple requests (via a multi-instance sending task) to the Hotel pool, thus creating many instances of this latter participant. The F. Casati et al. (Eds.): Proceedings of the Dissertation Award and Demonstration, Industrial Track at BPM 2018, CEUR-WS.org, 2018. Copyright © 2018 for this paper by its authors. Copying permitted for private and academic purposes. This volume is published and copyrighted by its editors. Canvas
Booking System collects the quotes for the Customer and sends them to him. Finally, the Customer communicates the selected hotel to the Booking System, which consequently gives a feedback to both selected and discarded Hotel instances. To model correctly this collaboration the diagram must contain enough information to correlate each message with the appropriate instance. For example, if messages are not properly delivered, an Hotel instance may save a transaction even if it has not been selected.
Model animation plays an important role to address the issues that may arise when
modelling multi-instance collaborations, such as the intricate management of the
correlation mechanism or the handling of data and messages. Animation can enhance the
understanding of business processes behaviour [
Therefore, we have developed MIDA (Multiple Instances and Data Animator), a novel animator tool. It supports designers in achieving a more precise understanding of the behaviour of a collaboration by means of the visualisation of the model execution, also in terms of the values evolution of data objects and messages. MIDA animation features result helpful both in educational contexts, for explaining the behaviour of BPMN elements, and in practical modelling activities, for debugging errors that can easily arise in multi-instance collaborations. 2
In this section, we present the MIDA modelling and animation features, specifically focussing on the interplay between data, messages and multiple instances.
Travel Info Travel Order
Travel m e t s y S ikgn Quote oo Management B
Customer
Choices Require Quotes
Receive and Combine Quotes
Travel Quotes
Selected Quote Send Quotes Customer
Quotes
Travel Quote if (Travel_destination != undefined
& Travel_period != undefined) { var TravelInfo_customer =Quotes vvaaCrrTCTaQrulrucTTaasuovrrvtlteaaaeeoltvvelmee_elldrIIeNnnsafftmooie__n;RdpQaTeeeurtcaosreivttiieeovslioennd;QSauet=onitdeon = TSrCaeovleenclfitQrmaFuRnoeaQdteeeucdoebitavecek
Receive Selected Quotes
Send Feedback
Quote
Managed MIDA: Multiple Instances and Data Animator
Feedback
Save yes Transaction Animation. The key characteristic of MIDA is the animation of collaboration models, supporting in particular the visualisation of data evolution and multiple instances execution. At any time, the animation can be paused by the user to check the distribution of
Fig. 4: MIDA Animation. tokens and the current state of data. From a practical point of view, this allows designers to debug their collaboration models. They can indeed detect undesired executions, where e.g. a control flow is blocked, and deduce the cause beyond them by possibly checking the values of the involved data. These debugging facilities are particularly useful in case of multi-instance collaborations, where data values are used to regulate the correlation of messages with instances. Anyway, like in software code debugging, the identification and fixing of bugs are still in charge of the human user.
By selecting the Animation Mode in the MIDA interface, a play button will appear
over each fireable start event. Once this button is clicked, one or more instances of the
desired process are activated, depending on the multi-instance information specified in
the model. This creates a new token labelled by a fresh instance identifier. Then, as
shown in Fig. 4, the token starts to cross the model according to the operational rules
induced by our formal semantics [
The Data Panel of the MIDA interface allows the user to monitor the evolution of data values. Fig. 6 shows how data values change after the execution of the task Require Quotes, which stores the values received via the Travel Info message in the fields destination, period and customer of the Customer Choices data object.
Fig. 5: Guard Violation. 3
In this section, we present how MIDA can effectively support designers in debugging their models. We resort to our case study in Fig. 1 to show how issues related to instance correlation can be detected.
Let us consider the effect of messages arrival into the Hotel pool. The arrival of a Travel Order message triggers the activation of a new process instance, while a Feedback message has to be routed to an already existing instance. Hence, in the latter case, the message needs to be properly correlated. In fact, after the choice performed by the Customer, only the selected Hotel instance has to receive a positive Feedback and then perform the Save Transaction task. To do that, each Feedback message contains the unique name of the Hotel that has to receive it. However, if the correlation check would be not properly specified in the receive task Receive Quote Feedback (e.g., only the travel period is used as correlation data or no correlation data is provided), the feedback messages would be not correctelly delivered. As consequence, the unwanted Hotel instances may book an overnight stay in vain, while the one selected by the Customer would not. This bug can be detected by MIDA as the Hotel chosen by the customer, stored in Customer Quotes, would not be the one stored in the Order Quote data object of the corresponding Hotel instance. To fix the bug, the designer can specify the conditional expression OrderQuote_hotelName === Feedback_hotelName of the receive task thus enabling the right message correlation. 4
The MIDA tool, as well as its source code, examples, user guide and screencast, are available at http://pros.unicam.it/mida/. In particular, the screencast shows a typical scenario where the user requires to model, animate, and debug a BPMN model. MIDA can be redistributed and/or modified under the terms of the MIT License.