Although much work in NLP has focused on measuring and mitigating stereotypical bias in semantic spaces, research addressing bias in computational argumentation is still in its infancy. In this paper, we address this research gap and conduct a thorough investigation of bias in argumentative language models. To this end, we introduce ABBA, a novel resource for bias measurement specifically tailored to argumentation. We employ our resource to assess the effect of argumentative fine-tuning and debiasing on the intrinsic bias found in transformer-based language models using a lightweight adapter-based approach that is more sustainable and parameter-efficient than full fine-tuning. Finally, we analyze the potential impact of language model debiasing on the performance in argument quality prediction, a downstream task of computational argumentation. Our results show that we are able to successfully and sustainably remove bias in general and argumentative language models while preserving (and sometimes improving) model performance in downstream tasks. We make all experimental code and data available at https://github.com/umanlp/FairArgumentativeLM.
Vyatka State University(维亚特卡国立大学);ITMO University(圣彼得堡国立信息技术机械与光学大学)
GPT-3、T5、Switch Transformer、ERNIE等大型、大语言模型的开发,显著提高了文本生成的性能。这一领域的一个重要研究方向是带论元的语篇生成。这个问题的解决方案可以用于商务会议、政治辩论、对话系统、学生论文的准备。这些应用的主要领域之一是经济领域。俄语论证文本生成的关键问题是缺乏标注的论证语料库。在本文中,我们使用的翻译版本的议论文,说服性的文章和UKP句子语料库微调RuBERT模型。最后,利用该模型对经济新闻语料进行了论证性标注。然后利用标注的语料库对ruGPT-3模型进行微调,生成论元文本。结果表明,与原始ruGPT-3模型相比,该方法将参数生成的准确性提高了20个百分点(63.2% vs. 42.5%)。
Evaluating LLM-Driven Summarisation of Parliamentary Debates with Computational Argumentation
School of Computer Science, University College Dublin(都柏林大学计算机科学学院);School of Politics and International Relations, University College Dublin(都柏林大学政治与国际关系学院);Department of Informatics, King’s College London(伦敦国王学院信息学院)
Toward Reasonable Parrots: Why Large Language Models Should Argue with Us by Design
University of Liverpool(利兹大学);University of Edinburgh(爱丁堡大学);University of Groningen(格罗宁根大学);Centrum Wiskunde & Informatica(数学与信息学研究中心);Airbus Central R&T(空客中央研发与技术部);University of Kassel(卡塞尔大学);University of Passau(帕绍大学);Universidad de Católica de la Santísima de Concepción(圣地亚哥圣母大学);University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);University of Dundee(邓迪大学);Leibniz University Hannover(汉诺威莱布尼茨大学)
Which Argumentative Aspects of Hate Speech in Social Media can be reliably identified?
Universidad de Buenos Aires(布宜诺斯艾利斯大学);Consejo Nacional de Investigaciones Científicas y Técnicas(阿根廷国家科学技术研究委员会);Universidad Nacional de Córdoba(科尔多瓦国立大学);Artificial Intelligence Research Institute (IIIA-CSIC)(人工智能研究所(IIIA-CSIC));Fundación Via Libre(自由之路基金会)
本文基于 Hateval 语料库和 Periodic Table of Arguments 标注仇恨言论的论证成分,评估各类论证特征的自动识别可靠性,并改进高错误类别。
Universidade de São Paulo Center for Artificial Intelligence (C4AI)(圣保罗大学人工智能中心(C4AI));Instituto Mauá de Tecnologia Núcleo de Sistemas Eletrônicos Embarcados (NSEE)(马乌阿技术研究所电子系统嵌入核(NSEE))
Attentive Interaction Model: Modeling Changes in View in Argumentation
Language Technologies Institute(语言技术研究所);Computer Science Department(计算机科学系);School of Computer Science(计算机科学学院);Carnegie Mellon University(卡内基梅隆大学)
我们提出了一个神经架构建模的争论性对话,明确建模意见持有者的(OH的)推理和挑战者的论点之间的相互作用,预测的目标,如果参数成功地改变了OH的看法。该模型有两个组成部分:(1)脆弱区域检测,一个注意力模型,它识别OH的推理中易于改变的部分,以及(2)交互编码,它识别OH的推理内容与挑战者的论点内容之间的关系。根据Reddit上Change My View论坛的讨论评估,这两个组件共同预测OH的视图变化,表现优于几个基线。事后分析表明,由注意力模型挑选出来的句子更容易被成功的论点所解决,而不是不成功的论点。
AMELIA: A Family of Multi-task End-to-end Language Models for Argumentation
Universite Claude Bernard Lyon 1(里昂1大学);CNRS(国家科学研究中心);Ecole Centrale de Lyon(里昂中央理工大学);INSA Lyon(里昂国立应用科学学院);Université Lumiere Lyon 2(里昂2大学);LIRIS(图像研究所)
On Theoretically-Driven LLM Agents for Multi-Dimensional Discourse Analysis
Laboratory of The New Ethos, Warsaw University of Technology, Poland(新伦理实验室,华沙理工大学,波兰);Faculty of Electronics and Information Technology, Warsaw University of Technology, Poland(电子与信息技术学院,华沙理工大学,波兰)