计算机科学 ›› 2026, Vol. 53 ›› Issue (4): 393-405.doi: 10.11896/jsjkx.250400050
辛奕辰1, 李时冲1, 陈斌1, 程章桃1, 李耶1,2, 周帆1
XIN Yichen1, LI Shichong1, CHEN Bin1, CHENG Zhangtao1, LI Ye1,2, ZHOU Fan1
摘要： 时序知识图谱以四元组(主体,关系,客体,时间)的形式动态记录事件知识,能够有效刻画现实世界中知识的动态演化特性,因此在推荐系统、大语言模型和知识问答等领域得到了广泛应用。然而,其固有的知识不完备性限制了进一步的拓展和应用。时序知识图谱推理任务旨在预测(补全)图谱中缺失的事件知识,因而受到学术界和工业界的高度关注。现有的时序知识图谱推理方法主要聚焦于挖掘图谱快照内部的结构信息以及快照之间的时序依赖,仍存在以下两个主要问题:1) 在建模过程中未充分考虑图谱快照中潜在的噪声和冗余信息;2) 过度依赖局部时间窗口内的序列模式,忽视了图谱全局时序依赖的建模。为解决上述问题,提出了一种融合图信息瓶颈理论与Transformer的时序知识图谱推理框架GIBformer。该框架首先引入图信息瓶颈理论,对时序知识图谱中的结构信息进行压缩,保留与下游预测任务强相关的关键信息,同时有效抑制噪声和冗余信息的干扰;其次,利用Transformer的多头注意力机制,捕捉跨图谱快照的全局时序依赖模式,并融合局部的时序演化信息,实现对缺失事件知识的精准预测。在4个主流基准数据集上进行的大量实验证明了该模型的有效性。
中图分类号:
| [1]CHEN J,CHEN J,YU Z.Incorporating structured common-sense knowledge in story completion[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2019:6244-6251. [2]WU S,SUN F,ZHANG W,et al.Graph neural networks in re-commender systems:a survey[J].ACM Computing Surveys,2022,55(5):1-37. [3]CHEN B,SHEN H,CHENG Z,et al.Unveiling cross-modalconsistency:Taming inter-and intra-modal noise for robust multi-modal knowledge graph completion[J].Information Processing & Management,2026,63(2):104472. [4]CHEN B,YANG Y,CHENG Z,et al.Semantic duality in hypergraphs:Uncertainty-aware bipolar evidence aggregation for temporal knowledge graph reasoning[J].Expert Systems with Applications,2026,299:130162. [5]CHEN M Q,ZHANG C L,YU H.Temporal knowledge graph reasoning model incorporating cross-time commonality features[J].Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition),2025,37(5):708-716. [6]SCHLICHTKRULL M,KIPF T N,BLOEM P,et al.Modeling relational data with graph convolutional networks[C]//The Semantic Web:15th International Conference.Springer,2018:593-607. [7]LIANG K,MENG L,LIU M,et al.A survey of knowledgegraph reasoning on graph types:Static,dynamic,and multi-modal[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2024,46(12):9456-9478. [8]CHEN B,WU J,LIU X,et al.Enhancing Temporal Knowledge Graph for Future Event Prediction with Long-Term Dense Graph[J].Tsinghua Science and Technology,2026,31(1):621-638. [9]JIN W,QU M,JIN X,et al.Recurrent Event Network:Autoregressive Structure Inferenceover Temporal Knowledge Graphs[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing(EMNLP).2020:6669-6683. [10]LIU Y,MA Y,HILDEBRANDT M,et al.Tlogic:Temporal lo-gical rules for explainable link forecasting on temporal knowledge graphs[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2022:4120-4127. [11]TISHBY N,PEREIRA F C,BIALEK W.The information bottleneck method[EB/OL].https://www.princeton.edu/~wbialek/our_papers/tishby+al_99.pdf. [12]VASWANI A,SHAZEER N,PARMAR N,et al.Attention is all you need[C]//Proceedings of the 31st International Confe-rence on Neural Information Processing Systems.2017:6000-6010. [13]CHEN B,XIAO C,ZHOU F.Natural evolution-based dual-level aggregation for temporal knowledge graph reasoning[C]//Findings of the association for computational linguistics:EMNLP 2024.2024:9274-9284. [14]LEBLAY J,CHEKOL M W.Deriving validity time in know-ledge graph[C]//Companion Proceedings of the Web Confe-rence 2018.2018:1771-1776. [15]BORDES A,USUNIER N,GARCIA-DURAN A,et al.Translating embeddings for modeling multi-relational data[C]//Proceedings of the 27th International Confefence on Neural Information Processing Systems.2013:2785-2795. [16]CHEN K,WANG Y,LI Y,et al.Rotateqvs:Representing temporal information as rotations in quaternion vector space for temporal knowledge graph completion[J].arXiv:2203.07993,2022. [17]DASGUPTA S S,RAY S N,TALUKDAR P.Hyte:Hyper-plane-based temporally aware knowledge graph embedding[C]//Proceedings of the 2018 Conference on Empirical Me-thods in Natural Language Processing.2018:2001-2011. [18]WANG J,WANG B,GAO J,et al.QDN:A quadruplet distributor network for temporal knowledge graph completion[J].IEEE Transactions on Neural Networks and Learning