US 12,079,860 B2
System and method for creating and analyzing a low-dimension representation of webpage sequence
Benjamin Hillel Myara, Tel Aviv (IL); and David Tolpin, Ashdod (IL)
Assigned to PAYPAL, INC., San Jose, CA (US)
Filed by PayPal, Inc., San Jose, CA (US)
Filed on Aug. 23, 2021, as Appl. No. 17/409,576.
Application 17/409,576 is a continuation of application No. 15/852,331, filed on Dec. 22, 2017, granted, now 11,100,568.
Prior Publication US 2021/0383459 A1, Dec. 9, 2021
This patent is subject to a terminal disclaimer.
Int. Cl. G06Q 30/00 (2023.01); G06F 40/30 (2020.01); G06Q 30/0204 (2023.01); G06Q 30/0251 (2023.01); G06Q 30/0601 (2023.01); H04L 9/40 (2022.01); H04L 67/02 (2022.01); H04L 67/14 (2022.01); H04L 67/145 (2022.01); H04L 67/50 (2022.01)
CPC G06Q 30/0641 (2013.01) [G06F 40/30 (2020.01); G06Q 30/0204 (2013.01); G06Q 30/0251 (2013.01); G06Q 30/0601 (2013.01); G06Q 30/0633 (2013.01); H04L 63/14 (2013.01); H04L 63/1425 (2013.01); H04L 67/02 (2013.01); H04L 67/14 (2013.01); H04L 67/145 (2013.01); H04L 67/535 (2022.05)] 20 Claims
OG exemplary drawing
 
1. A system, comprising:
a non-transitory memory; and
one or more hardware processors coupled to the non-transitory memory and configured to read instructions from the non-transitory memory to cause the system to perform operations comprising:
iteratively training a machine learning model on network traffic history data corresponding to a browsing behavior of a plurality of users visiting a plurality of webpages of a merchant website, wherein the machine learning model is iteratively trained at least in part by applying a word embedding algorithm to the machine learning model, and wherein executing the word embedding algorithm on the machine learning model causes the plurality of webpages to be represented as words and the browsing behavior of the plurality of users visiting the plurality of webpages to be represented as vectors;
generating, based on the words and the vectors a low-dimensional representation of the browsing behavior that associates an abandoned transaction or a fraudulent transaction with a type of user browsing behavior;
predicting, based on the low-dimensional representation of the browsing behavior, 1) a cart abandonment with respect to a first webpage of the merchant website or 2) the fraudulent transaction, wherein the predicting is performed at least in part by identifying a group of vectors corresponding to the abandoned transaction or the fraudulent transaction;
based on the predicting, determining that a user navigation from the first webpage to a second webpage of the merchant website is associated with a higher likelihood of the cart abandonment or the fraudulent transaction but that a user navigation from the first webpage to a third webpage of the merchant website is associated with a lower likelihood of the cart abandonment or the fraudulent transaction; and
based on the determining, performing an action to prevent the cart abandonment or the fraudulent transaction, wherein the action is performed at least in part by altering a layout of the first webpage of the merchant website, wherein the altering the layout comprises causing a link to the third webpage of the merchant website to be displayed more prominently than at least a link to the second webpage of the merchant website.