目录
6. UIT-HSE at WNUT-2020 Task 2: Exploiting CT-BERT for Identifying COVID-19 Information on the Twitter Social Network [PDF] 摘要
7. Team Alex at CLEF CheckThat! 2020: Identifying Check-Worthy Tweets With Transformer Models [PDF] 摘要
8. TransModality: An End2End Fusion Method with Transformer for Multimodal Sentiment Analysis [PDF] 摘要
11. UPB at SemEval-2020 Task 9: Identifying Sentiment in Code-Mixed Social Media Texts using Transformers and Multi-Task Learning [PDF] 摘要
12. UPB at SemEval-2020 Task 8: Joint Textual and Visual Modeling in a Multi-Task Learning Architecture for Memotion Analysis [PDF] 摘要
16. BANANA at WNUT-2020 Task 2: Identifying COVID-19 Information on Twitter by Combining Deep Learning and Transfer Learning Models [PDF] 摘要
17. Once Upon A Time In Visualization: Understanding the Use of Textual Narratives for Causality [PDF] 摘要
18. QiaoNing at SemEval-2020 Task 4: Commonsense Validation and Explanation system based on ensemble of language model [PDF] 摘要
19. MIDAS at SemEval-2020 Task 10: Emphasis Selection using Label Distribution Learning and Contextual Embeddings [PDF] 摘要
21. Accenture at CheckThat! 2020: If you say so: Post-hoc fact-checking of claims using transformer-based models [PDF] 摘要
摘要
1. COVCOR20 at WNUT-2020 Task 2: An Attempt to Combine Deep Learning and Expert rules [PDF] 返回目录
Ali Hürriyetoğlu, Ali Safaya, Nelleke Oostdijk, Osman Mutlu, Erdem Yörük
Abstract: In the scope of WNUT-2020 Task 2, we developed various text classification systems, using deep learning models and one using linguistically informed rules. While both of the deep learning systems outperformed the system using the linguistically informed rules, we found that through the integration of (the output of) the three systems a better performance could be achieved than the standalone performance of each approach in a cross-validation setting. However, on the test data the performance of the integration was slightly lower than our best performing deep learning model. These results hardly indicate any progress in line of integrating machine learning and expert rules driven systems. We expect that the release of the annotation manuals and gold labels of the test data after this workshop will shed light on these perplexing results.
摘要：WNUT-2020的任务2的范围内，我们开发了各种文本分类系统，采用使用语言告知规则深度学习模型和一个。虽然两者的深度学习系统的使用的语言告知规则跑赢大系统中，我们发现，通过三个系统有更好的表现可能会比每种方法的交叉验证设置独立的性能来达到（输出）的集成。然而，对测试数据整合的表现略低于我们最好的执行深度学习模式下。这些结果表明几乎没有在整合机器学习和驱动系统专家规则的行任何进展。我们预计，本次研讨会后，测试数据的标注手册和黄金标签的版本将阐明了这些令人费解的结果。
Ali Hürriyetoğlu, Ali Safaya, Nelleke Oostdijk, Osman Mutlu, Erdem Yörük
Abstract: In the scope of WNUT-2020 Task 2, we developed various text classification systems, using deep learning models and one using linguistically informed rules. While both of the deep learning systems outperformed the system using the linguistically informed rules, we found that through the integration of (the output of) the three systems a better performance could be achieved than the standalone performance of each approach in a cross-validation setting. However, on the test data the performance of the integration was slightly lower than our best performing deep learning model. These results hardly indicate any progress in line of integrating machine learning and expert rules driven systems. We expect that the release of the annotation manuals and gold labels of the test data after this workshop will shed light on these perplexing results.
摘要：WNUT-2020的任务2的范围内，我们开发了各种文本分类系统，采用使用语言告知规则深度学习模型和一个。虽然两者的深度学习系统的使用的语言告知规则跑赢大系统中，我们发现，通过三个系统有更好的表现可能会比每种方法的交叉验证设置独立的性能来达到（输出）的集成。然而，对测试数据整合的表现略低于我们最好的执行深度学习模式下。这些结果表明几乎没有在整合机器学习和驱动系统专家规则的行任何进展。我们预计，本次研讨会后，测试数据的标注手册和黄金标签的版本将阐明了这些令人费解的结果。
2. Why Not Simply Translate? A First Swedish Evaluation Benchmark for Semantic Similarity [PDF] 返回目录
Tim Isbister, Magnus Sahlgren
Abstract: This paper presents the first Swedish evaluation benchmark for textual semantic similarity. The benchmark is compiled by simply running the English STS-B dataset through the Google machine translation API. This paper discusses potential problems with using such a simple approach to compile a Swedish evaluation benchmark, including translation errors, vocabulary variation, and productive compounding. Despite some obvious problems with the resulting dataset, we use the benchmark to compare the majority of the currently existing Swedish text representations, demonstrating that native models outperform multilingual ones, and that simple bag of words performs remarkably well.
摘要：本文介绍了文本语义相似的第一个瑞典评价基准。基准是通过简单地运行通过谷歌机器翻译API的英语STS-B数据集的编译。本文讨论了使用这样一个简单的方法来编译瑞典的评价标准，包括翻译错误，词汇变化和生产复合潜在的问题。尽管与所得到的数据集的一些明显的问题，我们使用的基准来比较大多数现有的瑞典文交涉，表明本土车型超越多种语言的人，和词执行的简单袋子非常好。
Tim Isbister, Magnus Sahlgren
Abstract: This paper presents the first Swedish evaluation benchmark for textual semantic similarity. The benchmark is compiled by simply running the English STS-B dataset through the Google machine translation API. This paper discusses potential problems with using such a simple approach to compile a Swedish evaluation benchmark, including translation errors, vocabulary variation, and productive compounding. Despite some obvious problems with the resulting dataset, we use the benchmark to compare the majority of the currently existing Swedish text representations, demonstrating that native models outperform multilingual ones, and that simple bag of words performs remarkably well.
摘要：本文介绍了文本语义相似的第一个瑞典评价基准。基准是通过简单地运行通过谷歌机器翻译API的英语STS-B数据集的编译。本文讨论了使用这样一个简单的方法来编译瑞典的评价标准，包括翻译错误，词汇变化和生产复合潜在的问题。尽管与所得到的数据集的一些明显的问题，我们使用的基准来比较大多数现有的瑞典文交涉，表明本土车型超越多种语言的人，和词执行的简单袋子非常好。
3. Robust Spoken Language Understanding with RL-based Value Error Recovery [PDF] 返回目录
Chen Liu, Su Zhu, Lu Chen, Kai Yu
Abstract: Spoken Language Understanding (SLU) aims to extract structured semantic representations (e.g., slot-value pairs) from speech recognized texts, which suffers from errors of Automatic Speech Recognition (ASR). To alleviate the problem caused by ASR-errors, previous works may apply input adaptations to the speech recognized texts, or correct ASR errors in predicted values by searching the most similar candidates in pronunciation. However, these two methods are applied separately and independently. In this work, we propose a new robust SLU framework to guide the SLU input adaptation with a rule-based value error recovery module. The framework consists of a slot tagging model and a rule-based value error recovery module. We pursue on an adapted slot tagging model which can extract potential slot-value pairs mentioned in ASR hypotheses and is suitable for the existing value error recovery module. After the value error recovery, we can achieve a supervision signal (reward) by comparing refined slot-value pairs with annotations. Since operations of the value error recovery are non-differentiable, we exploit policy gradient based Reinforcement Learning (RL) to optimize the SLU model. Extensive experiments on the public CATSLU dataset show the effectiveness of our proposed approach, which can improve the robustness of SLU and outperform the baselines by significant margins.
摘要：口语理解（SLU）旨在提取结构从语音识别文本的语义表示（例如，槽 - 值对），从自动语音识别（ASR）的错误受到影响。为了减轻因ASR-错误的问题，以前的作品可以通过语音搜索最相似的考生把输入改编的语音识别文本，或纠正错误，ASR中的预测值。然而，这两种方法分别和独立地应用。在这项工作中，我们提出了一个新的强大的SLU框架来指导SLU输入适应了基于规则的值错误恢复模块。该框架包括一个槽标签模型和基于规则的值错误恢复模块。我们追求上的适配槽标签模型，其可以提取在ASR假说提到潜在时隙 - 值对，适合于现有的值错误恢复模块。值错误恢复后，我们就可以通过细化插槽 - 值对带注释的比较实现监控信号（奖励）。因为值错误恢复的操作是非微，我们利用政策梯度的强化学习（RL），以优化SLU模型。对公众CATSLU大量的实验数据集上我们提出的方法的有效性，从而提高SLU的鲁棒性和显著利润率跑赢基准。
Chen Liu, Su Zhu, Lu Chen, Kai Yu
Abstract: Spoken Language Understanding (SLU) aims to extract structured semantic representations (e.g., slot-value pairs) from speech recognized texts, which suffers from errors of Automatic Speech Recognition (ASR). To alleviate the problem caused by ASR-errors, previous works may apply input adaptations to the speech recognized texts, or correct ASR errors in predicted values by searching the most similar candidates in pronunciation. However, these two methods are applied separately and independently. In this work, we propose a new robust SLU framework to guide the SLU input adaptation with a rule-based value error recovery module. The framework consists of a slot tagging model and a rule-based value error recovery module. We pursue on an adapted slot tagging model which can extract potential slot-value pairs mentioned in ASR hypotheses and is suitable for the existing value error recovery module. After the value error recovery, we can achieve a supervision signal (reward) by comparing refined slot-value pairs with annotations. Since operations of the value error recovery are non-differentiable, we exploit policy gradient based Reinforcement Learning (RL) to optimize the SLU model. Extensive experiments on the public CATSLU dataset show the effectiveness of our proposed approach, which can improve the robustness of SLU and outperform the baselines by significant margins.
摘要：口语理解（SLU）旨在提取结构从语音识别文本的语义表示（例如，槽 - 值对），从自动语音识别（ASR）的错误受到影响。为了减轻因ASR-错误的问题，以前的作品可以通过语音搜索最相似的考生把输入改编的语音识别文本，或纠正错误，ASR中的预测值。然而，这两种方法分别和独立地应用。在这项工作中，我们提出了一个新的强大的SLU框架来指导SLU输入适应了基于规则的值错误恢复模块。该框架包括一个槽标签模型和基于规则的值错误恢复模块。我们追求上的适配槽标签模型，其可以提取在ASR假说提到潜在时隙 - 值对，适合于现有的值错误恢复模块。值错误恢复后，我们就可以通过细化插槽 - 值对带注释的比较实现监控信号（奖励）。因为值错误恢复的操作是非微，我们利用政策梯度的强化学习（RL），以优化SLU模型。对公众CATSLU大量的实验数据集上我们提出的方法的有效性，从而提高SLU的鲁棒性和显著利润率跑赢基准。
4. Uncovering the Corona Virus Map Using Deep Entities and Relationship Models [PDF] 返回目录
Kuldeep Singh, Puneet Singla, Ketan Sarode, Anurag Chandrakar, Chetan Nichkawde
Abstract: We extract entities and relationships related to COVID-19 from a corpus of articles related to Corona virus by employing a novel entities and relationship model. The entity recognition and relationship discovery models are trained with a multi-task learning objective on a large annotated corpus. We employ a concept masking paradigm to prevent the evolution of neural networks functioning as an associative memory and induce right inductive bias guiding the network to make inference using only the context. We uncover several import subnetworks, highlight important terms and concepts and elucidate several treatment modalities employed in related ailments in the past.
摘要：我们提取的通过采用新的实体和关系模型相关冠状病毒的文章语料库与COVID-19的实体和关系。实体识别和关系发现模型被训练与多任务学习目标上的大型注释的语料库。我们使用的一个概念掩蔽范式，以防止作为联想记忆功能的神经网络的发展，并引起右归纳偏置指导网络，使只使用情景推断。我们发现一些进口子网，突出重要术语和概念，并在过去相关疾病阐发采用几种治疗方式。
Kuldeep Singh, Puneet Singla, Ketan Sarode, Anurag Chandrakar, Chetan Nichkawde
Abstract: We extract entities and relationships related to COVID-19 from a corpus of articles related to Corona virus by employing a novel entities and relationship model. The entity recognition and relationship discovery models are trained with a multi-task learning objective on a large annotated corpus. We employ a concept masking paradigm to prevent the evolution of neural networks functioning as an associative memory and induce right inductive bias guiding the network to make inference using only the context. We uncover several import subnetworks, highlight important terms and concepts and elucidate several treatment modalities employed in related ailments in the past.