Systems,2024,35(10):14018-14030. [19]LI Z,JIN X,LI W,et al.Temporal knowledge graph reasoningbased on evolutional representation learning[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval.2021:408-417. [20]DENG S,RANGWALA H,NING Y.Dynamic knowledge graph based multi-event forecasting[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Disco-very & Data Mining.2020:1585-1595. [21]LI Y,SUN S,ZHAO J.TiRGN:Time-Guided Recurrent Graph Network with Local-Global Historical Patterns for Temporal Knowledge Graph Reasoning[C]//IJCAI.2022:2152-2158. [22]YANG L,WU F,ZHENG Z,et al.Heterogeneous Graph Information Bottleneck[C]//IJCAI.2021:1638-1645. [23]XU Y,OU J,XU H,et al.Temporal knowledge graph reasoning with historical contrastive learning[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2023:4765-4773. [24]HAN Z,CHEN P,MA Y,et al.Explainable subgraph reasoning for forecasting on temporal knowledge graphs[C]//Internatio-nal Conference on Learning Representations.2020. [25]TISHBY N,ZASLAVSKY N.Deep learning and the informa-tion bottleneck principle[C]//2015 IEEE Information Theory Workshop(ITW).IEEE,2015:1-5. [26]SUN Q,LI J,PENG H,et al.Graph structure learning with va-riational information bottleneck[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2022:4165-4174. [27]YU J,XU T,RONG Y,et al.Recognizing predictive substructures with subgraph information bottleneck[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2021,46(3):1650-1663. [28]WU T,REN H,LI P,et al.Graph information bottleneck[J].Advances in Neural Information Processing Systems,2020,33:20437-20448. [29]YUAN H,SUN Q,FU X,et al.Dynamic graph information bottleneck[C]//Proceedings of the ACM Web Conference 2024.2024:469-480. [30]HU Z,GUTIÉRREZ-BASULTO V,XIANG Z,et al.Trans-former-based entity typing in knowledge graphs[J].arXiv:2210.11151,2022. [31]XU Z,YE P,CHEN H,et al.Ruleformer:Context-aware rule mining over knowledge graph[C]//Proceedings of the 29th International Conference on Computational Linguistics.2022:2551-2560. [32]HAN Z,LIAO R,LIU B,et al.Enhanced temporal knowledgeembeddings with contextualized language representations[C]//ICLR.2022. [33]SUN H,GENG S,ZHONG J,et al.Graph hawkes transformer for extrapolated reasoning on temporal know-ledge graphs[C]//Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing.2022:7481-7493. [34]GAO Y,HE Y,KAN Z,et al.Learning joint structural and temporal contextualized knowledge embeddings for temporal know-ledge graph completion[C]//Findings of the Association for Computational Linguistics:ACL 2023.2023:417-430. [35]JANG E,GU S,POOLE B.Categorical reparameterization with gumbel-softmax[J].arXiv:1611.01144,2016. [36]ZHU C,CHEN M,FAN C,et al.Learning from history:Modeling temporal knowledge graphs with sequential copy-generation networks[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:4732-4740. [37]KAZEMI S M,GOEL R,EGHBALI S,et al.Time2vec:Lear-ning a vector representation of time[J].arXiv:1907.05321,2019. [38]SHANG C,TANG Y,HUANG J,et al.End-to-end structure-aware convolutional networks for knowledge base completion[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2019:3060-3067. [39]GARCIA-DURAN A,DUMANČIĆ S,NIEPERT M.Learning Sequence Encoders for Temporal Knowledge Graph Completion[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing.2018:4816-4821. [40]LEETARU K,SCHRODT P A.Gdelt:Global data on events,location,and tone,1979-2012[C]//ISA Annual Convention.Citeseer,2013:1-49. [41]LI Z,JIN X,LI W,et al.Temporal knowledge graph reasoning based on evolutional representation learning[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval.2021:408-417. [42]LI Z,GUAN S,JIN X,et al.Complex Evolutional PatternLearning for Temporal Knowledge Graph Reasoning[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics.2022:290-296. [43]DETTMERS T,MINERVINI P,STENETORP P,et al.Con-volutional 2d knowledge graph embeddings[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2018. [44]JIANG