Relation Extraction in 2019

This repo covers almost all the papers related to Relation Extraction in ACL, EMNLP, NAACL, AAAI, IJCAI in 2019. (also contains some papers from arxiv)

RE in 2018

⭐ is the recommended papers.

Use tags to search papers you like.

tags: |NRC | DSRE | PGM | Combining Direct Supervision | GNN | new perspective | new dataset | joint extraction of relations and entities | few shot | BERT | path | imbalance | trick | KBE | RL | cross bag | ML | GAN | false negative | BERT | ... |

DSRE: Distant Supervised Relation Extraction

NRC: Neural Relation Classification

KBE: Knowledge Base Embedding

RL: Reinforcement Learning

arxiv

1. ⭐ A Novel Hierarchical Binary Tagging Framework for Joint Extraction of Entities and Relations Zhepei Wei. Jianlin Su, Yue Wang, Yuan Tian, Yi Chang paper blog

   | joint extraction of relations and entities | BERT |

   This paper proposes a two-stage tagging scheme for joint extraction of entities and relations. The performance significant beats all of the previous works in NYT and WebNLG.

   my concern: the first stage, subject detection is more difficult than entity detection, so ...

NAACL 2019

1. Structured Minimally Supervised Learning for Neural Relation Extraction Fan Bai and Alan Ritter NAACL 2019 paper code

   | PGM | DSRE |

   This paper adds a PGM inference into training stage.

2. Combining Distant and Direct Supervision for Neural Relation Extraction Iz Beltagy, Kyle Lo and Waleed Ammar NAACL 2019 paper code

   | Combining Direct Supervision | DSRE |

   This paper combines direct supervision and distant supervision. It innovatively uses direct supervision for training sigmoid attention in a multi-task way. Further, when applying to the CNN backbone with different filter sizes, adding entity embedding as additional inputs is a useful trick, which performs comparable to RESIDE and better than PCNN-ATT. After combining the supervised sigmoid attention, this paper become a new sota.

3. Distant Supervision Relation Extraction with Intra-Bag and Inter-Bag Attentions Ye, Zhi-Xiu and Ling, Zhen-Hua NAACL 2019 paper code

   | DSRE | cross bag |

   In this paper, we have proposed a neural network with intra-bag and inter-bag attentions to cope with the noisy sentence and noisy bag problems in distant supervision relation extraction. First, relation-aware bag representations are calculated by a weighted sum of sentence embeddings where the noisy sentences are expected to have smaller weights. Further, an inter-bag attention module is designed to deal with the noisy bag problem by 2818 calculating the bag-level attention weights dynamically during model training.

4. A Richer-but-Smarter Shortest Dependency Path with Attentive Augmentation for Relation Extraction Duy-Cat Can, Hoang-Quynh Le, Quang-Thuy Ha, Nigel Collier NAACL 2019 paper code (code not available on 2019/06/28)

   | path | NRC |

   In this paper, we have presented RbSP, a novel representation of relation between two nominals in a sentence that overcomes the disadvantages of traditional SDP. Our RbSP is created by using multilayer attention to choose relevant information to augment a token in SDP from its child nodes.

5. Connecting Language and Knowledge with Heterogeneous Representations for Neural Relation Extraction Peng Xu and Denilson Barbosa NAACL 2019 paper code

   | KBE | DSRE |

   Multi-task learning: DSRE + KBE

6. ⭐ GAN Driven Semi-distant Supervision for Relation Extraction Pengshuai Li, Xinsong Zhang, Weijia Jia, Hai Zhao NAACL2019 paper

   | GAN | DSRE | false negative |

   To alleviate the effect of false negative instances, there are two possible ways. One is improving the accuracy of the automatically labeled dataset, and the other is properly leveraging unlabeled instances which cannot be labeled as positive or negative

   We assume that if an entity is relevant to another entity, its name is possibly mentioned in the description of the other entity.

7. Exploiting Noisy Data in Distant Supervision Relation Classification Kaijia Yang, Liang He, Xin-yu Dai, Shujian Huang, Jiajun Chen NAACL2019 paper

   | DSRE | RL | false negative |

   The methodology is suspicious. Not recommended.

8. ⭐ Long-tail Relation Extraction via Knowledge Graph Embeddings and Graph Convolution Networks Ningyu Zhang, Shumin Deng, Zhanlin Sun,Guanying Wang, Xi Chen, Wei Zhang, Huajun Chen∗paper

   | long tail | DSRE | GCN |

   This paper focus on the long tail relation problem in DSRE dataset (NYT). As there only has limited data for long tail relation, this paper leverage the pretrained KG embedding together with GCN for the long tail relations.

   This paper innovatively use a tree-like relation representation. Eg: /people -> /people/deceased_person ->/people/deceased_person/place_of_death

   Table 1 is a great design!

   I recommend this paper is because the long tail relation problem is very important. The method is not recommended.

ACL 2019

1. ⭐ Graph Neural Networks with Generated Parameters for Relation Hao Zhu and Yankai Lin and Zhiyuan Liu, Jie Fu, Tat-seng Chua, Maosong Sun ACL 2019 paper

   | GNN | new task | new perspective

   This paper considers multi-hop relation extraction, which constructs a fully-connected graph for all entities in a sentence. Experiments show that modeling entity-relation as a graph signifcantly improves the performance.

