报告主题:Aspect-Oriented Syntax Network for Aspect-Based Sentiment Analysis

报告摘要:Aspect-based sentiment analysis aims to determine the sentimental polarity towards a specific aspect in reviews or comments. Recent attempts mostly adopt attention-based mechanisms to link opinion words to their respective aspects in an implicit way. However, due to the tangle of multiple aspects or opinion words occurred in one sentence, the models often mix up the linkages. In this paper, we propose to encode sentence syntax explicitly to improve the effect of the linkages. We define an aspect-oriented dependency tree structure, which is reshaped and pruned from an ordinary parse tree, to express useful syntax information. The new tree is then encoded into a multifaceted syntax network, to be used in combination with attention-based models for prediction. Experimental results on three datasets from SemEval 2014 and Twitter show that, with our syntax network, the aspect-sentiment linkages can be better established and the attention-based models are substantially improved as a result.

嘉宾简介:权小军,教授,博士生导师。先后于中国科学技术大学计算机系、香港城市大学计算机系、美国罗格斯大学商学院、美国普渡大学计算机系、香港城市大学语言学与翻译系、新加坡科技研究局资讯通信研究院从事自然语言处理、文本挖掘和机器学习的研究工作,在国际知名期刊和会议如IEEE T-PAMI,ACM TOIS,ACL,IJCAI,SIGIR等发表论文30余篇。权小军2012年毕业于香港城市大学,获博士学位,回国前就职于新加坡科技研究局资讯通信研究院,任研究科学家,期间除从事相关方向的基础研究外,也同工业界紧密合作探索研究成果的应用。

成为VIP会员查看完整内容
Aspect-Oriented Syntax Network for Aspect-Based Sentiment Analysis -权小军.pdf
0
9

相关内容

狭义的情感分析(sentiment analysis)是指利用计算机实现对文本数据的观点、情感、态度、情绪等的分析挖掘。广义的情感分析则包括对图像视频、语音、文本等多模态信息的情感计算。简单地讲,情感分析研究的目标是建立一个有效的分析方法、模型和系统,对输入信息中某个对象分析其持有的情感信息,例如观点倾向、态度、主观观点或喜怒哀乐等情绪表达。

知识荟萃

精品入门和进阶教程、论文和代码整理等

更多

查看相关VIP内容、论文、资讯等

Target-Based Sentiment Analysis aims to detect the opinion aspects (aspect extraction) and the sentiment polarities (sentiment detection) towards them. Both the previous pipeline and integrated methods fail to precisely model the innate connection between these two objectives. In this paper, we propose a novel dynamic heterogeneous graph to jointly model the two objectives in an explicit way. Both the ordinary words and sentiment labels are treated as nodes in the heterogeneous graph, so that the aspect words can interact with the sentiment information. The graph is initialized with multiple types of dependencies, and dynamically modified during real-time prediction. Experiments on the benchmark datasets show that our model outperforms the state-of-the-art models. Further analysis demonstrates that our model obtains significant performance gain on the challenging instances under multiple-opinion aspects and no-opinion aspect situations.

0
8
下载
预览

Aspect-term level sentiment analysis (ATSA) is a fine-grained task in sentiment classification. It aims at extracting and summarizing the sentiment polarity towards a given aspect phrase from a sentence. Most existing studies combined various neural network models with a delicately carved attention mechanism to generate refined representations of sentences for better predictions. However, they were inadequate to capture correlations between aspects and sentiments. Moreover, the annotated aspect term might be unavailable in real-world scenarios which may challenge the existing methods to give correct forecasting. In this paper, we propose a capsule network based model named CAPSAR (CAPsule network with Sentiment-Aspect Reconstruction) to improve aspect-term level sentiment analysis. CAPSAR adopts a hierarchical structure of capsules and learns interactive patterns between aspects and sentiments through packaged sentiment-aspect reconstruction. Capsules in CAPSAR are capable of communicating with other capsules through a sharing-weight routing algorithm. Experiments on three ATSA benchmarks demonstrate the superiority of our model, and CAPSAR can detect the potential aspect terms from sentences by de-capsulizing the vectors in capsules when aspect terms are unknown.

