Название: Knowledge Graphs and Semantic Computing 10th China Conference, CCKS 2025, Fuzhou, China, September 19–21, 2025, Proceedings
Автор: Jiye Liang, Guolong Chen, Kang Liu, Jing Zhang, Zhichun Wang, Hongyu Lin, Yongbin Liu
Издательство: Springer
Серия: Communications in Computer and Information Science
Год: 2026
Страниц: 298
Язык: английский
Формат: pdf (true), epub
Размер: 67.6 MB

This book constitutes the proceedings of the 10th China Conference on Knowledge Graph and Semantic Computing, CCKS 2025, held in Fuzhou, China, during September 19–21, 2025.

The 22 full papers presented in this book were carefully reviewed and selected from 112 submissions. They were organized into the following topical sections: Knowledge Graph Construction and Integration; Large Models Enhanced by Knowledge Graphs; Applications of Knowledge Graphs and Large Models/Agents; Open Resources for Knowledge Graphs and Large Models; and Evaluations.

Knowledge Graphs (KGs) furnish reliable and structured information that is vital for several downstream applications, including information retrieval and recommendation system. However, the pervasive incompleteness inherent in KGs frequently constrains the efficacy of these applications. To mitigate this limitation, researchers have introduced the Knowledge Graph Completion (KGC) task, which aims to supplement missing facts within incomplete triples. In recent times, contrastive learning has been incorporated into the domain of KGC, yielding substantial enhancements to the discriminative power of KGC models and establishing new performance benchmarks. Nevertheless, current contrastive methodologies usually face the problems of insufficient generalization ability of sparse relations, poor understanding of importance differences towards heterogeneous relations, as well as information redundancy in single-view comparison. To overcome these challenges, this work proposes a novel cross-subgraph attention fusion and comparison method, consisting of oriented noise injection, cross-subgraph attention fusion and cross-subgraph contrastive loss. Especially, it helps to enhance the neighboring aggregation procedure as well as the comparative loss function in existing models, by fully utilizing beneficial and complementary semantic signals from different views in the given KG. Furthermore, this contribution can be regarded as a flexible and easily adaptable plug-in component, engineered for seamless compatibility with extant contrastive learning based KGC architectures.

Event prediction (EP), the accurate forecasting of future events, is vital for strategic planning and risk management in both governmental and business contexts. The rapid advancement of large language models (LLMs) has positioned AI-based automated prediction methods at the forefront of academic and industrial research. However, current LLM prediction systems exhibit several shortcomings. Firstly, their information retrieval mostly searches based on the question itself, failing to gather relevant data from multiple perspectives as human expert teams do. Secondly, their temporal analysis is inadequate, as the collected information often includes subjective opinions or speculations and lacks the ability to reconcile contradictory information across different time points during real-time prediction. To address these issues this paper introduces MAEPS (Multi-Agent Event Prediction System), which emulates the collaborative efforts of human expert teams through 12 specialized agents. Each agent collects data from a specific professional dimension. The system automatically identifies and resolves conflicting information, ensuring that predictions prioritize recent and consistent facts.

Contents: