US 11,669,915 B1
Systems and methods for making high value account recommendations
Alan Si, San Francisco, CA (US); Jialu Zhu, Mountain View, CA (US); Sourav Chatterji, Fremont, CA (US); and Brian Dolhansky, Seattle, WA (US)
Assigned to Meta Platforms, Inc., Menlo Park, CA (US)
Filed by META PLATFORMS, INC., Menlo Park, CA (US)
Filed on Sep. 27, 2017, as Appl. No. 15/717,898.
Int. Cl. G06Q 50/00 (2012.01); G06N 5/04 (2023.01); G06Q 30/0204 (2023.01); G06N 20/00 (2019.01)
CPC G06Q 50/01 (2013.01) [G06N 5/04 (2013.01); G06N 20/00 (2019.01); G06Q 30/0204 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
identifying, by a computing system, a set of accounts, each account of the set of accounts having a number of followers;
grouping, by the computing system, the set of accounts into a plurality of groups based on number of followers, wherein each group is associated with a value score;
training, by the computing system, a machine learning model using a set of training data, wherein
the training data includes a plurality of successful account recommendations and associated weights based on value scores,
the plurality of successful account recommendations are associated with a training set of accounts,
the training set of accounts are associated with a training plurality of groups,
each group in the training plurality of groups
i) includes a plurality of accounts,
ii) is associated with a predetermined range of numbers of followers of accounts in the group, wherein the plurality of groups comprises: a first group comprising accounts, each account in the first group having a number of followers lower than a first threshold; a second group comprising accounts, each account in the second group having a number of followers between the first threshold and a second threshold; and a third group comprising accounts, each account in the third group having a number of followers greater than the second threshold, and
iii) has a corresponding value score that is inversely related to an average number of followers for accounts in the group, wherein the first group has a value score greater than the second and third groups, and the second group has a value score greater than the third group, and
each successful account recommendation of the training data is assigned a weight based on a value score associated with a respective group with which the successful account recommendation is associated;
and
selecting, by the computing system, one or more accounts of the set of accounts to present as account recommendations based on the machine learning model.