计算机应用研究2025，Vol.42Issue(6)：1727-1733,7.DOI:10.19734/j.issn.1001-3695.2024.10.0463
融合局部-全局历史模式与历史知识频率的时序知识图谱补全方法
Time-series knowledge graph completion method combining local-global historical pattern and historical knowledge frequency
摘要
Abstract
TKGs are dynamic representations of evolving facts,and their completion task involves predicting future unknown facts based on historical data.The key lies in understanding historical data.However,existing models have limitations in cap-turing the features of historical events and cannot accurately extract useful information from timestamps.From the perspective of historical evolution,considering the sequence,frequency,and periodic patterns of historical facts comprehensively is bene-ficial for predicting future facts.Therefore,this paper proposed a temporal knowledge graph completion algorithm(LGH-HKF)that integrated local-global historical patterns and historical knowledge frequency.Specifically,it firstly used a local re-current graph encoder network to model the intrinsic associations and dynamic evolution of events at adjacent timestamps.Then,it used a global historical encoder network to consider all relevant facts at previous timestamps to avoid losing entities or relations that didn't appear at adjacent timestamps.Next,it learned the frequency scores of these facts through a historical knowledge frequency learning module to enrich the model's prediction basis.Finally,after balancing between the two en-coders,it used a periodic decoder to perform inference and completion.The paper used four benchmark datasets to evaluate the proposed method,and the experimental results prove that LGH-HKF is highly competitive compared to other current models in most cases.关键词
时序知识图谱/补全算法/局部循环图编码器/全局历史编码器/历史知识频率Key words
temporal knowledge graph/completion algorithm/local cyclic graph encoder/global history encoder/frequency of historical knowledge分类
信息技术与安全科学引用本文复制引用
贾凯,王阳萍,杨景玉,张希权..融合局部-全局历史模式与历史知识频率的时序知识图谱补全方法[J].计算机应用研究,2025,42(6):1727-1733,7.基金项目
国家自然科学基金资助项目(62067006,62367005) （62067006,62367005）
甘肃省知识产权计划资助项目(21ZSCQ013) （21ZSCQ013）
2024中央引导地方科技发展资金资助项目(332140068864) （332140068864）
甘肃省高校科研创新平台重大培育项目(2024CXPT-17) （2024CXPT-17）
甘肃省教育科技创新项目(2021jyjbgs-05) （2021jyjbgs-05）