摘要：我们提取的通过采用新的实体和关系模型相关冠状病毒的文章语料库与COVID-19的实体和关系。实体识别和关系发现模型被训练与多任务学习目标上的大型注释的语料库。我们使用的一个概念掩蔽范式，以防止作为联想记忆功能的神经网络的发展，并引起右归纳偏置指导网络，使只使用情景推断。我们发现一些进口子网，突出重要术语和概念，并在过去相关疾病阐发采用几种治疗方式。
5. TorchKGE: Knowledge Graph Embedding in Python and PyTorch [PDF] 返回目录
Armand Boschin
Abstract: TorchKGE is a Python module for knowledge graph (KG) embedding relying solely on PyTorch. This package provides researchers and engineers with a clean and efficient API to design and test new models. It features a KG data structure, simple model interfaces and modules for negative sampling and model evaluation. Its main strength is a very fast evaluation module for the link prediction task, a central application of KG embedding. Various KG embedding models are also already implemented. Special attention has been paid to code efficiency and simplicity, documentation and API consistency. It is distributed using PyPI under BSD license. Source code and pointers to documentation and deployment can be found at this https URL.
摘要：TorchKGE是完全嵌入在PyTorch依靠知识图（KG）的Python模块。这个软件包提供了研究人员和工程师提供了一种清洁，高效的API来设计和测试新的模式。它具有一个KG数据结构，负采样和模型评价的简单模型接口和模块。它的主要力量是链接预测任务非常快的评估模块，KG嵌入的核心应用。各种KG嵌入模型也已经实施。特别注意了代码的效率和简单性，文档和API的一致性。它是根据BSD许可使用的PyPI分布。源代码和指向文档和部署可以在此HTTPS URL中找到。
Armand Boschin
Abstract: TorchKGE is a Python module for knowledge graph (KG) embedding relying solely on PyTorch. This package provides researchers and engineers with a clean and efficient API to design and test new models. It features a KG data structure, simple model interfaces and modules for negative sampling and model evaluation. Its main strength is a very fast evaluation module for the link prediction task, a central application of KG embedding. Various KG embedding models are also already implemented. Special attention has been paid to code efficiency and simplicity, documentation and API consistency. It is distributed using PyPI under BSD license. Source code and pointers to documentation and deployment can be found at this https URL.
摘要：TorchKGE是完全嵌入在PyTorch依靠知识图（KG）的Python模块。这个软件包提供了研究人员和工程师提供了一种清洁，高效的API来设计和测试新的模式。它具有一个KG数据结构，负采样和模型评价的简单模型接口和模块。它的主要力量是链接预测任务非常快的评估模块，KG嵌入的核心应用。各种KG嵌入模型也已经实施。特别注意了代码的效率和简单性，文档和API的一致性。它是根据BSD许可使用的PyPI分布。源代码和指向文档和部署可以在此HTTPS URL中找到。
6. UIT-HSE at WNUT-2020 Task 2: Exploiting CT-BERT for Identifying COVID-19 Information on the Twitter Social Network [PDF] 返回目录
Khiem Vinh Tran, Hao Phu Phan, Kiet Van Nguyen, Ngan Luu-Thuy Nguyen
Abstract: Recently, COVID-19 has affected a variety of real-life aspects of the world and led to dreadful consequences. More and more tweets about COVID-19 has been shared publicly on Twitter. However, the plurality of those Tweets are uninformative, which is challenging to build automatic systems to detect the informative ones for useful AI applications. In this paper, we present our results at the W-NUT 2020 Shared Task 2: Identification of Informative COVID-19 English Tweets. In particular, we propose our simple but effective approach using the transformer-based models based on COVID-Twitter-BERT (CT-BERT) with different fine-tuning techniques. As a result, we achieve the F1-Score of 90.94\% with the third place on the leaderboard of this task which attracted 56 submitted teams in total.
摘要：近日，COVID-19已经影响了各种世界的现实生活方面，并导致可怕的后果。关于COVID-19越来越多的微博已经在Twitter上公开分享。然而，多个那些鸣叫的是不提供信息，这是有挑战性的建立自动化系统来检测有用AI应用的信息的。信息化COVID-19英语鸣叫的识别：在本文中，我们在W-NUT 2020共享任务2展示我们的成果。特别是，我们提出我们的简单，但使用基于COVID，Twitter的-BERT（CT-BERT）具有不同的微调技术基于变压器的模型有效的方法。其结果是，我们实现了90.94 \％，与这个任务吸引了56支球队提交了总排行榜的第三位F1-得分。
Khiem Vinh Tran, Hao Phu Phan, Kiet Van Nguyen, Ngan Luu-Thuy Nguyen
Abstract: Recently, COVID-19 has affected a variety of real-life aspects of the world and led to dreadful consequences. More and more tweets about COVID-19 has been shared publicly on Twitter. However, the plurality of those Tweets are uninformative, which is challenging to build automatic systems to detect the informative ones for useful AI applications. In this paper, we present our results at the W-NUT 2020 Shared Task 2: Identification of Informative COVID-19 English Tweets. In particular, we propose our simple but effective approach using the transformer-based models based on COVID-Twitter-BERT (CT-BERT) with different fine-tuning techniques. As a result, we achieve the F1-Score of 90.94\% with the third place on the leaderboard of this task which attracted 56 submitted teams in total.
摘要：近日，COVID-19已经影响了各种世界的现实生活方面，并导致可怕的后果。关于COVID-19越来越多的微博已经在Twitter上公开分享。然而，多个那些鸣叫的是不提供信息，这是有挑战性的建立自动化系统来检测有用AI应用的信息的。信息化COVID-19英语鸣叫的识别：在本文中，我们在W-NUT 2020共享任务2展示我们的成果。特别是，我们提出我们的简单，但使用基于COVID，Twitter的-BERT（CT-BERT）具有不同的微调技术基于变压器的模型有效的方法。其结果是，我们实现了90.94 \％，与这个任务吸引了56支球队提交了总排行榜的第三位F1-得分。
7. Team Alex at CLEF CheckThat! 2020: Identifying Check-Worthy Tweets With Transformer Models [PDF] 返回目录
Alex Nikolov, Giovanni Da San Martino, Ivan Koychev, Preslav Nakov
Abstract: While misinformation and disinformation have been thriving in social media for years, with the emergence of the COVID-19 pandemic, the political and the health misinformation merged, thus elevating the problem to a whole new level and giving rise to the first global infodemic. The fight against this infodemic has many aspects, with fact-checking and debunking false and misleading claims being among the most important ones. Unfortunately, manual fact-checking is time-consuming and automatic fact-checking is resource-intense, which means that we need to pre-filter the input social media posts and to throw out those that do not appear to be check-worthy. With this in mind, here we propose a model for detecting check-worthy tweets about COVID-19, which combines deep contextualized text representations with modeling the social context of the tweet. We further describe a number of additional experiments and comparisons, which we believe should be useful for future research as they provide some indication about what techniques are effective for the task. Our official submission to the English version of CLEF-2020 CheckThat! Task 1, system Team_Alex, was ranked second with a MAP score of 0.8034, which is almost tied with the wining system, lagging behind by just 0.003 MAP points absolute.
摘要：虽然误导和虚假信息已经在社交媒体蓬勃发展多年，出现了COVID-19大流行，政治和健康误传合并，从而提升问题，以一个全新的水平，从而引发了全球第一个infodemic 。针对此infodemic的斗争有许多方面，与事实查证和揭穿虚假和误导性声明的最重要的是当中。不幸的是，手动事实检查费时，自动事实检查是资源密集型的，这意味着我们需要预过滤器的输入社交媒体帖子，并扔掉那些不显得退房值得。考虑到这一点，我们在这里提出了检测围绕COVID-19，它结合了深刻的情境文本表示与造型鸣叫的社会背景检查值得鸣叫的典范。发明人还描述了一些额外的实验和比较，我们认为这为他们提供什么技术是有效的任务一些迹象应该是为今后的研究非常有用。我们正式提交英文版CLEF-2020 CheckThat的！任务1，系统Team_Alex，用MAP得分0.8034，这几乎与美酒系统捆绑，由刚刚0.003绝对MAP分落后排在第二位。
Alex Nikolov, Giovanni Da San Martino, Ivan Koychev, Preslav Nakov
Abstract: While misinformation and disinformation have been thriving in social media for years, with the emergence of the COVID-19 pandemic, the political and the health misinformation merged, thus elevating the problem to a whole new level and giving rise to the first global infodemic. The fight against this infodemic has many aspects, with fact-checking and debunking false and misleading claims being among the most important ones. Unfortunately, manual fact-checking is time-consuming and automatic fact-checking is resource-intense, which means that we need to pre-filter the input social media posts and to throw out those that do not appear to be check-worthy. With this in mind, here we propose a model for detecting check-worthy tweets about COVID-19, which combines deep contextualized text representations with modeling the social context of the tweet. We further describe a number of additional experiments and comparisons, which we believe should be useful for future research as they provide some indication about what techniques are effective for the task. Our official submission to the English version of CLEF-2020 CheckThat! Task 1, system Team_Alex, was ranked second with a MAP score of 0.8034, which is almost tied with the wining system, lagging behind by just 0.003 MAP points absolute.
摘要：虽然误导和虚假信息已经在社交媒体蓬勃发展多年，出现了COVID-19大流行，政治和健康误传合并，从而提升问题，以一个全新的水平，从而引发了全球第一个infodemic 。针对此infodemic的斗争有许多方面，与事实查证和揭穿虚假和误导性声明的最重要的是当中。不幸的是，手动事实检查费时，自动事实检查是资源密集型的，这意味着我们需要预过滤器的输入社交媒体帖子，并扔掉那些不显得退房值得。考虑到这一点，我们在这里提出了检测围绕COVID-19，它结合了深刻的情境文本表示与造型鸣叫的社会背景检查值得鸣叫的典范。发明人还描述了一些额外的实验和比较，我们认为这为他们提供什么技术是有效的任务一些迹象应该是为今后的研究非常有用。我们正式提交英文版CLEF-2020 CheckThat的！任务1，系统Team_Alex，用MAP得分0.8034，这几乎与美酒系统捆绑，由刚刚0.003绝对MAP分落后排在第二位。
8. TransModality: An End2End Fusion Method with Transformer for Multimodal Sentiment Analysis [PDF] 返回目录
Zilong Wang, Zhaohong Wan, Xiaojun Wan
Abstract: Multimodal sentiment analysis is an important research area that predicts speaker's sentiment tendency through features extracted from textual, visual and acoustic modalities. The central challenge is the fusion method of the multimodal information. A variety of fusion methods have been proposed, but few of them adopt end-to-end translation models to mine the subtle correlation between modalities. Enlightened by recent success of Transformer in the area of machine translation, we propose a new fusion method, TransModality, to address the task of multimodal sentiment analysis. We assume that translation between modalities contributes to a better joint representation of speaker's utterance. With Transformer, the learned features embody the information both from the source modality and the target modality. We validate our model on multiple multimodal datasets: CMU-MOSI, MELD, IEMOCAP. The experiments show that our proposed method achieves the state-of-the-art performance.
摘要：多模态的情感分析是通过预测从文本，视觉和听觉方式提取的特征说话者的情绪倾向的重要研究领域。的主要挑战是多峰信息的融合方法。各种融合方法已经被提出，但很少采用终端到终端的翻译模型矿方式之间的微妙关系。在机器翻译方面的变压器最近成功的启发，提出了一种新的融合方法，TransModality，以解决多模态情感分析的任务。我们假设模式有助于之间的翻译说话者的话语更好的联合代表。带变压器，学习特点体现无论是从源模式和目标模式的信息。我们验证了我们在多个数据集多式联运模式：CMU-MOSI，MELD，IEMOCAP。实验结果表明，我们所提出的方法实现国家的最先进的性能。
Zilong Wang, Zhaohong Wan, Xiaojun Wan
Abstract: Multimodal sentiment analysis is an important research area that predicts speaker's sentiment tendency through features extracted from textual, visual and acoustic modalities. The central challenge is the fusion method of the multimodal information. A variety of fusion methods have been proposed, but few of them adopt end-to-end translation models to mine the subtle correlation between modalities. Enlightened by recent success of Transformer in the area of machine translation, we propose a new fusion method, TransModality, to address the task of multimodal sentiment analysis. We assume that translation between modalities contributes to a better joint representation of speaker's utterance. With Transformer, the learned features embody the information both from the source modality and the target modality. We validate our model on multiple multimodal datasets: CMU-MOSI, MELD, IEMOCAP. The experiments show that our proposed method achieves the state-of-the-art performance.
