The Fifth International Conference on Business Process Management, BPM 2007, was held in Brisbane, Australia, on 24-29 September. I attended the main conference on 25-27 and the co-located event on BPM Governance, WoGo 2007, in which I gave a presentation of my paper “On the Lookout for Organizational Effectiveness — Requisite Control Structure in BPM Governance”.

The first day was kicked off by prof. Michael Rosemann, General Chair, Arun Sharma, Deputy Vice-Chancellor QUT, Michael Choi, MP, Parliamentary Secretary, Queensland Government, prof. Wil van der Aalst, BPM Conference Steering Committee Chair, and prof. Peter Dadam, Program Co-Chair. The brief greetings gave a breezy start to the jet-lagged day.

The keynote presentation by Simon Dale, Senior Vice-President and CTO, SAP Asia-Pacific was also an inspiring one. In his presentation “Holistic BPM: From Theory to Reality”, mr. Dale reviewed the three waves of BPM and related the technological development of SAP to the shift from improving operational efficiency through BPR and ERP to competitive differentiation through Business Network Transformation and Enterprise SOA, or from “BPM 1.0″ to “BPM 2.0″.

He argued that businesses have not fully leveraged “process” yet and that the next frontier will be reducing the “time to change”. On the other, he was (overly) optimistic about the industrialization of the software industry and maintained that systematic re-use across solutions brings about efficiency, shorter time-to-volume and unprecedented level of quality and scalability.

Mr. Dale recognized that automated BPM has layers of application: the human-centric process collaboration and human interaction management, packaged processes by SAP with embedded workflow and system-centric process automation and infrastructure for BAM. Enterprise modeling embracing business process analysis, simulation/optimization, process standardization and architectural planning and governance he considered as optional to BPM.

His roadmap of SAP NetWeaver looked really interesting. Next year would bring about composite business processes with full process lifecycle support, integrated composite developer experience, exposed application process flow models and Human Interaction Management with enhanced Business Task Management (UWL). Subsequent future releases later this decade would come with a common process layer, actionable process analysis providing drill-down into process instances and process design time, situational composites and advanced process collaboration. I later spoke with one of the SAP developers and he confirmed that SAP is entirely rewriting its platform to make process as the first citizen.

The first session featured three presentations on Business Process Maturity and Performance. The first one was given by prof. Dirk Deschoolmeester from Vlerick Leuven Gent Management School. In their paper, he and his colleagues defined a framework for identifying characteristics of Business Process Orientation and provided a tool for measuring BPO of an organization.

In the second presentation, Stephen Corea from Warwick Business School discussed “Challenges in Business Performance Measurement: The Case of a Corporate IT Function”. The exploratory case study by him and Andy Watters of SAP provided insights into the challenges of performance measurement. Based on their findings in the case organization, they suggested that the approach of learning bottom-up what factors actually drive results and should be concentrated on would be preferable to the prevailing practice of overarching top-down dashboard-based business performance measurement.

The third presentation of this session was given by Hajo Reijers from Eindhoven University of Technology. He and his colleagues had studied what kind of effect the geographical proximity of actors has on the performance of a workflow process. Based on their data mining of process logs, they concluded that there is a significant positive effect on process performance when workflow actors are geographically close — the technology does not remove geographical barriers.

After the lunch, the second session was about Business Process Modeling and moderated by prof. Arthur ter Hofstede from Queensland University of Technology. The first paper that was presented was “What makes process models understandable?” by Jan Mendling, Vienna University of Economics and Business Administration, Hajo Reijers, Eindhoven University of Technology, and Jorge Cardoso, University of Madeira. The microphone was given to Hajo again. He shared with us their results that the interviewed BPM students tend to exaggerate the differences in model understandability, that self-assessment of modeling competence appears to be invalid, and that the number of arcs in models has an important influence on understandability.

Christian Wolter and Andreas Schaad, SAP Research, had researched “Modeling of Task-Based Authorization in BPMN. Mr. Wolter presented the extensions that they propose for the Business Process Modeling Notation (BPMN) to express resource allocation patterns within the workflow model.

The last two presentation of the second session were about two short papers: “BPMN: How much does it cost? An incremental approach” by Matteo Magnani and Danilo Montesi from University of Bologna and “View-Based Process Visualization” by Ralph Bobrik, Institute of Databases and Information Systems, Ulm University, Manfred Reichert, Information Systems Group, University of Twente, and Thomas Bauer, DaimlerChrysler Group Research and Advanced Engineering, GR/EPD, Ulm.

