Bavaria supports dab research project on Process Mining
With Process Mining, companies can uncover many bottlenecks and problems in their digital processes and thus pave the way for improvements. However, the technology currently still has weaknesses, especially in the traceability of results and the identification of causes in complex business processes. Together with the University of Bamberg and the Technical University of Deggendorf, we want to develop an optimised AI and machine learning-based Process Mining approach. The funding commitment of the State of Bavaria from the "Information and Communication Technology" funding programme has now given the go-ahead for the research project.
Process Mining - Why is it worth using?
In recent years, Process Mining tools have found their way into more and more companies. With the progress of digital transformation, the importance of optimal digital business processes is increasing and this is exactly where Process Mining comes in.
The applications read digital event logs which systems generate in the course of business processes. Based on the data, they visualise the processes and thus make it easier for companies to identify problems and productivity hurdles. For example, the tools show which process steps take an unnecessarily long time or which processes are conspicuous for their high complexity.
Without question, Process Mining tools help optimising the efficiency and effectiveness of one's own processes. Instead of having to rely on assumptions and intuition, employees can use Process Mining to identify process obstacles much more quickly than before and make well-founded, data-based decisions. However, there are cases where the technology reaches its limits and does not yet provide satisfactory answers.
Valuable, but with limitations
Process Mining tools rely exclusively on event logs for their visualisations and analyses, which leads to a severe loss of information. This is because the event logs are essentially reduced to identification numbers for the respective process, names of the event and time stamps. User knowledge and contextual knowledge in the SAP system are not taken into account. How is this noticeable? Here are a few examples from practice:
- When analysing a production process, Process Mining tools cannot capture what triggered the production - A sales order? An event in the production plan? - and which ERP module the event is assigned to.
- Tools on the market have problems with more complex process hierarchies. For example, if an order is part of a production process, the presentation is currently not transparent enough for the user.
- If an event triggers several subsequent events, Process Mining tools reach their limits. For example, they do not always recognise when orders are delivered in different consignments of goods or display parallel deliveries sequentially.
- Some documents, especially in accounting, are automatically generated by the ERP system within fractions of a second. Process Mining tools do not capture this minimal time difference and assume simultaneity of events.
For humans, a glance is often enough to correctly assess causalities, hierarchies and chronologies. Technology - until now - lacks the contextual knowledge to do this. The problem: errors in the analysis results of the applications are not always obvious.
How do our customers benefit from this?
Together with the University of Bamberg and the Technical University of Deggendorf, we have therefore initiated a research project called "KIGA (KIGA - AI-supported Business Process Analysis)". With financial support, we will work on improving Process Mining algorithms and the transparency of their results over the next three years.
We want to bundle the background knowledge from ERP systems and from subject matter experts into knowledge graphs using methods of explainable artificial intelligence. By giving Process Mining tools access to this knowledge, they will be able to better "assess" the event logs and take contextual knowledge into account in their analysis. At the same time, interactive dashboards will make the results more transparent and easier to understand for the user.
We are pleased to face these challenges with two strong partners from science. Our goal is to achieve concrete, practical results that will ultimately benefit our customers. Will we be able to fulfil this claim? We will keep you up to date on our smaller and larger milestones in our blog.
If you would like to support our research and be one of the first companies to test the new technology in practice, get in touch with us. We would be pleased to have you join our research panel as a partner company.