Chunyan An, Zhihong Hao, Hui Yin, Qiang Yang, Zhixu Li · ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 2026
DOI: 10.1145/3799682.3840652
Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
Event knowledge graphs (EKGs) must evolve continuously as new events emerge, while preserving temporal integrity and structural consistency. However, existing triple-completion and embedding approaches assume closed-world settings and lack mechanisms for open-world alignment and interpretable updates. Furthermore, real-world streaming data often suffers from out-of-order arrivals and information fragmentation, which exacerbates update conflicts and causal inversion in continuous evolution. We define the task of Continuous Event Knowledge Graph (C-EKG) Updating and introduce HGNN-EAGE , a unified framework that addresses these challenges through adaptive stream scheduling and hypergraph neural alignment. Specifically, we propose an adaptive dependency-aware sorting strategy that dynamically prioritizes information-rich and temporally antecedent events to initialize the topological basis. For event alignment, HGNN-EAGE models events as hyperedges and learns numerically stable embeddings via a degree-regularized hypergraph neural network. It decouples semantic, structural, temporal, spatial, type-level, salient-entity, and consistency cues into orthogonal evidence dimensions, driving a hierarchical decision process that maps interpretable alignment outcomes to atomic, version-controlled graph operations (insert, supplement, link-subevent). Extensive experiments on EventKG and DuEE-fin benchmarks demonstrate that our framework significantly outperforms strong baselines in retrieval and classification accuracy. Moreover, the proposed streaming strategy substantially mitigates topological redundancy under high-load out-of-order data streams while maintaining near-linear scalability.
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
ACM International Conference on Information and Knowledge Management (CIKM) 2026 · 0 citations
No comments yet — start the discussion below.