IEEE CIS Task Force on Process Mining

Trace: 2011_09_14_bpm


|

Minutes and slides of IEEE Task Force on Process Mining meeting at BPM 2011

The slides, used in this meeting, can be found on taskforceinfo-meeting-at-bpm2011.pdf

Minutes Meeting IEEE Task Force on Process Mining at BPM 2011 Thursday September 1st 2011, Clermont-Ferrand, Campus des Cezeaux

The annual Task Force meeting was well attended (about 30 people), especially considering the fact that many Task Force members had to leave before the meeting (which was after the conference).

1. Introduction of participants and Task Force

Wil van der Aalst introduced some basic facts about the task force which was established in 2009. Members include representatives from software vendors (Pallas Athena, IDS Scheer/Software AG, Futura Process Intelligence, HP, IBM, Infosys, Fluxicon, Businesscape, Iontas, Fujitsu, Business Process Mining), consultancy (Some of the above and ProcessGold, Business Process Trends, Gartner, Deloitte, Rabobank), and u niversities (TU/e, University of Padua, University of Catalunya, New Mexico State University, Technical University of Lisbon, University of Calabria, Penn State University, University of Bari, Humboldt-Universität, Queensland University of Technology, Vienna University of Economics and Business, Stevens Institute of Technology, University of Haifa, Seoul National University of Technology, Cranfield University, K.U.Leuven, Tsinghua University, Innsbruck University).

2. Goals of the Task Force and New Initiatives

The goal of this Task Force is to promote the research, development, education and understanding of process mining. Several examples of past activities are given (see slides). Wil van der Aalst encourages all present to initiate new initiatives in the network and the contribute to IEEE events such as the IEEE World Congress On Computational Intelligence and IEEE Symposium on Computational Intelligence and Data Mining. Moreover, the workshops on Business Process Intelligence at BPM 2010 (New York) and BPM 2011 (Clermont-Ferrand) are examples of successful events co-organized by the IEEE Task Force on Process Mining. Another example was the First International Business Process Intelligence Challenge (BPIC'11). It would be good to repeat this contest to progress the developments and improve maturity of the field.

3. Standardization efforts: XES format

Only briefly discussed. Last year the Task Force agreed upon the XES format, see www.xes-standard.org and www.openxes.org. The format works well and only minor limitations have been encountered. Members are asked to promote, use and support the format.

4. Process Mining Manifesto

The main part of the meeting was devoted to discussing Version 2 of the Draft text. The manifesto serves as a public declaration of principles and intentions by the IEEE Task Force on Process Mining.

The draft lists 6 guiding principles:

GP1: Event Logs Should Be Treated as First-Class Citizens
GP2: Log Extraction Should Be Driven by Questions
GP3: Concurrency and Choice Should be Supported
GP4: Events Should Be Related to Model Elements
GP5: Models Should Be Viewed as Maps
GP6: Process Mining Is a Continuous Process

The draft also lists 10 challenges:

C01: Finding, Merging, and Cleaning Event Data
C02: Dealing with Event Logs of Different Sizes and at Different Levels of Granularity
C03: Creating Representative Benchmarks
C04: Dealing With Concept Drift
C05: Finding a Suitable Representational Bias
C06: Balancing Between Multiple Objectives
C07: Cross-Organizational Mining
C08: Providing Operational Support
C09: Combining Process Mining With Other Types of Analysis
C10: Improving Usability for Non-Experts

These and other parts of the text were discussed in detail. Some of the points discussed where (thanks to Boudewijn for summarizing the discussion)

GP1: The term “logs” is misleading here. We agreed on saying that “the recording of events should be treated as a first class citizen”.

GP3: Here, we should say “Concurrency, choice and other constructs should be supported by process mining algorithms.”

GP5: Since models can be different on various levels of abstractions and they basically are views on the data, we should say: “Models should be treated as purposeful abstractions of reality”. In the text, it should be mentioned that these models are not necessarily graphical.

GP6: “Process Mining should be a continuous process”

C2: “Dealing with complex event logs having various characteristics” The term “real-life” should be mentioned in the text, but not in the title, where it should be explained that real-life logs typically have specific characteristics, i.e. large size, many attributes, many event types, many different cases etc.

C6: The question was raised whether we should replace “objectives” with some concrete examples, of if we should rephrase this to state that a balance between quality metrics should be found, specific to your purpose.

C10: “Improving usability for non-expert users”, i.e. implementations should be fast and work for a large variety of logs,

C11: “Improving understandability for non-expert users” (added challenge), i.e. by making systems with easy-to-understand parameters, fast algorithms and semi-deterministic algorithms.

Glossary/positioning: It was agree that a glossary should be included. This could help to clarify terms (see the discussion at the end of the current Manifesto). There is consensus that the term process mining should be used a positioned well with respect to other terms such as automated business process discovery. A Venn diagram will be used to show relationships between terms also broadening the scope to the wide BI/ data mining area (thanks to Marlon).

Audience: The intended audience for the manifesto is mixed ranging from vendors and researchers to end users and consultant. The goal is to first get a full version ignoring formatting issues. After this we also make a “macified” version for industry, as well as a single page brochure that Marcello will distribute door to door in Australia :-).

Also a lot of input was generated via the e-mail (>50 people provide input), both from IEEE Task Force members and people other organizations working on process mining. The challenge is to integrate all the comments.

Planning:

  • Approx. Sept 20th: Wil van der Aalst will compile a “final draft”. This will be distributed to fine-tune the text (no major changes after this point).
  • October 1st: freezing the content of the Manifesto.
  • October 15th: various versions of ready for distribution.

5. Closing

Wil thanks the participants and closes the meeting around 19.30.