US 12,292,929 B2
Graph data processing method and apparatus, computer device, and storage medium
Zhitao Wang, Shenzhen (CN); Litao Hong, Shenzhen (CN); Weiyi Huang, Shenzhen (CN); and Hanjing Su, Shenzhen (CN)
Assigned to TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED, Shenzhen (CN)
Filed by TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED, Shenzhen (CN)
Filed on Jun. 12, 2023, as Appl. No. 18/333,530.
Application 18/333,530 is a continuation of application No. PCT/CN2022/113580, filed on Aug. 19, 2022.
Claims priority of application No. 202111151811.8 (CN), filed on Sep. 29, 2021.
Prior Publication US 2023/0334096 A1, Oct. 19, 2023
Int. Cl. G06F 16/901 (2019.01)
CPC G06F 16/9024 (2019.01) 20 Claims
OG exemplary drawing
 
1. A graph data processing method applied to a computer device, comprising:
based on object data, acquiring a target graph having M nodes for representing M objects, M being a positive integer, and oriented edges for representing presence of traffic relationships between objects corresponding to connected nodes, wherein pointing nodes of the oriented edges are initiators of the traffic relationships, and pointed nodes of the oriented edges are receivers of the traffic relationships; and
based on the M nodes and oriented edges between the M nodes, performing multiple iterations to obtain initiation embedding features and reception embedding features corresponding to the M nodes, the initiation embedding features being used for representing features of the initiators, and the reception embedding features being used for representing features of the receivers, wherein a zth iteration, z being a positive integer, of the multiple iterations comprises:
acquiring N nodes from the M nodes, N being a positive integer equal to or less than M;
based on oriented edges associated with the N nodes, determining a plurality of first neighborhood nodes associated with the N nodes and a plurality of second neighborhood nodes associated with the N nodes, the oriented edges pointing from the first neighborhood nodes pointing to the N nodes, and the oriented edges pointing from the N nodes pointing to the plurality of second neighborhood nodes; and
based on first embedding features of the N nodes, second embedding features of the plurality of first neighborhood nodes, and third embedding features of the plurality of second neighborhood nodes, determining fourth embedding features of the N nodes, the fourth embedding features of the N nodes representing the updated initiation embedding features and reception embedding features of the N nodes after the zth iteration.