US 12,271,870 B2
Rapid online clustering
Avishay Meron, Gush-Dan (IL); Xing Wang, Shanghai (CN); Adam Cohen, Ramat Gan (IL); Chunmao Ran, Shanghai (CN); and David Stein, Even Yehuda (IL)
Assigned to PAYPAL, INC., San Jose, CA (US)
Filed by PAYPAL, INC., San Jose, CA (US)
Filed on Nov. 16, 2022, as Appl. No. 17/988,015.
Application 17/988,015 is a continuation of application No. 17/120,893, filed on Dec. 14, 2020, granted, now 11,507,631.
Application 17/120,893 is a continuation of application No. 15/689,916, filed on Aug. 29, 2017, granted, now 10,866,995, issued on Dec. 15, 2020.
Prior Publication US 2023/0078312 A1, Mar. 16, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 16/30 (2019.01); G06F 16/28 (2019.01); G06F 16/906 (2019.01); G06F 16/951 (2019.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01); G06Q 20/10 (2012.01); G06Q 20/32 (2012.01); G06Q 20/40 (2012.01)
CPC G06Q 20/10 (2013.01) [G06F 16/285 (2019.01); G06F 16/906 (2019.01); G06F 16/951 (2019.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01); G06Q 20/322 (2013.01); G06Q 20/4016 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
accessing historical data pertaining to a plurality of events with respect to an electronically provided service, wherein the accessed historical data contains respective cluster labels and a plurality of feature values for a plurality of features associated with each of the plurality of events;
constructing a first table based on the accessed historical data;
reducing the first table into a plurality of reduced feature tables, wherein each of the reduced feature tables contains a plurality of keys that each corresponds to a different unique combination of multiple ones of the plurality of the features, and wherein each of the reduced feature tables further contains, for each event of the plurality of events, the feature values of features other than the different unique combination of features in the key corresponding to the event;
accessing data of a new event that is occurring or has occurred since the plurality of reduced feature tables has been constructed, the new event being associated with the electronically provided service;
electronically searching, using the data of the new event, at least a subset of the plurality of reduced feature tables in parallel; and
determining, based on a result of the electronically searching, which cluster label to assign to the new event.