US 12,141,209 B2
Machine learning and prediction using graph communities
Theodore D. Harris, San Francisco, CA (US); Craig O'Connell, San Mateo, CA (US); Terry Angelos, Mountain View, CA (US); Tatiana Korolevskaya, Mountain View, CA (US); Yue Li, San Mateo, CA (US); and Todd Sawyer, San Francisco, CA (US)
Assigned to Visa International Service Association, San Francisco, CA (US)
Filed by Visa International Service Association, San Francisco, CA (US)
Filed on Nov. 19, 2020, as Appl. No. 16/953,235.
Application 16/953,235 is a continuation of application No. 16/311,024, granted, now 10,872,298, previously published as PCT/US2017/041537, filed on Jul. 11, 2017.
Claims priority of provisional application 62/360,799, filed on Jul. 11, 2016.
Prior Publication US 2021/0073283 A1, Mar. 11, 2021
Int. Cl. G06F 16/9038 (2019.01); G06F 16/901 (2019.01); G06F 17/18 (2006.01); G06N 5/022 (2023.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01)
CPC G06F 16/9038 (2019.01) [G06F 16/9024 (2019.01); G06F 17/18 (2013.01); G06N 5/022 (2013.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
a) receiving, by one or more computers, interaction data for a plurality of known interactions between resource providers and users, the resource providers being merchants and the users purchasing goods and services at the merchants;
b) creating and training, by the one or more computers, a predictive machine learning model comprising a topological graph comprising nodes and edges, and based on the plurality of known interactions using an unsupervised machine learning algorithm and the interaction data, and by changing weights associated with the nodes and edges;
c) determining, by the one or more computers, a plurality of communities using the trained predictive machine learning model and assigning each community with a community identifier, wherein each community is determined by identifying densely connected nodes starting from a seed node and adding nodes with high interaction probability until a threshold is met, wherein the densely connected nodes are nodes that interact with a greater frequency than other connected nodes, and wherein the communities in the plurality of communities contain common nodes;
d) receiving, by the one or more computers, a request for a prediction;
e) applying the request to the predictive machine learning model, by the one or more computers, by identifying a second community in the plurality of communities by a community identifier based on an account identifier in the request for the prediction, the account identifier being in a first community, wherein nodes in the first community and the second community overlap;
f) determining, by the one or more computers, a node within the identified community by traversing a path starting from the account identifier to the node via one or more other nodes, the path including a node that is common to both the first community and the second community; and
g) providing, by the one or more computers, information regarding the node as the requested prediction.