We are working or have worked on the following research projects:
National Process Library
Decreasing budgets, demographic change, and costumer orientation of public services call for the modernisation of the German administration. Within this context the research project "National Process Library", funded by the German Ministry of Interior, and in cooperation with Humboldt University zu Berlin, aims at gathering and exchanging business processes of the public administration across national, state, city and municipality level. The Business Process Technology group develops a process plattform for the National Process Library providing a convenient tool to share, analyze, discuss and exchange processes among the different partners of the German administration.
Major challenges are the the abilitity to cope with the variety of different Business Process Notations used in the public sector as well as the different abstraction levels, structures and labels of these models.
Within the Design Thinking Research Program we have started a joined project with chair for Systems Analysis and Modeling
. Design Thinking is a set of methods and techniques to enable innovation processes. It is used in companies like IDEO
to create new products. In this project we want to explicitly support D-LABS in their software requirements engineering projects.
The goal is to integrate other types of models with BPM models. The main application will be the collection scenario descriptions which can be integrated with other information and ultimately refined to come to process models. We work on this project together with D-LABS.
Adaptive Services Grid (ASG)
The goal of the Adaptive Services Grid project is to develop an architectural blueprint and a proof-of-concept reference implementation of an open platform for the development and provision of adaptive services. The aimed platform offers functionality for the whole service lifecycle, starting from discovery of already existing services, continued by the composition and binding of these services, enactment, and monitoring of the composed service enactment in order to detect faults and handle them either by re-binding or re-composition. A key concept of ASG is the usage of semantic information about services to fulfil user requests. Services in ASG are thus described not only syntactically (the technical interface to invoke a service) but also semantically. These semantics are represented by formal semantic service specifications, which include functional and non-functional properties and rely on ontologies from industrial domains. ASG components make use of this information about what a service does in addition to how it does it for the already mentioned discovery and composition of services.
To achieve its goal, ASG addresses scientific and technological issues making use of the knowledge and expertise of major European research institutions with significant contributions from the software, telecommunications, and telematics industry. ASG provides the integration of its sub-projects in the context of an open platform, including tool development by small and medium sized enterprises.
ASG is an integrated project of the 6th frame work programm funded by the European Union. Our tasks in the project are:
- Scientific project coordination and the lead of prototype development
- Research of service matchmaking and automated composition approaches based and formally specified service semantics
For further information and publications on the project please refer to the ASG homepage
Process Family Engineering in Service-Oriented Applications (PESOA)
The objective of PESOA is to develop an approach for the product line based development of process oriented software (Product Family Engineering), which is validated by developing a platform prototype for Process Family Engineering. The project is funded by the German Federal Ministry of Education and Research. Our major roles in the project are:
- Project coordination
- Development of concepts and notations for modeling variant-rich UML and BPMN process models in the context of process family engineering
- Selection and application of appropriate process modeling techniques for PESOA application domains e-business and automative
- Development of prototypical applications for modeling variant-rich UML and BPMN process models
Compliance Management for Business Process Models.
Day to day, compliance requirements are forcing business to adapt and change business process in order to be aligned with these requirements. Challenges in the field of compliance checking are manifold. You need languages to express compliance rules with a balance between expressiveness and formal background. These languages have to address the major aspects of business processes namely, control flow, data flow, and resource assignment. A second challenge is to automate the identification of processes that are subject to checking once a rule/process model is added or changed. Giving intelligible feedback to the person of interest when a violation of a rule occurs is another challenge. An automated suggestion of correction to violating models is a wish. In this thesis, I am planning to have an environment where modeling of compliance rules takes place independent of modeling business processes. Later on, identification, validation, explanation, and correction would take place automatically. This vision is valid under the assumption that both process models and compliance rules modeling is aligned to a business-wide ontology.
In an age where more and more interactions between business partners are carried out electronically, choreographies, i.e. specifications of desired interaction behavior among partners, have increased importance. My research centers around the question of what a good choreography modeling language is. While favoring a modeling style where atomic interactions are the basic building blocks of choreographies, a number of challenges arise with this new modeling style. Therefore, the second half of my research centers around formal verification of interaction models.
As main research outcomes a number of choreography languages where proposed and studied, including Let's Dance, BPEL4Chor and iBPMN. Properties studied in this context are conformance, instance isolation, local enforceability and desynchronizability.
Process modeling has a major share in requirements analysis and documentation already. However, this is usually applied to BPM-Projects. Currently, we see lines bluring on what is requirements analysis in SOA projects and process analysis in BPM projects.
The work concentrates on processes as the central artifact to transport requirements. It integrates artifacts from requirements engineering with the processes.
Service Semantics for the Discovery, Management, and Composition of Services
The goal of service orientation is to align the IT infrastructure to the business goals of a company. Service composition is a fundamental concept of service orientation. Using service composition, existing functionality can be aggregated to form new, more complex functionality. Manual service composition is too inflexible to deal with today’s dynamic environment. It does not support reactions to changes in the service landscape, adaptation to changing business goal, or tailoring of compositions to individual requests.
