	US 11,972,470 B2
	Systems and methods for identifying item substitutions
Richard Downey, San Francisco, CA (US); Shirish Dhar, Boston, MA (US); Adam Whybrew, New South Whales (AU); and Jurgen Hanekom, Andross (AU)
Assigned to The Boston Consulting Group, Inc., Boston, MA (US)
Filed by The Boston Consulting Group, Inc., Boston, MA (US)
Filed on Mar. 29, 2023, as Appl. No. 18/128,148.
Application 18/128,148 is a continuation in part of application No. 17/806,385, filed on Jun. 10, 2022, granted, now 11,669,882.
Application 17/806,385 is a continuation in part of application No. 17/514,434, filed on Oct. 29, 2021.
Application 17/514,434 is a continuation of application No. 17/015,863, filed on Sep. 9, 2020, granted, now 11,321,763, issued on May 3, 2022.
Claims priority of provisional application 63/362,351, filed on Apr. 1, 2022.
Claims priority of provisional application 63/362,274, filed on Mar. 31, 2022.
Claims priority of provisional application 63/003,527, filed on Apr. 1, 2020.
Prior Publication US 2023/0306493 A1, Sep. 28, 2023
Int. Cl. G06Q 30/00 (2023.01); G06F 18/214 (2023.01); G06N 3/08 (2023.01); G06Q 30/0283 (2023.01); G06Q 30/0601 (2023.01)

CPC G06Q 30/0631 (2013.01) [G06F 18/214 (2023.01); G06N 3/08 (2013.01); G06Q 30/0283 (2013.01); G06Q 30/0633 (2013.01)]

34 Claims

1. A computer-implemented method comprising:

collecting history information, wherein the history information comprises one or more episodes from one or more customers, wherein each episode comprises one or more items of a collection of items;

transforming the history information into standardized attribute information comprising one or more attributes comprising a matrix of observed substitutions, wherein an observed substitution comprises an alternative item;

training a neural network on the matrix of observed substitutions to generate item embeddings receiving input comprising an item, wherein the neural network comprises a loss function for optimizing a machine learning algorithm;

identifying a substitution similarity between the item and another item based on the item embeddings;

automatically generating a message comprising at least one item of the collection of items based on a subset of the one or more attributes; and

transmitting the message to the one or more customers over a network;

wherein automatically generating a message based on the subset of the one or more attributes further comprises:

defining a message recipe based on the subset of the one or more attributes;

ranking the collection of items based on a subset of the one or more attributes;

defining a message action based on the message recipe and a message template; and

automatically generating the message comprising the message action and the at least one item of the collection of items based on the ranking.