Document-level relationship extraction based on evidence enhancement and multi-feature fusion

doi:10.11772/j.issn.1001-9081.2023101516

Abstract

Abstract:

Document-level Relationship Extraction （DocRE） aims at identifying all the relationships that exist between entity pairs in a document. Aiming at the problems of ineffective use of evidence sentences as well as document information， and multiple mentions of entities， a multi-feature fusion DocRE model named EMF （Evidence Multi-feature Fusion） was constructed based on evidence-enhanced contextual features. Firstly， entity types were added before and after entities， and relationship text features were associated with entity mentions to obtain relationship-specific entity features. Secondly， fragment representations were obtained through different convolutional kernels， and multi-granularity fragment-level features perceived by entity pairs were obtained through the attention mechanism. Meanwhile， contextual features highly correlated with the entity pairs were enhanced by using evidence distribution. Finally， the above features were fused for relationship classification， and during inference， the obtained evidence was composed into a pseudo-document and input into the classifier together with the original document for relationship classification. Experimental results on DocRED （Document-level Relation Extraction Dataset）， a DocRE dataset， show that when using BERT_base as the PLM encoder， compared with the state-of-the-art model EIDER （EvIDence-Enhanced DocRE）， the EMF model has the Ign F1 and F1 improved by 0.42 and 0.41 percentage points respectively， and the F1 reached 62.89%. It can be seen that the EMF model pays more attention to the parts that are related to entities and relationships， improves the extraction accuracy， and has a good interpretability.

Key words: document-level, relationship extraction, evidence, mention attention, fragment feature

摘要：

文档级关系抽取（DocRE）的目的是识别文档中实体对之间存在的所有关系。针对证据句子和文档信息未能被有效利用以及实体多提及的问题，在使用证据增强上下文特征的基础上，构建一种多特征融合的文档级关系抽取模型EMF（Evidence Multi-feature Fusion）。首先，在实体前后加上实体类型，将关系文本特征与实体提及进行关联，以获得特定于关系的实体特征。其次，通过不同卷积核获得片段表示，并通过注意力机制获得实体对感知的多粒度片段级特征；同时，利用证据分布增强与实体对高度相关的上下文特征。最后，融合以上特征进行关系分类，并在推理时将获得的证据组成伪文档与原文档一起输入分类器进行关系分类。在DocRE数据集DocRED（Document-level Relation Extraction Dataset）上的实验结果表明，使用BERT_base作为预训练语言模型编码器时，相较于先进模型EIDER（EvIDence-Enhanced DocRE），所提模型EMF的Ign F1和F1分别提高了0.42和0.41个百分点，F1达到了62.89%。EMF模型更关注与实体和关系相关的部分，可提高抽取的精度，并具有较好的可解释性。

关键词: 文档级, 关系抽取, 证据, 提及注意, 片段特征

CLC Number:

TP309.2

Xinyue YAN, Shuqun YANG, Yongbin GAO. Document-level relationship extraction based on evidence enhancement and multi-feature fusion[J]. Journal of Computer Applications, 2024, 44(11): 3379-3385.

颜新月, 杨淑群, 高永彬. 基于证据增强与多特征融合的文档级关系抽取[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3379-3385.

