Entity alignment method and device, electronic equipment, computer readable storage medium and computer program product
Through an unsupervised time-aware multi-fusion method that combines attribute similarity and time similarity in the timing knowledge graph, the problem of dynamic changes and insufficient generalization capabilities of entity alignment in the prior art is solved, and more accurate and robust Entity alignment effect.
Patent Information
- Application Number
- CN202411268302.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2025-01-14
AI Technical Summary
When handling dynamically changing timing knowledge graphs, existing entity alignment methods are difficult to accurately identify and merge entity representations in different knowledge graphs, and lack sufficient generalization capabilities, making it difficult to adapt to new or unseen entity alignment tasks.
Through the time-series knowledge graph entity alignment method based on unsupervised time-aware multi-fusion, combining attribute similarity and time similarity, the entity similarity is determined and entity alignment is performed. The graph structure data is processed using deep learning models such as graph neural networks and automatically extracted. Features for alignment.
Improve the accuracy and robustness of entity alignment, can more accurately capture the state changes of entities at different periods, enhance the ability to align entities in dynamic scenarios, reduce the labor cost of the alignment process, and adapt to knowledge in different fields. Graph alignment task.
Smart Images
Figure CN119311899A_ABST