基于行为序列建模与图神经网络的用户画像构建方法
By using behavioral sequence modeling and graph neural networks to construct user profiles, the dependency problem between user behavior sequences and interaction graphs is solved, dynamic user profiles are generated, and the accuracy and cold start capability of the recommendation system are improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHONGKE MICRO DOT TECH CO LTD
- Filing Date
- 2026-05-18
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to balance the temporal dependencies of user behavior sequences with the structural dependencies of user-item interaction graphs in complex business scenarios, resulting in low recommendation accuracy, difficulties in cold starts, and a lack of diversity in user profile representation.
By using behavior sequence modeling and graph neural networks, we obtain users' historical interaction behavior logs, construct a user-item interaction bipartite graph, and generate dynamic user profile vectors using multi-head self-attention networks and multi-layer graph attention networks, thus fusing behavioral temporal information and graph structure information.
It achieves high-fidelity extraction and dynamic expression of user interests, improves recommendation accuracy and cold start capability in complex scenarios, and provides more timely and robust personalized decision support.
Smart Images
Figure CN122221908B_ABST