The former paper was presented by Fabrizio Riguzzi, who was neither of the authors. Also in this study, some extensions to BPMN notation were proposed: this time to address the concept of cost in terms of cost intervals and average costs. The latter paper, presented by Ralph Bobrik, suggested a visualization approach, which allows to create personalized process views based on well-defined, parameterizable operations.

The third session of the day, moderated by prof. Peter Dadam, was started by Steen Brahe from Danske Bank who shared his experiences in adopting BPM and SOA at the bank in the last four years. His study showed that SOA and BPM have proven to be of business value. All participants agree it is the right way to go, but it is very difficult to adopt SOA and BPM, for it requires technical skills, in-house experience/experts, educated, trained organization, and development process and governance.

The research question of the next paper by Jürgen Moormann and Diana Heckl from Frankfurt School of Finance and Management was “How can the business processes of banks get aligned to the business processes of their customers?” In his presentation, Moormann pointed out that the products offered by banks do not reflect the requirements of their clients very well. Their model helps to transform the banks from product-oriented into process-driven organizations.

For the third presentation of the session, the floor was given to Hajo Reijers again. He presented a very interesting study by himself, Monique Jansen-Vullers, Eindhoven University of Technology, and Michael zur Muehlen and Winfried Appl, Stevens Institute of Technology, entitled “Workflow Management Systems + Swarm Intelligence = Emergency Management Systems”. They pointed out that the way in which work items are assigned in workflow management systems falls short in certain situations, such as emergency management services, where tasks need to be carried out by resources immediately available but not necessarily qualified. Their findings show that assignment strategies inspired by stimulus/response models derived from swarm intelligence outperform the traditional assignment of tasks in ad-hoc organizations. In the end of the presentation, prof. Michael zur Muehlen even demonstrated in a simulation how S-R model was the most robust mechanism for DTA.

The last presentation of the first day was given by Fabian Stäber, Siemens Corporate Technology. Together with his colleague Jörg P. Müller, Clausthal University of Technology, he had studied a case for a generic service layer between peer-to-peer overlay and business application. A number of service layer components was evaluated with respect to gathered requirements. The evaluation framework provides one step towards enabling collaborative business processes to benefit from the self-organization and resilience of decentralized peer-to-peer systems.

The second day started with a keynote by Steve Tieman of Estée Lauder. His presentation was entitled “Business Process Management: Closing The Gap Between Rigor And Relevance”. He pointed out that there is a gap between the rigor of academic research on BPM and its practical application in business and recounted the Strategic Modernization Initiative (SMI) at Estée Lauder, in which BPM and SOA are rigorously being employed on an enterprise-wide scale, thereby closing this gap beteen rigor and relevance.

The scope of their early BPM implementation of product lifecycle management was impressive indeed. Its organizational scope spanned 50 organizations, 25 brands, 36 affiliates and over 5600 concurrent projects. The undertaking had started with identifying a high level event network representing the level of understanding by the senior management. On the SOA side, major improvement in cost of goods was achieved through establishing an item master file and defining bills of material, recipes, component specifications, artwork mechanicals and direct material sourcing/contracts. For the BPM part, major improvement in speed to market was achieved through specifying 38 workflow-driven major sub-processes, common KPIs, common back-office organizations and common measurement and reporting platform. Major productivity and accuracy improvement was also achieved through automation.

Mr. Tieman identified a few challenges to further leverage BPM and SOA. Implementing BPM within creative organizations is challenging. Also, while the collaborative approach is promising, it is too slow based on the current organizational structure due to “hand offs”. Tieman posed a question, whether BPM could actually be used to drive changes to organizational structures.

The Strategic Modernization Initiative (SMI), currently underway at Estée Lauder, is a step change in their BPM/SOA endeavors. It involves ten thousand people and suppliers and is the key enabler to support sales growth to $10 billion and beyond. The SAP Netweaver based solution will be their first implementation of SAP. The project also makes use of ARIS as the modeling tool and WebMethods as the middleware.

The business drivers for the initiative are mainly to improve efficiency:
- reduce the number of hand-offs
- minimize time spent on compliance activities
- shorten time and effort by reusing “libraries” and standardized templates
- implement automated flow processes
- reduce waste and rework by implementing clear roles, responsibilities and processes
- minimize tactical tasks

In the project, five major business process groupings are being standardized. Existing leading practice models, e.g. SCOR, are used as the starting point and adapted to the company using ARIS. The adapted 114 primary processes are integrated to SAP using ARIS EPC methodology integrated with Solution Manager. These processes are then subdivided into sub-processes that, in turn, are normalized to services. Process steps in these 680 fundamental integrated processes are linked to transactions in Solution Manager and it is ensured that a transaction is limited to a single process step. Tieman also called for measuring KPIs on the transaction level.

