Predictive prompt generation by an automated prompt system

An apparatus for predictive prompt generation for an interaction between a party and an automated prompt system, the apparatus comprising at least one processor and at least one memory including computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to:

• receive metadata associated with the interaction;
• generating, based at least in part on the metadata and by utilizing a trained multi-label predictive model, a plurality of interaction probabilities, wherein:
• each interaction probability of the plurality of interaction probabilities is associated with a particular prompt question data object of a plurality of prompt question data objects and provides a likelihood that the particular prompt question data object is needed to be provided to the party during the interaction,
• generating the plurality of prompt question data objects comprises:
• extracting, by utilizing an extraction model, a plurality of extracted question data objects from one or more historical interactions,
• generating, by utilizing an embedding model, a vector representation for each extracted question data object of the plurality of extracted question data objects,
• generating, based at least in part on each vector representation for the plurality of extracted question data objects and by utilizing a clustering model, a plurality of extracted question data object clusters,
• for each extracted question data object cluster, determining a cluster-representative prompt question data object based at least in part on occurrence counts for a subset of the plurality of extracted question data objects whose vector representations fall within the extracted question data object cluster,
• generating the plurality of prompt question data objects to comprise each cluster-representative prompt question data object for the plurality of extracted question data object clusters, and
• training the multi-label predictive model is performed based at least in part on training data describing, for each respective historical interaction, historical metadata for the respective historical interaction and one or more identified prompt question data objects for the respective historical interaction that are selected from the plurality of prompt question data objects, wherein:
• (i) the multi-label predictive model is configured to generate, for each input interaction, a plurality of output interaction probabilities with respect to a plurality of labels, and (ii) the plurality of labels correspond to the plurality of prompt question data objects;
• identify one or more selected prompt question data objects of the plurality of prompt question data objects, where each interaction probability for a selected prompt question data object of the one or more selected prompt question data objects satisfies an interaction probability threshold;
• and
• cause the automated prompt system to provide at least one of the one or more selected prompt question data objects to the party during the interaction.

A non-transitory computer storage medium comprising instructions for predictive prompt generation for an interaction between a party and an automated prompt system, the instructions being configured to cause one or more processors to at least perform operations configured to:

• receive metadata associated with the interaction;
• generating, based at least in part on the metadata and by utilizing a trained multi-label predictive model, a plurality of interaction probabilities, wherein:
• each interaction probability of the plurality of interaction probabilities is associated with a particular prompt question data object of a plurality of prompt question data objects and provides a likelihood that the particular prompt question data object is needed to be provided to the party during the interaction,
• generating the plurality of prompt question data objects comprises:
• extracting, by utilizing an extraction model, a plurality of extracted question data objects from one or more historical interactions,
• generating, by utilizing an embedding model, a vector representation for each extracted question data object of the plurality of extracted question data objects,
• generating, based at least in part on each vector representation for the plurality of extracted question data objects and by utilizing a clustering model, a plurality of extracted question data object clusters,
• for each extracted question data object cluster, determining a cluster-representative prompt question data object based at least in part on occurrence counts for a subset of the plurality of extracted question data objects whose vector representations fall within the extracted question data object cluster,
• generating the plurality of prompt question data objects to comprise each cluster-representative prompt question data object for the plurality of extracted question data object clusters, and
• training the multi-label predictive model is performed based at least in part on training data describing, for each respective historical interaction, historical metadata for the respective historical interaction and one or more identified prompt question data objects for the respective historical interaction that are selected from the plurality of prompt question data objects, wherein:
• (i) the multi-label predictive model is configured to generate, for each input interaction, a plurality of output interaction probabilities with respect to a plurality of labels, and (ii) the plurality of labels correspond to the plurality of prompt question data objects;
• identify one or more selected prompt question data objects of the plurality of prompt question data objects, where each interaction probability for a selected prompt question data object of the one or more selected prompt question data objects satisfies an interaction probability threshold;
• and
• cause the automated prompt system to provide at least one of the one or more selected prompt question data objects to the party during the interaction.