0
4
下载
预览

报告主题:网络表示学习

报告摘要:数据特征的有效表示是机器学习任务中最为关键环节之一。网络数据(如社交网络、信息网络等)作为普适而广泛的数据呈现形式,对它的高效表示学习是近年来数据挖掘和机器学习领域的研究热点之一。本报告将重点围绕如下内容展开:(1)网络表示学习的基本概念;(2)几类新型网络表示学习方法,包括:网络Tag表示、域自适应表示、基于网络划分的表示以及内存自适应的表示方法等。

嘉宾简介:宋国杰,北京大学信息科学技术学院副教授。研究方向包括:网络大数据分析、机器学习&数据挖掘、社会网络分析和智能交通系统。主持了包括国家高技术研究发展计划(863计划)、国家科技支撑计划、国家自然科学基金等纵向课题10多项;主持了国际(内)科研机构合作课题、企业横向合作课题等20余项。国家级精品课程主讲教师,两度获得北京大学教学成果一等奖(2012、2009)。在包括国际顶级期刊TKDE、TPDS、TITS以及国际顶级会议KDD、IJCAI、AAAI等发表论文100余篇,是多个国际顶级会议(KDD、WWW、AAAI、IJCAI等)的程序委员。申请国家发明专利10项,软件著作权3项。研究成果获“2012年度中国公路学会科学技术奖一等奖”、“2012年度山西省科学技术奖二等奖”和“2013年度中国公路学会科学技术奖一等奖”。

成为VIP会员查看完整内容
0
21

报告主题:方面级别情感分析方法研究

报告摘要:方面级别情感分析(Aspect-level Sentiment Analysis)是更细粒度的情感分析任务,近年来受到了越来越多的关注和研究,本报告将介绍主讲人在方面级别情感分析任务上的几项研究工作。(1)提出特征增强注意力网络,融合单词内容特征、词性特征、位置特征得到特征增强的单词表示,并通过多视图共注意力机制从不同子空间中充分学习句子中内容词、方面词、情感词间的联系;(2)从人类认知角度出发,模拟人类阅读认知过程中预读、精读、后读三个阶段,提出包含词级别交互感知模块、目标感知语义蒸馏模块、语义反馈模块的类人类语义认知网络,以更贴近人类认知的方式解决方面级别情感分析问题;(3)提出一个人工标注的大规模方面级别情感分析数据集,显著提升了方面级别情感分析任务的难度,同时提出了一个新模型,在新的数据集中表现出良好的效果。

嘉宾简介:杨敏,中国科学院深圳先进技术研究院助理研究员,中科院深圳先进院得理法律人工智能联合实验室主任。2017年博士毕业于香港大学计算机科学专业,主要从事自然语言处理领域的研究,如情感分析、主题模型、文本摘要、智能问答等,在相关领域的知名学术期刊和会议(如AAAI, SIGIR, ACL, IJCAI, KDD, EMNLP, CIKM, TKDE,TOC, TMM)上发表论文50余篇。

成为VIP会员查看完整内容
方面级别情感分析方法研究-杨敏.pdf
0
27

论坛嘉宾:魏忠钰 复旦大学 副教授

报告主题:图卷积神经网络在计算金融等交叉学科领域的应用研究

报告摘要:基于图的模型能够描绘特定场景中的实体信息以及实体之间的关系,一直以来被各个学科的学者采用,在相关领域进行不同任务的建模和计算。近年来,图卷积神经网络在大规模图数据上的机器学习任务中有很好的性能表现,这也在交叉学科领域的学者中引起广泛的关注。本次报告将梳理图卷积神经网络在一些交叉学科进行表示学习以及标签预测的工作,并重点介绍报告人近期在计算金融等领域使用图卷积神经网络开展的应用研究工作。

