arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2207.06591cs.CLcs.AI
收藏

拉丁美洲自然语言处理中偏见和有害定型观念的特征化方法

A methodology to characterize bias and harmful stereotypes in natural language processing in Latin America

Universidad Nacional de Córdoba(国立科尔多瓦大学);Fundación Via Libre(自由之路基金会);Consejo Nacional de Investigaciones Científicas y Técnicas(国家科学技术研究委员会)
Laura Alonso Alemany, Luciana Benotti, Hernán Maina, Lucía González, Mariela Rajngewerc, Lautaro Martínez, Jorge Sánchez, Mauro Schilman, Guido Ivetta, Alexia H… 展开作者

Laura Alonso Alemany, Luciana Benotti, Hernán Maina, Lucía González, Mariela Rajngewerc, Lautaro Martínez, Jorge Sánchez, Mauro Schilman, Guido Ivetta, Alexia Halvorsen, Amanda Mata Rojo, Matías Bordone, Beatriz Busaniche

摘要:

自动决策系统,特别是那些基于自然语言处理的系统,在我们的生活中无处不在。它们不仅是我们日常使用的互联网搜索引擎的幕后推手,而且还扮演着更重要的角色:选择工作候选人,确定犯罪嫌疑人,诊断自闭症等等。这种自动化系统会产生错误,这在许多方面可能是有害的,无论是因为后果的严重性(如健康问题)还是因为它们影响的人数众多。当一个自动化系统所犯的错误对一个群体的影响大于其他群体时,我们称这个系统为\textit{biased}。大多数现代自然语言技术都是基于使用机器学习从大量文本中获得的工件,即语言模型和词嵌入。由于它们是通过应用子符号机器学习(主要是人工神经网络)创建的,它们是不透明的,实际上无法通过直接检查来解释,因此很难对其进行审计。在本文中,我们提出了一种方法,阐明了社会科学家,领域专家和机器学习专家如何合作探索词嵌入和大语言模型中的偏见和有害的刻板印象。我们的方法基于以下原则:* 关注歧视在词嵌入和语言模型上的语言表现,而不是模型的数学特性 * 减少歧视专家的技术障碍,无论是社会科学家,领域专家或其他 * 除了基于度量的方法外,还通过定性探索过程进行表征 * 将缓解作为培训过程的一部分,而不是事后的想法

英文摘要:

Automated decision-making systems, especially those based on natural language processing, are pervasive in our lives. They are not only behind the internet search engines we use daily, but also take more critical roles: selecting candidates for a job, determining suspects of a crime, diagnosing autism and more. Such automated systems make errors, which may be harmful in many ways, be it because of the severity of the consequences (as in health issues) or because of the sheer number of people they affect. When errors made by an automated system affect a population more than others, we call the system \textit{biased}. Most modern natural language technologies are based on artifacts obtained from enormous volumes of text using machine learning, namely language models and word embeddings. Since they are created by applying subsymbolic machine learning, mostly artificial neural networks, they are opaque and practically uninterpretable by direct inspection, thus making it very difficult to audit them. In this paper, we present a methodology that spells out how social scientists, domain experts, and machine learning experts can collaboratively explore biases and harmful stereotypes in word embeddings and large language models. Our methodology is based on the following principles: * focus on the linguistic manifestations of discrimination on word embeddings and language models, not on the mathematical properties of the models * reduce the technical barrier for discrimination experts%, be it social scientists, domain experts or other * characterize through a qualitative exploratory process in addition to a metric-based approach * address mitigation as part of the training process, not as an afterthought

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