Aspect Term Extraction (ATE)
文章平均质量分 87
主要存放ATE相关的论文
薰珞婷紫小亭子
这个作者很懒,什么都没留下…
展开
-
14. Bridge-Based Active Domain Adaptation for Aspect Term Extraction 阅读笔记
跨域ATE,武大,两种桥原创 2022-07-25 14:16:37 · 540 阅读 · 0 评论 -
31. Opinion-based Relational Pivoting forCross-domain Aspect Term Extraction 阅读笔记
AMTL原创 2022-07-24 22:30:53 · 661 阅读 · 0 评论 -
27. Dual Graph Convolutional Networks for Aspect-based Sentiment Analysis 阅读笔记
目录1. AbstractIntroduction2. Related Work模型简称:DualGCN Author,,, , , InstitutionsSchool of Artificial Intelligence,Beijing University of Posts and Telecommunications(北邮),China Language Technologies Institute,Carnegie Mellon University, USAA原创 2022-03-16 23:12:28 · 2717 阅读 · 0 评论 -
26.Graph Ensemble Learning over Multiple Dependency Trees for Aspect-level Sentiment Classification
目录26.Graph Ensemble Learning over Multiple Dependency Trees for Aspect-level Sentiment Classification阅读笔记AbstractIntroductionRelated WorksProposed Model3.1 Definition3.2 GraphMerge over Multiple Dependency3.2 RGAT over Ensemble Graph3.3..原创 2022-03-09 18:18:43 · 507 阅读 · 1 评论 -
3. Target-oriented Opinion Words Extraction with Target-fused Neural Sequence Labeling 读论文笔记
Target-oriented Opinion Words Extraction with Target-fused Neural Sequence Labeling(BIO标签)TOWE一、科学问题:属性抽取问题二、针对问题的方法设计动机:(1)针对目前已有的属性抽取很少将AT和OT成对抽取的现象,本文提出了一个新的模型TOWE(This paper propose a novel sequence labeling subtask for ABSA(Aspect-based Sentiment原创 2021-01-20 21:02:21 · 433 阅读 · 2 评论 -
15. Beyond Accuracy: Behavioral Testing of NLP Models with CHECKLIST阅读笔记
Beyond Accuracy: Behavioral Testing of NLP Models with CHECKLIST论文地址:https://arxiv.org/abs/2005.04118ACL 2020背景介绍这篇paper解决的是NLP领域的模型评测问题。目前最主流的评测方法是从已有数据中划分出一部分作为测试集,然后测试模型准确率。但这并不能全面地评估一个模型的好坏,还有很多意想不到的情况:测试集有部分数据和训练集相似度很高,模型如果overfit了也无法发现测试集存在原创 2021-08-30 21:14:22 · 317 阅读 · 0 评论 -
16.Modularized Interaction Network for Named Entity Recognition 阅读笔记
Modularized Interaction Network for Named Entity Recognition论文地址:https://aclanthology.org/2021.acl-long.17.pdfAuthor Information::Fei Li1, Zheng Wang2∗, Siu Cheung Hui2, Lejian Liao1, Dandan Song1∗,Jing Xu1, Guoxiu He3, Meihuizi Jia1Institutions Informa原创 2021-09-09 12:36:20 · 912 阅读 · 2 评论 -
19. Double Embeddings and CNN-based Sequence Labeling for Aspect Extraction
19. Double Embeddings and CNN-based Sequence Labeling for Aspect Extraction这篇论文,当时入门ATE的时候,就看过。因为需要,又重新回头看了一遍。现在读这篇Paper和之前的感受不一样了,所以记录下来。论文地址:https://arxiv.org/pdf/2106.04300.pdfAuthor Information::Hu Xu1, Bing Liu1, Lei Shu1 and Philip S. Yu1,2Instit原创 2021-09-16 17:07:27 · 334 阅读 · 0 评论 -
4. Latent Opinions Transfer Network for Target-Oriented Opinion Words Extraction
Latent Opinions Transfer Network for Target-Oriented Opinion Words Extraction AAAI 2020一、科学问题:细粒度情感分析的数据标注是复杂且耗时的,因为标注者需要标出评论中所有的目标对象以及它们对应的观点词和情感极性,这导致很难获取大规模的标注样本,数据不足严重制约了神经网络模型在 TOWE 任务上的性能。本文针对该团队2019年NAACL发表的TOWE论文的数据标注的弊端,提出了一种新的改进模型LOTN相比之下,我们很原创 2021-01-21 10:35:31 · 272 阅读 · 2 评论 -
7. Variational Graph Auto-Encoders论文阅读笔记
《Variational Graph Auto-Encoders》这篇论文是截止我博士一年级生涯为止,看到的最短的一篇论文,总页数3页,正文2页。截至今日,“他引”次数高达863次。废话不多数,一起来看下这篇论文吧论文地址:https://arxiv.org/pdf/1611.07308.pdf%5D本文模型简称为:VGAE在正式读这篇论文之前,默认您已经有了一定的VAE(Variational Auto-Encoder)、GCN(Graph Convolutional Network)等知识的积累原创 2021-04-08 21:21:08 · 1109 阅读 · 0 评论 -