T,LIU T,GE T,et al.Towards time-aware knowledge graph completion[C]//Proceedings of COLING 2016,the 26th International Conference on Computational Linguistics.2016:1715-1724. [45]HAN Z,DING Z,MA Y,et al.Learning neural ordinary equations for forecasting future links on temporal knowledge graphs[C]//Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing.2021:8352-8364. [46]LIU K,ZHAO F,XU G,et al.RETIA:relation-entity twin-interact aggregation for temporal knowledge graph extrapolation[C]//2023 IEEE 39th International Conference on Data Engineering(ICDE).IEEE,2023:1761-1774. [47]LIU Z,TAN L,LI M,et al.SiMFy:A Simple Yet Effective Approach for Temporal Knowledge Graph Reasoning[C]//Fin-dings of the Association for Computational Linguistics:EMNLP 2023.2023:3825-3836. |
| [1] | 陈涵, 徐泽锋, 蒋究, 樊凡, 章军建, 何楚, 王文伟. 基于大语言模型和深度网络的认知评估量表自动诊断 Large Language Model and Deep Network Based Cognitive Assessment Automatic Diagnosis 计算机科学, 2026, 53(3): 41-51. https://doi.org/10.11896/jsjkx.250600034 |
| [2] | 李泽群, 丁飞. 基于双分支融合与分段域适应迁移学习的疲劳驾驶检测 Fatigue Driving Detection Based on Dual-branch Fusion and Segmented Domain AdaptationTransfer Learning 计算机科学, 2026, 53(3): 78-87. https://doi.org/10.11896/jsjkx.250500025 |
| [3] | 喻定, 李章维. 基于Transformer架构的RNA二级结构预测方法 Prediction Method of RNA Secondary Structure Based on Transformer Architecture 计算机科学, 2026, 53(3): 375-382. https://doi.org/10.11896/jsjkx.250100005 |
| [4] | 李佳豪, 荆军昌, 徐茜, 刘栋. GTKT:融合联通主义学习和多层时序图Transformer的知识追踪模型 GTKT:Knowledge Tracing Model Integrating Connectivism Learning and Multi-layer TemporalGraph Transformer 计算机科学, 2026, 53(2): 78-88. https://doi.org/10.11896/jsjkx.250700188 |
| [5] | 潘建, 汪绪豪. 融合多尺度特征和注意力机制的时间序列预测模型 Time Series Forecasting Model Integrating Multi-scale Features and Attention Mechanism 计算机科学, 2026, 53(2): 180-186. https://doi.org/10.11896/jsjkx.250100113 |
| [6] | 刘晨红, 李凤莲, 阳佳, 王夙喆, 陈桂军. 聚焦边界和多尺度特征融合的脑卒中病灶分割 Boundary-focused Multi-scale Feature Fusion Network for Stroke Lesion Segmentation 计算机科学, 2026, 53(2): 264-272. https://doi.org/10.11896/jsjkx.250300137 |
| [7] | 王成, 金城. 基于KAN的无监督多元时间序列异常检测网络 KAN-based Unsupervised Multivariate Time Series Anomaly Detection Network 计算机科学, 2026, 53(1): 89-96. https://doi.org/10.11896/jsjkx.241200190 |
| [8] | 胡海龙, 许祥伟, 李雅倩. 基于动态病情建模的药物组合推荐模型 Drug Combination Recommendation Model Based on Dynamic Disease Modeling 计算机科学, 2025, 52(9): 96-105. https://doi.org/10.11896/jsjkx.250300033 |
| [9] | 邓佳燕, 田时瑞, 刘香丽, 欧阳红巍, 焦韵嘉, 段明星. 基于多阶段行人特征挖掘的轨迹预测方法 Trajectory Prediction Method Based on Multi-stage Pedestrian Feature Mining 计算机科学, 2025, 52(9): 241-248. https://doi.org/10.11896/jsjkx.250700138 |
| [10] | 丁政泽, 聂仁灿, 李锦涛, 苏华平, 徐航. MTFuse:基于Mamba和Transformer的红外与可见光图像融合网络 MTFuse:An Infrared and Visible Image Fusion Network Based on Mamba and Transformer 计算机科学, 2025, 52(8): 188-194. https://doi.org/10.11896/jsjkx.240600106 |
| [11] | 刘华咏, 徐明慧. 基于混合注意力与偏振非对称损失的哈希图像检索 Hash Image Retrieval Based on Mixed Attention and Polarization Asymmetric Loss 计算机科学, 2025, 52(8): 204-213. https://doi.org/10.11896/jsjkx.240600057 |
| [12] | 刘亚俊, 纪庆革. 基于运动模式与时频域融合的行人轨迹预测 Pedestrian Trajectory Prediction Based on Motion Patterns and Time-Frequency Domain Fusion 计算机科学, 2025, 52(7): 92-102. https://doi.org/10.11896/jsjkx.250200011 |
| [13] | 黄星宇, 王丽会, 唐堃, 程欣宇, 张健, 叶晨. EFormer:基于分频和广注意力的高效Transformer医学图像配准模型 EFormer:Efficient Transformer for Medical Image Registration Based on Frequency Division and Board Attention 计算机科学, 2025, 52(7): 151-160. https://doi.org/10.11896/jsjkx.240400159 |
| [14] | 王有康, 程春玲. 基于跨模态单向加权的多模态情感分析模型 Multimodal Sentiment Analysis Model Based on Cross-modal Unidirectional Weighting 计算机科学, 2025, 52(7): 226-232. https://doi.org/10.11896/jsjkx.240600066 |
| [15] | 龙肖, 黄巍, 胡凯. 基于双向多层级交互网络的肺部CT图像分类 Bi-MI ViT:Bi-directional Multi-level Interaction Vision Transformer for Lung CT ImageClassification 计算机科学, 2025, 52(6A): 240700183-6. https://doi.org/10.11896/jsjkx.240700183 |
|
||
首页