Research

Joos' research is on the topic of process mining in the group of prof. dr. ir. W.M.P. van der Aalst. His current research interests include:

  1. Process mining in healthcare: costs are rising in healthcare and increasing process efficiency is one of the key changes to make to make the costs manageable. Applying process mining on the abundance of data available in hospitals, clinics and other medical institutions allows process improvement based on facts and data. Joos is one of the founding members of the Process Mining 4 Health Care initiative in which several researchers collaborate to apply and promote process mining in the healthcare setting. Joos is working in the Philips flagship 'Data science for Health' where he supervises 2 PhD students. For his research Joos is also actively collaborating with Maxima Medical Center, Netherlands Comprehensive Cancer Organisation (IKNL), Maastricht University Medical Center, Antoni van Leeuwenhoek - Nederlands Kanker Instituut, Leiden University Medical Center and others.
  2. Learning analytics: analysing how students learn can be used to improve efficacy and efficiency of MOOCs, but also physical lectures supported by technology (aka the 'flipped classroom'). By applying process mining on learning data insights in the learning process can be obtained, which have proven to be a valuable addition to the data mining results. This research is funded by the European Data Science Academy (EDSA) project.

Next to these research topics Joos is also involved in MOOC creation on the topic of process mining. The first process mining MOOC is the Coursera MOOC "Process Mining: Data Science in Action". Another MOOC is the FutureLearn MOOC "Process mining with ProM", which is a hands-on MOOC where you learn how you can apply process mining on your own data! The third MOOC "Process mining in healthcare" aligns with Joos' main research interest.
Note that all three courses are available online for free, so join now!
All three MOOCs are sponsored by the European Data Science Academy (EDSA) project.

I'm interested in combining process mining with:

  1. Operations research management
  2. Stochastics / Statistics
  3. Data mining / machine learning in general
  4. Extracting event logs from real life systems;

And of course, the following topics are also interesting:

  1. Investigating the interplay of the four quality dimensions (replay fitness, precision, generalization and simplicity);
  2. Improving ways how to convert data sources to the event log format XES (see XESame);
  3. Brainstorming about new and crazy (process mining) ideas.

I'm currently supervising 2 PhDs as daily supervisor:

1. Bart Hompes: Bart is looking into tackling process complexity found in healthcare processes by not considering process models as a necessity. Bart for instance developed a trace clustering technique that clusters similar cases together. The key feature here is that the number of clusters does not need to be specified beforehand and that the clusters can be unbalanced. Bart also developed a technique to find relations between performance measures (e.g. 'case duration') and context (e.g. 'when Joos executes activity repair'). Currently Bart is working on verifying SLAs on event logs.

2. Alok Dixit: Alok is tackling process complexity by making process discovery semi-supervised. By assisting the user in building a (sound) process model, the user's knowledge is combined with data observations. Alok demonstrated that this approach is very powerful by winning the BPI 2017 Discovery contest. Currently Alok is working on speeding up process-event log alignment calculations, using domain knowledge to fix data issues in the event log, and of course on improving his interactive user interface.

3. I'm also involved in the daily supervision of Collin Drent who is looking into `Smart maintenance' with Philips and the Operations Research Management and Operations Research Management and Logistics research groups at TU/e.

4. In the near future I will also start supervising a third PhD in the TACTICS project with VU Amsterdam (Hajo Reijers) and Lunet zorg. If you're interested then please check the vacancy (if closed/unavailable then the vacancy is filled).

Joos worked full-time on the CoSeLoG project during his PhD (2010-2014):
The CoSeLoG project aims to create a cloud infrastructure for municipalities. Such a cloud would offer services for handling various types of permits, taxes, certificates, and licenses. Although municipalities are similar, their internal processes are typically different. Within the constraints of national laws and regulations, municipalities can differentiate because of differences in size, demographics, problems, and policies. Therefore, the cloud should provide configurable services such that products and processes can be customized while sharing a common infrastructure. The CoSeLoG project aims at the development and analysis of such services. (From http://www.win.tue.nl/coselog/wiki/start (September 2010))
Within this project he tries to use process mining for discovering the current processes within municipalities, discover and configure the configurable process model(s) and extend process mining to compare similar processes running in a cloud infrastructure.

Joos started on May 3rd, 2010 as a PhD Student on the CoSeLoG project. The first year of his PhD was mainly filled with visiting all 10 participating municipalities. Per municipality 5 processes were investigated using process mining. After getting almost 50 datasets and the subsequent analysis and reporting back to the municipalities, the first year was over. In the second year Joos shifted his focus to process discovery of comprehensible process models.

PhD Defense

A tool I created for my Masters' Project that helps to map data sources (single files or database tables) to elements in the XES event log format. More information can be found at the Processmining.org XESame page.

Minor Activities

I'm also the administrator of the ProM forum, I like to help people.

The growing list of publications I (co) produced can be found on this page and/or at Google Scholar.