摘要：多模态的情感分析是通过预测从文本，视觉和听觉方式提取的特征说话者的情绪倾向的重要研究领域。的主要挑战是多峰信息的融合方法。各种融合方法已经被提出，但很少采用终端到终端的翻译模型矿方式之间的微妙关系。在机器翻译方面的变压器最近成功的启发，提出了一种新的融合方法，TransModality，以解决多模态情感分析的任务。我们假设模式有助于之间的翻译说话者的话语更好的联合代表。带变压器，学习特点体现无论是从源模式和目标模式的信息。我们验证了我们在多个数据集多式联运模式：CMU-MOSI，MELD，IEMOCAP。实验结果表明，我们所提出的方法实现国家的最先进的性能。
9. E-BERT: A Phrase and Product Knowledge Enhanced Language Model for E-commerce [PDF] 返回目录
Denghui Zhang, Zixuan Yuan, Yanchi Liu, Fuzhen Zhuang, Hui Xiong
Abstract: Pre-trained language models such as BERT have achieved great success in a broad range of natural language processing tasks. However, BERT cannot well support E-commerce related tasks due to the lack of two levels of domain knowledge, i.e., phrase-level and product-level. On one hand, many E-commerce tasks require an accurate understanding of domain phrases, whereas such fine-grained phrase-level knowledge is not explicitly modeled by BERT's training objective. On the other hand, product-level knowledge like product associations can enhance the language modeling of E-commerce, but they are not factual knowledge thus using them indiscriminately may introduce noise. To tackle the problem, we propose a unified pre-training framework, namely, E-BERT. Specifically, to preserve phrase-level knowledge, we introduce Adaptive Hybrid Masking, which allows the model to adaptively switch from learning preliminary word knowledge to learning complex phrases, based on the fitting progress of two modes. To utilize product-level knowledge, we introduce Neighbor Product Reconstruction, which trains E-BERT to predict a product's associated neighbors with a denoising cross attention layer. Our investigation reveals promising results in four downstream tasks, i.e., review-based question answering, aspect extraction, aspect sentiment classification, and product classification.
摘要：预先训练语言模型，如BERT已经在广泛的自然语言处理任务，取得了巨大成功。然而，BERT不能很好地支持电子商务相关的任务，由于缺乏领域知识，即，短语级和产品级两个级别的。一方面，许多电子商务任务需要域短语的准确理解，而这种细粒度的短语层次的知识不明确BERT的训练目标建模。在另一方面，同类产品的关联产品层面的知识可以提升电子商务的建模语言，但他们并不因此使用它们可以不加区分地引进噪音事实性知识。为了解决这个问题，我们提出了一个统一的岗前培训框架，即E-BERT。具体来说，保持短语层次的知识，我们引入自适应混合屏蔽，这使得从学习初级字知识学习复杂的短语，基于两种模式的拟合进度模型自适应开关。要利用产品层次的知识，介绍邻居重建的产品，这列车E-BERT来预测产品相关联的邻居与降噪跨注意力层。我们的调查揭示了四个下行任务有希望的结果，即审查为基础的问答，提取方面，一方面情感分类和产品分类。
Denghui Zhang, Zixuan Yuan, Yanchi Liu, Fuzhen Zhuang, Hui Xiong
Abstract: Pre-trained language models such as BERT have achieved great success in a broad range of natural language processing tasks. However, BERT cannot well support E-commerce related tasks due to the lack of two levels of domain knowledge, i.e., phrase-level and product-level. On one hand, many E-commerce tasks require an accurate understanding of domain phrases, whereas such fine-grained phrase-level knowledge is not explicitly modeled by BERT's training objective. On the other hand, product-level knowledge like product associations can enhance the language modeling of E-commerce, but they are not factual knowledge thus using them indiscriminately may introduce noise. To tackle the problem, we propose a unified pre-training framework, namely, E-BERT. Specifically, to preserve phrase-level knowledge, we introduce Adaptive Hybrid Masking, which allows the model to adaptively switch from learning preliminary word knowledge to learning complex phrases, based on the fitting progress of two modes. To utilize product-level knowledge, we introduce Neighbor Product Reconstruction, which trains E-BERT to predict a product's associated neighbors with a denoising cross attention layer. Our investigation reveals promising results in four downstream tasks, i.e., review-based question answering, aspect extraction, aspect sentiment classification, and product classification.
摘要：预先训练语言模型，如BERT已经在广泛的自然语言处理任务，取得了巨大成功。然而，BERT不能很好地支持电子商务相关的任务，由于缺乏领域知识，即，短语级和产品级两个级别的。一方面，许多电子商务任务需要域短语的准确理解，而这种细粒度的短语层次的知识不明确BERT的训练目标建模。在另一方面，同类产品的关联产品层面的知识可以提升电子商务的建模语言，但他们并不因此使用它们可以不加区分地引进噪音事实性知识。为了解决这个问题，我们提出了一个统一的岗前培训框架，即E-BERT。具体来说，保持短语层次的知识，我们引入自适应混合屏蔽，这使得从学习初级字知识学习复杂的短语，基于两种模式的拟合进度模型自适应开关。要利用产品层次的知识，介绍邻居重建的产品，这列车E-BERT来预测产品相关联的邻居与降噪跨注意力层。我们的调查揭示了四个下行任务有希望的结果，即审查为基础的问答，提取方面，一方面情感分类和产品分类。
10. Duluth at SemEval-2020 Task 7: Using Surprise as a Key to Unlock Humorous Headlines [PDF] 返回目录
Shuning Jin, Yue Yin, XianE Tang, Ted Pedersen
Abstract: We use pretrained transformer-based language models in SemEval-2020 Task 7: Assessing the Funniness of Edited News Headlines. Inspired by the incongruity theory of humor, we use a contrastive approach to capture the surprise in the edited headlines. In the official evaluation, our system gets 0.531 RMSE in Subtask 1, 11th among 49 submissions. In Subtask 2, our system gets 0.632 accuracy, 9th among 32 submissions.
摘要：我们使用预训练的基于变压器的语言模型在SemEval-2020任务7：评估编辑新闻标题的Funniness。幽默的不协调理论的启发，我们用对比的方法来捕捉惊喜编辑的头条新闻。在官方的评价，我们的系统中获取任务1 0.531 RMSE，49个提交11。在子程序2，我们的系统得到0.632的准确性，32个提交第九。
Shuning Jin, Yue Yin, XianE Tang, Ted Pedersen
Abstract: We use pretrained transformer-based language models in SemEval-2020 Task 7: Assessing the Funniness of Edited News Headlines. Inspired by the incongruity theory of humor, we use a contrastive approach to capture the surprise in the edited headlines. In the official evaluation, our system gets 0.531 RMSE in Subtask 1, 11th among 49 submissions. In Subtask 2, our system gets 0.632 accuracy, 9th among 32 submissions.
摘要：我们使用预训练的基于变压器的语言模型在SemEval-2020任务7：评估编辑新闻标题的Funniness。幽默的不协调理论的启发，我们用对比的方法来捕捉惊喜编辑的头条新闻。在官方的评价，我们的系统中获取任务1 0.531 RMSE，49个提交11。在子程序2，我们的系统得到0.632的准确性，32个提交第九。
11. UPB at SemEval-2020 Task 9: Identifying Sentiment in Code-Mixed Social Media Texts using Transformers and Multi-Task Learning [PDF] 返回目录
George-Eduard Zaharia, George-Alexandru Vlad, Dumitru-Clementin Cercel, Traian Rebedea, Costin-Gabriel Chiru
Abstract: Sentiment analysis is a process widely used in opinion mining campaigns conducted today. This phenomenon presents applications in a variety of fields, especially in collecting information related to the attitude or satisfaction of users concerning a particular subject. However, the task of managing such a process becomes noticeably more difficult when it is applied in cultures that tend to combine two languages in order to express ideas and thoughts. By interleaving words from two languages, the user can express with ease, but at the cost of making the text far less intelligible for those who are not familiar with this technique, but also for standard opinion mining algorithms. In this paper, we describe the systems developed by our team for SemEval-2020 Task 9 that aims to cover two well-known code-mixed languages: Hindi-English and Spanish-English. We intend to solve this issue by introducing a solution that takes advantage of several neural network approaches, as well as pre-trained word embeddings. Our approach (multlingual BERT) achieves promising performance on the Hindi-English task, with an average F1-score of 0.6850, registered on the competition leaderboard, ranking our team 16th out of 62 participants. For the Spanish-English task, we obtained an average F1-score of 0.7064 ranking our team 17th out of 29 participants by using another multilingual Transformer-based model, XLM-RoBERTa.
摘要：情感分析是广泛使用在今天进行的民意采矿活动的过程。这种现象呈现在各个领域的应用，尤其是在收集相关的态度或关于特定主题的用户满意度的信息。然而，管理这样一个过程的任务变得时它倾向于两种语言，以表达自己的想法和思想文化的结合应用明显更加困难。通过从两种语言交织的话，用户可以很容易表达，但在制作远不如理解为那些谁不熟悉这项技术的文本，也为标准无保留意见挖掘算法的成本。在本文中，我们描述我们的团队SemEval-2020开发任务9的系统，其目的是覆盖两个著名的代码混合的语言：印地文，英语和西班牙语，英语。我们打算通过引入一个解决方案，需要几个神经网络的优势办法来解决这个问题，以及预先训练字的嵌入。我们的方法（multlingual BERT）实现承诺的印地文，英语任务绩效，与0.6850的平均F1-成绩，在竞争中领先登记，排名我们的团队16日从62人参加。对于西英的工作，我们通过其他多语种基于变压器的模型，XLM - 罗伯塔获得的0.7064的平均F1-成绩排名我们的团队17出29人参加。
George-Eduard Zaharia, George-Alexandru Vlad, Dumitru-Clementin Cercel, Traian Rebedea, Costin-Gabriel Chiru
Abstract: Sentiment analysis is a process widely used in opinion mining campaigns conducted today. This phenomenon presents applications in a variety of fields, especially in collecting information related to the attitude or satisfaction of users concerning a particular subject. However, the task of managing such a process becomes noticeably more difficult when it is applied in cultures that tend to combine two languages in order to express ideas and thoughts. By interleaving words from two languages, the user can express with ease, but at the cost of making the text far less intelligible for those who are not familiar with this technique, but also for standard opinion mining algorithms. In this paper, we describe the systems developed by our team for SemEval-2020 Task 9 that aims to cover two well-known code-mixed languages: Hindi-English and Spanish-English. We intend to solve this issue by introducing a solution that takes advantage of several neural network approaches, as well as pre-trained word embeddings. Our approach (multlingual BERT) achieves promising performance on the Hindi-English task, with an average F1-score of 0.6850, registered on the competition leaderboard, ranking our team 16th out of 62 participants. For the Spanish-English task, we obtained an average F1-score of 0.7064 ranking our team 17th out of 29 participants by using another multilingual Transformer-based model, XLM-RoBERTa.
摘要：情感分析是广泛使用在今天进行的民意采矿活动的过程。这种现象呈现在各个领域的应用，尤其是在收集相关的态度或关于特定主题的用户满意度的信息。然而，管理这样一个过程的任务变得时它倾向于两种语言，以表达自己的想法和思想文化的结合应用明显更加困难。通过从两种语言交织的话，用户可以很容易表达，但在制作远不如理解为那些谁不熟悉这项技术的文本，也为标准无保留意见挖掘算法的成本。在本文中，我们描述我们的团队SemEval-2020开发任务9的系统，其目的是覆盖两个著名的代码混合的语言：印地文，英语和西班牙语，英语。我们打算通过引入一个解决方案，需要几个神经网络的优势办法来解决这个问题，以及预先训练字的嵌入。我们的方法（multlingual BERT）实现承诺的印地文，英语任务绩效，与0.6850的平均F1-成绩，在竞争中领先登记，排名我们的团队16日从62人参加。对于西英的工作，我们通过其他多语种基于变压器的模型，XLM - 罗伯塔获得的0.7064的平均F1-成绩排名我们的团队17出29人参加。
12. UPB at SemEval-2020 Task 8: Joint Textual and Visual Modeling in a Multi-Task Learning Architecture for Memotion Analysis [PDF] 返回目录
George-Alexandru Vlad, George-Eduard Zaharia, Dumitru-Clementin Cercel, Costin-Gabriel Chiru, Stefan Trausan-Matu
Abstract: Users from the online environment can create different ways of expressing their thoughts, opinions, or conception of amusement. Internet memes were created specifically for these situations. Their main purpose is to transmit ideas by using combinations of images and texts such that they will create a certain state for the receptor, depending on the message the meme has to send. These posts can be related to various situations or events, thus adding a funny side to any circumstance our world is situated in. In this paper, we describe the system developed by our team for SemEval-2020 Task 8: Memotion Analysis. More specifically, we introduce a novel system to analyze these posts, a multimodal multi-task learning architecture that combines ALBERT for text encoding with VGG-16 for image representation. In this manner, we show that the information behind them can be properly revealed. Our approach achieves good performance on each of the three subtasks of the current competition, ranking 11th for Subtask A (0.3453 macro F1-score), 1st for Subtask B (0.5183 macro F1-score), and 3rd for Subtask C (0.3171 macro F1-score) while exceeding the official baseline results by high margins.