5. Conditional Augmentation for Aspect Term Extraction via Masked Sequence-to-Sequence Generation
论文阅读笔记 四《Conditional Augmentation for Aspect Term Extraction via Masked Sequence-to-Sequence Generation》论文地址:https://www.aclweb.org/anthology/2020.acl-main.631.pdf注:本文仅提出了一种数据增强的方法,并未提出任何新的ATE方法一、科学问题:现有的ATE任务最大的挑战是缺乏足够的已经标注好的数据集。因此,本文是一个数据增强的论文二原创 2021-01-23 22:12:02 · 494 阅读 · 1 评论 -
24. Progressive Self-Training with Discriminator for Aspect Term Extraction阅读笔记
Progressive Self-Training with Discriminator for Aspect Term Extraction**AuthorAuthor Information: Qianlong Wang1 , Zhiyuan Wen1 , Qin Zhao1 , Min Yang2, Ruifeng Xu1,3* Institutions Information:1.Harbin Institute of Technology (Shenzhen), China2.Shenzh原创 2022-01-03 21:42:34 · 333 阅读 · 0 评论 -
25. Deep Weighted MaxSAT for Aspect-based Opinion Extraction 阅读笔记
25. Deep Weighted MaxSAT for Aspect-based Opinion Extraction 阅读笔记Author Information::Meixi Wu* , Wenya Wang, Sinno Jialin PanInstitutions Information:1 Nanyang Technological University, Singapore EMNLP 2021 模型简写:DeepWMaxSAT 序列标注:BIO摘要深度学习在各项NLP任务上取原创 2022-01-22 14:40:30 · 870 阅读 · 0 评论 -
23.Position-Aware Tagging for Aspect Sentiment Triplet Extraction阅读笔记
23.Position-Aware Tagging for Aspect Sentiment Triplet Extraction阅读笔记Author Information::Lu Xu* 1, 2 , Hao Li* 1, 3, Wei Lu1, Lidong Bing2Institutions Information:1 StatNLP Research Group, Singapore University of Technology and Design2 DAMO Academy, Al原创 2021-11-18 23:12:42 · 1394 阅读 · 2 评论 -
10.Unsupervised aspect term extraction with b-lstm & crf using automatically labelled datasets阅读笔记
Unsupervised aspect term extraction with b-lstm & crf using automatically labelled datasets一、科学问题:ATE (Aspect Term Extraction)是ABSA的一个子任务,其旨在提取评论文本中的属性词。本文针对监督属性抽取缺乏数据集的现状,提出了一个自动构造数据集的方法。并且,提出了一个既可以适用于监督属性抽取,又可以适用于无监督属性抽取的方法。二、针对问题的方法设计动机:用于监督ATE的原创 2021-07-07 20:24:03 · 266 阅读 · 0 评论 -
12.Coupling Global and Local Context for Unsupervised Aspect Extraction阅读笔记
Coupling Global and Local Context for Unsupervised Aspect Extraction3.1 Input and Output 在接触细节以揭示本文的模型如何工作之前,首先描述本文的输入和输出。输入给定一个有|C|个句子的数据集C,{X1,X2,……,X|c|}。将每一个句子X表示成两种形式,词序列形式Xseq和词袋形式Xbow。将Xseq输入局部上下文模型,习得单词如何与其邻居共同出现。而将Xbow作为全局上下文模型的输入,捕获句子级的单词共现规律原创 2021-07-19 16:11:27 · 316 阅读 · 0 评论 -
6. Joint Aspect Extraction and Sentiment Analysis with Directional Graph Convolutional Networks
《Jointly Learning Aspect-Focused and Inter-Aspect Relations with Graph Convolutional Networks for Aspect Sentiment Analysis》1.Introduction2. The approach2.1 Graph Convolutional Networks2.2 Directional Graph Convolutional Networks2.3 Tagging with Directiona原创 2021-03-19 16:28:19 · 762 阅读 · 2 评论 -
20. Learning to Perturb Word Embeddings for Out-of-distribution QA 阅读笔记
20. Learning to Perturb Word Embeddings for Out-of-distribution QA论文地址:https://arxiv.org/pdf/2105.02692.pdfAuthor Information::Seanie Lee1,* Minki Kang1,*Juho Lee1, Sung Ju Hwang1,2Institutions Information: 1. KAIST, South Kor原创 2021-09-29 17:18:48 · 198 阅读 · 0 评论 -
2. Don’t Eclipse Your Arts Due to Small Discrepancies Boundary Repositioning with a Pointer Network
Don’t Eclipse Your Arts Due to Small Discrepancies Boundary Repositioning with a Pointer Network for Aspect Extraction解读2020 ACL属性抽取论文《Don’t Eclipse Your Arts Due to Small Discrepancies:Boundary Repositioning with a Pointer Network for Aspect Extraction原创 2021-01-15 16:04:10 · 285 阅读 · 0 评论 -
