一种基于双曲动态神经网络的知识感知推荐方法
By using a knowledge-aware recommendation method based on hyperbolic dynamic neural networks, high-order collaborative signals are generated and fused, solving the problem of low accuracy of recommendation information in existing technologies and achieving more accurate recommendation results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING UNIV OF TECH
- Filing Date
- 2023-11-16
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies ignore higher-order collaborative signals of items, resulting in low accuracy of recommendation information and an inability to accurately recommend the knowledge or products that users want.
A knowledge-aware recommendation method based on hyperbolic dynamic neural networks is adopted. By encoding users, items and entities, an enhanced knowledge graph and user-item interaction graph are generated. High-order collaborative signals are generated by knowledge aggregation and collaborative aggregation, and then input into a bilateral memory network for fusion. Finally, cosine similarity comparison is used for recommendation.
It improves the accuracy and efficiency of recommendation systems, enabling a better understanding of user needs and preferences, and achieving complex, fine-grained relationship modeling, thus enhancing the accuracy and efficiency of recommendations.
Smart Images
Figure CN117688254B_ABST