---
title: "Extracting Relational Facts by an End-to-End Neural Model with Copy Mechanism — Zeng 2018"
description: "The proposed model is an end-to-end Seq2Seq network with a copy mechanism. 16 references, cited by 88 papers here."
authors: ["Xiangrong Zeng", "Daojian Zeng", "Shizhu He", "Kang Liu", "Jun Zhao"]
year: 2018
venue: "Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/P18-1047/"
doi: "10.18653/v1/p18-1047"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/zeng-2018-extracting-relational-facts-end-end/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zeng-2018-extracting-relational-facts-end-end/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2018](https://md.research.elamur.ai/free/en/papers/2018/) / Extracting Relational Facts by an End-to-End Neural Model with Copy Mechanism

# Extracting Relational Facts by an End-to-End Neural Model with Copy Mechanism

**Authors:** Xiangrong Zeng, [Daojian Zeng](https://md.research.elamur.ai/free/en/authors/daojian-zeng/), [Shizhu He](https://md.research.elamur.ai/free/en/authors/shizhu-he/), [Kang Liu](https://md.research.elamur.ai/free/en/authors/kang-liu/) and [Jun Zhao](https://md.research.elamur.ai/free/en/authors/jun-zhao/) · **Year:** 2018 · **Venue:** Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/P18-1047/) · [PDF](https://aclanthology.org/P18-1047.pdf) · [DOI 10.18653/v1/p18-1047](https://doi.org/10.18653/v1/p18-1047) · [OpenAlex](https://openalex.org/W2798734500)

**Citations:** 519 citations (OpenAlex, 2026-10-14) · [88](https://md.research.elamur.ai/free/en/papers/zeng-2018-extracting-relational-facts-end-end/cited-by/) in our library, of the 140,884 papers whose bibliographies we read

## Summary
The paper addresses joint extraction of relational triplets (two entities plus a relation) from raw sentences when triplets overlap. The authors sort sentences into three types: Normal (no shared entities), EntityPairOverlap (EPO, two triplets share an entity pair) and SingleEntityOverlap (SEO, triplets share only one entity). Earlier tagging-based joint models assign a single tag to each word, so they cannot handle overlap. The proposed model is an end-to-end Seq2Seq network with a copy mechanism. A bi-directional RNN encodes the sentence. The decoder then produces triplets in three steps: it predicts a relation, copies the head entity from the input, and copies the tail entity. Because an entity can be copied any number of times, it can take part in several triplets. Two decoding variants are compared: OneDecoder, a single unified decoder for all triplets, and MultiDecoder, a separate decoder for each triplet. On the NYT and WebNLG datasets, both variants clearly beat the NovelTagging baseline, and MultiDecoder performs best. This matters for building knowledge bases from text, where overlapping facts are common.

## Key points
- **Reported gains:** the models beat the prior state of the art (NovelTagging, Zheng et al., 2017) by 39.8% on NYT and 31.1% on WebNLG. NovelTagging has higher precision but much lower recall, while the new models are more balanced.
- **Decoding scheme:** each triplet takes three decoder steps (relation, then first entity, then second entity). Attention uses a selu activation. Generation stops when the decoder outputs an NA triplet.
- **Entity generation (Table 3):** MultiDecoder reaches F1 0.862 on NYT and 0.821 on WebNLG, versus 0.858 and 0.745 for OneDecoder.
- **Relation generation (Table 4):** the two variants are comparable, with F1 0.874 vs 0.870 on NYT and 0.759 vs 0.751 on WebNLG.
- **Effect of triplet count:** the NYT test set was split by the number of triplets per sentence (1, 2, 3, 4, ≥5). NovelTagging is best only for single-triplet sentences and degrades sharply as the count grows. The proposed models lose some accuracy on the Normal class and score lower overall on EPO and SEO sentences than on Normal ones.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

Fields: [Natural language processing](https://md.research.elamur.ai/free/en/shelves/nlp/)

Topics: [🔎 Information extraction](https://md.research.elamur.ai/free/en/topics/information-extraction/)

