US 11,886,509 B2Granted

Predictive prompt generation by an automated prompt system

AssigneeOptum Services (Ireland) Limited

InventorsGregory Buckley, Jack Sullivan, Damian Kelly, Mariah Sonja Pereira Penha, Bruno Ohana

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods are configured for predictive prompt generation for an interaction between a party and an automated prompt system. In various embodiments, metadata is received on the interaction and provided as input to a multi-label predictive model to generate interaction probabilities for a plurality of prompt information data objects. Each probability generated by the predictive model provides a likelihood a particular information data object needs to be provided to the first party during the interaction. Accordingly, one or more of the prompt information data objects are identified based on the probability of each piece found in the one or more prompt information data objects that satisfy a set threshold and one or more notifications are provided so that the automated prompt system provides at least one of the prompt information data objects to the first party during the interaction.

Claims20

Dependent claims

A computer-implemented method comprising:

generating, by one or more processors, based at least in part on metadata associated with an interaction between a party and an automated prompt system and 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
generating the plurality of prompt question data objects comprises:
generating, by the one or more processors, based at least in part on a plurality of question data objects associated with one or more historical interactions, and utilizing an embedding model, a vector representation for each question data object of the plurality of question data objects;
generating, by the one or more processors, based at least in part on each vector representation for the plurality of question data objects and utilizing a clustering model, a plurality of question data object clusters;
for one or more selected question data object clusters of the plurality of question data object clusters, generating, by the one or more processors, a cluster-representative prompt question data object based at least in part on occurrence counts for a subset of the plurality of question data objects whose vector representations fall within the one or more selected question data object clusters;
and
generating, by the one or more processors, the plurality of prompt question data objects to comprise each cluster-representative prompt question data object for the one or more selected question data object clusters;
identifying, by the one or more processors, one or more selected prompt question data objects of the plurality of prompt question data objects, wherein 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
providing, by the one or more processors, at least one of the one or more selected prompt question data objects to the party during the interaction.

8 more elements

A system comprising memory and one or more processors communicatively coupled to the memory, the one or more processors configured to:

generate, based at least in part on metadata associated with an interaction between a party and an automated prompt system and 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
generating the plurality of prompt question data objects comprises:
generating, based at least in part on a plurality of question data objects associated with one or more historical interactions and utilizing an embedding model, a vector representation for each question data object of the plurality of question data objects;
generating, based at least in part on each vector representation for the plurality of question data objects and utilizing a clustering model, a plurality of question data object clusters;
for one or more selected question data object clusters of the plurality of question data object clusters, generating a cluster-representative prompt question data object based at least in part on occurrence counts for a subset of the plurality of question data objects whose vector representations fall within the one or more selected question data object clusters;
and
generating the plurality of prompt question data objects to comprise each cluster-representative prompt question data object for the one or more selected question data object clusters;
identify one or more selected prompt question data objects of the plurality of prompt question data objects, wherein 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
provide at least one of the one or more selected prompt question data objects to the party during the interaction.

8 more elements

One or more non-transitory computer-readable storage media including instructions that, when executed by one or more processors, cause the one or more processors to:

generate, based at least in part on metadata associated with an interaction between a party and an automated prompt system and 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
generating the plurality of prompt question data objects comprises:
generating, based at least in part on a plurality of question data objects associated with one or more historical interactions and utilizing an embedding model, a vector representation for each question data object of the plurality of question data objects;
generating, based at least in part on each vector representation for the plurality of question data objects and utilizing a clustering model, a plurality of question data object clusters;
for one or more selected question data object clusters of the plurality of question data object clusters, generating a cluster-representative prompt question data object based at least in part on occurrence counts for a subset of the plurality of question data objects whose vector representations fall within the one or more selected question data object clusters;
and
generating the plurality of prompt question data objects to comprise each cluster-representative prompt question data object for the one or more selected question data object clusters;
identify one or more selected prompt question data objects of the plurality of prompt question data objects, wherein 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
provide at least one of the one or more selected prompt question data objects to the party during the interaction.

8 more elements

Dependencies are inferred from the claim text.

Facts

Application: 17/818,558Filed 9 Aug 2022 · continuation
Priority: 25 Jun 2020
Published: US 2023/0153356 A118 May 2023
Granted: 30 Jan 2024
Term ends: Not estimatedThe source records don’t support an estimate.

Classification15 codes

G06F 16/90332
Query formulation: natural language query formulation or dialogue systems
G06N 3/0442
Recurrent networks, e.g. Hopfield networks: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
G06N 3/049
Architecture, e.g. interconnection topology: temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
G06N 3/09
Learning methods: supervised learning

Family

Application16/911,766Parent of this continuation · filed 25 Jun 2020
US 2023/0153356 A1Pre-grant publication · published 18 May 2023
US 11,455,340 B2Parent of this continuation
US 2021/0406316 A1Published application
US 11,886,509 B2This patent

From the US filing history. Foreign family members aren’t listed.