He concluded that existing SAP implemetations are poorly documented and increased BPM rigor is anticipated as a requirement for success. According to him, it is difficult to get senior management’s “buy in” for longer than a “one time” effort; rigorous BPM implementation may require another “Y2K” or “SOX”.

After the keynote presentation and the coffee break, I changed to the practitioners day track to hear two presentations. The first one was given by Bryan Frew of Commonwealth Bank Australia, Sydney, who illustrated a hypothetical day in the life of a process owner in the desired future state of the bank. He maintained that the critical processes of the bank, running across product/functional lines, must be managed end-to-end by senior, experienced and capable staff to achieve their Customer Service aspirations — this will not be achieved through process improvement projects. Real-time access is needed to process performance metrics to maintain excellent levels of customer service. Objectives and strategies must be linked to the processes that will deliver. These processes must be proactively and collaboratively managed. To this end, executives and staff need to be coached in the principles and practices of BPM until they become self-sufficient.

The second presentation in this session was probably the best one in the entire conference. Gavin Keeley, Executive GM, Solutions, Suncorp, talked about “Tricks, Traps and Cul-de-sacs — Lessons from the Front Line”. He started with a provocative statement: “SOA is dead.” He maintained that there can not be a one-size-fits-all central orchestrator, but at least two orchestrators are needed: one for business orchestration and one for technology orchestration. Also, more than one process engine should co-exist in an organization based on its LOB infrastructure.

Keeley called for version control of processes, support for subroutines, adequate integration facilities, high performance and ability to expose the processes to the outside world. These features have markedly been lacking from the early BPM engines, yet things are improving. As “advanced traps” he mentioned race conditions, voting and partial completion, and correlated events sequencing all of which need to be taken into account in full-fledge BPM implementations.

He also emphasized the important role of people in processes. There is no “Commit Universe — Rollback Everything” mechanism, and so parts of compensation need to be taken care of by what he calls Human Process Engine (HPE).

In the end of his presentation, Keeley underlined the importance of “managing the white space”, as cross-organizational processes inherently cause political friction. Reward systems should be aligned with the processes rather than functions to overcome this friction. Despite his critical remarks on BPM, he concluded on a somewhat optimistic note that there is a prize to be gained through BPM, which is agility and distinction between process and IT. Yet “buyer beware”.

After the lunch break, I moved back on the main conference track. Session 5 was about Process Configuration and Execution and moderated by Joerg Desel. The session featured these four papers:

“BPEL Light” by Joerg Nitzsche, Tammo van Lessen, Dimka Karastoyanova, and Frank Leymann, Institute of Architecture of Application Systems, University of Stuttgart. The authors introduced a “WSDL-less” interaction model for BPEL in which the process logic is decoupled from the activity definition with the mechanism. As any interface description language (WSDL, WSMO, OWL-S…) is allowed, reusability and flexibility are increased. Also, mapping between BPMN and BPEL is eased.

“An Enactment-Engine Based On Use Cases” by Avner Ottensooser and Alan Fekete, School of Information Technologies, University of Sydney. The paper showed how one can control a workflow enactment engine based on the information which is available in written use cases.

“Requirements-Driven Design and Configuration Management of Business Processes” by Alexei Lapouchnian, University of Toronto, Yijun Yu, The Open University and John Mylopoulos, University of Toronto. They proposed a systematic requirements-driven approach for BP design and configuration management that uses requirements goal models to capture alternative process configurations.

“SAP WebFlow Made Configurable: Unifying Workflow Templates into a Configurable Model” (short paper) by Florian Gottschalk, Wil van der Aalst and Monique H. Jansen-Vullers, Eindhoven University of Technology. The authors suggested to combine different workflow templates into a single configurable workflow template and showed how a configurable workflow modeling language can be created by identifying the configurable elements in the original language.

After the break, John Deeb, Oracle, gave the afternoon keynote entitled “Business Process Blueprints — the Next Generation of Process Analysis to Execution”.