Figures/Tables 9

References 29

1	PENG N， POON H， QUIRK C， et al. Cross-sentence n-ary relation extraction with graph LSTMs［J］. Transactions of the Association for Computational Linguistics， 2017， 5： 101-115.
2	VERGA P， STRUBELL E， McCALLUM A. Simultaneously self-attending to all mentions for full-abstract biological relation extraction［C］// Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics： Human Language Technologies， Volume 1 （Long Papers）. Stroudsburg， PA： ACL， 2018：872-884.
3	YAO Y， YE D， LI P， et al. DocRED： a large-scale document-level relation extraction dataset［C］// Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. Stroudsburg， PA： ACL， 2019：764-777.
4	YE D， LIN Y， DU J， et al. Coreferential reasoning learning for language representation［C］// Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Stroudsburg， PA： ACL， 2020： 7170- 7186.
5	XU B， WANG Q， LYU Y， et al. Entity structure within and throughout： modeling mention dependencies for document-level relation extraction［C］// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto， CA： AAAI Press， 2021： 14149-14157.
6	LI J， XU K， LI F， et al. MRN： a locally and globally mention-based reasoning network for document-level relation extraction［C］// Findings of the Association for Computational Linguistics： ACL-IJCNLP 2021. Stroudsburg， PA： ACL， 2021： 1359-1370.
7	ZHOU W， HUANG K， MA T， et al. Document-level relation extraction with adaptive thresholding and localized context pooling［C］// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto， CA： AAAI Press， 2021： 14612-14620.
8	CAI R， ZHANG X， WANG H. Bidirectional recurrent convolutional neural network for relation classification［C］// Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics （Volume 1： Long Papers）. Stroudsburg， PA： ACL， 2016： 756-765.
9	ZHANG N， DENG S， SUN Z， et al. Relation adversarial network for low resource knowledge graph completion［C］// Proceedings of the Web Conference 2020. New York： ACM， 2020： 1-12.
10	QUIRK C， POON H. Distant supervision for relation extraction beyond the sentence boundary［C］// Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics： Volume 1， Long Papers. Stroudsburg， PA： ACL， 2017： 1171-1182.
11	ZENG S， XU R， CHANG B， et al. Double graph based reasoning for document-level relation extraction［C］// Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Stroudsburg， PA： ACL， 2020： 1630-1640.
12	XU W， CHEN K， ZHAO T. Document-level relation extraction with reconstruction［C］// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto， CA： AAAI Press， 2021： 14167-14175.
13	ZENG S， WU Y， CHANG B. SIRE： separate intra-and inter-sentential reasoning for document-level relation extraction［C］// Findings of the Association for Computational Linguistics： ACL-IJCNLP 2021. Stroudsburg， PA： ACL， 2021： 524-534.
14	VASWANI A， SHAZEER N， PARMAR N， et al. Attention is all you need［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook， NY： Curran Associates Inc.， 2017：6000-6010.
15	ZHANG N， CHEN X， XIE X， et al. Document-level relation extraction as semantic segmentation［C］// Proceedings of the 30th International Joint Conference on Artificial Intelligence. California： IJCAI， 2021： 3999-4006.
16	袁泉，徐雲鹏，唐成亮. 基于路径标签的文档级关系抽取方法［J］. 计算机应用， 2023， 43（4）：1029-1035.
	YUAN Q， XU Y P， TANG C L. Document level-relationship extraction method based on path labels［J］. Journal of Computer Applications， 2023， 43（4）： 1029-1035.
17	HUANG K， QI P， WANG G， et al. Entity and evidence guided document-level relation extraction［C］// Proceedings of the 6th Workshop on Representation Learning for NLP. Stroudsburg， PA： ACL， 2021： 307-315.
18	HUANG Q， ZHU S， FENG Y， et al. Three sentences are all you need： local path enhanced document relation extraction［C］// Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing （Volume 2： Short Papers）. Stroudsburg， PA： ACL， 2021： 998-1004.
19	TAN Q， HE R， BING L， et al. Document-level relation extraction with adaptive focal loss and knowledge distillation［C］// Findings of the Association for Computational Linguistics： ACL 2022. Stroudsburg， PA： ACL， 2022： 1672-1681.
20	DEVLIN J， CHANG M W， LEE K， et al. BERT： pre-training of deep bidirectional Transformers for language understanding［C］// Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics： Human Language Technologies， Volume 1 （Long and Short Papers）. Stroudsburg， PA： ACL， 2019： 4171-4186.
21	XIE Y， SHEN J， LI S， et al. EIDER： empowering document-level relation extraction with efficient evidence extraction and inference-stage fusion［C］// Findings of the Association for Computational Linguistics： ACL 2022. Stroudsburg， PA： ACL， 2022： 257-268.
22	LI J， SUN Y， JOHNSON R J， et al. BioCreative V CDR task corpus： a resource for chemical disease relation extraction［J］. Database， 2016， 2016： No.baw068.
23	WU Y， LUO R， LEUNG H C M， et al. RENET： a deep learning approach for extracting gene-disease associations from literature［C］// Proceedings of the 2019 International Conference on Research in Computational Molecular Biology， LNCS 11467. Cham： Springer， 2019： 272-284.
24	WOLF T， DEBUT L， SANH V， et al. Transformers： state-of-the-art natural language processing［C］// Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing： System Demonstrations. Stroudsburg， PA： ACL， 2020： 38-45.
25	LIU Y， OTT M， GOYAL N， et al. RoBERTa： a robustly optimized BERT pretraining approach［EB/OL］. ［2023-07-09］..
26	NAN G， GUO Z， SEKULIĆ I， et al. Reasoning with latent structure refinement for document-level relation extraction［C］// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg， PA： ACL， 2020： 1546-1557.
27	DU Y， MA T， WU L， et al. Improving long tailed document-level relation extraction via easy relation augmentation and contrastive learning［EB/OL］. ［2023-08-12］..
28	CHRISTOPOULOU F， MIWA M， ANANIADOU S. Connecting the dots：document-level neural relation extraction with edge-oriented graphs［C］// Proceedings of 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg， PA： ACL， 2019： 4925-4936.
29	BELTAGY I， LO K， COHAN A. SciBERT： a pretrained language model for scientific text［C］// Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg， PA： ACL， 2019： 3615-3620.
	TFoundation:his work is partially supported by Shanghai Local Capacity Building Project （21010501500）； Shanghai “Science and Technology Innovation Action Plan” Social Development Science and Technology Research Project （21DZ1204900）.

