Process Mining

DELIVERY:

Last updated May 14, 2025

Selecting the process mining type

You can select the process mining type for the process model to be used for the app.

Attention:

Discover process model and Import BPMN model introduce detailed semantic events for exclusive, parallel, and loop behaviors that enhance process precision but may affect app scalability. As a result, this might impact the maximum number of real events supported in a Process mining app.

1. Select the process mining type for your process app.

2. Select Next.

Directly follows graph

The "directly follows" relationship represents the order in which activities occur in a process. It is based on event logs that contain records of activities performed during the execution of a process. By analyzing the event logs, Process Mining techniques can identify which activities are typically executed one after another. Between these two activities, an edge is defined as A➝B.

If, for example, paths A→B and C→D in a process can be performed in parallel, the activities of the two paths can occur in the event log in any order. In the "directly follows graph" approach, these events will not be considered to run in parallel, and the resulting process flow can be the following: A➝C➝B➝D.

A Discover Process Model or a BPMN process model however will take parallel structures into account.

Discover process model

Discovering a process model allows you to have a better understanding of your process structure. By analyzing the whole process (the whole event log) with advanced process mining techniques, activities that happen in parallel, are part of a decision, or are part of a more complex loop are automatically discovered. These relationships are called the process semantics.

The following table describes the process semantic types.

Semantic node type	Description	Visual representation
Parallel gateway (AND)	Represents two or more branches that happen in parallel.
Exclusive choice gateway (XOR)	Represents an exclusive choice (decision point) between two or more branches.
Loop gateway	Represents a repeatable sub-process with one loop body path and one or more rework paths.

Import BPMN model

Select Import BPMN model if you want to import your own BPMN 2.0 model and use it as a process mining type for your process app.

Note:

With the BPMN model, advanced process algorithms will map your event log data on top of the BPMN model, allowing you to analyze how the BPMN model and your data relate.

Drag and drop the .bpmn file that contains the BPMN 2.0 model you want to import.
When the file is successfully imported, the Next button will be enabled.

Tip:

If you want to use a different BPMN model after you created the process app, you can import a BPMN model in Process manager.

BPMN model requirements

The following sections describe the requirements to successfully import a BPMN 2.0 model for use in Process Mining.

General requirements for the BPMN model

The BPMN must adhere to the BPMN 2.0 standard.
The BPMN must contain a single process definition.
The BPMN can only contain supported elements.
All nodes in the BPMN should have an id.
The BPMN model should contain no more than 999 nodes.

Supported BPMN elements

The following BPMN elements ares supported:

Start events
End events
Tasks
Exclusive choice gateways
Parallel gateways
Sequence flows

Task and flow requirements

Requirement	Example
A start event must be defined, and there can only be one start event.
An end event must be defined, and there can only be one end event.
At least one activity task must be defined.
There must be at least two sequence flows defined for the BMPM model.
The tasks must have unique labels.
All sequence flows must have a source and a target.
Each task must have a single incoming, and a single outgoing sequence flow.
The start event only has a single outgoing edge.
The end event only has a single incoming edge.
All nodes and sequence flows in the BPMN are connected in a single model.

Gateway requirements

The gateways in the BPMN must form pairs. Each pair has an opening gateway, which splits the process flow into multiple process flows. Each pair has a closing gateway, which joins the split process flows into a single flow. This structure is called a block. The flows inside a block may contain other, nested blocks. However, the only flows to enter and exit the flows within a block, are via the split and join gateways.

The following illustration shows an example of a BPMN model with gateway pairs, forming blocks. The blocks are highlighted.

Requirement	Example
Each gateway is either a split or a join gateway.
A split gateway has a single incoming edge, and at least 2 outgoing edges.
A join gateway has at least 2 incoming edges, and a single outgoing edge.
For every split gateway, there is a corresponding join gateway of the same type, and vice versa.
Every split-join gateway pair and the elements between them are a single-entry-single-exit component: a model component that has only a single entry edge and a single exit edge.
Each path from a split gateway its corresponding join gateway must also be a single-entry-single-exit.
There cannot be any direct sequence flows between corresponding splitting and joining parallel gateways.
Every exclusive choice split-join gateway pair that describes a loop flow to a previous point in the process cannot have an empty main path and an empty redo path, as this allows for indefinite looping without a task being executed.