Speed up data engineering for process mining in practice

You’ve learned about process mining during your courses, but how much do you know about creating the event log for process mining? In business, creating the event log required for process mining is one of the most time-intensive, most complex parts of a process mining project. At Konekti, we’ve built a platform that simplifies and speeds up the data preparation for process mining. In this context, we are offering graduation projects to investigate new techniques and methodologies to make the data preparation for process mining even faster and easier.

Who are we?

We’re an Eindhoven-based start-up, driven by entrepreneurs who have applied process mining in practice for more than five years and are now eager to bring the process mining world to a new level. To realize our vision, we need to grow the team with more people that understand process mining and are eager to apply their knowledge in practice.

Why work @ Konekti?

By joining Konekti, you become part of an ambitious team with a restless drive to make process mining scalable in business. You will be part of a small team with a strong vision, working on challenges that we’ve faced in the industry. At Konekti, you will get the opportunity to work on challenges that directly impact our core product. As an added bonus, you’ll be a stone-throw away from the university.

Possible Graduation Project Topics

Presented below are several prospective topics for graduation theses. The specific details of the graduation project and scope will be further developed in mutual agreement.

  1. Smart Sub-sampling for Enhanced Efficiency
    In light of the expansive data volumes employed in process mining and the complex data models that users build, we are always looking for ways to make Konekti run faster. This topic explores the strategic application of sampling techniques to reduce query processing times while ensuring the detection of data quality issues and modelling errors affecting event log quality.
  2. Measuring Data Quality on an Object-Centric Data Model
    Konekti leverages an object-centric data model to facilitate the data preparation process, offering enhanced flexibility. However, the complexities of these data models pose challenges for swift and accurate validation. This research topic investigates methodologies to assist users in monitoring the data quality of their object-centric data models.
  3. Crawling Relevant Event Data for OCED
    Schema information is not always available for data that a user wants to load into Konekti, especially when data is combined over multiple systems. This topic explores the possibility of ‘crawling’ through this data to find relevant timestamps and related tables efficiently, and how to employ this information to suggest subsequent actions to users.
  4. Integration of High- and Low-Level Granularity Event Data in OCED
    Traditional process mining assumes that the activities are roughly from the same level of granularity, but across different domains, we observe that this is not always the case. Consider, for example, data for task mining and process mining, or combining machine data with ERP data in the context of predictive maintenance. This topic explores how users can be guided in creating data models with different levels of granularity.
  5. Advanced Modelling Patterns for Objects and Activities in OCED
    In certain instances, the interpretation of an activity depends on the activities occurring (almost) at the same time. In other instances, the demarcation between object instances relies on specific activities. For traditional event logs, these patterns have been researched. This topic explores how to classify advanced modelling patterns for OCED, and how they can be used within Konekti to support users in modelling these advanced patterns.
  6. Leveraging LLM’s for Data Exploration in Process Mining
    Many prominent ERP systems have abundant online documentation that can be leveraged to identify the storage locations of pertinent process data.  Ask ChatGPT how EKPO and RSEG are related in SAP ECC and it will give a result, although not necessarily a correct one. This topic investigates the potential of LLMs to guide users in finding the relevant process data while maintaining a reasonable level of accuracy.
  7. Autogenerated Documentation of OCED Models
    Data preparation for process mining is currently a ‘black box’, only understandable for the data engineer that has built it. This topic investigates how to automatically create a readable report explaining the rationale behind the data model, such that other users can understand and review the model created.  


For more information, contact Dirk Fahland or Boudewijn van Dongen.


Leave a Reply