嘉宾简介:魏忠钰,复旦大学大数据学院副教授,香港中文大学博士,美国德州大学达拉斯分校博士后,中文信息学会社交媒体处理专委会通讯委员,中国中文信息学会青年工作委员会委员。主要研究领域为自然语言处理,机器学习和社会媒体处理,专注于自动化文本生成(Text Generation)和论辩挖掘(Argumentation Mining)的研究,在相关领域在国际会议、期刊如CL,ACL,SIGIR,EMNLP,AAAI,IJCAI, Bioinformatics等发表学术论文40余篇。担任多个重要的国际会议或者期刊评审,入选2017年度上海市青年科技英才扬帆计划。

成为VIP会员查看完整内容
SMP 2019 - 表示学习论坛 - 魏忠钰 - 复旦大学.pptx.pdf
0
26

论坛嘉宾:沈华伟 中国科学院计算技术研究所 研究员

报告主题:图卷积神经网络及其应用

报告摘要:卷积神经网络在处理图像、语音、文本等具有较好空间结构的数据时展现出了很好的优势。然而,卷积神经网络不能直接应用于图(Graph)这类空间结构不规则的数据上。近年来,研究人员开始研究如何将卷积神经网络迁移到图数据上,涌现出ChevNet、MoNet、GraphSAGE、GCN、GAT等一系列方法,在基于图的半监督分类和图表示学习等任务中表现出很好的性能。报告首先梳理和回顾该方向的主要研究进展和发展趋势,进而介绍报告人近期在图卷积神经网络方面的一些研究工作(ICLR’19; IJCAI’19)。

嘉宾简介:沈华伟,博士,中国科学院计算技术研究所研究员,中国中文信息学会社会媒体处理专委会副主任。主要研究方向:社交网络分析、网络数据挖掘。先后获得过CCF优博、中科院优博、首届UCAS-Springer优博、中科院院长特别奖、入选首届中科院青年创新促进会、中科院计算所“学术百星”。2013年在美国东北大学进行学术访问。2015年被评为中国科学院优秀青年促进会会员。获得国家科技进步二等奖、北京市科学技术二等奖、中国电子学会科学技术一等奖、中国中文信息学会钱伟长中文信息处理科学技术一等奖。出版个人专/译著3部,在网络社区发现、信息传播预测、群体行为分析等方面取得了系列研究成果,发表论文100余篇。担任PNAS、IEEE TKDE、ACM TKDD等10余个学术期刊审稿人和KDD、WWW、SIGIR、AAAI、IJCAI、CIKM、WSDM等20余个国际学术会议的程序委员会委员。

成为VIP会员查看完整内容
0
44

Aspect-based sentiment analysis (ABSA) is to predict the sentiment polarity towards a particular aspect in a sentence. Recently, this task has been widely addressed by the neural attention mechanism, which computes attention weights to softly select words for generating aspect-specific sentence representations. The attention is expected to concentrate on opinion words for accurate sentiment prediction. However, attention is prone to be distracted by noisy or misleading words, or opinion words from other aspects. In this paper, we propose an alternative hard-selection approach, which determines the start and end positions of the opinion snippet, and selects the words between these two positions for sentiment prediction. Specifically, we learn deep associations between the sentence and aspect, and the long-term dependencies within the sentence by leveraging the pre-trained BERT model. We further detect the opinion snippet by self-critical reinforcement learning. Especially, experimental results demonstrate the effectiveness of our method and prove that our hard-selection approach outperforms soft-selection approaches when handling multi-aspect sentences.

0
5
下载
预览

Aspect-based sentiment analysis (ABSA), which aims to identify fine-grained opinion polarity towards a specific aspect, is a challenging subtask of sentiment analysis (SA). In this paper, we construct an auxiliary sentence from the aspect and convert ABSA to a sentence-pair classification task, such as question answering (QA) and natural language inference (NLI). We fine-tune the pre-trained model from BERT and achieve new state-of-the-art results on SentiHood and SemEval-2014 Task 4 datasets.

