US 12,406,773 B2
Transferring information through knowledge graph embeddings
Rory McGrath, Kildare Town (IE); Xu Zheng, Dublin (IE); and Jeremiah Hayes, Dublin (IE)
Filed by Accenture Global Solutions Limited, Dublin (IE)
Filed on Aug. 24, 2022, as Appl. No. 17/821,910.
Prior Publication US 2024/0071629 A1, Feb. 29, 2024
Int. Cl. G16H 70/40 (2018.01); G06N 5/022 (2023.01); G16C 20/70 (2019.01)
CPC G16H 70/40 (2018.01) [G06N 5/022 (2013.01); G16C 20/70 (2019.02)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
receiving, by a device, a knowledge graph representing information and simplified molecular-input line-entry (SMILE) data identifying compounds;
training, by the device, embeddings based on the knowledge graph;
generating, by the device, graph embeddings for the SMILE data based on the embeddings;
encoding, by the device, the SMILE data into a latent space;
combining, by the device, the graph embeddings and the latent space to generate a combined latent-embedding space;
decoding, by the device, the combined latent-embedding space to generate decoded SMILE data;
utilizing, by the device, the decoded SMILE data to train an encoder and to generate a trained encoder;
processing, by the device, source SMILE data, with the trained encoder, to generate a source combined latent-embedding space;
searching, by the device, the source combined latent-embedding space to identify new SMILE data associated with new compounds;
decoding, by the device, the new SMILE data to generate decoded new SMILE data; and
evaluating, by the device, the decoded new SMILE data to identify particular SMILE data associated with a new compound.