He started by saying that BPM pressures IT foundations as processes span people, systems and partners. Both human workflow and systems integration are required; the former is about automation of relationship and longer running workflows, whereas the latter is about automation of routine tasks and short running system tasks. In many ways, SOA provides the foundation for effective BPM: BPM, human-workflow, exception handling, collaboration between business and IT, complexity and organizational budget.

Because of the highly dynamic environment and constantly changing external influences, there is the need to change processes on a daily or weekly basis. To achieve such agility, BPM components must work together. Oracle’s strategy is to close the lifecycle loop with standards-based innovation and best-of-breed products to provide a complete BPM offering embracing Business Process Analysis (BPA), Business Process Execution (BPE) and Business Activity Monitoring (BAM).

Of these components, Deeb discussed BPE and BAM. Oracle’s BPEL Process Manager is a standards-based process modeling and execution environment with “comprehensive and native BPEL implementation”, “superior human-workflow functionality” and “industry’s most scalable and reliable BPM engine”. The market-leading process monitoring, for its part, allows monitoring of a single process as well as process aggregation with averages, KPIs, SLAs and identification of bottlenecks, and supports complex event processing.

The second day ended with a panel discussion “BPM 2010: Between Rigor and Relevance” attended by Steve Tieman, Barry Norton, Barbara Pernici, Michael zur Muehlen and Jamie Cornes (of Suncorp). When asked how BPM will look like in 2010, the most interesting answers were along similar lines:

Cornes: Ability to change business model in an agile way.

Norton: Business experts mash up the enteprise: rapid reconfiguration supported by semantic BPM

Pernici: Unpredictable business processes.

In the evening, there was a river tour and a conference dinner at Vino’s. The day ended on a pleasant note as lively conversation, good food and great wines took away my jet lag fatigue.

Thursday’s keynote presentation “Accelerating Scientific Knowledge Discovery through Scientific Workflow” was given by Shawn Bowers, Data and Knowledge Systems, UC Davis Genome Center. It was based on his joint work with Bertram Ludäscher and Timothy McPhillips.

He pointed out that as science is moving from few data and lots of thinking to lots of data and lots of analysis, there is a pronounced need for advanced technology for managing data and knowledge. So called scientific workflows are needed, because “raw” data is easier to collect, but harder to analyze. Collaboration is increasingly important, as complex analyses is today (mostly) performed manually.

Scientific workflows provide visual description and design of analytical processes with analytical, data types and data flow. They wire together diverse applications and tools for automation: data management, data processing, data visualization. Workflow components and entire workflows can be reused and repurposed.

Some characteristics of scientific workflows include that they are inherently complex, heterogeneous, semantically rich and “messy”. They are exploratory and dynamic, or collaborative, in nature — tasks between workflows often cannot be automated and next workflow to run may not be known (or exist) a priori. Computation models are data-driven; data is streamed, pipelined and shipped between multiple hosts, while maintaining data dependencies. This puts a lot of emphasis on design.

Bowers reviewed some typical models of computation used in scientific workflows:

PN (Process Network)
- actors are independent process threads
- buffered, unidirectional channels
- pipelined, asynchronous execution
- unbounded buffers (e.g. fast producers, slow consumers)
- loops ok (but beware of deadlocks)

SDF (Synchronous Data-Flow)
- actors declare token consumption and production rates
- static “firing” schedule (pos. integer solutions to balance equations)
- fixed buffer size
- loops ok (deadlocks statically detectable)

DAG (Directed Acyclic Graph)
- limited, special case of SDF (token rate is 1-in, 1-out)
- each actor fires exactly once (topological sort yields schedule)
- no loops

He also presented a new approach called Collection-Oriented Modeling & Design (COMAD) that embraces the assembly line metaphor. The data are tagged nested collections that are represented as flattened, pipelined (XML) token streams. Collection-oriented workflows explicitly group data tokens in the data flow. With a completely automatic pipeline, they enable concurrency with fewer shims and adapters taking scientific workflows back to clear, conceptual designs that are more change-resilient in the face of handling nested data collections.

Bowers concluded his presentation quoting E.O. Wilson’s notion of consilience — the unity of knowledge: “Literally jumping together of knowledge by the linking of facts and fact-based theory across disciplines to create a common framework for explanation.” Scientific workflows are crucial for “high-throughput” scientific analysis. They come with unique challenges from data-driven nature, complexity of scientific data, computation models, provenance and reproducibility. Opportunities can be found for new and applied research in the quest for the “right” scientific workflow design languages and supporting systems.

After the coffee break, prof. Wil van der Aalst moderated a session on “Formal Foundations of BPM”.