模型	PLM	开发集			测试集
模型	PLM	Ign F1	F1	Evi F1	Ign F1	F1
LSR	BERT_base	52.43	59.00	—	56.97	59.05
GAIN		59.14	61.22	—	59.00	61.24
BERT		—	54.16	—	—	53.20
SSAN		57.03	59.19	—	55.84	58.16
ATLOP		59.22	61.09	—	59.31	61.30
E2GRE		55.22	58.79	47.12	—	—
DocuNet		59.86	61.83	—	59.93	61.86
ERA		59.72	61.80		59.08	61.36
EIDER		60.51	62.48	51.27	60.96	62.77
EMF		60.93	62.89	52.35	61.10	62.88
Core	RoBERTa_large	57.35	59.43	—	57.90	60.25
SSAN		60.25	62.08	—	59.47	61.42
ATLOP		61.32	63.18	—	61.39	63.40
EMF		62.30	64.20	54.12	62.12	64.42

模型	PLM	开发集			测试集
模型	PLM	Ign F1	F1	Evi F1	Ign F1	F1
LSR	BERT_base	52.43	59.00	—	56.97	59.05
GAIN		59.14	61.22	—	59.00	61.24
BERT		—	54.16	—	—	53.20
SSAN		57.03	59.19	—	55.84	58.16
ATLOP		59.22	61.09	—	59.31	61.30
E2GRE		55.22	58.79	47.12	—	—
DocuNet		59.86	61.83	—	59.93	61.86
ERA		59.72	61.80		59.08	61.36
EIDER		60.51	62.48	51.27	60.96	62.77
EMF		60.93	62.89	52.35	61.10	62.88
Core	RoBERTa_large	57.35	59.43	—	57.90	60.25
SSAN		60.25	62.08	—	59.47	61.42
ATLOP		61.32	63.18	—	61.39	63.40
EMF		62.30	64.20	54.12	62.12	64.42

模型	F1
模型	CDR	GDA
BRAN	62.1	—
EGO	63.6	81.5
LSR	64.8	82.2
ATLOP-SciBERT	69.4	83.9
EMF-SciBERT	75.8	84.9

模型	F1
模型	CDR	GDA
BRAN	62.1	—
EGO	63.6	81.5
LSR	64.8	82.2
ATLOP-SciBERT	69.4	83.9
EMF-SciBERT	75.8	84.9

模型	Ign F1	F1	Evi F1
EMF	60.93	62.89	52.35
-evidence	59.60	61.82	42.79
-relation	58.72	61.40	42.73
-n-gram	58.05	61.08	42.75
-cls	57.55	60.22	42.80