摘要：从在线环境中的用户可以创建表达自己的想法，意见或娱乐的概念不同的方式。网络爆红了针对这些情况专门创建。他们的主要目的是为发射的想法用图片和文字等，他们将创造一定的状态受体，取决于米姆有发送消息的组合。这些职位可以与各种情况或事件，从而增加了有趣的一面我们的世界坐落在任何情况下，在本文中，我们描述我们的团队SemEval-2020任务8开发的系统：Memotion分析。更具体地讲，我们引入一个新的系统来分析这些职位，多模式多任务学习架构，结合伟业与VGG-16的文字编码图像表示。通过这种方式，我们表明，他们背后的信息可以适当披露。我们的方法实现对每个当前竞争的三个子任务的良好表现，排名第11位的子任务A（0.3453宏观F1-分），第一个为子任务B（0.5183宏观F1-得分），和第3的子任务C（0.3171宏观F1 -score），而超过高利润率的官方基准结果。
George-Alexandru Vlad, George-Eduard Zaharia, Dumitru-Clementin Cercel, Costin-Gabriel Chiru, Stefan Trausan-Matu
Abstract: Users from the online environment can create different ways of expressing their thoughts, opinions, or conception of amusement. Internet memes were created specifically for these situations. Their main purpose is to transmit ideas by using combinations of images and texts such that they will create a certain state for the receptor, depending on the message the meme has to send. These posts can be related to various situations or events, thus adding a funny side to any circumstance our world is situated in. In this paper, we describe the system developed by our team for SemEval-2020 Task 8: Memotion Analysis. More specifically, we introduce a novel system to analyze these posts, a multimodal multi-task learning architecture that combines ALBERT for text encoding with VGG-16 for image representation. In this manner, we show that the information behind them can be properly revealed. Our approach achieves good performance on each of the three subtasks of the current competition, ranking 11th for Subtask A (0.3453 macro F1-score), 1st for Subtask B (0.5183 macro F1-score), and 3rd for Subtask C (0.3171 macro F1-score) while exceeding the official baseline results by high margins.
摘要：从在线环境中的用户可以创建表达自己的想法，意见或娱乐的概念不同的方式。网络爆红了针对这些情况专门创建。他们的主要目的是为发射的想法用图片和文字等，他们将创造一定的状态受体，取决于米姆有发送消息的组合。这些职位可以与各种情况或事件，从而增加了有趣的一面我们的世界坐落在任何情况下，在本文中，我们描述我们的团队SemEval-2020任务8开发的系统：Memotion分析。更具体地讲，我们引入一个新的系统来分析这些职位，多模式多任务学习架构，结合伟业与VGG-16的文字编码图像表示。通过这种方式，我们表明，他们背后的信息可以适当披露。我们的方法实现对每个当前竞争的三个子任务的良好表现，排名第11位的子任务A（0.3453宏观F1-分），第一个为子任务B（0.5183宏观F1-得分），和第3的子任务C（0.3171宏观F1 -score），而超过高利润率的官方基准结果。
13. Romanian Diacritics Restoration Using Recurrent Neural Networks [PDF] 返回目录
Stefan Ruseti, Teodor-Mihai Cotet, Mihai Dascalu
Abstract: Diacritics restoration is a mandatory step for adequately processing Romanian texts, and not a trivial one, as you generally need context in order to properly restore a character. Most previous methods which were experimented for Romanian restoration of diacritics do not use neural networks. Among those that do, there are no solutions specifically optimized for this particular language (i.e., they were generally designed to work on many different languages). Therefore we propose a novel neural architecture based on recurrent neural networks that can attend information at different levels of abstractions in order to restore diacritics.
摘要：变音符号的恢复是充分处理罗马尼亚文强制性的一步，而不是一个简单的一个，因为你一般需要以正确地还原人物背景。这是实验的变音符号的罗马尼亚恢复以往大多数方法不使用神经网络。在那些做，有没有专门针对这种特殊的语言（即，它们一般设计工作在许多不同的语言）优化的解决方案。因此，我们建议基于能够以恢复变音符号参加不同层次的抽象信息的递归神经网络一种新型的神经结构。
Stefan Ruseti, Teodor-Mihai Cotet, Mihai Dascalu
Abstract: Diacritics restoration is a mandatory step for adequately processing Romanian texts, and not a trivial one, as you generally need context in order to properly restore a character. Most previous methods which were experimented for Romanian restoration of diacritics do not use neural networks. Among those that do, there are no solutions specifically optimized for this particular language (i.e., they were generally designed to work on many different languages). Therefore we propose a novel neural architecture based on recurrent neural networks that can attend information at different levels of abstractions in order to restore diacritics.
摘要：变音符号的恢复是充分处理罗马尼亚文强制性的一步，而不是一个简单的一个，因为你一般需要以正确地还原人物背景。这是实验的变音符号的罗马尼亚恢复以往大多数方法不使用神经网络。在那些做，有没有专门针对这种特殊的语言（即，它们一般设计工作在许多不同的语言）优化的解决方案。因此，我们建议基于能够以恢复变音符号参加不同层次的抽象信息的递归神经网络一种新型的神经结构。
14. SemEval-2020 Task 11: Detection of Propaganda Techniques in News Articles [PDF] 返回目录
G. Da San Martino, A. Barrón-Cedeño, H. Wachsmuth, R. Petrov, P. Nakov
Abstract: We present the results and the main findings of SemEval-2020 Task 11 on Detection of Propaganda Techniques in News Articles. The task featured two subtasks. Subtask SI is about Span Identification: given a plain-text document, spot the specific text fragments containing propaganda. Subtask TC is about Technique Classification: given a specific text fragment, in the context of a full document, determine the propaganda technique it uses, choosing from an inventory of 14 possible propaganda techniques. The task attracted a large number of participants: 250 teams signed up to participate and 44 made a submission on the test set. In this paper, we present the task, analyze the results, and discuss the system submissions and the methods they used. For both subtasks, the best systems used pre-trained Transformers and ensembles.
摘要：我们目前的结果和SemEval-2020任务11对新闻文章宣传的检测技术的主要结论。任务有两位子任务。子任务SI约为跨度识别：给定的纯文本文件，斑含有宣传特定文本片段。子任务TC约为技术分类：给定一个具体的文字片段，在一个完整的文档的情况下，确定它使用的宣传手法，从14个可能的宣传技术清单进行选择。任务吸引了大量的参与者：250个团队报名参加，并取得44在测试组提交。在本文中，我们目前的任务，分析结果，并讨论了系统提交和他们使用的方法。对于这两个子任务，最好的系统中使用预先训练变压器和合唱团。
G. Da San Martino, A. Barrón-Cedeño, H. Wachsmuth, R. Petrov, P. Nakov
Abstract: We present the results and the main findings of SemEval-2020 Task 11 on Detection of Propaganda Techniques in News Articles. The task featured two subtasks. Subtask SI is about Span Identification: given a plain-text document, spot the specific text fragments containing propaganda. Subtask TC is about Technique Classification: given a specific text fragment, in the context of a full document, determine the propaganda technique it uses, choosing from an inventory of 14 possible propaganda techniques. The task attracted a large number of participants: 250 teams signed up to participate and 44 made a submission on the test set. In this paper, we present the task, analyze the results, and discuss the system submissions and the methods they used. For both subtasks, the best systems used pre-trained Transformers and ensembles.
摘要：我们目前的结果和SemEval-2020任务11对新闻文章宣传的检测技术的主要结论。任务有两位子任务。子任务SI约为跨度识别：给定的纯文本文件，斑含有宣传特定文本片段。子任务TC约为技术分类：给定一个具体的文字片段，在一个完整的文档的情况下，确定它使用的宣传手法，从14个可能的宣传技术清单进行选择。任务吸引了大量的参与者：250个团队报名参加，并取得44在测试组提交。在本文中，我们目前的任务，分析结果，并讨论了系统提交和他们使用的方法。对于这两个子任务，最好的系统中使用预先训练变压器和合唱团。
15. Automatic Dialect Adaptation in Finnish and its Effect on Perceived Creativity [PDF] 返回目录
Mika Hämäläinen, Niko Partanen, Khalid Alnajjar, Jack Rueter, Thierry Poibeau
Abstract: We present a novel approach for adapting text written in standard Finnish to different dialects. We experiment with character level NMT models both by using a multi-dialectal and transfer learning approaches. The models are tested with over 20 different dialects. The results seem to favor transfer learning, although not strongly over the multi-dialectal approach. We study the influence dialectal adaptation has on perceived creativity of computer generated poetry. Our results suggest that the more the dialect deviates from the standard Finnish, the lower scores people tend to give on an existing evaluation metric. However, on a word association test, people associate creativity and originality more with dialect and fluency more with standard Finnish.
摘要：我们提出了适应用标准芬兰语不同的方言文本的新方法。我们与人物等级NMT车型都采用了多方言和转移学习方法进行实验。该机型有超过20个不同的方言进行测试。结果似乎有利于转移的学习，虽然不是强在多方言的做法。我们研究的影响方言适应对计算机生成的诗歌感知创造力。我们的研究结果表明，更多的从标准芬兰方言偏离时，得分较低的人往往给人在现有的评价指标。然而，就一个字联想测试，人把创造力和独创性多用方言和流畅更标准的芬兰。
Mika Hämäläinen, Niko Partanen, Khalid Alnajjar, Jack Rueter, Thierry Poibeau
Abstract: We present a novel approach for adapting text written in standard Finnish to different dialects. We experiment with character level NMT models both by using a multi-dialectal and transfer learning approaches. The models are tested with over 20 different dialects. The results seem to favor transfer learning, although not strongly over the multi-dialectal approach. We study the influence dialectal adaptation has on perceived creativity of computer generated poetry. Our results suggest that the more the dialect deviates from the standard Finnish, the lower scores people tend to give on an existing evaluation metric. However, on a word association test, people associate creativity and originality more with dialect and fluency more with standard Finnish.
摘要：我们提出了适应用标准芬兰语不同的方言文本的新方法。我们与人物等级NMT车型都采用了多方言和转移学习方法进行实验。该机型有超过20个不同的方言进行测试。结果似乎有利于转移的学习，虽然不是强在多方言的做法。我们研究的影响方言适应对计算机生成的诗歌感知创造力。我们的研究结果表明，更多的从标准芬兰方言偏离时，得分较低的人往往给人在现有的评价指标。然而，就一个字联想测试，人把创造力和独创性多用方言和流畅更标准的芬兰。
16. BANANA at WNUT-2020 Task 2: Identifying COVID-19 Information on Twitter by Combining Deep Learning and Transfer Learning Models [PDF] 返回目录
Tin Van Huynh, Luan Thanh Nguyen, Son T. Luu
Abstract: The outbreak COVID-19 virus caused a significant impact on the health of people all over the world. Therefore, it is essential to have a piece of constant and accurate information about the disease with everyone. This paper describes our prediction system for WNUT-2020 Task 2: Identification of Informative COVID-19 English Tweets. The dataset for this task contains size 10,000 tweets in English labeled by humans. The ensemble model from our three transformer and deep learning models is used for the final prediction. The experimental result indicates that we have achieved F1 for the INFORMATIVE label on our systems at 88.81% on the test set.