My research focus is on making the creation of service composition simpler. One approach is to develop a fully-automated, run-time service composition approach. Service compositions are created on-demand for individual requests. This approach is based on heuristic search algorithms from automated planning. A second approach assists human modelers in creating service compositions by suggesting services, automatically creating process parts, and detecting errors in service compositions.
The world wide web has introduced a platform for large-scale application to everybody. The underlying architecture of the world wide web is resource-oriented and differs in several aspects from previous attempts at large-scale distributed environments.
My research focuses on the applicability of resource orientation for business process management. This includes modeling resources in a process-oriented fashion as well as resource-oriented modeling of processes
Business Process Modeling and particularly modeling business processes as collaborations is one of the challenges of today enterprise software development. The industry of software development becomes more and more expensive making crucial a correct translation from business idea to implementation to allow for a complete understanding and complete exchange of information between development participants. Particularly, large established software infrastructures are critical with respect to the integration of new components.
My research envisions validation of business processes using Agent-based Simulations.
In Business Process Management
by Mathias Weske
(page 7) was argued that "business process models are the main artifacts for implementing business processes". In addition in Business Process Management
by Mathias Weske
was defined a business process management system as "a generic software system that is driven by explicit process representations to coordinate the enactment of business processes." On the other side in An Introduction to MultiAgent Systems
by Michael Wooldridge
was argued that "an agent is a computer system that is capable of independent action on behalf of its user or owner". In this context at least semantically it is obvious that the generic software system in the form of a business process management system can be in fact a multi agent system.
According to Business Process Management
by Mathias Weske
"a business process consists of a set of activities that are performed in coordination in an organizational and technical environment" (page 5). Moreover, Service Oriented Architectures (SOA) and business process modeling go hand in hand. Very often the activities modeled within BPMN models are typically implemented as services. The relationship between agent-based simulation environments and services is a critical one since may help understanding organizational models as services.
Software engineering went a long way from low level programming languages to process models (both defining control flow, but on different abstraction levels). The driving force for such an evolution was (and still remains) reuse. On one hand, engineers want to operate with coarse grained statements when describing algorithms, on the other, stay flexible when capturing an arbitrary system. Achievement of the equilibrium point is impossible, approaching it is research. At the end, it is about the choice of abstraction level to operate at.
My research concentrates on the questions regarding structural aggregation of control flow logic, i.e., searching for meaningful process fragments suitable for generalization. Assuming the existence of executable processes, the questions to be answered are: how to extract coarse grained fragments of process logic, can the system be expressed in less lines of code or modeling constructs, what are the methodologies and implications. In order to answer these questions I reuse and improve the existing, but also develop new techniques of process model decomposition.
Finally, in order to achieve flexibility in describing special kind of processes that capture ad-hoc scenarios (instance rich processes), I develop and shape the technique of Flexible Process Graphs (FPG). Instead of structural generalization, FPG proposes generalization of the control flow logic. The core process modeling construct - a sequential execution constraint between two activities, is replaced in FPG by the relation on two sets of activities. Once all the prerequisite activities are accomplished, another set can be enabled for execution.
Business process analysis requires multiple perspectives of one business process, varying in the level of details. On one hand there are process executors, who want to know every single detail about the process; on the other – managers, who are interested in the “big picture”. This demand results in a necessity to maintain several models for one business process, which is a laborious and error prone task. Meanwhile, companies often possess detailed models, capturing all the peculiarities of the processes. Given this, it is possible to abstract
these models and derive high level process overviews.
In my thesis I would like to analyze several aspects of the business process abstraction task. The first question which arises is which model elements are insignificant. To answer this question analysis of nonfunctional properties of a model and model elements can be used. Semantic annotation of process model elements can be another option. Within the work algorithms for identification of insignificant elements using this information will be developed.
When insignificant elements are identified, they should be abstracted. This task assumes that there is a model transformation that results in a correct
abstracted model. Furthermore, extra requirements can be imposed, e.g., the ordering constraints of the initial model should be preserved and the abstraction should have minimal side effects. Model transformation is based on decomposition of a model into fragments and subsequent replacement of fragments with nodes. This part of the work assumes improvement of existing model decomposition techniques (PST, RPST, SPQR) and introduction of new algorithms for model decomposition.
The theoretical results of this work have to be validated by means of implementation in Oryx
– a Web 2.0 application for modeling business processes.
It is a common observation that a business scenario is captured by multiple process models on different abstraction levels. Depending on the purpose of the model, not only the level of abstraction, but also the adopted perspective is different between these models. Therefore, alignment of these models is of key importance but far from trivial.
My research aims at improving such a model alignment. In particular, my research centers around the semantics of links between models of different abstraction levels and different perspectives. That includes a categorization of common mismatches and the definition of methods to detect them. I also investigate, what kind of consistency can still be guaranteed for such models.