0
6
下载
预览

Aspect based sentiment analysis (ABSA) can provide more detailed information than general sentiment analysis, because it aims to predict the sentiment polarities of the given aspects or entities in text. We summarize previous approaches into two subtasks: aspect-category sentiment analysis (ACSA) and aspect-term sentiment analysis (ATSA). Most previous approaches employ long short-term memory and attention mechanisms to predict the sentiment polarity of the concerned targets, which are often complicated and need more training time. We propose a model based on convolutional neural networks and gating mechanisms, which is more accurate and efficient. First, the novel Gated Tanh-ReLU Units can selectively output the sentiment features according to the given aspect or entity. The architecture is much simpler than attention layer used in the existing models. Second, the computations of our model could be easily parallelized during training, because convolutional layers do not have time dependency as in LSTM layers, and gating units also work independently. The experiments on SemEval datasets demonstrate the efficiency and effectiveness of our models.

0
12
下载
预览

While existing machine learning models have achieved great success for sentiment classification, they typically do not explicitly capture sentiment-oriented word interaction, which can lead to poor results for fine-grained analysis at the snippet level (a phrase or sentence). Factorization Machine provides a possible approach to learning element-wise interaction for recommender systems, but they are not directly applicable to our task due to the inability to model contexts and word sequences. In this work, we develop two Position-aware Factorization Machines which consider word interaction, context and position information. Such information is jointly encoded in a set of sentiment-oriented word interaction vectors. Compared to traditional word embeddings, SWI vectors explicitly capture sentiment-oriented word interaction and simplify the parameter learning. Experimental results show that while they have comparable performance with state-of-the-art methods for document-level classification, they benefit the snippet/sentence-level sentiment analysis.

0
11
下载
预览
小贴士
相关VIP内容
相关资讯
计算机 | EMNLP 2019等国际会议信息6条
Call4Papers
10+阅读 · 2019年4月26日
学会期刊丨《中国人工智能学会通讯》2019年 第9卷 第02期
中国人工智能学会
6+阅读 · 2019年2月28日
CAAI-AIDL 第六期《自然语言处理》丨 京东何晓冬,清华大学唐杰
美国化学会 (ACS) 北京代表处招聘
知社学术圈
3+阅读 · 2018年9月4日
人工智能类 | 国际会议/SCI期刊专刊信息9条
Call4Papers
4+阅读 · 2018年7月10日
COLING 2018-最新论文最全分类-整理分享
深度学习与NLP
6+阅读 · 2018年7月6日
笔记 | Sentiment Analysis
黑龙江大学自然语言处理实验室
7+阅读 · 2018年5月6日
报名 | 知识图谱前沿技术课程(苏州大学站)
PaperWeekly
8+阅读 · 2017年11月27日
相关论文
Jointly Modeling Aspect and Sentiment with Dynamic Heterogeneous Graph Neural Networks
Shu Liu,Wei Li,Yunfang Wu,Qi Su,Xu Sun
8+阅读 · 2020年4月14日
Mengting Hu,Shiwan Zhao,Honglei Guo,Renhong Cheng,Zhong Su
5+阅读 · 2019年9月25日
Comprehensive Analysis of Aspect Term Extraction Methods using Various Text Embeddings
Łukasz Augustyniak,Tomasz Kajdanowicz,Przemysław Kazienko
5+阅读 · 2019年9月11日
Chi Sun,Luyao Huang,Xipeng Qiu
6+阅读 · 2019年3月22日
Hwiyeol Jo,Jeong Ryu
8+阅读 · 2018年6月3日
Wei Xue,Tao Li
12+阅读 · 2018年5月18日
Sven Schmit,Carlos Riquelme
5+阅读 · 2018年3月28日
Hongwei Wang,Fuzheng Zhang,Xing Xie,Minyi Guo
3+阅读 · 2018年1月25日
Shuai Wang,Mianwei Zhou,Geli Fei,Yi Chang,Bing Liu
11+阅读 · 2018年1月18日
Top