Each node data can be a group (i.e. a node "container") or not (i.e. an "atomic" node). It is assumed that a node data cannot dynamically switch between being a group or not. Each node data can have a list of references to "member" nodes. All node references are by node data key value.

This kind of model assumes the grouping is implicit from these references. At this time we do not support a model where the grouping information is explicitly provided by data structures different from the node data.

There are three categories of methods: updating, navigation, and modification.

The updating methods need to be called when there has been a change to the data, so that the model can be kept up-to-date. These methods include changes to the collection of "member" node keys: DoMemberNodeKeyAdded, DoMemberNodeKeyRemoved, DoMemberNodeKeysChanged.

The navigation methods support examining and traversing the graph. One such method is actually defined in the base interface, FindNodeByKey. This interface adds the following methods: GetIsGroupForNode, GetMemberNodesForGroup, IsMember, and IsMemberValid.

The modification methods are used to alter the graph. A number of these methods are defined in IDiagramModel, such as AddNode and RemoveNode. But this interface adds several methods: AddMemberNodeKey, RemoveMemberNodeKey, SetMemberNodeKeys.

Note that this interface is universal, because it can only assume the node data is of type System.Object. The corresponding methods in the generic model classes operate on and return a specific node data type.

Reference