US 12,086,823 B2
Computer-based systems including machine learning models trained on distinct dataset types and methods of use thereof
Abdelkader M'Hamed Benkreira, New York, NY (US); Michael Mossoba, Arlington, VA (US); and Joshua Edwards, Philadelphia, PA (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Feb. 28, 2022, as Appl. No. 17/683,287.
Application 17/683,287 is a continuation of application No. 16/781,718, filed on Feb. 4, 2020, granted, now 11,263,486, issued on Mar. 1, 2022.
Prior Publication US 2022/0188574 A1, Jun. 16, 2022
Int. Cl. G06Q 30/02 (2023.01); G06F 16/95 (2019.01); G06F 16/951 (2019.01); G06F 16/9537 (2019.01); G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06N 20/00 (2019.01); G06Q 30/0201 (2023.01); G06Q 30/06 (2023.01); G06Q 30/0601 (2023.01)
CPC G06Q 30/0206 (2013.01) [G06F 16/951 (2019.01); G06F 16/9537 (2019.01); G06F 18/2148 (2023.01); G06N 20/00 (2019.01); G06Q 30/0643 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
configuring, by at least one processor, a web browser of a user computing device to install a recommendation add-on to the web browser that cause the web browser to automatically determine and present a forecasted price of at least one product to at least one user associated with the user computing device upon the at least one user selecting to view at least one webpage associated with the at least one product;
wherein the recommendation add-on is configured to cause the web browser to automatically detect the at least one product on the at least one webpage;
obtaining, by the at least one processor, upon the recommendation add-on detecting the at least one product on the at least one webpage and via at least one programmed call, historical product information for the at least one product;
wherein the product information comprises:
i) one or more identifying attributes, and
ii) a price across a plurality of dates;
obtaining, by the at least one processor, upon the recommendation add-on detecting the at least one product on the at least one webpage and via the at least one programmed call, historical event information from one or more online sources;
wherein the historical event information comprises:
a date associated with each historical event of a plurality of historical events, and
at least one content descriptive of each historical event;
utilizing, by the at least one processor, at least one similarity measurement algorithm to determine a similarity measure indicative of a similarity between the at least one content descriptive of each historical event and the one or more identifying attributes of the at least one product so as to determine a similarity between each historical event and the at least one product;
determining, by the at least one processor, at least one historical event associated with the at least one product based at least in part on the similarity measure exceeding a predetermined threshold similarity;
determining, by the at least one processor, at least one product impact of the at least one product based at least in part on:
the price having an impact comprising a fluctuation exceeding a predetermined impact fluctuation value over a duration exceeding a predetermined impact fluctuation duration and
the predetermined impact fluctuation duration being within the at least one historical event based at least in part on the date associated with each historical event of the plurality of historical events;
generating, by the at least one processor, at least one event-dependent products training dataset based at least in part on the product information of the at least one product impact and the historical event information;
wherein the at least one event-dependent products training dataset comprises the at least one product impact and the at least one historical event associated with the at least one product impact;
determining, by the at least one processor, an impact period for each historical event of the plurality of historical events based on the predetermined impact fluctuation duration;
training, by the at least one processor, an attribute prediction machine learning model based at least in part on the at least one event-dependent products training dataset to obtain a trained attribute prediction machine learning model;
wherein the trained attribute prediction machine learning model comprises a regression layer formed of model parameters trained to correlate the at least one historical event to:
the impact on of the at least one product, and
ii) the impact period of the at least one product;
receiving, by the at least one processor, upon the recommendation add-on detecting the at least one product on the at least one webpage and via the at least one programmed call, current event information for at least one current event, the current event information comprising at least one current content associated with the at least one product;
utilizing, by the at least one processor, upon the recommendation add-on detecting the at least one product on the at least one webpage, the attribute prediction machine learning model to output a forecasted price of the at least one product based on an input of the at least one current content based at least in part on the regression layer formed of the model parameters; and
causing to display, by the at least one processor via the recommendation add-on, the at least the forecasted price of the at least one product and an interface element in at least one user interface rendered on a display of the user computing device associated with the user, wherein the interface element is configured to enable the user to purchase the at least one product.