A graph embedding-based dynamic update method for intelligence knowledge graphs
Yong Chen, Enhong Chen, Wenjie Liu, Nuo Chen, Xiaoning Wu, Tong Xu, Zhi Zheng
Frontiers of Computer Science
Problems Identified (2)
Dynamic KG update inaccuracy: Dynamic updating of intelligence knowledge graphs faces inaccuracy due to complex incremental data and noise interference.
Dynamic KG update inaccuracy: Dynamic updating of intelligence knowledge graphs faces inaccuracy due to complex incremental data and noise interference.
Proposed Solutions (5)
GEDUM graph embedding update: The paper proposes GEDUM, a graph embedding-based dynamic update method for intelligence knowledge graphs that models dynamic evolution and optimizes updating through embedding networks.
Local-to-global feature aggregation: The method includes L2GFAM to learn global graph embeddings by exploring and optimizing intrinsic node and edge features.
Attention-guided weighted fusion: The method includes AWFS to merge and update embeddings of local subgraphs and newly added graph components using correlations between new and existing data.
GEDUM graph embedding update: The paper proposes GEDUM, a graph embedding-based dynamic update method for intelligence knowledge graphs that models dynamic evolution and optimizes updating through embedding networks.
Local-to-global feature aggregation: The method includes L2GFAM to learn global graph embeddings by exploring and optimizing intrinsic node and edge features.
Results (2)
Superior dynamic update performance:
Superior dynamic update performance:
Research Domain
Intelligence knowledge graphs; dynamic knowledge graph updating; graph embeddings