基于深度学习的关系抽取研究进展

doi:10.16180/j.cnki.issn1007-7820.2025.07.006

Abstract

Abstract:

In natural language processing, as the core task, the research direction of entity relation extraction task has gradually shifted from rule-based learning and traditional machine learning to deep learning. At present, deep learning relationship extraction models widely use convolutional neural networks, recurrent neural networks and graph neural networks. This study summarizes the excellent relationship extraction models in each neural network, shows the evolution direction of each model by tracing the development history and trend of the model, and makes a comparative analysis of each method and model. Due to the continuous improvement of attention mechanism and other methods, the semantic analysis ability of relational extraction model has been significantly enhanced. In this study, the relevant improvement methods are reviewed, and the characteristics and experimental results of each method are described. This study introduces the common data sets in the field of relational extraction, and summarizes and compares the models with the best performance on each data set. The challenges in relation extraction are summarized and the solutions are proposed.

Key words: relation extraction, deep learning, neural network, information extraction, knowledge graph, convolution, Transformer, attention

CLC Number:

TP391.1

SHEN Yining, WANG Yiran, WU Cong. Research Progress of Relation Extraction Based on Deep Learning[J].Electronic Science and Technology, 2025, 38(7): 40-49.

Figures/Tables 11

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Table 1.

Table 2.

Table 3.

