Refined process mining framework

Hi and welcome back. In today’s lecture, I would like to explain the refined process mining framework. So we can use this to categorize the different process mining activities that we can actually execute. And you’ve already seen the more generic process mining framework that relates the world, how it interacts with an information system, how the information system is modeled by using process models, and the data created by this information system, and then how process mining bridges the data with the models. However, we can detail this a bit further. So on the top of the refined process mining framework, we still have the world’s information system and the data that’s created.

But in the middle, we have three categories of process mining activities that we can execute, and on the bottom we have the process models that are used both to configure the information system but also that we can discover using process mining. And there are actually two types of process models. So let’s go in a bit more detail and explain how all these things are related. So on the top of the refined process mining framework we have the world, humans, all the services, and machines that interact with our information system. An information system has to record certain activities and events being executed. And we can distinguish two types of event data, historic event data and current event data.

So historic event data contains the cases that are completed, and you could call them archived. Current data are those cases that are still running and that we could still influence using our analysis. Then on the bottom of the refined process mining framework are the models. And we can divide them into two categories, the de facto models and the de jure models. The de facto models are descriptive, so we could have discovered them or observed them in another way. And we can distinguish them into three types of models.

We can have control flow models, relating how the activities should be executed, we can have data or rules– data rules– that distinguish or mention how particular data can influence other aspects, or we can discover a resource or organizational perspective on the process. Then we also have the same types of models as de jure models, which are actually the normative or prescriptive models. These were made when configuring the system. These are rules and regulations that need to be adhered to. So using these two types of models, we can do performance checking or compliance checking, for instance, or auditing. Then in the middle of this refined process mining framework are the several categories of activities that we can execute.

One of these categories is cartography where you describe history. So you’re taking the historic data and you’re creating de facto models. And cartography contains three types of activity. You can discover. So using the data alone, the output is a de facto model on either the control flow data or resource perspective. Using enhancement, we can enrich a model, which is either a de facto or a de jure model, using the historic data. And that, again, creates a new de facto model. Lastly, we can also diagnose or analyze a model in a bit more detail. The second category is auditing where you’re actually confronting the models with reality. And here we see four types of activities.

Detection uses current data and de jure models to provide online alerts, and this is important because there you can still act. It’s current data– active cases– so things can still be fixed. Compliance checking uses the historic data with the de jure models to see where rules are violated and also maybe why. Comparing uses both de jure and de facto models to compare them– where do they differ. And finally, promote means using de facto models. So the models that you discovered– you can update the de jure models. So change your rules and regulations to update with your new findings. So you could recommend a particular resource to work on a particular case because it’s close to the deadline.

So these are three of the main process mining categories that we distinguish. And you see how they relate to the different types of data that can be recorded and the different types of process models that exist. And I hope this gives you an understanding of how different processes mining activities exist and the different types of things they can do. However, this is not the final list of process mining activities that exist. There can be some more that don’t fit this framework, but it gives you an overview of the general idea. In the next lecture, I’ll show you what you need for particular process mining activities. So I hope to see again in the next lecture.