Methods and apparatus for electronically providing item advertisement recommendations

Patent metadata

Field	Value
Publication number	US-11455656-B2
Application number	US-201916686577-A
Country	US
Kind code	B2
Filing date	Nov 18, 2019
Priority date	Nov 18, 2019
Publication date	Sep 27, 2022
Grant date	Sep 27, 2022

Abstract

This application relates to apparatus and methods for automatically determining and providing item advertisements to customers. In some examples, a computing device obtains transaction data identifying in-store and/or online transactions. The computing device determines a distribution of purchased items over a plurality of item categories based on the transaction data. The computing device generates factorization matrices based on applying a machine learning process to the distribution, and generates relevancy scores for the plurality of item categories based on the factorization. The computing device may then select or generate item advertisements for items associated with the item categories based on the generated relevancy scores. The selected item advertisements may be displayed to a customer, for example, on a website.

First claim

What is claimed is:

1. A system comprising:
a communications interface configured to communicate, over one or more communication networks, with a server configured to host an online platform that is accessible by a computing device of each of a plurality of customers of the online platform, and cause a computing device of one or more of the plurality of customers of the online platform to display one or more of a plurality of item advertisement recommendations;
a database storing transaction data of each customer of the plurality of customers based on data obtained from the computing device of each of the plurality of customers;
a memory having instructions stored thereon; and
one or more processors coupled to the communications interface, the database and the memory, the one or more processors being configured to execute the instructions to:
obtain, over one or more communication networks and from the server, online session data of the plurality of customers;
obtain, over one or more networks and from the database, transaction data of each customer of the plurality of customers, the transaction data characterizing a plurality of purchases and a category associated with each of the plurality of purchases;
for each customer of the plurality of customers, obtain the plurality of item advertisement recommendations based on transaction data of the customer;
train a machine learning process with a first portion of the online session data and a first portion of the transaction data and implement, by applying the trained machine learning process to at least a second portion of the transaction data, a first set of operations that generate, for each customer of the plurality of customers, mask data, the first set of operations including:
generating, for each customer of the plurality of customers, first data identifying (i) a plurality of categories associated with the plurality of purchases, and (ii) for each of the plurality of categories, a corresponding frequency of purchases, based on the second portion of the transaction data associated with each customer of the plurality of customers;
generating, for each customer, a plurality of relevancy scores for pairs of the plurality of categories based on the first data;
generating, for each customer, the mask data identifying a subset of categories from the plurality of categories based on the generated plurality of relevancy scores;

for each customer and based on the mask data of each customer, implement a set of item advertisement recommendation operations, the set of item advertisement recommendation operations including:
for a corresponding plurality of item advertisement recommendations, determining an associated item category;
for the corresponding plurality of item advertisement recommendations, determining whether the associated item category matches at least one of the categories of the subset of categories by comparing the associated item category to at least the portion of the corresponding mask data;
selecting each of the obtained plurality of item advertisement recommendations that are determined to have the associated item category that matches at least one of the categories of the subset of categories; and
in response to a corresponding computing device of the customer accessing the online platform, causing, by communicating with the server over the one or more communication networks, the online platform displayed on a corresponding computing device of the customer to include at least one of the selected item advertisement recommendations based on data indicating the selected one or more item advertisement recommendations.

2. The system of claim 1 , wherein the transaction data identifies in-store transactions.

3. The system of claim 1 , wherein the relevancy score for the pairs of the plurality of categories is based on a cosine similarity between each of the pairs of the plurality of categories.

4. The system of claim 1 , wherein generating, for each customer, the mask data identifying the subset of categories from the plurality of categories comprises determining whether each of the plurality of relevancy scores for the pairs of the plurality of categories is beyond a threshold, wherein the mask data identifies the pairs of the plurality of categories that have relevancy scores beyond the threshold.

5. The system of claim 1 , wherein generating the first data comprises generating a category matrix, where each entry of the category matrix identifies a distribution of purchases in a category of the plurality of categories, and wherein the computing device is further configured to apply a non-negative matrix factorization to the category matrix.

6. The system of claim 5 , wherein the one or more processors are configured to execute the instructions further to:
apply a smoothing algorithm to each entry of the category matrix.

7. The system of claim 1 , wherein the one or more processors are configured to execute the instructions further to:
transmit the mask data to a second computing device.

8. The system of claim 1 , wherein to the one or more item advertisement recommendation operations includes:
determining at least one item recommendation advertisement based on the mask data;
receiving online session data for a first customer; and
determining a performance value based on the at least one item recommendation advertisement and the online session data.

9. A computer-implemented method comprising:
obtaining, by a processor of a server, over one or more communication networks, and from a database storing transaction data of each customer of a plurality of customers based on data obtained from a computing device of each of the plurality of customers of an online platform, the transaction data characterizing a plurality of purchases and a category associated with each of the plurality of purchases;
obtain, by the processor of the server, over one or more communication networks, and from a second server, online session data of the plurality of customers;
for each customer of the plurality of customers, obtaining a plurality of item advertisement recommendations based on transaction data of the customer;
training, by the processor of the server, a machine learning process with a first portion of the online session data and a second portion of the transaction data and implementing, by the processor of the server and by applying the trained machine learning process to at least the transaction data, a first set of operations that generate, for each customer of the plurality of customers, mask data, the first set of operations including:
generating, by the processor of the server and for each customer of the plurality of customers, first data identifying (i) a plurality of categories associated with the plurality of purchases, and (ii) for each of the plurality of categories, a corresponding frequency of purchases, based on the second portion of the transaction data associated with each customer of the plurality of customers;
generating, by the processor of the server and for each customer, a plurality of relevancy scores for pairs of the plurality of categories based on the first data;

generating, by the processor of the server and for each customer, the mask data identifying a subset of categories from the plurality of categories based on the plurality of relevancy scores;
for each customer and based on the mask data of each customer, implementing, by the processor of the server, a set of item advertisement recommendation operations, the one or more item advertisement recommendation operations including:
for a corresponding plurality of item advertisement recommendations, determining an associated item category;
for the corresponding plurality of item adver

Assignees

• Walmart Apollo Llc

Inventors

• Ma Luyi
• Cho Hyun Duk
• Kumar Sushant
• Achan Kannan

Classifications

• G06Q30/0255 Primary

  based on user history · CPC title

• G06Q30/0251 Primary

  Targeted advertisements · CPC title

• G06Q30/0242

  Determining effectiveness of advertisements · CPC title

• G06Q30/0631

  Recommending goods or services · CPC title