US 12,259,928 B2
Named entity recognition in search queries
Xiang Cheng, Atlanta, GA (US); Mitchell Bowden, Irving, TX (US); Priyanka Goyal, Atlanta, GA (US); and Bhushan Ramesh Bhange, Irving, TX (US)
Assigned to Home Depot Product Authority, LLC, Atlanta, GA (US)
Filed by Home Depot Product Authority, LLC, Atlanta, GA (US)
Filed on Jan. 3, 2024, as Appl. No. 18/403,230.
Application 18/403,230 is a continuation of application No. 17/173,937, filed on Feb. 11, 2021, granted, now 11,880,411.
Claims priority of provisional application 62/975,538, filed on Feb. 12, 2020.
Prior Publication US 2024/0184830 A1, Jun. 6, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 16/903 (2019.01); G06F 16/954 (2019.01); G06F 18/214 (2023.01); G06F 40/295 (2020.01); G06N 3/045 (2023.01)
CPC G06F 16/90335 (2019.01) [G06F 16/954 (2019.01); G06F 18/2148 (2023.01); G06F 40/295 (2020.01); G06N 3/045 (2023.01)] 17 Claims
OG exemplary drawing
 
1. A method for responding to a search query, the method comprising:
defining a machine learning algorithm to create a trained model, the machine learning algorithm configured to receive a user search query as input and to output zero or more named entities of one or more entity types in the user search query, each of the one or more entity types comprising a respective plurality of values;
receiving the user search query;
defining a first data set according to user behavior data, the first data set comprising a plurality of first data pairs, each data pair comprising the user search query and a one or more defined named entity values in the user search query;
defining a second data set by creating a plurality of artificial second data pairs, each second data pair comprising an artificial search query and one or more defined named entity values in the artificial search query;
recognizing one or more named entity values in the user search query based on the first data set and the second data set; and
outputting a response to the user according to the one or more recognized named entity values,
wherein the machine learning algorithm comprises:
a character-to-word layer comprising a bidirectional long short-term memory network, and
a word-to-label layer comprising a bidirectional gated recurrent unit network.