Terminologies in IBM Process Mining

Process

A business process is a collection of activities or tasks that are related to a specific service or product to serve an established goal of a particular customer or customers.

Depending on its complexity, a process can be categorized as follows:

• Flat: When the activities are related to one main process entity, for example, an order or a ticket.
• Multilevel: When the activities are related to multiple main process entities, with possible complex many-to-many relations between each other. For example, in a P2P process, an order can refer to multiple requisitions or might be received in multiple receipts, an invoice can refer to multiple orders.

IBM Process Mining automatically derives a process model from the data. When you create a process in IBM Process Mining, the following mandatory fields are required at the Data Mapping step:

• Process ID
  The Process ID acts as a distinctive identifier of every new instance. Depending on the process type, you can choose one element to be the Process ID, for example, it can be an order number or a ticket number. In a multilevel process, you can insert multiple Process IDs.
• Activity
  The Activity indicates the list of activities that are related to the process IDs.
• Time
  The information on time for each activity, for example, the datetime when each activity is completed.

You can also provide information on the following parameters for more insight on the process:

• Activity end time
  Activity end time helps the system to provide an analysis of the activity service time, for example, the time that is required for an activity to complete.
• Resource and role
  You can include the information so that the system automatically provides an activity map and a social network analysis of the resources.
• Relevant data
  IBM Process Mining can use the information to extend and compute advanced statistics, provide root-cause analyses, and prompt the decision rules of the process.

Case

A case is an instance of a process.

In flat processes, each different process-id defines a new case. For example, in a ticketing process, you can relate case to a specific ticket number.

In multilevel processes, each different combination of process-ids defines a new case. For example, consider a process in which two orders are received at two different moments but are registered in a unique invoice, then it is considered as one single case.

Activity and event

An event occurs every time an activity of the process is performed.

Process variant

A process variant is a unique path that a case takes to cross the process (from the start to the end). Being able to analyze different variants is important to identify, for example, common deviations (regarding a normative model) and inefficiencies.