Systems and methods for mitigating display of non-compliant information

Patent metadata

Field	Value
Publication number	US-12106350-B2
Application number	US-202117496676-A
Country	US
Kind code	B2
Filing date	Oct 7, 2021
Priority date	Oct 7, 2021
Publication date	Oct 1, 2024
Grant date	Oct 1, 2024

Abstract

Systems and methods including 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, perform: receiving user search queries and product items, the product items including metadata corresponding to product types; determining a data relationship between the user search queries and the product items based on a weighting factor, the data relationship including user search query nodes and product item nodes; determining a first vector representation for first ones of the product item nodes in the data relationship; determining a second vector representation for second ones of the product item nodes in the data relationship; grouping the first vector representation for first ones of the product item nodes and the second vector representation for second ones of the product item nodes into a product type group based on a proximity search; and modifying the metadata corresponding to the product types based on the product type group to mitigate a computing system from displaying non-compliant material to a 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, perform:
receiving user search queries and product items, the product items including metadata corresponding to product types, wherein receiving the product items comprises receiving the product items from a cataloging engine, the product types comprising errors from the cataloging engine, and the errors corresponding to one or more of the product items including a non-compliant product type;
determining a data relationship between the user search queries and the product items based on a weighting factor, the data relationship including user search query nodes and product item nodes;
determining a first vector representation for first ones of the product item nodes in the data relationship;
determining a second vector representation for second ones of the product item nodes in the data relationship;
grouping the first vector representation for first ones of the product item nodes and the second vector representation for second ones of the product item nodes into a product type group based on a proximity search;
modifying the metadata corresponding to the product types based on the product type group to mitigate a computing system from displaying non-compliant material to a user; and
in response to a product type of the product type group being the non-compliant product type, automatically performing a computer vision prediction on an image of a product corresponding to the product type to detect a misclassification in the product type group.

2. The system of claim 1 , wherein the weighting factor is based on user interaction information, the user interaction information comprising one or more of: product items selected by a user, product items added-to-cart by a user, and product items purchased by a user.

3. The system of claim 2 , wherein determining the data relationship comprises linking the user search query nodes and the product item nodes in a bipartite graph based on the user interaction information.

4. The system of claim 3 , wherein linking the user search query nodes and the product item nodes in the bipartite graph is further based on:
identifying a user search query node of the user search query nodes; and
linking one or more of the product item nodes to the user search query node based on a user interaction with the one or more of the product item nodes during a user session associated with a user search query corresponding to the user search query node.

5. The system of claim 3 , wherein determining the first vector representation further comprises:
identifying a user search query node of the user search query nodes;
determining a vector representation of product item nodes in the bipartite graph by vector generation; and
determining a centroid node of product item nodes with a same product type to represent product type, the first vector representation of the centroid node determined by averaging the vector representation of all product item nodes with the same product type.

6. The system of claim 1 , wherein the proximity search is at least one of a k-nearest neighbor search, or a cosine similarity search.

7. The system of claim 6 , wherein grouping the first vector representation for the first ones of the product item nodes and the second vector representation for the second ones of the product item nodes into the product type group comprises grouping the first vector representation and the second vector representation that are within a threshold cosine similarity.

8. The system of claim 1 , wherein modifying the metadata corresponding to the product types based on the product type group comprises:
identifying a product item from the product items, the product item including metadata corresponding to a first product type of the product types;
verifying the first product type based on the product type group; and
modifying the metadata to replace the first product type from the product item with the product type group.

9. The system of claim 8 , wherein replacing the first product type from the product item with the product type group comprises removing the product item from being displayed in response to a search query that corresponds to the first product type.

10. A method implemented via execution of computing instructions configured to run at one or more processors and configured to be stored at non-transitory computer-readable media, the method comprising:
receiving user search queries and product items, the product items including metadata corresponding to product types, wherein receiving the product items comprises receiving the product items from a cataloging engine, the product types comprising errors from the cataloging engine, and the errors corresponding to one or more of the product items including a non-compliant product type;
determining a data relationship between the user search queries and the product items based on a weighting factor, the data relationship including user search query nodes and product item nodes;
determining a first vector representation for first ones of the product item nodes in the data relationship;
determining a second vector representation for second ones of the product item nodes in the data relationship;
grouping the first vector representation for first ones of the product item nodes and the second vector representation for second ones of the product item nodes into a product type group based on a proximity search;
modifying the metadata corresponding to the product types based on the product type group to mitigate a computing system from displaying non-compliant material to a user; and
in response to a product type of the product type group being the non-compliant product type, automatically performing a computer vision prediction on an image of a product corresponding to the product type to detect a misclassification in the product type group.

11. The method of claim 10 , wherein the weighting factor is based on user interaction information, the user interaction information comprising one or more of: product items selected by a user, product items added-to-cart by a user, and product items purchased by a user.

12. The method of claim 11 , wherein determining the data relationship comprises linking the user search query nodes and the product item nodes in a bipartite graph based on the user interaction information.

13. The method of claim 12 , wherein linking the user search query nodes and the product item nodes in the bipartite graph is further based on:
identifying a user search query node of the user search query nodes; and
linking one or more of the product item nodes to the user search query node based on a user interaction with the one or more of the product item nodes during a user session associated with a user search query corresponding to the user search query node.

14. The method of claim 12 , wherein determining the first vector representation further comprises:
identifying a user search query node of the user search query nodes;
determining a vector representation of product item nodes in the bipartite graph by vector generation; and
determining a centroid node of product item nodes with a same product type to represent product type, the first vector representation of the centroid node determined by averaging the vector representation of all product item nodes with the same product type.

15. The method of claim 10 , wherein the proximity search is at least one of a k-nearest neighbor search, or a cosine si

Assignees

• Walmart Apollo Llc

Inventors

• Sun Ming
• Tong Weiqi
• Pattaniyil Nidhin
• Moretti Antonio Khalil
• Zaman Tanwir
• Huang Yiyi
• Yang Guang
• Yan Zheng

Classifications

• G06Q30/0627 Primary

  by specifying product or service characteristics, e.g. product dimensions · CPC title