22. Knowing What, How and Why: A Near Complete Solution for Aspect-based Sentiment Analysis阅读笔记
22. Knowing What, How and Why: A Near Complete Solution for Aspect-based Sentiment Analysis阅读笔记论文地址:Click HereAuthor Information::Haiyun Peng1 , Lu Xu∗1,2, Lidong Bing1 , Fei Huang1, Wei Lu2 , Luo Si1Institutions Information:1 DAMO Academy, Alibaba Gro原创 2021-11-11 19:13:17 · 1524 阅读 · 2 评论 -
9. Enhancing Aspect Term Extraction with Soft Prototypes论文阅读笔记
Enhancing Aspect Term Extraction with SoftBy 武汉大学译文:使用Prototypes软模版SoftProto框架增强属性抽取EMNLP20201. 科学问题该工作提出基于软模板的 SoftProto 框架来增强方面词抽取任务,旨在解决评论文本中属性词和环境词具有长尾分布(long-tail distribution)的问题。SoftProto 框架几乎可以与所有的序列标注器进行结合。在多个 SemEval 数据集上的实验表明,软模板的引入大幅度地提升了几原创 2021-06-28 21:21:34 · 394 阅读 · 1 评论 -
13. Enhancing Aspect Extraction Hindi 阅读笔记
Enhance Aspect term extraction Hindi ACL 2021 作者:Arghya Bhattacharya, Alok Debnath and Manish Shrivastava 单位:International Institute of Information Technology, Hyderabad (IIIT-H)原创 2021-08-07 15:24:25 · 170 阅读 · 0 评论 -
14 Bridge-Based Active Domain Adaptation for Aspect Term Extraction阅读笔记
Bridge-Based Active Domain Adaptation for Aspect Term ExtractionACL 2021Author: Zhuang Chen, Tieyun Qian∗Unit: School of Computer Science, Wuhan University, China本文的核心思想:为了缓解标注数据带 来的巨大压力,有学者尝试使用跨领域传输知识的域适应方法解决该问题.但大多数属性词都是基于某些特定领域,它们之间并不能直接传递.Chen(202原创 2021-08-09 16:45:01 · 239 阅读 · 0 评论 -
11. Coupled multilayer attentions for co-extraction of aspect and opinion terms阅读笔记
Title: Coupled multilayer attentions for co-extraction of aspect and opinion terms简称:CMLAAAAI 2020标签:BIO2020论文地址:一、科学问题:旨在解决属性抽取和情感抽取问题。二、针对问题的方法设计动机:已有的属性抽取方式可以分为三大类:(1)rules-based:基于规则的(2)feature-engineering-based:基于特征工程的(3)deeping learning-ba原创 2021-07-17 21:26:40 · 532 阅读 · 2 评论 -
8.A Multi-task Learning Model for Chinese-oriented Aspect Polarity Classification and Aspect T论文阅读笔记
论文下载地址:https://arxiv.org/pdf/1912.07976.pdf《A Multi-task Learning Model for Chinese-oriented Aspect Polarity Classification and Aspect Term Extraction》 模型名称:LCF-ATEPC 标签:BIO一、科学问题:在以往的研究中,主要关注的APC的精度,而忽略了对于ATE的研究,因此,在基于方面的情感分析进行迁移学习时,往往会因为缺少aspect t原创 2021-05-01 19:53:27 · 1100 阅读 · 0 评论 -
17. A Unified Generative Framework for Aspect-Based Sentiment Analysis
A Unified Generative Framework for Aspect-Based Sentiment Analysis论文地址:https://arxiv.org/pdf/2106.04300.pdfAuthor Information::Hang Yan1,∗, Junqi Dai1,∗, Tuo Ji1, Xipeng Qiu1,2†, Zheng Zhang3Institutions Information: 1. Shanghai原创 2021-09-13 23:12:13 · 1268 阅读 · 0 评论 -
21.Selective Attention Based Graph Convolutional Networks for Aspect-Level Sentiment Classification阅
21.Selective Attention Based Graph Convolutional Networks for Aspect-Level Sentiment Classification阅读笔记论文地址:https://arxiv.org/pdf/1910.10857.pdfAuthor Information::Xiaochen Hou* , Jing Huang, Guangtao Wang,Xiaodong He, Bowen ZhouInstitutions Information原创 2021-10-21 20:17:44 · 474 阅读 · 3 评论 -
1. Exploring Sequence-to-Sequence Learning in Aspect Term Extraction
Exploring Sequence-to-Sequence Learning in Aspect Term Extraction解读注:本文采用BIO标签,且默认读这篇论文的人对属性抽取任务有一定的了解。一、科学问题:ATE是从一个句子中识别出所有的Aspect Term,并经常被看作是序列标注问题。二、针对问题的方法设计动机:针对目前已有的序列标注方法不能够充分的利用整个句子的意义,以及在处理标签之间的依存关系受限的现象。(sequence labeling based methods can原创 2021-01-15 15:56:16 · 477 阅读 · 0 评论