US 12,423,516 B2
Recommendation generation using machine learning model trained on multi-step performance data
Pierre Cottin, Paris (FR); Martin Dufour, Paris (FR); and Martin Tournier, Paris (FR)
Assigned to SAP SE, Walldorf (DE)
Filed by SAP SE, Walldorf (DE)
Filed on Nov. 23, 2022, as Appl. No. 17/993,236.
Prior Publication US 2024/0169146 A1, May 23, 2024
Int. Cl. G06F 40/194 (2020.01); G06Q 10/1053 (2023.01)
CPC G06F 40/194 (2020.01) [G06Q 10/1053 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A system comprising:
a memory storing processor-executable program code; and
a processing unit to execute the processor-executable program code to cause the system to:
train a machine learning model to determine a probability associated with each of two or more posting platforms based on, for each of a plurality of posts, a first numerical representation of the text of the post, a number of candidates reaching each of a plurality of steps of a hiring process for each of the two or more posting platforms, and an indicator for each of the two or more posting platforms indicating whether a candidate was hired from the posting platform;
for each of the plurality of posts, input the first numerical representation of the text of the post and the number of candidates reaching each of the plurality of steps of a hiring process for each of the two or more posting platforms into the trained machine learning model to generate a probability associated with each posting platform;
receive a first post comprising text;
identify one or more prior posts similar to the first post from a plurality of prior posts based on the text of the first post and text of the plurality of prior posts;
determine the probabilities associated with each of two or more posting platforms which were generated for each of the identified one or more prior posts;
for each of the identified one or more posts, determine a weighted probability associated with each of two or more posting platforms based on a similarity of the identified post to the first post and the probabilities associated with each of the two or more posting platforms which were generated for the identified post;
determine one or more of the two or more posting platforms based on the weighted probabilities; and
transmit the first post to the determined one or more posting platforms.