After a selection tree is built, using the tree editor, it is
submitted for compilation and execution. Execution can be done
against a cross section of the customer data base for testing
or to get representative counts or it can be run against the
entire customer data base to select people for promotion. Execution
is controlled by a selection driver program called the Data
Presentation Layer (DPL). The DPL is built for each database.
It is responsible for gathering together, the data pertaining
to a customer and presenting it as a micro-database to the executable
form of the selection tree. Output can be generated by the selection
tree or by the DPL. It is possible for information to be passed
from the selection tree back to the DPL. This gives the user
considerable flexibility in integrating WinTree into existing
Figure 2 shows the contents of the node called RecentDecline.
This node's rule
tests if the person has declined a promotion since a specified
date. If so, the person becomes part of the set and will not
be promoted for this campaign. The action,
ELIMINATE, is a key word in the selection language. It makes
the customer unavailable to any other nodes that share an
Exclusive ancestor node with the node called RecentDecline.
An Exclusive node is a type of node. It helps to control the
flow of records through the selection tree. The comments
section holds both comments and local variable declarations.
Many selection trees may be combined
together to run as a single super tree. The combining together
is done by a higher level tree known as a control tree. The
terminal nodes of a control tree LINK to promotion trees. The
control tree may contain logic that governs how the individual
promotion trees are allowed to interact. Thus, with one pass
of the data, the selections for many promotions can be done.
Running promotions at the same time makes it easier to share
names without over-using them. Click here to view an
interactive simulation of WinTree