Skip to content

Improving process discovery

Study level

Master of Philosophy

Honours

Faculty/Lead unit

Science and Engineering Faculty

School of Information Systems

Topic status

We're looking for students to study this topic.

Supervisors

Dr Sander Leemans
Position
Lecturer in Business Process Management
Division / Faculty
Science and Engineering Faculty

Overview

Nowadays, commercial and governmental organisations store lots of data related to their business processes. From such recorded data, process mining aims to gain insights in order for the process to be improved.

A first step in typical process mining projects is to discover a process model automatically from the recorded event log. Leading process discovery techniques currently provide several guarantees, such as soundness of the model and a good trade-off between including and excluding behaviour of the log in/from the model.

In this project, we aim to improve on these existing techniques for process discovery. That is, current techniques search for the most important behaviour in an event log (that is, sequence, exclusive choice, concurrency or loops) by considering which activities follow one another directly.

The most important behaviour is identified using graph techniques. In particular, the activities of the event log (that is, the different process steps) are divided in two subsets.

However, a smarter way to do this would be to consider all possible subsets and to see which division is the "best". In this project, you will research what "best" means and develop ways to use that in discovery algorithms.

The project will run in collaboration with RWTH Aachen University, Germany.

Research activities

In this project, we aim to:

  • determine ways to measure the "best" division of activities into subsets
  • implement the idea in a prototype
  • evaluate the idea on real-life event logs to establish whether the prototype can discover better process models than existing techniques.

Outcomes

All of the mentioned research activities are on the edge of current scientific knowledge. Therefore, we expect to publish the performed research in top conferences and journal articles.

Skills and experience

For this project, you will need basic knowledge of:

  • event logs
  • process models
  • process mining
  • programming (preferably in Java, but this is not essential).

Scholarships

You may be able to apply for a research scholarship in our annual scholarship round.

Annual scholarship round

Keywords

Contact

Contact the supervisor for more information.