References 60

[1]	Ji S, Pan S, Cambria E, et al. A survey on knowledge graphs:Representation,acquisition and applications[J]. IEEE Transactions on Neural Networks and Learning Systems, 2022, 33(2):494-514.
[2]	Nguyen T H, Grishman R. Relation extraction:Perspective from convolutional neural networks[C]. Colorado: Proceedings of the First Workshop on Vector Space Modeling for Natural Language Processing, 2015:39-48.
[3]	Zeng D, Liu K, Lai S, et al. Relation classification via convolutional deep neural network[C]. Dublin:Proceedings of COLING,the Twenty-fifth International Conference on Computational Linguistics:Technical Papers, 2014: 2335-2344.
[4]	Zeng D, Liu K, Chen Y, et al. Distant supervision for relation extraction via piecewise convolutional neuralnetworks[C]. Lisbon: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2015:1753-1762.
[5]	Shen Y, Huang X. Attention-based convolutional neural network for semantic relation extraction[C]. Osaka:Proceedings of COLING,the Twenty-sixth International Conference on Computational Linguistics: Technical Papers, 2016:2526-2536.
[6]	Ji G, Liu K, He S, et al. Distant supervision for relation extraction with sentence-level attention and entity descriptions[C]. San Francisco: Proceedings of the AAAI Conference on Artificial Intelligence, 2017:897-905.
[7]	Ren Y, Zhao Y, Jin G, et al. Joint entity relation extraction model based on relationship box[C]. Taizhou: The Sixteenth International Congress on Image and Signal Processing,BioMedical Engineering and Informatics, 2023:1-6.
[8]	Hashimoto K, Miwa M, Tsuruoka Y, et al. Simple customization of recursive neural networks for semantic relation classification[C]. Seattle: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2013:1372-1376.
[9]	Socher R, Huval B, Manning C D, et al. Semantic compositionality through recursive matrix-vector spaces[C]. Jeju Island: Proceedings of the Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, 2012:1201-1211.
[10]	Xu Y, Mou L, Li G, et al. Classifying relations via long short term memory networks along shortest dependency paths[C]. Lisbon: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2015:1785-1794.
[11]	Wang H, Qin K, Lu G, et al. Direction-sensitive relation extraction using Bi-SDP attention model[J]. Knowledge-Based Systems, 2020, 198(6):105928-105930.
[12]	Rao S K. Nary relation extraction using web-scraped data[C]. Bangalore: International Conference on Advances in Electronics, Communication, Computing and Intelligent Information Systems, 2023:611-615.
[13]	Schlichtkrull M, Kipf T N, Bloem P, et al. Modeling Relational Data with Graph Convolutional Networks[C]. Cham: The Semantic Web, 2018: 593-607.
[14]	Guo Z, Zhang Y, Lu W. Attention guided graph convolutional networks for relation extraction[C]. Florence: Proceedings of the Fifty-seventh Annual Meeting of the Association for Computational Linguistics, 2019:241-251.
[15]	Sun Y, Wang J, Lin H, et al. Knowledge guided attention and graph convolutional networks for chemical-disease relation extraction[J]. IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2023, 20(1):489-499.
[16]	Zhang Z, Meng F, Liu X, et al. An improved relation extraction method based on information control injection and attention-guided densely connected graph convolutional network[C]. Gold Coast: International Joint Conference on Neural Networks, 2023:1-7.
[17]	Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[EB/OL]. (2023-08-01)[2024-01-18]. http://arxiv.org/abs/1706.03762.
[18]	Xiao Y, Jin Y, Cheng R, et al. Hybrid attention-based transformer block model for distant supervision relation extraction[J]. Neurocomputing, 2022, 470(1):29-39.
[19]	Li Z, Chen H, Qi R, et al. DocR-BERT:Document-level R-BERT for chemical-induced disease relation extraction via gaussian probability distribution[J]. IEEE Journal of Biomedical and Health Informatics, 2022, 26(3):1341-1352.
[20]	Wu Y, Bamman D, Russell S. Adversarial training for relation extraction[C]. Copenhagen: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2017:1778-1783.
[21]	Goodfellow I, J Pouget-Abadie J, Mirza M, et al. Generative adversarial networks[J]. Communications of the ACM, 2020, 63(11):139-144.
[22]	Qin P, Xu W, Wang W Y. DSGAN:Generative adversarial training for distant supervision relation extraction[C]. Melbourne: Proceedings of the Fifty-sixth Annual Meeting of the Association for Computational Linguistics, 2018:496-505.
[23]	Lyu L, Shen Y, Zhang S. The advance of reinforcement learning and deep reinforcement learning[C]. Changchun: IEEE International Conference on Electrical Engineering, Big Data and Algorithms, 2022: 644-648.
[24]	Zhu Z, Lin K, Jain A K, et al. Transfer learning in deep reinforcement learning:A survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(11):13344-13362.
[25]	Qin P, Xu W, Wang W Y. Robust distant supervision relation extraction via deep reinforcement learning[C]. Melbourne: Proceedings of the Fifty-sixth Annual Meeting of the Association for Computational Linguistics, 2018:2137-2147.
[26]	Zhou K, Luo X, Wang H, et al. Multi-task learning for relation extraction[C]. Portland: IEEE Thirty-first International Conference on Tools with Artificial Intelligence, 2019:1480-1487.
[27]	Elallaly E, Sarrouti M, En-Nahnahi N, et al. MTTLADE:A multi-task transfer learning-based method for adverse drug events extraction[J]. Information Processing and Management, 2021, 58(3):102473-102475.
[28]	Yao Y, Ye D, Li P, et al. DocRED:A large-scale document-level relation extraction dataset[C]. Florence: Proceedings of the Fifth-seventh Annual Meeting of the Association for Computational Linguistics, 2019:764-777.
[29]	Hendrickx I, Kim S N, Kozareva Z, et al. SemEval-2010 Task 8:Multi-way classification of semantic relations between pairs of nominals[C]. Uppsala: Proceedings of the Fifth International Workshop on Semantic Evaluation, 2010:33-38.
