华侨大学学报（自然科学版）2025，Vol.46Issue(5)：539-550,12.DOI:10.11830/ISSN.1000-5013.202508020
面向可信推理与语义补全的区块链知识图谱共享框架
Knowledge Graph Sharing Framework With Blockchain for Trustworthy Inference and Semantic Completion
摘要
Abstract
To address challenges faced by current blockchain-driven knowledge sharing architectures in heter-ogeneous knowledge graph integration,trusted interaction,and system efficiency,a blockchain-based knowl-edge graph trusted reasoning framework(BKTRF)is proposed.First,the trusted transmission of on-chain and off-chain reasoning results are achieved through decentralized oracles,and the BLS signature algorithm based on a threshold mechanism is employed to enhance the efficiency and fault tolerance of multi-node collabo-rative signatures while reducing on-chain computational and storage overhead.Then,a large language model-based knowledge graph completion method is introduced to automatically generate high-quality logic rules for semantic completion and reasoning optimization.The results show that compared to the standard BLS signa-ture algorithm,the threshold BLS signature algorithm can improve signing efficiency by up to 53.5％.In addi-tion,BKTRF demonstrates superior completion performance and generalization capability across three data-sets.关键词
区块链/门限签名算法/大语言模型/知识图谱推理/协同优化Key words
blockchain/threshold signature algorithm/large language model/knowledge graph reasoning/collaborative optimization分类
信息技术与安全科学引用本文复制引用
黄慧玲,郭荣新,施一帆..面向可信推理与语义补全的区块链知识图谱共享框架[J].华侨大学学报（自然科学版）,2025,46(5):539-550,12.基金项目
国家自然科学青年基金资助项目(62306122) （62306122）
福建省科技项目引导性项目(2023H0012) （2023H0012）
