Year: 2025
Type: conference-paper
Abstract: Aiming at the problems of multi-source heterogeneous fusion, cross-domain security sharing and whole-process anomaly detection in government service data governance, this paper proposes a GEME (Graph Entropy Maximization with Enhanced Attention) model that combines graph entropy maximization and attention mechanism. This method innovatively extends the Körner graph entropy theory to the representation learning of government service graph, captures the topological features of government business process by maximizing the entropy value under the constraint of orthogonal representation, and realizes the dynamic weighting of sensitive nodes by combining the dual channel attention mechanism. This study provides the core algorithm support for the key research and development project of the autonomous region 's key research and development plan “Key Technologies for Government Data Lifecycle Governance in an Open Environment ”, and effectively solves the contradiction between privacy protection and structural interpretability in cross-departmental data sharing.
Authors: Haohui Zhou, Junyu Zhu, Shumei Bao, Enguang Zuo
Institutions: Xinjiang University, Tsinghua University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4413290223