carrie_0307的博客

ACL2021放榜啦！具体请看这里！小编对其中事件抽取以及事件相关的论文整理如下:事件检测(ED)Unleash GPT-2 Power for Event DetectionOntoED: Low-resource Event Detection with Ontology EmbeddingFew-Shot Event Detection with Prototypical Amortized Conditional Random Field事件抽取(EE)Text2Event: C

Paper: ACL2020-A Contextual Hierarchical Attention Network with Adaptive Objective for Dialogue State TrackingLink: https://www.aclweb.org/anthology/2020.acl-main.563.pdfCode: https://github.com/ictnlp/CHAN-DST这是一篇关于 对话状态追踪(DST) 任务的论文，在 Intro 部分，作者提出了当.

Paper: [ACL2020]SAS: Dialogue State Tracking via Slot Attention and Slot Information SharingLink: https://www.aclweb.org/anthology/2020.acl-main.567.pdf对话状态追踪(Dialogue State Tracking, DST) 的目的，在于从历史的对话记录中推断出当前的对话状态(一般以槽-值对的形式表示)，基于对话状态得到用户的真实目的。在这里，借用这篇.

Paper: Document-level Relation Extraction with Dual-tier Heterogeneous GraphLink: https://www.aclweb.org/anthology/2020.coling-main.143.pdf这是一篇进行**篇章级关系抽取(DRE)**的工作。以下图为例，如果想得到radycardia和 ramipril 之间的关系，需要先radycardiaand ramipril, one has to first gr.

Paper: Neural Extractive Summarization with Hierarchical Attentive Heterogeneous Graph NetworkLink: https://www.aclweb.org/anthology/2020.emnlp-main.295.pdfCode: https://github.com/coder352/HAHSum这是一篇关于抽取式摘要任务的工作，抽取式摘要的任务目标是: 在给定的篇章中，选出其中关键的个句子作为这篇文章.

Paper: Is Killed More Significant than Fled? A Contextual Model for Salient Event DetectionLink: https://www.aclweb.org/anthology/2020.coling-main.10.pdf这是一篇关于Salient Event Detection(重要事件检测)任务的文章，其具体任务内容是，在一篇标注了若干事件记录的篇章中，检测出其中显著的、重要的事件有哪些。例如下图包含两篇篇章，标红的

Paper: (EMNLP2020)Hierarchical Graph Network for Multi-hop Question AnsweringLink: https://www.aclweb.org/anthology/2020.emnlp-main.710/Code: https://github.com/yuwfan/HGN这是一篇关于多跳问答(multi-hop QA)任务的论文，不同于单跳QA，multi-hop QA任务下问题的答案需要对多个段落或篇章进行多跳推理。然而，.

Paper: (ICLR2017) BI-DIRECTIONAL ATTENTION FLOW FOR MACHINE COMPREHENSIONLink: https://arxiv.org/pdf/1611.01603.pdf大家在读MRC论文时可能都会频繁注意到Bi-attention Layer的使用，那么今天我们就来读一读提出Bi-attention Layer的这篇论文~Bi-attention Flow提出于ICLR2017的一篇进行机器阅读任务(MRC)的论文中。MRC任务要求根据问.

2017TACL2017: Cross-sentence n-ary relation extraction with graph LSTMs paperEACL2017: Distant supervision for relation extraction beyond the sentence boundary paper2018NAACL2018: Simultaneously Self-Attending to All Mentions forFull-Abstract B

Paper: Global-to-Local Neural Networks for Document-Level Relation ExtractionLink: https://www.aclweb.org/anthology/2020.emnlp-main.303.pdfCode: https://github.com/nju-websoft/GLRE这是一篇进行 篇章级关系抽取(Doc-RE) 任务的论文。在Intro部分，作者首先提出了现有的通过在篇章级别构图进行Doc-RE工作的缺点:

Paper:(EMNLP2020) Is Graph Structure Necessary for Multi-hop Question Answering?Link: https://www.aclweb.org/anthology/2020.emnlp-main.583/今天推荐的论文来自科大讯飞和哈工大SCIR的联合出品，他们为我们提出一道思考题: 在Multi-hop QA中，图结构是必要的吗？在当前进行Multi-hop QA的工作中，将实体、句子和段落建模为一个图结构，然后引入图网络来

Paper: (EMNLP2020)Double Graph Based Reasoning for Document-level Relation ExtractionLink: https://www.aclweb.org/anthology/2020.emnlp-main.127.pdfCode: https://github.com/PKUnlp-icler/GAIN这是一篇进行篇章级关系抽取任务的工作。在Intro部分，作者首先提出了篇章级关系抽取任务的难点:同一三元组中的头实体和尾实

Paper: [EMNLP2020]Two are Better Than One: Joint Entity and Relation Extraction with Table-Sequence Encoderslink: https://www.aclweb.org/anthology/2020.emnlp-main.133.pdfcode: https://github.com/LorrinWWW/two-are-better-than-one这是一篇关于实体关系联合抽取的工作。关于现有的.

Paper: EMNLP2020-Findings: How Does Context Matter? On the Robustness of Event Detection with Context-Selective Mask GeneralizationLink: : https://www.aclweb.org/anthology/2020.findings-emnlp.229Authors: Jian Liu, Yubo Chen, Kang Liu, Yantao Jia, Zhiche.

在NLP相关的论文中，我们常常回看到含有 Global Information 字眼相关的论文；那么，这些文中提到的 Global Information， 到底是什么、与 local information 有什么不同、是如何计算得到的呢？； 基于这几个问题，小编以 EMNLP2019的若干篇论文为例，进行了一点点简单粗暴小小整理，希望能够淡淡地解开Global Information的神秘面纱~~~Coupling Global and Local Context for Unsupervised A

EMNLP2019: Fine-Grained Entity Typing via Hierarchical Multi Graph Convolutional Networks链接: https://www.aclweb.org/anthology/D19-1502/代码: https://github.com/SIGKDD/HMGCN介绍Motivation在知识库中，一些实体的...

EMNLP2019: A Boundary-aware Neural Model for Nested Named Entity Recognition论文链接: https://www.aclweb.org/anthology/D19-1034.pdf论文代码: https://github.com/thecharm/boundary-aware-nested-ner简介Motiva...

AMNLP2019: Dependency-Guided LSTM-CRF for Named Entity Recognition论文链接: https://www.aclweb.org/anthology/D19-1399.pdfhttp://www.statnlp.org/research/information-extraction一. 介绍Motivation长距离的...

论文链接: https://www.aclweb.org/anthology/N19-1323.pdf论文代码: https://github.com/billy-inn/HRERE

[论文链接](Heterogeneous Graph Attention Networks for Semi-supervised Short Text Classification)[]Connecting Language and Knowledge with Heterogeneous Representations for Neural Relation ExtractionMu...

论文链接论文代码一. 介绍Motivation作者首先介绍了lexical word information对中文NER的重要性，并提出 lexical word 实际包含了两重信息: 一是词语的边界信息， 二是词语的语义信息。基于此结合现有工作，作者提出了两项挑战：如何融合self-matched words.一个字符的 self-matched lexical words ...

论文链接[论文代码](https: //github.com/RowitZou/LGN)

原文链接：https://www.ijcai.org/Proceedings/2019/0692.pdf代码: https://github.com/guitaowufeng/LR-CNN一. 介绍Motivation本文模型提出的动机是，尽管现有的融合了lexicon信息的RNN模型在中文NER任务中取得了一定的成功，但这类模型存在两处缺点效率限制: 基于RNN的模型...