Abstract: Propensities of entities, comprising human users and virtual assistants (VAs), for various behaviors can be determined and used to facilitate managing interactions between entities. An interaction management component (IMC) can determine an aggregate propensity metric relating to a propensity of an entity to engage in a behavior based on a cross-correlation of respective propensity metrics relating to respective propensities of the entity to engage in respective behaviors. During an interaction between entities, including the entity, IMC can determine an action to perform to interact with the entity based on the aggregate propensity metric and a context determined for the interaction. The action can be one that is predicted to elicit a defined action by the entity in response to the action. During (or after) the interaction, IMC can update behavior attributes, context, and/or aggregate propensity metric associated with the entity based on actions performed during the interaction.

Abstract: Propensities of entities, comprising human users and virtual assistants (VAs), for various behaviors can be determined and used to facilitate managing interactions between entities. An interaction management component (IMC) can determine an aggregate propensity metric relating to a propensity of an entity to engage in a behavior based on a cross-correlation of respective propensity metrics relating to respective propensities of the entity to engage in respective behaviors. During an interaction between entities, including the entity, IMC can determine an action to perform to interact with the entity based on the aggregate propensity metric and a context determined for the interaction. The action can be one that is predicted to elicit a defined action by the entity in response to the action. During (or after) the interaction, IMC can update behavior attributes, context, and/or aggregate propensity metric associated with the entity based on actions performed during the interaction.