时空融合的时序知识图谱多跳推理模型

打开文本图片集

关键词：时序知识图谱；多跳推理；三联体分配器；时空注意力；分层强化学习框架中图分类号：TP391 文献标志码：A 文章编号：1001-3695（2026）04-006-1013-08doi：10.19734/j.issn.1001-3695.2025.08.0296

Spatiotemporal fusion multi-hop reasoning model for temporal knowledge graphs

Ma Handa†，Fei Fan （SchoolofComputer Scienceand Communication，Jiangsu University，ZhenjiangJiangsu 212o13，China）

Abstract：Toaddress the isues ofsemanticdisconnectioncaused byseparated entity-relation embedding spaces and limited temporal expressiveness in existing temporalknowledgegraphmulti-opreasoning models，this paper proposedaspatiotemporalfusionmulti-hopreasoning model（SF-MR）.Themodelincorporatedatripledistrbutorwithdual-pathresidualconections andspatialconvolutiontocapturecros-spacesemanticdependenciesbetweenentitiesandrelations.Itintroducedaspatiotemporalatentionmechanismtojointlymodelentitytemporal evoutionandspatialcorelations，dynamicallyfusedviaagatednetwork．Ahierarchicalreiforcementleaningframework decoupledreasoningintorelation-levelandentity-leveldecisions，alleviatingaction space explosion.Experiments onfour benchmark datasets （ICEWS14，ICEWS18，WIKI，YAGO）demonstrate that SF-MR outperforms state-of-the-art baselinesacrossmultiple metrics.Specificall，onICEWS14，SF-MR improves MRR， hits@3 ，and hits@10 by 1. 10% ， 1.53% ，and 2.69% ，respectively，over the best baseline.Consistent improvements of 0.79% to 1.01% are observed on WIKI and YAGO.Ablation studies confirm the effectiveness of the triple distributor and spatiotemporal attention in enhancing semantic interaction and temporal modeling.

Key words：temporal knowledgegraph（TKG）；multi-hopreasoning；tripletdistributor;spatiotemporalatention；hierarchi calreinforcementlearningframework

0 引言

随着社会活动的动态演进，知识图谱（knowledgegraph，KG）的时序建模能力成为支撑智能决策的关键技术。（剩余22566字）