2. ⭐ Entity-Relation Extraction as Multi-turn Question Answering Xiaoya Li, Fan Yin, Zijun Sun, Xiayu Li Arianna Yuan, Duo Chai, Mingxin Zhou and Jiwei Li ACL2019 paper

   | new dataset | new perspective| joint extraction of relations and entities

   In this paper, we propose a new paradigm for the task of entity-relation extraction. We cast the task as a multi-turn question answering problem, i.e., the extraction of entities and relations is transformed to the task of identifying answer spans from the context. This multi-turn QA formalization comes with several key advantages: firstly, the question query encodes important information for the entity/relation class we want to identify; secondly, QA provides a natural way of jointly modeling entity and relation; and thirdly, it allows us to exploit the well developed machine reading comprehension (MRC) models. Additionally, we construct a newly developed dataset RESUME in Chinese, which requires multi-step reasoning to construct entity dependencies, as opposed to the single-step dependency extraction in the triplet exaction in previous datasets.

3. ⭐ Matching the Blanks: Distributional Similarity for Relation Learning Livio Baldini Soares, Nicholas FitzGerald, Jeffrey Ling, Tom Kwiatkowski ACL2019 paper

   | few shot | sota | BERT |

   In this paper we study the problem of producing useful relation representations directly from text. We describe a novel training setup, which we call matching the blanks, which relies solely on entity resolution annotations. When coupled with a new architecture for fine-tuning relation representations in BERT, our models achieves state-of-the-art results on three relation extraction tasks, and outperforms human accuracy on few-shot relation matching. In addition, we show how the new model is particularly effective in low-resource regimes, and we argue that it could significantly reduce the amount of human effort required to create relation extractors.

4. Exploiting Entity BIO Tag Embeddings and Multi-task Learning for Relation Extraction with Imbalanced Data Wei Ye1*, Bo Li*, Rui Xie, Zhonghao Sheng, Long Chen and Shikun Zhang1 ACL2019 paper

   | NRC | imbalance | trick | ranking loss |

   This paper detailed analyze the impact of imbalanced data (other relation) to the final performance. Incorporating BIO tagging to the embedding layer is an important trick!

5. GraphRel: Modeling Text as Relational Graphs for Joint Entity and Relation Extraction Tsu-Jui Fu, Peng-Hsuan Li and Wei-Yun Ma ACL2019 paper

   | joint extraction of relations and entities |

   They use GCN and argue they are the new sota.

6. ⭐ DocRED: A Large-Scale Document-Level Relation Extraction Dataset Yuan Yao, Deming Ye, Peng Li, Xu Han, Yankai Lin, Zhenghao Liu, Zhiyuan Liu, Lixin Huang, Jie Zhou, Maosong Sun ACL2019 paper code

   | document level re | new task |

   The proposed dataset contain both supervised and distant supervised data for training.

   The inputs contain: [sentence], entity_canditdate: (sent_id, pos_in_sent, ner_type). The prediction should contain (r, h, t) triplets. And the prediction of evidence is optional.

7. Attention Guided Graph Convolutional Networks for Relation Extraction Zhijiang Guo*, Yan Zhang* and Wei Lu

   | GCN | cross sentence re |

   This paper proposed a dependency based model by leveraging novel GCN. The experiments contains both cross sentence re and sentence re.

8. Neural Relation Extraction for Knowledge Base Enrichment Bayu Distiawan Trisedya, Gerhard Weikum, Jianzhong Qi, Rui Zhang

   | KB enrichment |

   This paper map the entities and relation to the ID in KB directly. The method is a end-to-end manner thus getting out of the error propagation problem.

9. Joint Type Inference on Entities and Relations via Graph Convolutional Networks Changzhi Sun, Yeyun Gong, Yuanbin Wu, Ming Gong, Daxin Jiang, Man Lan, Shiliang Sun, Nan Duan

   | GCN | joint extraction of relations and entities |

   GCN, ACE05

AAAI 2019

1. Hybrid Attention-based Prototypical Networks for Noisy Few-Shot Relation Classification Tianyu Gao*, Xu Han*, Zhiyuan Liu, Maosong Sun. (* means equal contribution) AAAI2019 paper code

   | few shot |

   instance level and feature-level attention schemes based on prototypical networks

2. A Hierarchical Framework for Relation Extraction with Reinforcement Learning Takanobu, Ryuichi and Zhang, Tianyang and Liu, Jiexi and Huang, Minlie AAAI2019 paper code

   | joint extraction of relations and entities | RL |

   This paper use RL based multi-pass tagging to tackle relation overlapping problem.

3. Kernelized Hashcode Representations for Biomedical Relation Extraction Sahil Garg, Aram Galstyan, Greg Ver Steeg Irina Rish, Guillermo Cecchi, Shuyang Gao AAAI2019 paper code code not released on 07/05/2019

   | ML |

   Here we propose to use random subspaces of KLSH codes for efficiently constructing an explicit representation of NLP structures suitable for general classification methods. Further, we propose an approach for optimizing the KLSH model for classification problems by maximizing an approximation of mutual information between the KLSH codes (feature vectors) and the class labels

4. Cross-relation Cross-bag Attention for Distantly-supervised Relation Extraction Yujin Yuan, Liyuan Liu, Siliang Tang, Zhongfei Zhang, Yueting Zhuang, Shiliang Pu, Fei Wu, Xiang Ren AAAI2019 paper

   | DSRE | cross bag |

   see title

EMNLP 2019

1. ⭐ Self-Attention Enhanced CNNs and Collaborative Curriculum Learning for Distantly Supervised Relation Extraction Yuyun Huang, Jinhua Du EMNLP2019 paper

   | Curriculum Learning | DSRE |

   This paper uses two student network, PCNN-ONE and PCNN-ATT, to collaboratively denoise wrong labels from the DSRE dataset. Curriculum Learning is adopted to train the model. If the two student predict the same, then let them learn the labels while minimizing the discrepancy between them; otherwise, only minimizing their discrepancy to encourage them reach an agreement.