摘要：爆发COVID-19病毒引起的人遍布世界各地的健康有显著的影响。因此，必须有一块约和大家一起不断的疾病和准确的信息。本文介绍了WNUT-2020任务2我们的预测系统：信息化COVID-19英语鸣叫的识别。此任务的数据集包含大小的英文万个微博由人类标记。从我们的三个变压器和深厚的学习模型的集成模型用于最终的预测。实验结果表明，我们已经在88.81％，在测试组已达到F1对我们的系统翔实的标签。
Tin Van Huynh, Luan Thanh Nguyen, Son T. Luu
Abstract: The outbreak COVID-19 virus caused a significant impact on the health of people all over the world. Therefore, it is essential to have a piece of constant and accurate information about the disease with everyone. This paper describes our prediction system for WNUT-2020 Task 2: Identification of Informative COVID-19 English Tweets. The dataset for this task contains size 10,000 tweets in English labeled by humans. The ensemble model from our three transformer and deep learning models is used for the final prediction. The experimental result indicates that we have achieved F1 for the INFORMATIVE label on our systems at 88.81% on the test set.
摘要：爆发COVID-19病毒引起的人遍布世界各地的健康有显著的影响。因此，必须有一块约和大家一起不断的疾病和准确的信息。本文介绍了WNUT-2020任务2我们的预测系统：信息化COVID-19英语鸣叫的识别。此任务的数据集包含大小的英文万个微博由人类标记。从我们的三个变压器和深厚的学习模型的集成模型用于最终的预测。实验结果表明，我们已经在88.81％，在测试组已达到F1对我们的系统翔实的标签。
17. Once Upon A Time In Visualization: Understanding the Use of Textual Narratives for Causality [PDF] 返回目录
Arjun Choudhry, Mandar Sharma, Pramod Chundury, Thomas Kapler, Derek W.S. Gray, Naren Ramakrishnan, Niklas Elmqvist
Abstract: Causality visualization can help people understand temporal chains of events, such as messages sent in a distributed system, cause and effect in a historical conflict, or the interplay between political actors over time. However, as the scale and complexity of these event sequences grows, even these visualizations can become overwhelming to use. In this paper, we propose the use of textual narratives as a data-driven storytelling method to augment causality visualization. We first propose a design space for how textual narratives can be used to describe causal data. We then present results from a crowdsourced user study where participants were asked to recover causality information from two causality visualizations--causal graphs and Hasse diagrams--with and without an associated textual narrative. Finally, we describe CAUSEWORKS, a causality visualization system for understanding how specific interventions influence a causal model. The system incorporates an automatic textual narrative mechanism based on our design space. We validate CAUSEWORKS through interviews with experts who used the system for understanding complex events.
摘要：因果关系可视化可以帮助人们了解事件的时间链，如在历史冲突，或随着时间的推移政治行动者之间的相互作用在分布式系统中，因果发送的消息。然而，随着这些事件序列的规模和复杂性的增长，甚至这些可视化可以成为压倒使用。在本文中，我们建议使用文本叙述的是一个数据驱动的讲故事的方法来增强因果关系可视化。我们首先提出了如何文本叙事可以用来描述因果数据的设计空间。我们从一个众包的用户研究，参与者被要求从两个因果关系的可视化恢复因果关系的信息。然后现在的结果 - 因果图和哈斯图 - 与不相关的文字叙述。最后，我们描述CAUSEWORKS，为了解具体的干预措施如何影响一个因果模型因果关系可视化系统。该系统采用基于我们的设计空间的自动文本叙事机制。我们通过与专家谁使用的系统理解复杂事件的采访验证CAUSEWORKS。
Arjun Choudhry, Mandar Sharma, Pramod Chundury, Thomas Kapler, Derek W.S. Gray, Naren Ramakrishnan, Niklas Elmqvist
Abstract: Causality visualization can help people understand temporal chains of events, such as messages sent in a distributed system, cause and effect in a historical conflict, or the interplay between political actors over time. However, as the scale and complexity of these event sequences grows, even these visualizations can become overwhelming to use. In this paper, we propose the use of textual narratives as a data-driven storytelling method to augment causality visualization. We first propose a design space for how textual narratives can be used to describe causal data. We then present results from a crowdsourced user study where participants were asked to recover causality information from two causality visualizations--causal graphs and Hasse diagrams--with and without an associated textual narrative. Finally, we describe CAUSEWORKS, a causality visualization system for understanding how specific interventions influence a causal model. The system incorporates an automatic textual narrative mechanism based on our design space. We validate CAUSEWORKS through interviews with experts who used the system for understanding complex events.
摘要：因果关系可视化可以帮助人们了解事件的时间链，如在历史冲突，或随着时间的推移政治行动者之间的相互作用在分布式系统中，因果发送的消息。然而，随着这些事件序列的规模和复杂性的增长，甚至这些可视化可以成为压倒使用。在本文中，我们建议使用文本叙述的是一个数据驱动的讲故事的方法来增强因果关系可视化。我们首先提出了如何文本叙事可以用来描述因果数据的设计空间。我们从一个众包的用户研究，参与者被要求从两个因果关系的可视化恢复因果关系的信息。然后现在的结果 - 因果图和哈斯图 - 与不相关的文字叙述。最后，我们描述CAUSEWORKS，为了解具体的干预措施如何影响一个因果模型因果关系可视化系统。该系统采用基于我们的设计空间的自动文本叙事机制。我们通过与专家谁使用的系统理解复杂事件的采访验证CAUSEWORKS。
18. QiaoNing at SemEval-2020 Task 4: Commonsense Validation and Explanation system based on ensemble of language model [PDF] 返回目录
Pai Liu
Abstract: In this paper, we present language model system submitted to SemEval-2020 Task 4 competition: "Commonsense Validation and Explanation". We participate in two subtasks for subtask A: validation and subtask B: Explanation. We implemented with transfer learning using pretrained language models (BERT, XLNet, RoBERTa, and ALBERT) and fine-tune them on this task. Then we compared their characteristics in this task to help future researchers understand and use these models more properly. The ensembled model better solves this problem, making the model's accuracy reached 95.9% on subtask A, which just worse than human's by only 3% accuracy.
摘要：在本文中，我们目前语言模型提交给SemEval-2020任务4赛制：“常识验证和说明”。我们参加了两项子的子任务：确认和子任务B：说明。我们使用预训练的语言模型（BERT，XLNet，罗伯塔和ALBERT）和微调他们对这个任务转移的学习来实现。然后，我们比较了它们在这项任务特点，帮助未来的研究者理解和更恰当地使用这些模型。整体模型较好地解决了这一问题，使得模型的准确度只有3％的准确率达到了上子任务95.9％，其中仅比人类差。
Pai Liu
Abstract: In this paper, we present language model system submitted to SemEval-2020 Task 4 competition: "Commonsense Validation and Explanation". We participate in two subtasks for subtask A: validation and subtask B: Explanation. We implemented with transfer learning using pretrained language models (BERT, XLNet, RoBERTa, and ALBERT) and fine-tune them on this task. Then we compared their characteristics in this task to help future researchers understand and use these models more properly. The ensembled model better solves this problem, making the model's accuracy reached 95.9% on subtask A, which just worse than human's by only 3% accuracy.
摘要：在本文中，我们目前语言模型提交给SemEval-2020任务4赛制：“常识验证和说明”。我们参加了两项子的子任务：确认和子任务B：说明。我们使用预训练的语言模型（BERT，XLNet，罗伯塔和ALBERT）和微调他们对这个任务转移的学习来实现。然后，我们比较了它们在这项任务特点，帮助未来的研究者理解和更恰当地使用这些模型。整体模型较好地解决了这一问题，使得模型的准确度只有3％的准确率达到了上子任务95.9％，其中仅比人类差。
19. MIDAS at SemEval-2020 Task 10: Emphasis Selection using Label Distribution Learning and Contextual Embeddings [PDF] 返回目录
Sarthak Anand, Pradyumna Gupta, Hemant Yadav, Debanjan Mahata, Rakesh Gosangi, Haimin Zhang, Rajiv Ratn Shah
Abstract: This paper presents our submission to the SemEval 2020 - Task 10 on emphasis selection in written text. We approach this emphasis selection problem as a sequence labeling task where we represent the underlying text with various contextual embedding models. We also employ label distribution learning to account for annotator disagreements. We experiment with the choice of model architectures, trainability of layers, and different contextual embeddings. Our best performing architecture is an ensemble of different models, which achieved an overall matching score of 0.783, placing us 15th out of 31 participating teams. Lastly, we analyze the results in terms of parts of speech tags, sentence lengths, and word ordering.
摘要：本文介绍了我们提交SemEval 2020 - 对书面文字的重点选择任务10。我们处理这个重点的选择问题，因为我们代表了不同的上下文嵌入模型基础文本序列标签制作任务。我们还采用标签分发学习账户注释分歧。我们与模型架构层的可训练性，不同语境的嵌入的选择实验。我们表现最好的建筑是不同型号的集合，它实现了0.783的整体匹配分数，把我们的31支参赛队伍15日进行。最后，我们分析在语音标签，句子的长度，和字顺序的部分所获得的成果。
Sarthak Anand, Pradyumna Gupta, Hemant Yadav, Debanjan Mahata, Rakesh Gosangi, Haimin Zhang, Rajiv Ratn Shah
Abstract: This paper presents our submission to the SemEval 2020 - Task 10 on emphasis selection in written text. We approach this emphasis selection problem as a sequence labeling task where we represent the underlying text with various contextual embedding models. We also employ label distribution learning to account for annotator disagreements. We experiment with the choice of model architectures, trainability of layers, and different contextual embeddings. Our best performing architecture is an ensemble of different models, which achieved an overall matching score of 0.783, placing us 15th out of 31 participating teams. Lastly, we analyze the results in terms of parts of speech tags, sentence lengths, and word ordering.
摘要：本文介绍了我们提交SemEval 2020 - 对书面文字的重点选择任务10。我们处理这个重点的选择问题，因为我们代表了不同的上下文嵌入模型基础文本序列标签制作任务。我们还采用标签分发学习账户注释分歧。我们与模型架构层的可训练性，不同语境的嵌入的选择实验。我们表现最好的建筑是不同型号的集合，它实现了0.783的整体匹配分数，把我们的31支参赛队伍15日进行。最后，我们分析在语音标签，句子的长度，和字顺序的部分所获得的成果。
20. Bio-inspired Structure Identification in Language Embeddings [PDF] 返回目录
Hongwei, Zhou, Oskar Elek, Pranav Anand, Angus G. Forbes
Abstract: Word embeddings are a popular way to improve downstream per-formances in contemporary language modeling. However, the un-derlying geometric structure of the embedding space is not wellunderstood. We present a series of explorations using bio-inspiredmethodology to traverse and visualize word embeddings, demon-strating evidence of discernible structure. Moreover, our modelalso produces word similarity rankings that are plausible yet verydifferent from common similarity metrics, mainly cosine similarityand Euclidean distance. We show that our bio-inspired model canbe used to investigate how different word embedding techniquesresult in different semantic outputs, which can emphasize or obscureparticular interpretations in textual data.
摘要：Word中的嵌入是提高当代语言建模下游的每一个formances流行的方式。然而，嵌入空间的未derlying几何结构不是wellunderstood。我们提出了一系列采用生物inspiredmethodology遍历和可视化探索的嵌入字，可辨别结构的恶魔strating证据。此外，我们的modelalso产生单词类似的排名是似是而非然而，从普通的相似性指标，主要是余弦similarityand欧氏距离verydifferent。我们证明了我们的仿生模型canbe用于研究词不同的语义输出嵌入techniquesresult如何不同，它可以强调或文本数据obscureparticular解释。
Hongwei, Zhou, Oskar Elek, Pranav Anand, Angus G. Forbes
Abstract: Word embeddings are a popular way to improve downstream per-formances in contemporary language modeling. However, the un-derlying geometric structure of the embedding space is not wellunderstood. We present a series of explorations using bio-inspiredmethodology to traverse and visualize word embeddings, demon-strating evidence of discernible structure. Moreover, our modelalso produces word similarity rankings that are plausible yet verydifferent from common similarity metrics, mainly cosine similarityand Euclidean distance. We show that our bio-inspired model canbe used to investigate how different word embedding techniquesresult in different semantic outputs, which can emphasize or obscureparticular interpretations in textual data.