The first presentation of the session, “Behavioral Constraints for Services”, was given by Niels Lohmann, Institut für Informatik, Humboldt-Universität zu Berlin, on his work together with Peter Massuthe, Humboldt-Universität, and Karsten Wolf, Institut für Informatik, Universität Rostock. They provided a formal approach to express intended behavior of services as behavioral constraints with which unintended partners can be “filtered” yielding a customized operating guideline. These customized operating guidelines can be applied to validate a service and for service discovery.

The next paper was selected as the best paper of the conference: Kamal Bhattacharya, IBM T.J. Watson Research Center, Cagdas Gerede, UC Santa Barbara, Richard Hull, Bell Labs, Rong Liu, IBM and Jianwen Su, UC Santa Barbara: “Towards Formal Analysis of Artifact-Centric Business Process Models”. It was eloquently and vociferously presented by Jianwen Su. The paper presented a formal model for artifact-centric business processes coupling control and data, precursoring the development of a framework and tools that enable automated construction of processes.

The third presentation was given by Gero Decker, Hasso-Plattner-Institute, University of Potsdam. Together with Mathias Weske, he had studied “Local Enforceability in Interaction Petri Net”. Their paper introduced Interaction Petri Nets that represent global interaction models and provided algorithms for deriving the behavioral interface for each partner and for local enforceability checking.

The last paper presented in this session was the short paper “Modeling with History-Dependent Petri Nets” by Kees M. van Hee, Alexander Serebrenik, Natalia Sidorova, Marc Voorhoeve and Jan Martijn E.M. van der Werf (who also presented the paper), all from Eindhoven University of Technology. They proposed and evaluated a methodology for modelling processes by history-dependent Petri nets and showed how these nets can combine modelling comfort with analysability.

After the lunch break, Mathias Weske moderated a session on “Business Process Mining”.

Christian W. Günther presented a paper by him and Wim van der Aalst, Eindhoven University of Technology, entitled “Fuzzy Mining — Adaptive Process Simplification Based on Multi-Perspective Metrics”. They proposed a new process mining approach that provides meaningful configurable abstractions of operational processes to overcome the “spaghetti” of details in the discovered models.

The second paper was “Inducing Declarative Logic-Based Models from Labeled Traces” by Evelina Lamma (1), Paola Mello (2), Marco Montali (2), Fabrizio Riguzzi (1) and Sergio Storari (1) from University of Ferrara (1) and University of Bologna (2). It was presented by Fabrizio Riguzzi. They proposed an approach for the automatic discovery of logic-based models starting from a set of process execution traces.

Diogo Ferreira (1,3), Marielba Zacarias (2,3), Miguel Malheiros (3) and Pedro Ferreira (3) from Technical University of Lisbon (1), University of the Algarve (2) and Organizational Engineering Center, Lisbon (3), had authored “Approaching Process Mining with Sequence Clustering: Experiments and Findings”. They applied sequence clustering, a technique of bioinformatics, to process mining in cases where it is not known which event belongs to a given process and process instance.

The fourth paper was a short one by Sebastian Mauser, Robin Bergenthum, Jörg Desel and Robert Lorenz, Department of Applied Computer Science, Catholic University of Eichstätt-Ingolstadt: “Process Mining Based on Regions of Languages”. They gave an overview how to apply region based methods for the synthesis of Petri nets from languages to process mining.

The jetlag got better of me and I skipped the final afternoon session to get some rest before the river cruise in the evening. This was also a very pleasant event and again the food catered was really delicious.

On Friday 28th was the day of co-located events. I was presenting my paper “On the lookout for organizational effectiveness –- Requisite Control Structure in BPM governance” at the Workshop on BPM Governance, WoGo 2007. The workshop was chaired by Jude Fernandez, Infosys, and attended by about 15 people. Before my presentation, Fernandez welcomed the delegates and Tonia de Bruin gave a presentation on her research on BPM Governance at Queensland University of Technology.

My presentation could have gone a lot better. Very short sleep on the previous night and the lasting jetlag taxed my energies to the very limits. However, I got my message across and the few comments I got were not all that discouraging.

After my presentation, Jyoti Bhat, Infosys, gave a presentation on how BPM Governance has been employed at Infosys and Anibal Bustillo-Leal, EDS, suggested what role Enterprise Architecture and IT Governance would play in successful BPM Governance. After the lunch, there was also a panel discussion on BPM Governance, its challenges and way forward moderated by Jude Fernandez.