Defining arbitration

Each decision can produce zero or more outcomes (true facts). The collection of outcomes is unordered.

Depending on your business process you can order them, limit the number returned, and add or modify their attributes. Typically used to assign a machine learning score to an outcome. Arbitration allows you to apply these types of processes in any order. For example, you want to order the outcomes based on an attribute you have defined named priority. Select the highest 5 outcomes and use a formula to add a score attribute to the selected items. Then sort them based on score and return the top 2 outcomes. Arbitration can be as simple as you need or as complex as your business process requires. If no arbitration is specified then all outcomes are returned in the order in which they were processed.

As your arbitration increases in complexity, you can end up with an empty set of outcomes.

Each arbitration function can have zero or more parameters that can be configured. Some parameters may reference data dictionary fields or outcome attributes. In-context help provides details on how to configure each function.