US 12,412,569 B2
Disambiguation of requests for digital assistant task execution
James Harrison Creager, Raleigh, NC (US)
Assigned to Truist Bank, Charlotte, NC (US)
Filed by Truist Bank, Charlotte, NC (US)
Filed on Nov. 30, 2022, as Appl. No. 18/072,558.
Prior Publication US 2024/0177709 A1, May 30, 2024
Int. Cl. G10L 15/18 (2013.01); G06F 3/0482 (2013.01); G06F 40/30 (2020.01); G10L 15/22 (2006.01); G10L 15/30 (2013.01)
CPC G10L 15/1815 (2013.01) [G06F 3/0482 (2013.01); G06F 40/30 (2020.01); G10L 15/22 (2013.01); G10L 15/30 (2013.01); G10L 2015/223 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A disambiguation system associated with a digital assistant, the disambiguation system comprising:
a processing device; and
a memory device that includes instructions executed by the processing device for causing the processing device to perform operations comprising:
receiving an utterance comprising an intent from a user, the intent defining a task executable by the digital assistant in response to the utterance;
determining, based on user data, a time frame during which the user has been in contact with the digital assistant;
determining that the time frame exceeds a time frame threshold;
in response to determining that the time frame exceeds the time frame threshold, adjusting a confidence score threshold to increase the confidence score threshold;
transmitting the utterance and the increased confidence score threshold to a natural language processing system;
receiving a plurality of predicted intents associated with confidence scores that exceed the increased confidence score threshold and a plurality of slot values determined by the natural language processing system based on the utterance, the plurality of predicted intents defining tasks executable by the digital assistant and the plurality of slot values providing information usable by the digital assistant to execute the tasks;
determining, based on a hierarchy model, at least one common ancestor for the plurality of predicted intents; and
generating a response for display to the user using the plurality of predicted intents, the plurality of slot values, and the at least one common ancestor, the response usable by the digital assistant to determine the intent.