摘要：Word中的嵌入是提高当代语言建模下游的每一个formances流行的方式。然而，嵌入空间的未derlying几何结构不是wellunderstood。我们提出了一系列采用生物inspiredmethodology遍历和可视化探索的嵌入字，可辨别结构的恶魔strating证据。此外，我们的modelalso产生单词类似的排名是似是而非然而，从普通的相似性指标，主要是余弦similarityand欧氏距离verydifferent。我们证明了我们的仿生模型canbe用于研究词不同的语义输出嵌入techniquesresult如何不同，它可以强调或文本数据obscureparticular解释。
21. Accenture at CheckThat! 2020: If you say so: Post-hoc fact-checking of claims using transformer-based models [PDF] 返回目录
Evan Williams, Paul Rodrigues, Valerie Novak
Abstract: We introduce the strategies used by the Accenture Team for the CLEF2020 CheckThat! Lab, Task 1, on English and Arabic. This shared task evaluated whether a claim in social media text should be professionally fact checked. To a journalist, a statement presented as fact, which would be of interest to a large audience, requires professional fact-checking before dissemination. We utilized BERT and RoBERTa models to identify claims in social media text a professional fact-checker should review, and rank these in priority order for the fact-checker. For the English challenge, we fine-tuned a RoBERTa model and added an extra mean pooling layer and a dropout layer to enhance generalizability to unseen text. For the Arabic task, we fine-tuned Arabic-language BERT models and demonstrate the use of back-translation to amplify the minority class and balance the dataset. The work presented here was scored 1st place in the English track, and 1st, 2nd, 3rd, and 4th place in the Arabic track.
摘要：介绍了CLEF2020 CheckThat使用埃森哲团队的策略！实验室任务1，英语和阿拉伯语。评估社交媒体文本的索赔是否应当认真事实上这个共享任务检查。一个记者，声明提出的事实，这将是感兴趣的大量受众，传播需要专业的前事实检查。我们使用BERT和罗伯塔模型，以确定在社交媒体上的文字要求专业的事实检查应审查和排名这些优先顺序的事实检查。对于英语的挑战，我们微调一个罗伯塔模型，并增加了额外的平均汇聚层和辍学层，以加强推广到看不见的文本。对于阿拉伯语的任务，我们微调阿拉伯语BERT模型和演示使用回译的放大少数类和平衡的数据集。这里介绍的工作在阿拉伯语赛道拿下了中英文曲目第一名，和第1，第2，第3和第4位。
Evan Williams, Paul Rodrigues, Valerie Novak
Abstract: We introduce the strategies used by the Accenture Team for the CLEF2020 CheckThat! Lab, Task 1, on English and Arabic. This shared task evaluated whether a claim in social media text should be professionally fact checked. To a journalist, a statement presented as fact, which would be of interest to a large audience, requires professional fact-checking before dissemination. We utilized BERT and RoBERTa models to identify claims in social media text a professional fact-checker should review, and rank these in priority order for the fact-checker. For the English challenge, we fine-tuned a RoBERTa model and added an extra mean pooling layer and a dropout layer to enhance generalizability to unseen text. For the Arabic task, we fine-tuned Arabic-language BERT models and demonstrate the use of back-translation to amplify the minority class and balance the dataset. The work presented here was scored 1st place in the English track, and 1st, 2nd, 3rd, and 4th place in the Arabic track.
摘要：介绍了CLEF2020 CheckThat使用埃森哲团队的策略！实验室任务1，英语和阿拉伯语。评估社交媒体文本的索赔是否应当认真事实上这个共享任务检查。一个记者，声明提出的事实，这将是感兴趣的大量受众，传播需要专业的前事实检查。我们使用BERT和罗伯塔模型，以确定在社交媒体上的文字要求专业的事实检查应审查和排名这些优先顺序的事实检查。对于英语的挑战，我们微调一个罗伯塔模型，并增加了额外的平均汇聚层和辍学层，以加强推广到看不见的文本。对于阿拉伯语的任务，我们微调阿拉伯语BERT模型和演示使用回译的放大少数类和平衡的数据集。这里介绍的工作在阿拉伯语赛道拿下了中英文曲目第一名，和第1，第2，第3和第4位。
22. Recent Trends in the Use of Deep Learning Models for Grammar Error Handling [PDF] 返回目录
Mina Naghshnejad, Tarun Joshi, Vijayan N. Nair
Abstract: Grammar error handling (GEH) is an important topic in natural language processing (NLP). GEH includes both grammar error detection and grammar error correction. Recent advances in computation systems have promoted the use of deep learning (DL) models for NLP problems such as GEH. In this survey we focus on two main DL approaches for GEH: neural machine translation models and editor models. We describe the three main stages of the pipeline for these models: data preparation, training, and inference. Additionally, we discuss different techniques to improve the performance of these models at each stage of the pipeline. We compare the performance of different models and conclude with proposed future directions.
摘要：语法错误处理（GEH）是在自然语言处理（NLP）的一个重要课题。 GEH包括语法错误检测和语法错误校正。在计算系统的最新进展，促进了NLP问题，如GEH使用深度学习（DL）的模型。在本次调查中，我们重点关注两个主要DL为GEH方法：神经机器翻译模型和编辑模式。我们描述了管道的三个主要阶段对这些模型：数据准备，训练和推理。此外，我们还讨论了不同的技术来改善这些模型在流水线的每一个阶段的表现。我们比较不同车型的性能和提出未来发展方向的结论。
Mina Naghshnejad, Tarun Joshi, Vijayan N. Nair
Abstract: Grammar error handling (GEH) is an important topic in natural language processing (NLP). GEH includes both grammar error detection and grammar error correction. Recent advances in computation systems have promoted the use of deep learning (DL) models for NLP problems such as GEH. In this survey we focus on two main DL approaches for GEH: neural machine translation models and editor models. We describe the three main stages of the pipeline for these models: data preparation, training, and inference. Additionally, we discuss different techniques to improve the performance of these models at each stage of the pipeline. We compare the performance of different models and conclude with proposed future directions.
摘要：语法错误处理（GEH）是在自然语言处理（NLP）的一个重要课题。 GEH包括语法错误检测和语法错误校正。在计算系统的最新进展，促进了NLP问题，如GEH使用深度学习（DL）的模型。在本次调查中，我们重点关注两个主要DL为GEH方法：神经机器翻译模型和编辑模式。我们描述了管道的三个主要阶段对这些模型：数据准备，训练和推理。此外，我们还讨论了不同的技术来改善这些模型在流水线的每一个阶段的表现。我们比较不同车型的性能和提出未来发展方向的结论。
23. Measuring Massive Multitask Language Understanding [PDF] 返回目录
Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, Jacob Steinhardt
Abstract: We propose a new test to measure a text model's multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more. To attain high accuracy on this test, models must possess extensive world knowledge and problem solving ability. We find that while most recent models have near random-chance accuracy, the very largest GPT-3 model improves over random chance by almost 20 percentage points on average. However, on every one of the 57 tasks, the best models still need substantial improvements before they can reach human-level accuracy. Models also have lopsided performance and frequently do not know when they are wrong. Worse, they still have near-random accuracy on some socially important subjects such as morality and law. By comprehensively evaluating the breadth and depth of a model's academic and professional understanding, our test can be used to analyze models across many tasks and to identify important shortcomings.
摘要：本文提出了一种新的测试来衡量一个文本模型的多任务准确性。该测试涵盖57项任务，包括初等数学，美国历史，计算机科学，法律，等等。为了达到这个测试精度高，模型必须具备广泛的全球知识和解决问题的能力。我们发现，尽管最新的车型有近随机机会精度，规模非常大的GPT-3型平均提高近20个百分点，改进了随机的机会。然而，在的57项任务的每一个，最好的车型仍然需要实质性的改进才可以达到人类水平的精确度。型号也有不平衡的表现，经常不知道什么时候他们错了。更糟的是，他们仍然有一些重要社会意义的科目，如道德和法律附近随机准确性。通过综合评价模型的学术和专业认识的广度和深度，我们的测试可以用来分析横跨许多任务模型，并确定重要的缺点。
Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, Jacob Steinhardt
Abstract: We propose a new test to measure a text model's multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more. To attain high accuracy on this test, models must possess extensive world knowledge and problem solving ability. We find that while most recent models have near random-chance accuracy, the very largest GPT-3 model improves over random chance by almost 20 percentage points on average. However, on every one of the 57 tasks, the best models still need substantial improvements before they can reach human-level accuracy. Models also have lopsided performance and frequently do not know when they are wrong. Worse, they still have near-random accuracy on some socially important subjects such as morality and law. By comprehensively evaluating the breadth and depth of a model's academic and professional understanding, our test can be used to analyze models across many tasks and to identify important shortcomings.
摘要：本文提出了一种新的测试来衡量一个文本模型的多任务准确性。该测试涵盖57项任务，包括初等数学，美国历史，计算机科学，法律，等等。为了达到这个测试精度高，模型必须具备广泛的全球知识和解决问题的能力。我们发现，尽管最新的车型有近随机机会精度，规模非常大的GPT-3型平均提高近20个百分点，改进了随机的机会。然而，在的57项任务的每一个，最好的车型仍然需要实质性的改进才可以达到人类水平的精确度。型号也有不平衡的表现，经常不知道什么时候他们错了。更糟的是，他们仍然有一些重要社会意义的科目，如道德和法律附近随机准确性。通过综合评价模型的学术和专业认识的广度和深度，我们的测试可以用来分析横跨许多任务模型，并确定重要的缺点。
24. KoSpeech: Open-Source Toolkit for End-to-End Korean Speech Recognition [PDF] 返回目录
Soohwan Kim, Seyoung Bae, Cheolhwang Won
Abstract: We present KoSpeech, an open-source software, which is modular and extensible end-to-end Korean automatic speech recognition (ASR) toolkit based on the deep learning library PyTorch. Several automatic speech recognition open-source toolkits have been released, but all of them deal with non-Korean languages, such as English (e.g. ESPnet, Espresso). Although AI Hub opened 1,000 hours of Korean speech corpus known as KsponSpeech, there is no established preprocessing method and baseline model to compare model performances. Therefore, we propose preprocessing methods for KsponSpeech corpus and a baseline model for benchmarks. Our baseline model is based on Listen, Attend and Spell (LAS) architecture and ables to customize various training hyperparameters conveniently. By KoSpeech, we hope this could be a guideline for those who research Korean speech recognition. Our baseline model achieved 10.31% character error rate (CER) at KsponSpeech corpus only with the acoustic model. Our source code is available here.
摘要：我们提出KoSpeech，一个开放源代码的软件，它是基于深学习库PyTorch模块化和可扩展的端至端自动韩文语音识别（ASR）工具包。一些自动语音识别的开源工具包已经被释放，但所有的人都应对非韩国语言，如英语（例如ESPnet，咖啡）。虽然AI中心开业千小时称为KsponSpeech韩国语料库的，没有确定的预处理方法和基准模型进行对比模型表演。因此，我们提出了预处理语料KsponSpeech和基准的基准模型的方法。我们的基本模型是基于听，出席法术（LAS）架构和ABLES方便地定制各种培训的超参数。通过KoSpeech，我们希望这可以为那些谁研究的韩国语音识别的准则。我们的基准预测模型在KsponSpeech语料达到10.31％字符错误率（CER）只与声学模型。我们的源代码可以在这里找到。
Soohwan Kim, Seyoung Bae, Cheolhwang Won
Abstract: We present KoSpeech, an open-source software, which is modular and extensible end-to-end Korean automatic speech recognition (ASR) toolkit based on the deep learning library PyTorch. Several automatic speech recognition open-source toolkits have been released, but all of them deal with non-Korean languages, such as English (e.g. ESPnet, Espresso). Although AI Hub opened 1,000 hours of Korean speech corpus known as KsponSpeech, there is no established preprocessing method and baseline model to compare model performances. Therefore, we propose preprocessing methods for KsponSpeech corpus and a baseline model for benchmarks. Our baseline model is based on Listen, Attend and Spell (LAS) architecture and ables to customize various training hyperparameters conveniently. By KoSpeech, we hope this could be a guideline for those who research Korean speech recognition. Our baseline model achieved 10.31% character error rate (CER) at KsponSpeech corpus only with the acoustic model. Our source code is available here.
