Developing sound business process models is not trivial, even for experts. Although new tools, such as the BPMN Analyzer 2.0, are evolving, the majority of them are based on model checking or state-space exploration leading to known drawbacks in software testing such as fault blocking and fault masking. This paper presents BPMNinvest as an extension of the bpmn.io framework that provides fault feedback on every change of the model through structural approaches of the state of the art. This feedback contains a detailed explanation and localization of the fault and the possibility to manifest the fault as error during execution. For this reason, it is also suitable for teaching and quality training purposes. Currently, it detects 11 kinds of flaws and faults and was developed with the challenges of coverage, immediacy, and consumability in mind.
(a) Fault masking: Unbounded number of tokens
produced by the left loop hinders a deadlock at the right
structure. The browser crashes when to analyze this
model with BPMN Analyzer 2.0.
(b) Fault blocking: BPMN Analyzer 2.0 detects the
deadlock situation at the left-most parallel gateway
but not the lack of synchronization at the right-most
exclusive gateway.
(a) Fault illusion: BPMN Analyzer 2.0 detects a deadlock
situation at the right-most parallel gateway but it is caused by the
previous lack of synchronization at the exclusive gateway.
(b) Although BPMN Analyzer 2.0 states that
this model is not guaranteed to terminate, no
visual feedback is provided.
(an unsound behavior, such as deadlock, prevents other unsound behaviors from being reached). To verify
such disadvantages, example cases in [
This paper describes our new tool BPMNinvest, which orientates on the usability aspects and features
of BPMN Analyzer 2.0. After any change of a process model, the detected faults are illustrated directly
in the model without any noticeable delay (cf. Figure 3 (a)). The methodology of loop decomposition
in [
The rest of this paper is structured as follow: Section 2 describes the innovative features of BPMNinvest whereas Section 3 gives a short report of its maturity. The paper is concluded by Section 4.
Regarding the main innovations of an instantaneous and comprehensible control oflw error detection of BPMN Analyzer 2.0, BPMNinvest extend those innovations with a broad coverage of BPMN elements, an analysis on change, and a very detailed fault feedback.
BPMNinvest covers most elements available in the bpmn.io modeler except complex gateways and
message flows . It fully supports inclusive gateways regarding [
Although Fahland et al. [
Instead of showing errors, which occur during execution (as returned by a state-space exploration
approach), BPMNinvest collects 11 kinds of flaws and faults in the entire model, i. e., fault masking, fault
illusion, and fault blocking are not a problem. Regarding best practices, e. g., stated in [
Figure 3 (a) illustrates BPMNinvest how it highlights BPMN elements with a fault. Overall, informative hints are highlighted blue (such as implicit starts and ends), wrongly structured gateways are highlighted yellow, and faults are colored red. If the user hovers the elements, a short explanation of the fault/flaw is displayed. Clicking on the element and on the question mark opens a detailed fault explanation panel. Furthermore, the fault with the related BPMN elements are focused in the model (as illustrated in Figure 3 (b)). It gives a general explanation of the fault, a model-oriented explanation, some repair suggestions, and references to the literature if more detailed information is required. Most flaws and faults provide a simulation, i. e., the attempt to manifest the fault as an error during execution. This is not trivial as a fault is not directly associated to a trace of execution or a similar concept. Since fault blocking, fault illusion, and fault masking may appear during simulation, forcing the error is not guaranteed but usually works ifne. The simulation is done via a token game being based on the bpmn.io Token simulation extension. If the error is reached, the simulation automatically pauses and provides a further hint about the error.
BPMNinvest is the result of ongoing refinements on the algorithms of our business process analysis tool
Mojo. Those algorithms have been tested multiple times and were leading with regard to other approaches
in both diagnostic information and performance regarding running time [
Modeling correct business process models is not trivial. BPMNinvest supports users during modeling by giving feedback on change directly in the model with a detailed explanation and a simulation of the faults. Currently, it detects 11 kinds of flaws and faults. As it is based on structural instead of behavioral analysis, it is able to detect faults behind other faults and remove difficulties such as fault masking, illusion, and blocking. BPMNinvest addresses coverage, immediacy, and consumability of the analysis results.
In the future, we plan to extend BPMNinvest to handle further flaws and faults such as loops without entry, loops without exits, complex gateways, and message flows. Furthermore, BPMNinvest will get a “reachability checker” allowing to ask if a specific state can be reached from an initial state.
The author(s) have not employed any Generative AI tools.