System and method for post transaction seasonal item recommendations

Patent metadata

Field	Value
Publication number	US-12482019-B2
Application number	US-202218092062-A
Country	US
Kind code	B2
Filing date	Dec 30, 2022
Priority date	Dec 30, 2022
Publication date	Nov 25, 2025
Grant date	Nov 25, 2025

Abstract

Systems and methods for post transaction seasonal item recommendations are disclosed. In some embodiments, a current seasonal time window associated with a seasonal event and some seasonal product types is determined. Based on historical transaction data of the seasonal product types, a first seasonal index score is computed for each item, and a second seasonal index score is computed for each product type including one or more items. A seasonal rank score is generated for each item based on the first seasonal index score and the second seasonal index score, such that the items in the historical transaction data are ranked based on their respective seasonal rank scores. Based on the ranked items and a transaction order from a user, a list of recommended items is generated and displayed to the user.

First claim

What is claimed is:

1 . A system, comprising:
a processor; and a non-transitory memory storing instructions that, when executed, cause the processor to:
based on a first frequency, generate a plurality of seasonal time windows each of which represents a time period associated with a respective seasonal event, wherein the plurality of seasonal time windows are generated by:
initializing, within a database, a set of cluster centroids based on at least one of the plurality of seasonal time windows and a search query; and
executing an iterative clustering algorithm that operates on each of the set of cluster centroids and converges each one of the set of cluster centroids to find a start bound and an end bound for each of a set of time bounds for each of the at least one of the plurality of seasonal time windows;

store the at least one of the plurality of seasonal time windows in thea database;
for each of the plurality of seasonal time windows, based on a second frequency, wherein the second frequency is greater than the first frequency:
obtain from the database catalog data for a plurality of products;
based on the catalog data, determine a plurality of seasonal product types that corresponds to a different seasonal time window of the at least one of the plurality of seasonal time windows;
obtain from the database historical transaction data, from a plurality of user transaction data stored in the database, of the seasonal product types;

execute an algorithm to operate on the corresponding seasonal time window, each item associated with the seasonal product types and the historical transaction data, and generate a first seasonal index score for each item in the historical transaction data;
execute the algorithm to operate on the corresponding seasonal time window, the seasonal product types, and the historical transaction data, and generate, for each product type in the historical transaction data, a second seasonal index score, wherein each product type includes one or more items in the historical transaction data;
generate a seasonal rank score for each item in the historical transaction data based on the first seasonal index score and the second seasonal index score, and store the seasonal rank score in the database;
generate a ranked item set of the one or more items in the historical transaction data based on their respective seasonal rank scores; and
store the ranked item set in the database, where in the ranked item set is associated with the corresponding seasonal time window;
during a current user session, receive a transaction order from a user device of a user, and store the transaction order in the database; responsive to receiving the transaction order:
determine a current seasonal time window based on the plurality of seasonal time windows and a current time;
extract, from the transaction order, a value of items in the transaction order;
determine a consideration intent of the user based on the value of items in the transaction order, the consideration intent identifying one of a plurality of intent groups;
select at least a portion of the ranked item set associated with the current seasonal time window as a final list of recommended items, personalized to the user, based on the transaction order and the consideration intent; and
transmit, to the web server in an acceptable data format, instructions that cause inclusion of one or more items from the final list of recommended items for display on at least one post transaction webpage to the user device;
during the current user session, receive an updated transaction order from the web server for the user device of the user, the updated transaction order comprising a subset of recommended items selected by the user from the final list of recommended items displayed to the user; and
in response to receiving the updated transaction order, update the transaction order within the database by adding the subset of recommended items into the transaction order, wherein the transaction order is executed to purchase the subset of recommended items.

2 . The system of claim 1 , wherein the processor executes the instructions to:
obtain historical interaction data including transaction data and search related data of users from at least last two years; filter the historical interaction data based on keywords, each of which is predetermined to be associated with at least one seasonal event; generate a frequency distribution of interactions across time based on the filtered historical interaction data; and for each of a plurality of seasonal events, determine a respective seasonal time window based on the frequency distribution of interactions and a pre-known date of the seasonal event.

3 . The system of claim 2 , wherein each of the plurality of seasonal events is at least one of: a season, a holiday, a festival, a sporting event, or a social event.

4 . The system of claim 1 , wherein the processor executes the instructions to:
aggregate the historical transaction data by week to generate weekly transaction data of the seasonal product types, wherein the historical transaction data is obtained from at least last one year; compute a first weekly seasonal index score for each item and each week based on the weekly transaction data, wherein each first weekly seasonal index score measures a percentage of transactions of a respective item within a respective week among total transactions of the respective item over one year; compute a second weekly seasonal index score for each product type and each week based on the weekly transaction data, wherein each second weekly seasonal index score measures a percentage of transactions of items in a respective product type within a respective week among total transactions of the items in the respective product type over one year; compute a weighted sum of the first weekly seasonal index score and the second weekly seasonal index score; and generate the seasonal rank score based on the weighted sum.

5 . The system of claim 4 , wherein the processor executes the instructions to:
compute a first weight associated with the first weekly seasonal index score for each item and each week, wherein the first weight represents a percentage of transactions of a respective item among total transactions of all items within a respective week; compute a second weight associated with the second weekly seasonal index score for each item and each week, wherein the second weight represents a percentage of transactions of a respective product type among total transactions of all product types within a respective week; and compute the weighted sum of the first weekly seasonal index score and the second weekly seasonal index score, based on the first weight and the second weight.

6 . The system of claim 4 , wherein the processor executes the instructions to:
compute a seasonal theme score for each item and each week based on search related data of users; and generate the seasonal rank score based on a weighted sum of: the first weekly seasonal index score, the second weekly seasonal index score and the seasonal theme score.

7 . The system of claim 1 , wherein the processor executes the instructions to:
filter items in each product type of the historical transaction data, to remove items whose transaction quantity is less than a threshold, wherein the threshold is determined based on a median quantity of transactions of the one or more items in the historical transaction data; and determine, based on the filtered items, top items ranked based on their respective seasonal rank scores.

8 . The system of claim 7 , wherein the processor executes the instructions to:
filter the top items to generate a first filtered list of items th

Assignees

• Walmart Apollo Llc

Inventors

• Vajge Parth Ramesh
• Ma Luyi
• Cho Hyun Duk
• Kumar Sushant
• Achan Kannan
• Lin Lawrence David

Classifications

• G06Q30/0277 Primary

  Online advertisement · CPC title

• G06Q30/0256

  User search · CPC title

• G06Q30/0631

  Recommending goods or services · CPC title

• G06Q30/0252 Primary

  based on events or environment, e.g. weather or festivals · CPC title