摘要：我们提出KoSpeech，一个开放源代码的软件，它是基于深学习库PyTorch模块化和可扩展的端至端自动韩文语音识别（ASR）工具包。一些自动语音识别的开源工具包已经被释放，但所有的人都应对非韩国语言，如英语（例如ESPnet，咖啡）。虽然AI中心开业千小时称为KsponSpeech韩国语料库的，没有确定的预处理方法和基准模型进行对比模型表演。因此，我们提出了预处理语料KsponSpeech和基准的基准模型的方法。我们的基本模型是基于听，出席法术（LAS）架构和ABLES方便地定制各种培训的超参数。通过KoSpeech，我们希望这可以为那些谁研究的韩国语音识别的准则。我们的基准预测模型在KsponSpeech语料达到10.31％字符错误率（CER）只与声学模型。我们的源代码可以在这里找到。
25. Adversarial Watermarking Transformer: Towards Tracing Text Provenance with Data Hiding [PDF] 返回目录
Sahar Abdelnabi, Mario Fritz
Abstract: Recent advances in natural language generation have introduced powerful language models with high-quality output text. However, this raises concerns about the potential misuse of such models for malicious purposes. In this paper, we study natural language watermarking as a defense to help better mark and trace the provenance of text. We introduce the Adversarial Watermarking Transformer (AWT) with a jointly trained encoder-decoder and adversarial training that, given an input text and a binary message, generates an output text that is unobtrusively encoded with the given message. We further study different training and inference strategies to achieve minimal changes to the semantics and correctness of the input text. AWT is the first end-to-end model to hide data in text by automatically learning -- without ground truth -- word substitutions along with their locations in order to encode the message. We show that our model is effective in largely preserving text utility and decoding the watermark while hiding its presence against adversaries. Additionally, we demonstrate that our method is robust against a range of local changes and denoising attacks.
摘要：在自然语言生成的最新进展，介绍了功能强大的语言模型具有高品质的输出文本。然而，这引起了人们对此类车型的用于恶意目的可能被滥用的担忧。在本文中，我们研究自然语言水印作为一个防御，以帮助更好地标记和跟踪文字的出处。我们用共同训练编码器 - 解码器和对抗训练，鉴于输入文本和二进制消息，生成被不显眼地与给定的消息中编码的输出文本介绍的对抗性水印变压器（AWT）。我们进一步研究不同的训练和推理战略，以实现最小的更改输入文本的语义和正确性。 AWT是第一个终端到高端型号以文本隐藏数据通过自动学习 - 无地面真相 - 他们为了对邮件进行编码位置沿字替代。我们表明，我们的模型是有效很大程度上保持了文字工具，并同时隐藏其对对手的存在解码的水印。此外，我们证明了我们的方法是针对一系列的局部变化和去噪攻击强劲。
Sahar Abdelnabi, Mario Fritz
Abstract: Recent advances in natural language generation have introduced powerful language models with high-quality output text. However, this raises concerns about the potential misuse of such models for malicious purposes. In this paper, we study natural language watermarking as a defense to help better mark and trace the provenance of text. We introduce the Adversarial Watermarking Transformer (AWT) with a jointly trained encoder-decoder and adversarial training that, given an input text and a binary message, generates an output text that is unobtrusively encoded with the given message. We further study different training and inference strategies to achieve minimal changes to the semantics and correctness of the input text. AWT is the first end-to-end model to hide data in text by automatically learning -- without ground truth -- word substitutions along with their locations in order to encode the message. We show that our model is effective in largely preserving text utility and decoding the watermark while hiding its presence against adversaries. Additionally, we demonstrate that our method is robust against a range of local changes and denoising attacks.
摘要：在自然语言生成的最新进展，介绍了功能强大的语言模型具有高品质的输出文本。然而，这引起了人们对此类车型的用于恶意目的可能被滥用的担忧。在本文中，我们研究自然语言水印作为一个防御，以帮助更好地标记和跟踪文字的出处。我们用共同训练编码器 - 解码器和对抗训练，鉴于输入文本和二进制消息，生成被不显眼地与给定的消息中编码的输出文本介绍的对抗性水印变压器（AWT）。我们进一步研究不同的训练和推理战略，以实现最小的更改输入文本的语义和正确性。 AWT是第一个终端到高端型号以文本隐藏数据通过自动学习 - 无地面真相 - 他们为了对邮件进行编码位置沿字替代。我们表明，我们的模型是有效很大程度上保持了文字工具，并同时隐藏其对对手的存在解码的水印。此外，我们证明了我们的方法是针对一系列的局部变化和去噪攻击强劲。
26. Ambiguity Hierarchy of Regular Infinite Tree Languages [PDF] 返回目录
Alexander Rabinovich, Doron Tiferet
Abstract: An automaton is unambiguous if for every input it has at most one accepting computation. An automaton is k-ambiguous (for k>0) if for every input it has at most k accepting computations. An automaton is boundedly ambiguous if there is k, such that for every input it has at most k accepting computations. An automaton is finitely (respectively, countably) ambiguous if for every input it has at most finitely (respectively, countably) many accepting computations. The degree of ambiguity of a regular language is defined in a natural way. A language is k-ambiguous (respectively, boundedly, finitely, countably ambiguous) if it is accepted by a k-ambiguous (respectively, boundedly, finitely, countably ambiguous) automaton. Over finite words, every regular language is accepted by a deterministic automaton. Over finite trees, every regular language is accepted by an unambiguous automaton. Over $\omega$-words every regular language is accepted by an unambiguous Büchi automaton and by a deterministic parity automaton. Over infinite trees, Carayol et al. showed that there are ambiguous languages. We show that over infinite trees there is a hierarchy of degrees of ambiguity: For every k>1 there are k-ambiguous languages which are not k-1 ambiguous; and there are finitely (respectively countably, uncountably) ambiguous languages which are not boundedly (respectively finitely, countably) ambiguous.
摘要：自动机是明确的，如果每个输入它最多有一个接受的计算。的自动机为k歧义（对于k> 0），如果为每一个输入具有至多为k接受计算。的自动机是boundedly暧昧如果有K，使得对于每一个输入具有至多为k接受计算。的自动机是有限（分别为可数）暧昧如果为每一个输入具有至多有限（分别为可数）许多接受计算。正规语言的模糊性的程度以自然的方式来定义。语言是K-暧昧（分别为boundedly，有限，可数不明确的），如果它是由K-暧昧（分别为boundedly，有限，可数不明确的）自动机接受。在有限的话，每一个正规语言是由确定性自动接受。在有限的树，每一个正规语言是一个明确的自动机接受。超过$ \ $欧米茄每-words正规语言是一个明确的Büchi自动和确定性奇偶自动机接受。在无限的树木，卡拉约尔等。数据显示，目前模棱两可的语言。我们发现，在无限的树木有度歧义的层次：对于每一个K> 1有K-暧昧的语言中没有K-1不明确;和有有限（分别可数，不可数）模棱两可的语言不属于boundedly（分别为有限，可数）不明确的。
Alexander Rabinovich, Doron Tiferet
Abstract: An automaton is unambiguous if for every input it has at most one accepting computation. An automaton is k-ambiguous (for k>0) if for every input it has at most k accepting computations. An automaton is boundedly ambiguous if there is k, such that for every input it has at most k accepting computations. An automaton is finitely (respectively, countably) ambiguous if for every input it has at most finitely (respectively, countably) many accepting computations. The degree of ambiguity of a regular language is defined in a natural way. A language is k-ambiguous (respectively, boundedly, finitely, countably ambiguous) if it is accepted by a k-ambiguous (respectively, boundedly, finitely, countably ambiguous) automaton. Over finite words, every regular language is accepted by a deterministic automaton. Over finite trees, every regular language is accepted by an unambiguous automaton. Over $\omega$-words every regular language is accepted by an unambiguous Büchi automaton and by a deterministic parity automaton. Over infinite trees, Carayol et al. showed that there are ambiguous languages. We show that over infinite trees there is a hierarchy of degrees of ambiguity: For every k>1 there are k-ambiguous languages which are not k-1 ambiguous; and there are finitely (respectively countably, uncountably) ambiguous languages which are not boundedly (respectively finitely, countably) ambiguous.
摘要：自动机是明确的，如果每个输入它最多有一个接受的计算。的自动机为k歧义（对于k> 0），如果为每一个输入具有至多为k接受计算。的自动机是boundedly暧昧如果有K，使得对于每一个输入具有至多为k接受计算。的自动机是有限（分别为可数）暧昧如果为每一个输入具有至多有限（分别为可数）许多接受计算。正规语言的模糊性的程度以自然的方式来定义。语言是K-暧昧（分别为boundedly，有限，可数不明确的），如果它是由K-暧昧（分别为boundedly，有限，可数不明确的）自动机接受。在有限的话，每一个正规语言是由确定性自动接受。在有限的树，每一个正规语言是一个明确的自动机接受。超过$ \ $欧米茄每-words正规语言是一个明确的Büchi自动和确定性奇偶自动机接受。在无限的树木，卡拉约尔等。数据显示，目前模棱两可的语言。我们发现，在无限的树木有度歧义的层次：对于每一个K> 1有K-暧昧的语言中没有K-1不明确;和有有限（分别可数，不可数）模棱两可的语言不属于boundedly（分别为有限，可数）不明确的。
27. Any-to-Many Voice Conversion with Location-Relative Sequence-to-Sequence Modeling [PDF] 返回目录
Songxiang Liu, Yuewen Cao, Disong Wang, Xixin Wu, Xunying Liu, Helen Meng
Abstract: This paper proposes an any-to-many location-relative, sequence-to-sequence (seq2seq) based, non-parallel voice conversion approach. In this approach, we combine a bottle-neck feature extractor (BNE) with a seq2seq based synthesis module. During the training stage, an encoder-decoder based hybrid connectionist-temporal-classification-attention (CTC-attention) phoneme recognizer is trained, whose encoder has a bottle-neck layer. A BNE is obtained from the phoneme recognizer and is utilized to extract speaker-independent, dense and rich linguistic representations from spectral features. Then a multi-speaker location-relative attention based seq2seq synthesis model is trained to reconstruct spectral features from the bottle-neck features, conditioning on speaker representations for speaker identity control in the generated speech. To mitigate the difficulties of using seq2seq based models to align long sequences, we down-sample the input spectral feature along the temporal dimension and equip the synthesis model with a discretized mixture of logistic (MoL) attention mechanism. Since the phoneme recognizer is trained with large speech recognition data corpus, the proposed approach can conduct any-to-many voice conversion. Objective and subjective evaluations shows that the proposed any-to-many approach has superior voice conversion performance in terms of both naturalness and speaker similarity. Ablation studies are conducted to confirm the effectiveness of feature selection and model design strategies in the proposed approach. The proposed VC approach can readily be extended to support any-to-any VC (also known as one/few-shot VC), and achieve high performance according to objective and subjective evaluations.
摘要：本文提出了一种任何一对多相对位置，序列到序列（seq2seq）基于非平行语音转换方法。在这种方法中，我们结合一个瓶颈特征提取（BNE）与基于seq2seq合成模块。在训练阶段，编码器 - 解码器基于混合联结 - 时间分类注意力（CTC-注意）音素识别被训练，其编码器具有瓶颈层。甲BNE从音素识别获得和被利用来提取说话者无关的，致密的和从光谱特征丰富的语言表示。然后多扬声器的相对位置，基于关注合成seq2seq模型被训练以重构从瓶颈特征光谱特征，在扬声器的表示，在所生成的语音发言者身份控制调节。为了减轻使用基于模型seq2seq对齐长序列的困难，我们下采样沿着时间维度的输入光谱特征和装备合成模型的物流（MOL）注意机制的离散混合物。由于音位识别与大型语音识别数据语料库的训练，该方法可以进行任何一对多语音转换。客观和主观评价表明，提出的任何一对多的方式有两种自然和扬声器相似性出色的语音转换性能。消融研究以确认该方法的特征选择和模型设计策略的有效性。所提出的方法的VC可以容易地扩展，以支持任何到任何VC（也称为一个/几个触发VC），并且实现根据客观和主观评价高性能。
Songxiang Liu, Yuewen Cao, Disong Wang, Xixin Wu, Xunying Liu, Helen Meng
Abstract: This paper proposes an any-to-many location-relative, sequence-to-sequence (seq2seq) based, non-parallel voice conversion approach. In this approach, we combine a bottle-neck feature extractor (BNE) with a seq2seq based synthesis module. During the training stage, an encoder-decoder based hybrid connectionist-temporal-classification-attention (CTC-attention) phoneme recognizer is trained, whose encoder has a bottle-neck layer. A BNE is obtained from the phoneme recognizer and is utilized to extract speaker-independent, dense and rich linguistic representations from spectral features. Then a multi-speaker location-relative attention based seq2seq synthesis model is trained to reconstruct spectral features from the bottle-neck features, conditioning on speaker representations for speaker identity control in the generated speech. To mitigate the difficulties of using seq2seq based models to align long sequences, we down-sample the input spectral feature along the temporal dimension and equip the synthesis model with a discretized mixture of logistic (MoL) attention mechanism. Since the phoneme recognizer is trained with large speech recognition data corpus, the proposed approach can conduct any-to-many voice conversion. Objective and subjective evaluations shows that the proposed any-to-many approach has superior voice conversion performance in terms of both naturalness and speaker similarity. Ablation studies are conducted to confirm the effectiveness of feature selection and model design strategies in the proposed approach. The proposed VC approach can readily be extended to support any-to-any VC (also known as one/few-shot VC), and achieve high performance according to objective and subjective evaluations.
