US 12,271,803 B2
Systems and methods for predicting and providing automated online chat assistance
Yu-Hsuan Kuo, San Jose, CA (US); and Venkata Ramana Nadimpalli, Fremont, CA (US)
Assigned to PAYPAL, INC., San Jose, CA (US)
Filed by PAYPAL, INC., San Jose, CA (US)
Filed on Feb. 27, 2023, as Appl. No. 18/175,194.
Application 18/175,194 is a continuation of application No. 16/665,709, filed on Oct. 28, 2019, granted, now 11,593,608.
Prior Publication US 2023/0222316 A1, Jul. 13, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. H04L 51/04 (2022.01); G06N 3/04 (2023.01); G06N 3/08 (2023.01); G06N 5/04 (2023.01); H04L 51/02 (2022.01); H04L 65/1069 (2022.01)
CPC G06N 3/04 (2013.01) [G06N 3/08 (2013.01); G06N 5/04 (2013.01); H04L 51/02 (2013.01); H04L 51/04 (2013.01); H04L 65/1069 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system, comprising:
a non-transitory memory; and
one or more hardware processors coupled with the non-transitory memory and configured to read instructions from the non-transitory memory to cause the system to perform operations comprising:
receiving, via a chat module, an utterance from a device during an online chat session;
providing the utterance to a first prediction model of the chat module, wherein the first prediction model is configured to classify the utterance as one of a first plurality of intent classifications based on a plurality of words in the utterance;
determining that the first prediction model fails to classify the utterance based on a first output of the first prediction model being associated with a confidence level below a threshold;
in response to determining that the first prediction model fails to classify the utterance, classifying, using an output of a second prediction model configured to classify the utterance as one of a second plurality of intent classifications based on the plurality of words, the utterance as a particular intent classification of the second plurality of intent classifications, wherein the first plurality of intent classifications and the second plurality of intent classifications is at least partially overlapping, and wherein the particular intent classification is not included in the first plurality of intent classifications; and
in response to determining that the second prediction model has succeeded in classifying the utterance that the first prediction model failed to classify, re-training the first prediction model of the chat module using training data generated based on the utterance and the output of the second prediction model.