The offerTieBreakMethod property defines the behavior of offer serving when two offers have equivalent (tied) scores. If you set this property to its default value of Random,
Interact presents a random choice from among the offers that have equivalent scores. If you set this configuration to Newer Offer,
Interact serves up the newer offer (based on having a higher offer ID) ahead of the older offer (lower offer ID) in the case where the scores among the offers are the same.
Interact has an optional feature that allows the administrator to configure the system to return the offers in random order independent of the score, by setting the
percentRandomSelection option (
Campaign | partitions | [partition_number] | Interact | learning | percentRandomSelection). The
offerTieBreakMethod property described here is used only when
percentRandomSelection is set to zero (disabled).
The optimizationType property defines whether
Interact uses a learning engine to assist with offer assignments. If set to
NoLearning,
Interact does not use learning. If set to
BuiltInLearning,
Interact uses the baysean learning engine built with
Interact. If set to
ExternalLearning,
Interact uses a learning engine you provide. If you select
ExternalLearning, you must define the
externalLearningClass and
externalLearningClassPath properties.