摘要：本文提出了一种任何一对多相对位置，序列到序列（seq2seq）基于非平行语音转换方法。在这种方法中，我们结合一个瓶颈特征提取（BNE）与基于seq2seq合成模块。在训练阶段，编码器 - 解码器基于混合联结 - 时间分类注意力（CTC-注意）音素识别被训练，其编码器具有瓶颈层。甲BNE从音素识别获得和被利用来提取说话者无关的，致密的和从光谱特征丰富的语言表示。然后多扬声器的相对位置，基于关注合成seq2seq模型被训练以重构从瓶颈特征光谱特征，在扬声器的表示，在所生成的语音发言者身份控制调节。为了减轻使用基于模型seq2seq对齐长序列的困难，我们下采样沿着时间维度的输入光谱特征和装备合成模型的物流（MOL）注意机制的离散混合物。由于音位识别与大型语音识别数据语料库的训练，该方法可以进行任何一对多语音转换。客观和主观评价表明，提出的任何一对多的方式有两种自然和扬声器相似性出色的语音转换性能。消融研究以确认该方法的特征选择和模型设计策略的有效性。所提出的方法的VC可以容易地扩展，以支持任何到任何VC（也称为一个/几个触发VC），并且实现根据客观和主观评价高性能。
28. Visually Analyzing Contextualized Embeddings [PDF] 返回目录
Matthew Berger
Abstract: In this paper we introduce a method for visually analyzing contextualized embeddings produced by deep neural network-based language models. Our approach is inspired by linguistic probes for natural language processing, where tasks are designed to probe language models for linguistic structure, such as parts-of-speech and named entities. These approaches are largely confirmatory, however, only enabling a user to test for information known a priori. In this work, we eschew supervised probing tasks, and advocate for unsupervised probes, coupled with visual exploration techniques, to assess what is learned by language models. Specifically, we cluster contextualized embeddings produced from a large text corpus, and introduce a visualization design based on this clustering and textual structure - cluster co-occurrences, cluster spans, and cluster-word membership - to help elicit the functionality of, and relationship between, individual clusters. User feedback highlights the benefits of our design in discovering different types of linguistic structures.
摘要：本文介绍了视觉分析由深基于神经网络的语言模型制作语境的嵌入的方法。我们的做法是通过对自然语言处理，其中任务的目的是探测语言结构的语言模型，如零件的词性和命名实体语言探头启发。这些方法在很大程度上是验证，然而，只有使用户能够测试用于已知的先验信息。在这项工作中，我们避开监管探测任务，并倡导无监督探头，加上视觉的勘探技术，评估什么是语言模型的经验教训。具体来说，我们聚集从大量文本语料库产生情境嵌入物，并介绍了基于此的聚类和文本结构的可视化设计 - 之间的帮助引起的功能性和关系 - 集群共现，集群跨度和集群字成员，个别群。用户的反馈意见强调了在发现不同类型的语言结构的我们的设计带来的好处。
Matthew Berger
Abstract: In this paper we introduce a method for visually analyzing contextualized embeddings produced by deep neural network-based language models. Our approach is inspired by linguistic probes for natural language processing, where tasks are designed to probe language models for linguistic structure, such as parts-of-speech and named entities. These approaches are largely confirmatory, however, only enabling a user to test for information known a priori. In this work, we eschew supervised probing tasks, and advocate for unsupervised probes, coupled with visual exploration techniques, to assess what is learned by language models. Specifically, we cluster contextualized embeddings produced from a large text corpus, and introduce a visualization design based on this clustering and textual structure - cluster co-occurrences, cluster spans, and cluster-word membership - to help elicit the functionality of, and relationship between, individual clusters. User feedback highlights the benefits of our design in discovering different types of linguistic structures.
摘要：本文介绍了视觉分析由深基于神经网络的语言模型制作语境的嵌入的方法。我们的做法是通过对自然语言处理，其中任务的目的是探测语言结构的语言模型，如零件的词性和命名实体语言探头启发。这些方法在很大程度上是验证，然而，只有使用户能够测试用于已知的先验信息。在这项工作中，我们避开监管探测任务，并倡导无监督探头，加上视觉的勘探技术，评估什么是语言模型的经验教训。具体来说，我们聚集从大量文本语料库产生情境嵌入物，并介绍了基于此的聚类和文本结构的可视化设计 - 之间的帮助引起的功能性和关系 - 集群共现，集群跨度和集群字成员，个别群。用户的反馈意见强调了在发现不同类型的语言结构的我们的设计带来的好处。
29. Video Moment Retrieval via Natural Language Queries [PDF] 返回目录
Xinli Yu, Mohsen Malmir, Cynthia He, Yue Liu, Rex Wu
Abstract: In this paper, we propose a novel method for video moment retrieval (VMR) that achieves state of the arts (SOTA) performance on R@1 metrics and surpassing the SOTA on the high IoU metric (R@1, IoU=0.7). First, we propose to use a multi-head self-attention mechanism, and further a cross-attention scheme to capture video/query interaction and long-range query dependencies from video context. The attention-based methods can develop frame-to-query interaction and query-to-frame interaction at arbitrary positions and the multi-head setting ensures the sufficient understanding of complicated dependencies. Our model has a simple architecture, which enables faster training and inference while maintaining . Second, We also propose to use multiple task training objective consists of moment segmentation task, start/end distribution prediction and start/end location regression task. We have verified that start/end prediction are noisy due to annotator disagreement and joint training with moment segmentation task can provide richer information since frames inside the target clip are also utilized as positive training examples. Third, we propose to use an early fusion approach, which achieves better performance at the cost of inference time. However, the inference time will not be a problem for our model since our model has a simple architecture which enables efficient training and inference.
摘要：在本文中，我们提出一种用于视频的时刻检索（VMR），其实现了艺术上R（SOTA）性能@ 1个指标和超过在高IOU度量（R的SOTA @ 1的状态的新方法，IOU = 0.7 ）。首先，我们建议使用一个多头的自我关注机制，进一步交叉注意方案来捕捉视频/互动查询和远程查询依赖从视频背景。注意基于方法可以在任意位置开发框架 - 查询互动，查询到帧的互动和多头设置确保复杂的依赖关系的充分理解。我们的模型有一个简单的架构，实现了更高的训练和推理，同时保持。其次，我们还建议使用多个任务的培训目标由时刻分割的任务，开始/结束分布预测和开始/结束位置回归任务。我们已验证的开始/结束的预测是嘈杂由于标注的分歧，并与时刻分割任务联合训练可以提供更丰富的信息，因为目标剪辑中的帧也用作正例。第三，我们建议使用早期的融合方法，这在推理时间成本获得更好的性能。然而，推理时间将不会对我们的模型一个问题，因为我们的模式有一个简单的架构，它能够有效的训练和推理。
Xinli Yu, Mohsen Malmir, Cynthia He, Yue Liu, Rex Wu
Abstract: In this paper, we propose a novel method for video moment retrieval (VMR) that achieves state of the arts (SOTA) performance on R@1 metrics and surpassing the SOTA on the high IoU metric (R@1, IoU=0.7). First, we propose to use a multi-head self-attention mechanism, and further a cross-attention scheme to capture video/query interaction and long-range query dependencies from video context. The attention-based methods can develop frame-to-query interaction and query-to-frame interaction at arbitrary positions and the multi-head setting ensures the sufficient understanding of complicated dependencies. Our model has a simple architecture, which enables faster training and inference while maintaining . Second, We also propose to use multiple task training objective consists of moment segmentation task, start/end distribution prediction and start/end location regression task. We have verified that start/end prediction are noisy due to annotator disagreement and joint training with moment segmentation task can provide richer information since frames inside the target clip are also utilized as positive training examples. Third, we propose to use an early fusion approach, which achieves better performance at the cost of inference time. However, the inference time will not be a problem for our model since our model has a simple architecture which enables efficient training and inference.
摘要：在本文中，我们提出一种用于视频的时刻检索（VMR），其实现了艺术上R（SOTA）性能@ 1个指标和超过在高IOU度量（R的SOTA @ 1的状态的新方法，IOU = 0.7 ）。首先，我们建议使用一个多头的自我关注机制，进一步交叉注意方案来捕捉视频/互动查询和远程查询依赖从视频背景。注意基于方法可以在任意位置开发框架 - 查询互动，查询到帧的互动和多头设置确保复杂的依赖关系的充分理解。我们的模型有一个简单的架构，实现了更高的训练和推理，同时保持。其次，我们还建议使用多个任务的培训目标由时刻分割的任务，开始/结束分布预测和开始/结束位置回归任务。我们已验证的开始/结束的预测是嘈杂由于标注的分歧，并与时刻分割任务联合训练可以提供更丰富的信息，因为目标剪辑中的帧也用作正例。第三，我们建议使用早期的融合方法，这在推理时间成本获得更好的性能。然而，推理时间将不会对我们的模型一个问题，因为我们的模式有一个简单的架构，它能够有效的训练和推理。
30. FANG: Leveraging Social Context for Fake News Detection Using Graph Representation [PDF] 返回目录
Van-Hoang Nguyen, Kazunari Sugiyama, Preslav Nakov, Min-Yen Kan
Abstract: We propose Factual News Graph (FANG), a novel graphical social context representation and learning framework for fake news detection. Unlike previous contextual models that have targeted performance, our focus is on representation learning. Compared to transductive models, FANG is scalable in training as it does not have to maintain all nodes, and it is efficient at inference time, without the need to re-process the entire graph. Our experimental results show that FANG is better at capturing the social context into a high fidelity representation, compared to recent graphical and non-graphical models. In particular, FANG yields significant improvements for the task of fake news detection, and it is robust in the case of limited training data. We further demonstrate that the representations learned by FANG generalize to related tasks, such as predicting the factuality of reporting of a news medium.
摘要：本文提出时事新闻图（方），一种新型图形的社会情境表征和假新闻的检测学习框架。不像有针对性的性能比上情境模型中，我们的重点是学习的表示。相比于直推式模型，方是可扩展的训练，因为它不具备维护所有节点，它是在推理时间效率，而无需重新处理整个图形。我们的实验结果表明，方是捕捉社会背景成高保真表现更好，相比于近期的图形和非图形模型。具体而言，方为产生假新闻的检测任务显著的改善，并且它是在有限的训练数据的情况下强劲。我们进一步表明，交涉学会了方推广到相关的任务，如预测新闻媒体报告的真实性。
Van-Hoang Nguyen, Kazunari Sugiyama, Preslav Nakov, Min-Yen Kan
Abstract: We propose Factual News Graph (FANG), a novel graphical social context representation and learning framework for fake news detection. Unlike previous contextual models that have targeted performance, our focus is on representation learning. Compared to transductive models, FANG is scalable in training as it does not have to maintain all nodes, and it is efficient at inference time, without the need to re-process the entire graph. Our experimental results show that FANG is better at capturing the social context into a high fidelity representation, compared to recent graphical and non-graphical models. In particular, FANG yields significant improvements for the task of fake news detection, and it is robust in the case of limited training data. We further demonstrate that the representations learned by FANG generalize to related tasks, such as predicting the factuality of reporting of a news medium.
摘要：本文提出时事新闻图（方），一种新型图形的社会情境表征和假新闻的检测学习框架。不像有针对性的性能比上情境模型中，我们的重点是学习的表示。相比于直推式模型，方是可扩展的训练，因为它不具备维护所有节点，它是在推理时间效率，而无需重新处理整个图形。我们的实验结果表明，方是捕捉社会背景成高保真表现更好，相比于近期的图形和非图形模型。具体而言，方为产生假新闻的检测任务显著的改善，并且它是在有限的训练数据的情况下强劲。我们进一步表明，交涉学会了方推广到相关的任务，如预测新闻媒体报告的真实性。
注：中文为机器翻译结果！封面为论文标题词云图！