## Institutions

- [Chinese Academy of Sciences](https://md.research.elamur.ai/free/en/government/chinese-academy-of-sciences/) — Government body, Beijing, China · 51 works · 17,412 citations
- [Institute of Automation](https://md.research.elamur.ai/free/en/labs/institute-of-automation/) — Research institute, Beijing, China · 17 works · 13,299 citations
- [University of Chinese Academy of Sciences](https://md.research.elamur.ai/free/en/universities/university-of-chinese-academy-of-sciences/) — University, Beijing, China · 24 works · 6,312 citations
- [Changsha University of Science and Technology](https://md.research.elamur.ai/free/en/universities/changsha-university-of-science-and-technology/) — University, Changsha, China · 2 works · 716 citations

## Cite

Cite as: Xiangrong Zeng, Daojian Zeng, Shizhu He, Kang Liu and Jun Zhao. 2018. Extracting Relational Facts by an End-to-End Neural Model with Copy Mechanism. Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{zeng2018extracting,
  title = {{Extracting Relational Facts by an End-to-End Neural Model with Copy Mechanism}},
  author = {Xiangrong Zeng, Daojian Zeng, Shizhu He, Kang Liu and Jun Zhao},
  booktitle = {Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2018},
  doi = {10.18653/v1/p18-1047},
  url = {https://doi.org/10.18653/v1/p18-1047}
}
```

- [← previous Higher-Order Coreference Resolution with Coarse-to-Fine Inference](https://md.research.elamur.ai/free/en/papers/lee-2018-higher-order-coreference-resolution-coarse/)
- [next → Linguistically-Informed Self-Attention for Semantic Role Labeling](https://md.research.elamur.ai/free/en/papers/strubell-2018-linguistically-informed-self-attention-semantic/)
- [↑ back to papers of 2018](https://md.research.elamur.ai/free/en/papers/2018/)

## References

16 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Yee Seng Chan and Dan Roth. 2011. Exploiting syntactico-semantic structures for relation extraction. In Proceedings of ACL, pages 551–560. [find](https://scholar.google.com/scholar?q=Yee+Seng+Chan+and+Dan+Roth%2E+2011%2E+Exploiting+syntactico%2Dsemantic+structures+for+relation+extraction%2E+In+Proceedings+of+ACL%2C+pages+551%E2%80%93560%2E)
2. Sebastian Riedel, Limin Yao, and Andrew McCallum. 2010. Modeling relations and their mentions without labeled text. In Proceedings of ECML PKDD, pages 148–163. [find](https://scholar.google.com/scholar?q=Sebastian+Riedel%2C+Limin+Yao%2C+and+Andrew+McCallum%2E+2010%2E+Modeling+relations+and+their+mentions+without+labeled+text%2E+In+Proceedings+of+ECML+PKDD%2C+pages+148%E2%80%93163%2E)
3. Junyoung Chung, Caglar Gulcehre, KyungHyun Cho, and Yoshua Bengio. 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555. [find](https://scholar.google.com/scholar?q=Junyoung+Chung%2C+Caglar+Gulcehre%2C+KyungHyun+Cho%2C+and+Yoshua+Bengio%2E+2014%2E+Empirical+evaluation+of+gated+recurrent+neural+networks+on+sequence+modeling%2E+arXiv+preprint+arXiv%3A1412%2E3555%2E)
4. Kun Xu, Yansong Feng, Songfang Huang, and Dongyan Zhao. 2015a. Semantic relation classification via convolutional neural networks with simple negative sampling. In Proceedings of EMNLP, pages 536–540. [find](https://scholar.google.com/scholar?q=Kun+Xu%2C+Yansong+Feng%2C+Songfang+Huang%2C+and+Dongyan+Zhao%2E+2015a%2E+Semantic+relation+classification+via+convolutional+neural+networks+with+simple+negative+sampling%2E+In+Proceedings+of+EMNLP%2C+pages+536%E2%80%93540%2E)
5. Li Dong and Mirella Lapata. 2016. Language to logical form with neural attention. In Proceedings of ACL, pages 33–43. [our card](https://md.research.elamur.ai/free/en/papers/dong-2016-language-logical-form-neural-attention/)