[30]	Doddington G, Mitchell A, Przybocki M, et al. The automatic content extraction program-tasks,data and evaluation[C]. Lisbon: Proceedings of the Fourth International Conference on Language Resources and Evaluation, 2004:33-38.
[31]	Riedel S, Yao L, McCallum A. Modeling relations and their mentions without labeled text[C]. Berlin: Machine Learning and Knowledge Discovery in Databases, 2010:148-163.
[32]	Ma Y, Wang A, Okazaki N. DREEAM:Guiding attention with evidence for improving document-level relation extraction[C]. Dubrovnik: Proceedings of the Seventeenth Conference of the European Chapter of the Association for Computational Linguistics, 2023:1971-1983.
[33]	Tan Q, He R, Bing L, et al. Document-level relation extraction with adaptive focal loss and knowledge distillation[C]. Dublin: Findings of the Association for Computational Linguistics:ACL, 2022:1672-1681.
[34]	Xu B, Wang Q, Lyu Y, et al. Entity structure within and throughout:Modeling mention dependencies for document-level relation extraction[C]. Online: Proceedings of the AAAI Conference on Artificial Intelligence, 2021:14149-14157.
[35]	Xiao Y, Zhang Z, Mao Y, et al. SAIS:Supervising and augmenting intermediate steps for Document-Level Relation Extraction[C]. Seattle:Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2022:2395-2409.
[36]	Xie Y, Shen J, Li S, et al. Eider:Empowering document-level relation extraction with efficient evidence extraction and inference-stage fusion[C]. Dublin: Findings of the Association for Computational Linguistics, 2022:257-268.
[37]	Li B, Yu D, Ye W, et al. Sequence generation with label augmentation for relation extraction[EB/OL]. (2023-02-09)[2024-01-18]. http://arxiv.org/abs/2212.14266.
[38]	Li C, Tian Y. Downstream model design of pre-trained language model for relation extraction task[EB/OL]. (2020-04-07)[2024-01-18]. http://arxiv.org/abs/2004.03786.
[39]	Li J, Katsis Y, Baldwin T, et al. SPOT: Knowledge-enhanced language representations for information extraction[EB/OL]. (2022-10-23) [2024-01-18]. http://arxiv.org/abs/2208.09625.
[40]	Li B, Ye W, Zhang J, et al. Reviewing labels:Label graph network with top-k prediction set for relation extraction[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37(11):13051-13058.
[41]	Zhao K, Xu H, Cheng Y, et al. Representation iterative fusion based on heterogeneous graph neural network forjoint entity and relation extraction[J]. Knowledge-Based Systems, 2021, 219(5):106888-106889.
[42]	Ye D, Lin Y, Li P, et al. Packed levitated marker for entity and relation extraction[C]. Dublin: Proceedings of the Sixtieth Annual Meeting of the Association for Computational Linguistics, 2022:4904-4917.
[43]	Liu T, Jiang Y E, Monath N, et al. Autoregressive structured prediction with language models[C]. Abu Dhabi: Findings of the Association for Computational Linguistics, 2022:993-1005.
[44]	Sainz O, García-Ferrero I, Agerri R, et al. GoLLIE: Annotation guidelines improve zero-shot information-extraction[EB/OL]. (2023-12-11)[2024-01-18]. http://arxiv.org/abs/2310.03668.
[45]	Zhong Z, Chen D. A frustratingly easy approach for entity and relation extraction[C]. Online:Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2021:50-61.
[46]	Wang J, Lu W. Two are better than one:Joint entity and relation extraction with table-sequence encoders[C]. Online: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2020: 1706-1721.
[47]	Tang W, Xu B, Zhao Y, et al. UniRel:Unified representation and interaction for joint relational triple extraction[C]. Abu Dhabi: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2022:7087-7099.
[48]	Huguet Cabot P L, Navigli R. REBEL:Relation extraction by end-to-end language generation[C]. Punta Cana: Findings of the Association for Computational Linguistics, 2021:2370-2381.
[49]	Zhao F, Jiang Z, Kang Y, et al. Adjacency list oriented relational fact extraction via adaptive multi-task learning[C]. Online: Findings of the Association for Computational Linguistics, 2021: 3075-3087.
[50]	Sui D, Zeng X, Chen Y, et al. Joint entity and relation extraction with set prediction networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 12(4):1-12.
[51]	Wei Z, Guo W, Zhang Y, et al. Bidirectional matching and aggregation network for few-shot relation extraction[J]. PeerJ Computer Science, 2023, 9(17):1272-1273.
[52]	Cao Y, Kuang J, Gao M, et al. Learning relation prototype from unlabeled texts for long-tail relation extraction[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(2):1761-1774.
[53]	Wu H, He Y, Chen Y, et al. Improving few-shot relation extraction through semantics-guided learning[J]. Neural Networks, 2024, 169(1):453-461.
[54]	Dong Y, Xu X. Relational distance and document-level contrastive pre-training based relation extraction model[J]. Pattern Recognition Letters, 2023, 167(3):132-140.
[55]	Wang N, Chen T, Ren C, et al. Document-level relation extraction with multi-layer heterogeneous graphattention network[J]. Engineering Applications of Artificial Intelligence, 2023, 123(8):106212-106213.
[56]	Gharagozlou H, Mohammadzadeh J, Bastanfard A, et al. Semantic relation extraction:A review of approaches, datasets and evaluation methods with looking at the methods and datasets in the persian language[J]. ACM Transactions on Asian and Low-Resource Language Information Processing, 2023, 22(7):1-29.
[57]	Song Z, Wan L. Research of Chinese relation extraction based on BERT[C]. Shenyang: IEEE the Third International Conference on Power, Electronics and Computer Applications, 2023:841-845.
[58]	Tuo M, Yang W. Review of entity relation extraction[J]. Journal of Intelligent and Fuzzy Systems:Applications in Engineering and Technology, 2023, 44(5):7391-7405.
[59]	Lin Y, Lu K, Yu S, et al. Multimodal learning on graphs for disease relation extraction[J]. Journal of Biomedical Informatics, 2023, 143(7):104415-104416.
[60]	Zhang Z, Fang M, Wu R, et al. Large-scale biomedical relation extraction across diverse relation types:Model development and usability study on COVID-19[J]. Journal of Medical Internet Research, 2023, 25(9): 48115-48116.

