Granted Patent B2

US 12,536,556 · App. 18/105,483 · Granted Jan 27, 2026

Inventor: SenthilKumar Gopal (Fremont, CA)

Assignee: EBAY INC.

G06Q30/0202G06N5/022G06Q10/087

Abstract

Claims (78)

1 . A method comprising:

training, by a training component of a machine learning system on a first server, a time series-based machine learning (ML) model that forecasts sales based on stored transaction histories;

determining, by an evaluation component of the machine learning system, one or more potential categories for a user based on custom preferences and previous analytic queries of the user;

applying, by the evaluation component, high demand items in the one or more potential categories to the ML model, the ML model outputting a predicted sales count and a probability of predicted sales for each of the high demand items in the one or more potential categories;

determining, by the evaluation component, items having a potential inventory gap by cross-checking current inventory with a list of one or more items having a probability outputted by the ML model that satisfies a probability threshold;

for each item that has a potential inventory gap, causing presentation, on a user interface, of competitor sales data, the predicted sales count and the probability derived from the ML model, and editable fields displaying automatic listing thresholds that trigger automatic generation of a corresponding listing;

based on the automatic listing thresholds being satisfied, triggering, by the evaluation component, a listing generator of a publication system on a second server to automatically generate the corresponding listing;

continually storing new transactions, the new transactions indicating an actual sales count;

retraining the ML model based on the new transactions and the actual sales count to account for ongoing trends;

validating, by the evaluation component, predictive accuracy of the ML model by comparing a predicted sales count outputted by the ML model with the actual sales count; and

tuning, by the evaluation component, one or more hyperparameters of the ML model based on the validating to improve generalization and the predictive accuracy of the ML model.

2 . The method of claim 1 , wherein the editable fields include a price margin field that indicates a price margin listing threshold.

3 . The method of claim 1 , wherein the editable fields includes an inventory count field that indicates an inventory amount listing threshold the user is willing to stock and sell.

4 . The method of claim 1 , wherein:

the training the ML model comprises:

accessing the transaction histories; and

windowing time series events for a predetermined time interval; and

the retraining the ML model occurs at the predetermined time interval.

5 . The method of claim 1 , further comprising:

receiving, via one of the editable fields, an adjustment to at least one of the automatic listing thresholds; and

updating the custom preferences based on the adjustment.

6 . The method of claim 1 , further comprising:

receiving, via the user interface, an indication to switch an automatic listing creation and publication opt-in option between automatic and manual; and

updating the custom preferences based on the indication.

7 . The method of claim 1 , further comprising:

automatically publishing, by the publication system, the corresponding listing to an electronic marketplace at a time determined by the machine learning system.

8 . The method of claim 1 , further comprising:

in response to generating the corresponding listing, generating and transmitting a notification to the user that includes inventory stocking instructions.

9 . The method of claim 1 , further comprising:

accessing updated search traffic, wherein the retraining the ML model is further based on the updated search traffic.

10 . The method of claim 1 , further comprising:

determining the high demand items in the one or more potential categories, the determining the high demand items comprising:

accessing buyer demand data and search traffic; and

filtering the high demand items from the one or more potential categories based on the buyer demand data and the search traffic.

11 . A system comprising:

one or more hardware processors; and

a memory storing instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising:

training, by a machine learning system on a first server, a time series-based machine learning (ML) model that forecasts sales based on stored transaction histories, determining, by the machine learning system, one or more potential categories for a user based on custom preferences and previous analytic queries of the user;

applying, by the machine learning system, high demand items in the one or more potential categories to the ML model, the ML model outputting a predicted sales count and a probability of predicted sales for each of the high demand items in the one or more potential categories;

determining, by the machine learning system, items having a potential inventory gap by cross-checking current inventory with a list of one or more items having a probability outputted by the ML model that satisfies a probability threshold;

for each item that has a potential inventory gap, causing presentation, on a user interface, of competitor sales data, the predicted sales count and the probability derived from the ML model, and editable fields displaying automatic listing thresholds that trigger automatic generation of a corresponding listing;

based on the automatic listing thresholds being satisfied, triggering, by the machine learning system, a publication system on a second server to automatically generate the corresponding listing;

continually storing new transactions, the new transactions indicating an actual sales count;

retraining the ML model based on the new transactions and the actual sales count to account for ongoing trends;

validating predictive accuracy of the ML model by comparing a predicted sales count outputted by the ML model with the actual sales count; and

tuning one or more hyperparameters of the ML model based on the validating to improve generalization and the predictive accuracy of the ML model.

12 . The system of claim 11 , wherein the editable fields include a price margin field that indicates a price margin listing threshold.

13 . The system of claim 11 , wherein the editable fields includes an inventory count field that indicates an inventory amount listing threshold the user is willing to stock and sell.

14 . The system of claim 11 , wherein:

the training the ML model comprises:

accessing the transaction histories; and

windowing time series events for a predetermined time interval; and

the retraining the ML model occurs at the predetermined time interval.

15 . The system of claim 11 , wherein the operations further comprise:

receiving, via one of the editable fields, an adjustment to at least one of the automatic listing thresholds; and

updating the custom preferences based on the adjustment.

16 . The system of claim 11 , wherein the operations further comprise:

receiving, via the user interface, an indication to switch an automatic listing creation and publication opt-in option between automatic and manual; and

updating the custom preferences based on the indication.

17 . The system of claim 11 , wherein the operations further comprise:

automatically publishing, by the publication system, the corresponding listing to an electronic marketplace at a time determined by the machine learning system.

18 . The system of claim 11 , wherein the operations further comprise:

in response to generating the corresponding listing, generating and transmitting a notification to the user that includes inventory stocking instructions.

19 . The system of claim 11 , wherein the operations further comprise:

determining the high demand items in the one or more potential categories, the determining the high demand items comprising:

accessing buyer demand data and search traffic; and

filtering the high demand items from the one or more potential categories based on the buyer demand data and the search traffic.

20 . A machine-storage medium comprising instructions which, when executed by one or more hardware processors of a machine, cause the machine to perform operations comprising:

training, by a machine learning system on a first server, a time series-based machine learning (ML) model that forecasts sales based on stored transaction histories;

determining, by the machine learning system, one or more potential categories for a user based on custom preferences and previous analytic queries of the user;

applying, by the machine learning system, high demand items in the one or more potential categories to the ML model, the ML model outputting a predicted sales count and a probability of predicted sales for each of the high demand items in the one or more potential categories;

determining, by the machine learning system, items having a potential inventory gap by cross-checking current inventory with a list of one or more items having a probability outputted by the ML model that satisfies a probability threshold;

for each item that has a potential inventory gap, causing presentation, on a user interface, of competitor sales data, the predicted sales count and the probability derived from the ML model, and editable fields displaying automatic listing thresholds that trigger automatic generation of a corresponding listing;

based on the automatic listing thresholds being satisfied, triggering, by the machine learning system, a publication system on a second server to automatically generate the corresponding listing;

continually storing new transactions, the new transactions indicating an actual sales count;

retraining the ML model based on the new transactions and the actual sales count to account for ongoing trends;

validating predictive accuracy of the ML model by comparing a predicted sales count outputted by the ML model with the actual sales count; and

tuning one or more hyperparameters of the ML model based on the validating to improve generalization and the predictive accuracy of the ML model.

Assignments (1)

Continuity (1)

References Cited (20)