US 11,727,137 B2
Predictive update of profile database with update logic
Hongqin Song, Austin, TX (US); and Yu Gu, Austin, TX (US)
Assigned to VISA INTERNATIONAL SERVICE ASSOCIATION, San Francisco, CA (US)
Filed by Visa International Service Association, San Francisco, CA (US)
Filed on Mar. 31, 2022, as Appl. No. 17/710,536.
Application 17/710,536 is a continuation of application No. 16/380,715, filed on Apr. 10, 2019, granted, now 11,314,881.
Prior Publication US 2022/0229926 A1, Jul. 21, 2022
Int. Cl. G06F 21/62 (2013.01); G06N 3/08 (2023.01); G06N 20/00 (2019.01); G06F 21/31 (2013.01)
CPC G06F 21/6218 (2013.01) [G06F 21/316 (2013.01); G06N 3/08 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A risk determination method comprising performing, by a computer system:
receiving, at an authentication platform of the computer system, an access request for a resource, the access request from a user of a plurality of users using a client computer of a plurality of client computers;
determining a risk of providing access to the resource by retrieving, from a distributed storage, a profile corresponding to the user and analyzing the profile;
storing, in the distributed storage, raw data from a plurality of historical access requests corresponding to the access request from the plurality of client computers;
transmitting, by the authentication platform to a prediction engine, request data from a set of access requests from the user and a plurality of different users of the plurality of users, wherein the prediction engine uses the request data to calculate and assign a rank to each profile in a plurality of profiles, wherein each calculated rank corresponds to a respective probability that a next time interval will include a future access request for a corresponding profile of the plurality of profiles;
selecting a first set of the profiles from a profile repository, the first set corresponding to the highest ranking profiles from the profile repository according to the calculated ranks of the plurality of profiles; and
for each profile in the first set of the profiles:
selecting a respective profile;
retrieving previously stored raw data corresponding to the respective profile from the distributed storage;
using the retrieved raw data to determine one or more updated value for at least one statistical value;
updating the respective profile with the one or more determined updated values; and
storing the updated respective profile in the profile repository,
whereby a next risk determination corresponding to a next access request has an increased probability of being determined from the corresponding profile updated with corresponding one or more determined updated values more recently based on the corresponding assigned rank than if the corresponding profile were updated in an order of an index to thereby increase accuracy of the next risk determination.