方法	优点	缺点	计算复杂度
CNN	能够捕捉局部语义特征,参数共享,适用于固定输入任务	全局信息捕捉困难,不适用于序列任务	文本处理时,卷积核较小,计算复杂度较低
RNN	能够捕捉顺序信息,建立长距离依赖关系,处理不同长度的输入序列	存在梯度消失及梯度爆炸,计算效率低	时间步的计算复杂度为线性,与序列长度成正比
LSTM	在RNN基础上,克服了梯度消失,缓解了梯度爆炸	训练缓慢,参数量较高	计算复杂度高于RNN,与序列长度成正比
GCN	能够处理不同大小图结构数据且参数共享,可捕获实体上下文信息	全局信息捕捉较少,训练缓慢,计算量大	在每一层需要对N个节点的邻域进行聚合,计算复杂度为O(N)
Attention	能够实现全局信息建模,灵活性强,适应不定长序列	处理复杂关系时性能降低,训练缓慢,计算量大	自注意力与多头注意力的计算复杂度分别为O(n²)和O(h×n²),h为注意力头数

模型	准确率/%	召回率/%	F1/%	数据集	模型	准确率/%	召回率/%	F1/%	数据集
CDNN	-	-	82.70	SemEval-2010 Task 8	Bi-SDP-ATT	-	-	85.10	SemEval-2010 Task 8
PCNN	78.30	-	-	NYT	AGGCN	-	-	85.70	SemEval-2010 Task 8
ACNN	-	-	85.90	SemEval-2010 Task 8	KGAGN	-	-	73.30	BioCreative-V CDR
APCNN	81.30	-	-	NYT	ICI-ATT-GCN	-	-	86.70	NYT
JREMRB	-	-	74.88	CoNLL04	Trans-SA	81.50	-	-	NYT
MV-RNN	-	-	82.40	SemEval-2010 Task 8	Doc-BERT	-	-	85.10	SemEval-2010 Task 8
SDP-LSTM	-	-	83.70	SemEval-2010 Task 8

数据集	模型
DocRED	DREEAM^[32]	67.5
	KD-Rb-I^[33]	67.2
	SSAN-RoBERTa-large+Adaptation^[34]	65.9
	SAIS-RoBERTa-large^[35]	65.1
	Eider-RoBERTa-large^[36]	64.8
SemEval-2010 Task8	RIFRE^[37]	91.3
	REDN^[38]	91.0
	SPOT^[39]	90.6
	KLG^[40]	90.5
	RELA^[41]	90.4
ACE 2005	PL-Marker^[42]	73.0
	ASP+T5-3B^[43]	72.7
	GoLLIE^[44]	70.1
	PURE^[45]	69.4
	Table-Sequence^[46]	67.6
NYT	UniRel^[47]	93.7
	REBEL^[48]	93.4
	REBEL(no pre-training)^[48]	93.1
	DIRECT^[49]	92.5
	SPN^[50]	92.5