6. Yan Xu, Lili Mou, Ge Li, Yunchuan Chen, Hao Peng, and Zhi Jin. 2015b. Classifying relations via long short term memory networks along shortest dependency paths. In Proceedings of EMNLP, pages 1785–1794. [our card](https://md.research.elamur.ai/free/en/papers/xu-2015-classifying-relations-long-short-term/)
7. Claire Gardent, Anastasia Shimorina, Shashi Narayan, and Laura Perez-Beltrachini. 2017. Creating training corpora for nlg micro-planners. In Proceedings of ACL, pages 179–188. [our card](https://md.research.elamur.ai/free/en/papers/gardent-2017-creating-training-corpora-nlg-micro/)
8. Jiatao Gu, Zhengdong Lu, Hang Li, and Victor O.K. Li. 2016. Incorporating copying mechanism in sequence-to-sequence learning. In Proceedings of ACL, pages 1631–1640. [our card](https://md.research.elamur.ai/free/en/papers/gu-2016-incorporating-copying-mechanism-sequence-sequence/)
9. Xiaofeng Yu and Wai Lam. 2010. Jointly identifying entities and extracting relations in encyclopedia text via a graphical model approach. In Proceedings of COLING, pages 1399–1407. [find](https://scholar.google.com/scholar?q=Xiaofeng+Yu+and+Wai+Lam%2E+2010%2E+Jointly+identifying+entities+and+extracting+relations+in+encyclopedia+text+via+a+graphical+model+approach%2E+In+Proceedings+of+COLING%2C+pages+1399%E2%80%931407%2E)
10. Pankaj Gupta, Hinrich Schtze, and Bernt Andrassy. 2016. Table filling multi-task recurrent neural network for joint entity and relation extraction. In Proceedings of COLING, pages 2537–2547. [find](https://scholar.google.com/scholar?q=Pankaj+Gupta%2C+Hinrich+Schtze%2C+and+Bernt+Andrassy%2E+2016%2E+Table+filling+multi%2Dtask+recurrent+neural+network+for+joint+entity+and+relation+extraction%2E+In+Proceedings+of+COLING%2C+pages+2537%E2%80%932547%2E)
11. Dmitry Zelenko, Chinatsu Aone, and Anthony Richardella. 2003. Kernel methods for relation extraction. J. Mach. Learn. Res., 3:1083–1106. [our card](https://md.research.elamur.ai/free/en/papers/zelenko-2002-kernel-methods-relation-extraction/)
12. Daojian Zeng, Kang Liu, Siwei Lai, Guangyou Zhou, and Jun Zhao. 2014. Relation classification via convolutional deep neural network. In Proceedings of COLING, pages 2335–2344. [our card](https://md.research.elamur.ai/free/en/papers/zeng-2014-relation-classification-convolutional-deep-neural/)
13. Shizhu He, Cao Liu, Kang Liu, and Jun Zhao. 2017. Generating natural answers by incorporating copying and retrieving mechanisms in sequence-to-sequence learning. In Proceedings of ACL, pages 199–208. [find](https://scholar.google.com/scholar?q=Shizhu+He%2C+Cao+Liu%2C+Kang+Liu%2C+and+Jun+Zhao%2E+2017%2E+Generating+natural+answers+by+incorporating+copying+and+retrieving+mechanisms+in+sequence%2Dto%2Dsequence+learning%2E+In+Proceedings+of+ACL%2C+pages+199%E2%80%93208%2E)
14. Meishan Zhang, Yue Zhang, and Guohong Fu. 2017. End-to-end neural relation extraction with global optimization. In Proceedings of EMNLP, pages 1730– 1740. [find](https://scholar.google.com/scholar?q=Meishan+Zhang%2C+Yue+Zhang%2C+and+Guohong+Fu%2E+2017%2E+End%2Dto%2Dend+neural+relation+extraction+with+global+optimization%2E+In+Proceedings+of+EMNLP%2C+pages+1730%E2%80%93+1740%2E)
15. Iris Hendrickx, Su Nam Kim, Zornitsa Kozareva, Preslav Nakov, Diarmuid Ó Séaghdha, Sebastian Padó, Marco Pennacchiotti, Lorenza Romano, and Stan Szpakowicz. 2010. Semeval-2010 task 8: Multi-way classification of semantic relations between pairs of nominals. In Proceedings of the 5th International… [our card](https://md.research.elamur.ai/free/en/papers/hendrickx-2010-semeval-2010-task-8-multi/)
16. Suncong Zheng, Feng Wang, Hongyun Bao, Yuexing Hao, Peng Zhou, and Bo Xu. 2017. Joint extraction of entities and relations based on a novel tagging scheme. In Proceedings of ACL, pages 1227–1236. [our card](https://md.research.elamur.ai/free/en/papers/zheng-2017-joint-extraction-entities-relations-based/)