Systems and methods for personalizing search engine recall and ranking using machine learning techniques

Patent metadata

Field	Value
Publication number	US-12182215-B2
Application number	US-202318222807-A
Country	US
Kind code	B2
Filing date	Jul 17, 2023
Priority date	Jan 30, 2021
Publication date	Dec 31, 2024
Grant date	Dec 31, 2024

Abstract

A system can include one or more processors and one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising: receiving, at a search engine, a user search query submitted by a user; generating, using a recall personalization model, a simulated query that supplements the user search query with a feature vector reflecting personalization preferences of the user; and generating, using the search engine, search results for the user search query based, at least in part, on the simulated query that accounts for the personalization preferences of the user. Other embodiments are disclosed herein.

First claim

What is claimed is:

1. A system comprising:
one or more processors; and
one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors, cause the one or more processors to perform operations comprising:
receiving, at a search engine, a user search query submitted by a user;
generating, using a recall personalization model, a simulated query that supplements the user search query with a feature vector reflecting personalization preferences of the user, wherein the recall personalization model utilizes an attribute selection component associated with the feature vector; and
the recall personalization model is a machine learning model configured to convert the user search query into the simulated query; and

generating, using the search engine, search results for the user search query based, at least in part, on the simulated query that accounts for the personalization preferences of the user.

2. The system of claim 1 , wherein:
the user search query identifies an item type category without additional descriptors, wherein the user search query comprises a generic search query; and
the search engine further supplements the user search query with contextual features.

3. The system of claim 1 , wherein the operations further comprise:
generating, using the recall personalization model, the feature vector comprising contextual features associated with the user, wherein generating the feature vector comprises:
appending the feature vector with one or more attribute keys associated with a brand attribute.

4. The system of claim 1 , wherein the operations further comprise:
generating, using the recall personalization model, the feature vector comprising contextual features associated with the user, wherein generating the feature vector comprises:
appending the feature vector with one or more attribute keys associated with a price band attribute.

5. The system of claim 1 , wherein:
the attribute selection component is configured to determine a number of attribute keys to be incorporated into the feature vector comprising contextual features associated with the user; and
the simulated query, as converted, is configured to account for the personalization preferences of the user, wherein the contextual features associated with the user are derived from historical data associated with the user, and wherein the simulated query comprises a simulated narrowing query.

6. The system of claim 5 , wherein the attribute selection component comprises a statistical model that is configured to determine the number of attribute keys to be incorporated into the feature vector associated with the user.

7. The system of claim 6 , wherein the statistical model determines the number of attribute keys based, at least in part, on a number of the personalization preferences of the user.

8. The system of claim 6 , wherein the statistical model determines the number of attribute keys based, at least in part, on preference scores associated with attribute values of the number of attribute keys.

9. The system of claim 1 , wherein:
a plurality of attribute keys are pre-computed for the user;
each of the plurality of attribute keys comprises a respective attribute value for the user and a respective preference score for the user;
the respective preference score indicates a respective affinity of the user for the respective attribute value; and
the plurality of attribute keys are incorporated into the feature vector associated with the user.

10. The system of claim 1 , wherein:
the search engine is configured to communicate with a personalized ranking model; and
the personalized ranking model is configured to sort a recall set of the search results based on the personalization preferences of the user.

11. A method implemented via execution of computing instructions configured to run at one or more processors and stored at non-transitory computer-readable media, the method comprising:
receiving, at a search engine, a user search query submitted by a user;
generating, using a recall personalization model, a simulated query that supplements the user search query with a feature vector reflecting personalization preferences of the user, generating, using a recall personalization model, a simulated query that supplements the user search query with a feature vector reflecting personalization preferences of the user, wherein the recall personalization model utilizes an attribute selection component associated with the feature vector; and the recall personalization model is a machine learning model configured to convert the user search query into the simulated query; and
generating, using the search engine, search results for the user search query based, at least in part, on the simulated query that accounts for the personalization preferences of the user.

12. The method of claim 11 , wherein:
the user search query identifies an item type category without additional descriptors, wherein the user search query comprises a generic search query; and
the search engine further supplements the user search query with contextual features.

13. The method of claim 11 further comprising:
generating, using the recall personalization model, the feature vector comprising contextual features associated with the user, wherein generating the feature vector comprises:
appending the feature vector with one or more attribute keys associated with a brand attribute.

14. The method of claim 11 further comprising:
generating, using the recall personalization model, the feature vector comprising contextual features associated with the user, wherein generating the feature vector comprises:
appending the feature vector with one or more attribute keys associated with a price band attribute.

15. The method of claim 11 , wherein:
the attribute selection component is configured to determine a number of attribute keys to be incorporated into the feature vector comprising contextual features associated with the user; and
the simulated query, as converted, is configured to account for the personalization preferences of the user, wherein the contextual features associated with the user are derived from historical data associated with the user, and wherein the simulated query comprises a simulated narrowing query.

16. The method of claim 15 , wherein the attribute selection component comprises a statistical model that is configured to determine the number of attribute keys to be incorporated into the feature vector associated with the user.

17. The method of claim 16 , wherein the statistical model determines the number of attribute keys based, at least in part, on a number of the personalization preferences of the user.

18. The method of claim 16 , wherein the statistical model determines the number of attribute keys based, at least in part, on preference scores associated with attribute values of the number of attribute keys.

19. The method of claim 11 , wherein:
a plurality of attribute keys are pre-computed for the user;
each of the plurality of attribute keys comprises a respective attribute value for the user and a respective preference score for the user;
the respective preference score indicates a respective affinity of the user for the respective attribute value; and
the plurality of attribute keys are incorporated into the feature vector associated with the user.

20. The method of claim 11 , wherein:
the search engine is configured to communicate with a personaliz

Assignees

• Walmart Apollo Llc

Inventors

• Cho Hyun Duk
• Kumar Sushant
• Achan Kannan
• Sinha Nimesh
• Inan Aysenur

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06Q10/067

  Enterprise or organisation modelling · CPC title

• G06Q30/0633

  Managing shopping lists, e.g. compiling or processing purchase lists (shipping orders G06Q10/083; order filling G06Q10/087) · CPC title

• G06N20/00

  Machine learning · CPC title