Research Progress of Relation Extraction Based on Deep Learning

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 11

References 60

Related Articles 15

Metrics

Comments

Recommended 0

[1]	NI Tingxuan, SONG Yan. A Non-Iterative Routing CapsNet Based on Agglomerator Feature Extraction [J]. Electronic Science and Technology, 2025, 38(8): 27-32.
[2]	YANG Yuanxing, DING Xudong, WANG Junchao, WU Dong. Research on BP Neural Network PID Control Algorithm of Refrigeration System Based on Smith Predictor [J]. Electronic Science and Technology, 2025, 38(8): 42-48.
[3]	CHEN Qingjiang, LI Zongying. Underwater Image Enhancement Combining Parallel Transformer and Residual U-Net Networks [J]. Electronic Science and Technology, 2025, 38(8): 57-65.
[4]	WANG Lin, WANG Linxian, LI Weishuo, WANG Tao, LUO Zhiheng, ZHU Jiajun. Short-Term Photovoltaic Power Generation Forecasting Based on Similar Weather Fluctuation Patterns [J]. Electronic Science and Technology, 2025, 38(8): 79-86.
[5]	CHENG Weihao, YANG Xiaocheng, WU Lin, YAN Jingye, JIANG Mingfeng, WEI Bo. Reconstruction Algorithm of Synthetic Aperture Interferometer Radiometer Based on Improved Transformer [J]. Electronic Science and Technology, 2025, 38(8): 94-100.
[6]	ZHANG Bin, DING Haibing, WANG Jing, XUE Qianzhong. Dual-Band U-Slot Patch Antenna Optimization Using Neural Network Model [J]. Electronic Science and Technology, 2025, 38(7): 34-39.
[7]	WANG Junchao, DING Xudong, YANG Yuanxing, LIU Yuting, YANG Yuping. Research on Decoupling Control of Refrigeration System Based on Neural Network Inverse Model [J]. Electronic Science and Technology, 2025, 38(7): 82-88.
[8]	TANG Yuliang, ZHANG Xuanxiong. A Parking Space Detection Method Based on Deep Learning [J]. Electronic Science and Technology, 2025, 38(6): 23-29.
[9]	CHEN Qingjiang, YANG Shuang. A Dual Branch Image Dehazing Algorithm Based on Multi-Layer Feature Enhancement [J]. Electronic Science and Technology, 2025, 38(6): 30-38.
[10]	LIU Xueli, DU Hongbo, YUAN Xuefeng, ZHU Lijun. Ink Transfer Algorithm Based on CBAM Residual Block Combined with Texture Sampler [J]. Electronic Science and Technology, 2025, 38(6): 65-73.
[11]	SHEN Tingqian, LU Yujun, XIN Hao, WU Hanchao, WANG Shinan. Steel Plate Surface Defect Detection Based on Improved YOLOv5 [J]. Electronic Science and Technology, 2025, 38(6): 82-88.
[12]	CHEN Manman, YU Lianzhi. Deep BiGRU and DPCS Sentiment Analysis Model Combined with Coattention Network [J]. Electronic Science and Technology, 2025, 38(5): 22-30.
[13]	SUN Kuiyi, WANG Jingmei. Research on Active Clamping Technique for Flyback Converter [J]. Electronic Science and Technology, 2025, 38(5): 76-82.
[14]	ZHANG Haibo, PAN Pengcheng, ZHENG Feng. Corrosion Rate Prediction of Grounding Grid Based on Improved Whale Algorithm Optimization [J]. Electronic Science and Technology, 2025, 38(5): 8-14.
[15]	HU Jinlei, TIAN Engang, QU Feng. Dual-Path Real-Time Semantic Segmentation Network with Channel-Level Global Attention Mechanism [J]. Electronic Science and Technology, 2025, 38(5): 83-88.