A process-aware information system (PAIS) will be not accepted by end users if its software clients do not support their native work ows or are too complex for them. When implementing business processes based on process management technology important issues are, therefore, how end-users can participate in the execution of the processes and how this can be accomplished as intuitively as possible. This becomes extremely important if high exibility demands need to be ful lled during process execution, while PAIS robustness and error safety need to be assured. In this software demonstration we show how the AristaFlow BPM Suite { an adaptive process management system developed by us { was applied to challenging applications in domains like healthcare, logistics, disaster management, and software development. The implementation of adaptive software clients in these di erent applications particularly proves the bene ts provided by an open application programming interface (API) as o ered by AristaFlow.
In many domains IT support can bene t from BPM technologies and the PAIS based on them. However, in many cases still specialized application systems are developed for accomplishing tasks being similar to process management functions. One reason for this is that available PAIS often cannot be adapted to domain-speci c problems. This leads to in exible systems with hard-wired processes where small changes in the underlying process often require changes to the whole system. Especially in domains with high exibility demands, existing BPM technology is not used due to its brittleness and in exibility. Another side e ect of these specialized systems is that processes are only known implicitly and are not modeled explicitly, making it hard to analyze and improve them.
During the last decade we developed the ADEPT2 process management technology [1{3]. Due to the high interest companies had in ADEPT2 we transferred it into an industrial-strength process management system called AristaFlow BPM Suite [4{6]. One of our basic goals is to enable robust and exible PAIS for a large number of processes from di erent domains. Recently, together with partners, we applied the AristaFlow BPM Suite to a variety of challenging application domains like healthcare, disaster management, logistics, and software engineering. In our software demo we will show some of the tools and software clients developed in this context using the AristaFlow Open API.
Section 2 presents results from domain-speci c projects in which we applied the AristaFlow BPM Suite. Section 3 describes the AristaFlow Open API. We conclude with a short summary and outlook in Section 4. 2
Applying AristaFlow to Advanced Applications Generally, domain experts only have little or no IT knowledge. Therefore, the standard clients as o ered by existing BPM systems are inappropriate for them and domain-speci c tools and PAIS interfaces have to be developed. This means that working habits of end-users need to be analyzed and speci c tools be developed. These tools must provide exactly that functionality as required by the respective domain expert and it must present this functionality in a way the user can easily handle, i.e., the respective systems needs to be tailored to the users' needs and the given application domain.
Recently, AristaFlow BPM Suite was applied to a number of challenging domains. In each project sophisticated BPM tools were implemented supporting end-users in their work as best as possible. 2.1
Domain. The project for process-aware, cooperative emergency management of
water infrastructures [
Applying AristaFlow. AristaFlow was used to manage and control the procedures and tasks during ood events as well as the corresponding information ow. Thus, it supported the responders in planning and executing ood response operations. Fig. 1 shows a domain-speci c PAIS user interface that was implemented using the AristaFlow Open API. In particular, users may order resources and deploy them to emergency locations.
Discussion. AristaFlow supports responders in accomplishing their operations in a coordinated, but exible way. One important aspect was to provide the necessary exibility, while ensuring robustness and error safety of the PAIS. 2.2
Domain. Healthcare and logistics are both characterized by high exibility
demands. Additionally, both require tools that are easy to use since domain
specialists have no IT knowledge. By supporting domain-speci c views on processes
(e.g., clinical pathways) and services, the SPOT project [
Applying AristaFlow. The SPOT project team selected AristaFlow BPM Suite as implementation platform for several reasons: Besides its correctness-byconstruction principle, its strict modular design and its service-oriented architecture were basic points in favor of AristaFlow. This enabled the service-oriented provision of advanced process support features and simpli ed the integration of existing application systems. Fig. 2 shows a mobile client for patient assistance in a hospital being connected to an AristaFlow process server. For representing clinical pathways the SPOT project uses editable tree structures, which are automatically mapped to (block-structured) AristaFlow process models. In particular, adaptations within such tree structure are translated into corresponding changes of the underlying process model and can be automatically applied to the considered instance using ad-hoc changes. This empowered domain experts (e.g. nurses) to change processes at a high level of abstraction.
Discussion. Again, the AristaFlow Open API facilitated the integration of our process engine into the overall architecture of the SPOT system. Additionally, features like robust process execution and user assistance in connection with adhoc process instance changes where considered being extremely useful features. 2.3
Domain. Due to the dynamic nature and high degree of collaboration,
communication, and coordination inherent in software engineering projects, automated
work ows can assist overburdened software engineers by providing orientation
and guidance. Yet, since there are so many di erent kinds of issues with
ambiguous and subjective delineation, it is di cult and burdensome to universally
and correctly model them in advance. This will also lead to work ows of
considerable size and complexity. The Q-Advice project [
Applying AristaFlow. The Q-Advice project uses the AristaFlow BPM Suite
as its process module. Thereby it makes heavy use of the AristaFlow API for
automatically constructing and adapting process models as well as for
developing specialized activity components. Based on context information an issue
work ow is automatically, dynamically, and uniquely constructed for every
software engineering issue. The activities of the work ow are then distributed to the
responsible users based on the organization model maintained by AristaFlow.
Fig. 3 shows part of the Q-Advice user interface where in the lower section the
user can see his current task as well as the next upcoming tasks. In the upper
section additional information is provided by the framework.
Discussion. Q-Advice makes it possible to provide situational and tailored
support and guidance for software engineers. In particular the work ows
resulting from the Q-Advice approach are much simpler than pre-modeled work ows
would be. Thereby Q-Advice makes use of the advanced change facilities of
AristaFlow and integrates them into its framework. By using AristaFlow BPM
Suite it became possible to hide the inherent complexity of process-orientation,
dynamic process changes, and exible task management from users; i.e., all
complex things are performed \beneath the surface" within the AristaFlow system.
Over the last years, more than 20 other groups from academia and industry
applied the ADEPT/AristaFlow process management technology in the context of
research projects. In the E-Commerce domain, for example, CONSENSUS [
The AristaFlow Open API Due to its Open API as well as its strict modular and service-oriented design (see Fig. 4), AristaFlow can be easily applied and adapted to di erent application domains. This way we enable integration of advanced process support functions into domain-speci c PAIS as well as provision of domain-speci c client, service and activity implementations.
All services of the upper layers of the overall architecture of the AristaFlow
BPM Suite depicted in Figure 4 can be utilized by domain-speci c client
applications and thus be easily integrated using the AristaFlow Open API. A functional
overview of the di erent services is provided in [
The development of the platform itself was driven by the development of its API. Even the built-in client applications of AristaFlow (including, e.g., the process editor) are just implementations of this Open API. The structure of the API also enables easy integration of di erent execution platforms and client types (e.g., rich-clients, web-clients, or mobile devices) and further enables integration into specialized applications. Last but not least, ad-hoc changes cannot only be applied by system administrators, but also by end-users. Thereby one or more low-level change operations can be combined to form higher-level change patterns in order to perform domain-speci c operations.
ExecutionManager
RuntimeManager
ChangeOperations
Execution layer WorklistManager
Application OrgModelManager
DataManager
ProcessManager LogManager
Persistence (DBMS)
User interaction layer Basic services layer Low-level services layer
Fig. 4. Basic architecture of the AristaFlow BPM Suite
Domain-speci c solutions can be easily developed based on the AristaFlow Open
API. This API is not limited to solely high-level client applications, but includes
all levels of the architecture; e.g. new services can be integrated easily and even
the process and application meta models themselves can be extended (see [