Process mining comes with a small vocabulary. Once you know these terms, every screen in the feature makes sense. This article defines the core concepts in plain language. For the bigger picture, see the "What is process mining?" article.
Process
A process is a discovered workflow - for example order handling or ticket resolution. You create one by giving it a name, choosing what to track, and pointing it at the inboxes whose conversations it should analyze. Tekst then mines that data and keeps the process up to date as new conversations arrive.
Case
A case is a single instance flowing through the process - usually one conversation from start to finish. When you choose how to track a process, you decide what counts as a case: each conversation, or each business entity you extract from your messages. The number of cases that take a particular route is its case frequency.
Activity
An activity is a single step in the process, such as "Order received" or "Refund issued". Each activity has a name and a frequency (how often it occurs). The map always begins at a Start point and finishes at a Stop point, with your activities in between.
Activities come in two kinds, classification and system. The "Classification vs system activities" article explains the difference.
The process map
The process map is the diagram at the heart of the feature. It shows:
- Nodes - the activities, drawn as boxes.
- Edges - the arrows between activities, representing a transition from one step to the next.
Each edge can display a metric, such as how long that transition typically takes or how many cases follow it. The map is interactive: you can select nodes and edges, zoom, and open a side panel with details and example cases.
Variant (path)
A variant, also called a path, is one distinct route through the process from Start to Stop. Real workflows rarely follow a single route, so a process usually has many variants - the "happy path" plus every variation. You can rank variants by how common or how fast they are, and use the process coverage control to focus on the variants that account for most of your cases.
Throughput time
Throughput time is how long things take. It is measured both per edge (the time for one transition) and overall (the time from a case's first step to its last). You can set a throughput time target for a process and track how many cases finish within it.
Where to go next
- To build a process, see the "Set up your first process" article.
- To read the diagram, see the "Read the process map" article.
- To work with routes, see the "Explore process variants" article.
0 comments
Please sign in to leave a comment.