US 12,141,536 B1
Chatbot utterance routing in a provider network
Sopan Khosla, Sunnyvale, CA (US); and Rashmi Gangadharaiah, San Jose, CA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Mar. 16, 2023, as Appl. No. 18/122,301.
Int. Cl. G06F 40/284 (2020.01); G06F 40/126 (2020.01); G06F 40/35 (2020.01); G06N 20/00 (2019.01); H04L 51/02 (2022.01)
CPC G06F 40/35 (2020.01) [G06F 40/126 (2020.01); G06F 40/284 (2020.01); G06N 20/00 (2019.01); H04L 51/02 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
jointly training a service classifier, a set of auxiliary classifiers, and an utterance encoder based on a mixed service set of labeled chatbot utterance training examples to yield a trained service classifier; wherein the service classifier, the set of auxiliary classifiers, and the utterance encoder are jointly trained according to a masked cross-entropy loss and a contrastive loss;
receiving a particular chatbot user utterance;
using the trained service classifier to determine that the particular chatbot user utterance is in-scope and is intended for a particular service of a plurality of services in a provider network;
sending the particular chatbot user utterance to the particular service;
receiving a response to the particular chatbot user utterance generated by the particular service; and
sending the response.