The UML activity diagram records the logic of a single operation or method, a single user case, or a single business flow. To create a UML activity diagram, perform the following steps repeatedly.
Step 1: define the scope of the activity diagram. First, define the modeling target. Individual user case? Part of a user case? A business process that contains multiple user cases? A single method of a class? Once you define the scope of your drawing, you should add a label at the top of the graph to indicate the title and unique identifier of the graph. You may also want to include the time of the graph or the author name.
Step 2: Add start points and end points. Each activity chart has a start point and end point. Therefore, you must add them immediately. In UML essence (UML Distilled) (see references), Fowler and Scott consider the end point to be optional. Sometimes an activity is just a simple end. In this case, it is harmless to indicate that the only change is to an end point. As a result, when someone else reads your chart, he or she knows that you have considered how to exit these activities.
Step 3: add an activity if you are modeling a User case, introduce an activity to the main steps that each role (actor) sends (this activity may include the start step, add any steps for the system response to the start step ). If you are modeling a high-level business process, introduce an activity for each major process, typically a User case or user case package. Finally, if you are modeling a method, it is common to introduce this activity.
Step 4: change between adding activities my style should always exit an activity, even if it is switched to an end point. Once there are multiple transformations in an activity, you must mark each transformation accordingly.
Step 5: Add a decision point. Sometimes, you need to make a decision for the logic you are modeling. Some transactions may need to be checked or compared. Note that the use of decision points is optional.
Step 6: Find out where parallel activities are available. When there is no direct connection between the two activities and they must all end before the start of the third activity, they can run in parallel.