StereoKG: Data-Driven Knowledge Graph Construction For Cultural Knowledge and Stereotypes

StereoKG: Data-Driven Knowledge Graph Construction For Cultural Knowledge and Stereotypes
复制标题

StereoKG:数据驱动的文化知识和刻板印象知识图谱构建

DOI:
--
复制
发表时间:
2022
期刊:
WOAH
影响因子:
--
通讯作者:
D. Klakow
D. Klakow
中科院分区:
--
文献类型:
--
作者:
Awantee V. Deshpande;Dana Ruiter;Marius Mosbach;D. Klakow

文献摘要

参考文献

被引文献

相似文献

分析种族或宗教偏见对于提高自然语言处理模型的公平性、问责性和透明度非常重要。然而,许多技术依赖于人工编制的偏见术语列表,这些列表的创建成本很高,而且覆盖范围有限。在这项研究中,我们提出了一个完全数据驱动的管道,用于生成文化知识和刻板印象的知识图(KG)。我们得到的KG涵盖了5个宗教团体和5个民族,可以很容易地扩展到更多的实体。我们的人类评估显示,大多数(59.2%)的非单例条目是连贯的和完整的刻板印象。我们进一步表明,在语言化的KG上执行中间掩面语言模型训练会导致模型中更高水平的文化意识,并且有可能在相关任务(即仇恨言论检测)上提高对知识关键样本的分类性能。
Analyzing ethnic or religious bias is important for improving fairness, accountability, and transparency of natural language processing models. However, many techniques rely on human-compiled lists of bias terms, which are expensive to create and are limited in coverage. In this study, we present a fully data-driven pipeline for generating a knowledge graph (KG) of cultural knowledge and stereotypes. Our resulting KG covers 5 religious groups and 5 nationalities and can easily be extended to more entities. Our human evaluation shows that the majority (59.2%) of non-singleton entries are coherent and complete stereotypes. We further show that performing intermediate masked language model training on the verbalized KG leads to a higher level of cultural awareness in the model and has the potential to increase classification performance on knowledge-crucial samples on a related task, i.e., hate speech detection.
黑人之于罪犯就像白人之于警察:检测和消除词嵌入中的多类偏差
DOI: --
发表时间: 2019
期刊: 2019 Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL
影响因子: --
作者:
Manzini, Thomas;Lim, Yao Chong;Tsvetkov, Yulia;Black, Alan W
通讯作者: Black, Alan W
DOI: 10.1162/tacl_a_00290
发表时间: 2019-01-01
影响因子: 10.9
作者:
Warstadt, Alex;Singh, Amanpreet;Bowman, Samuel R.
通讯作者: Bowman, Samuel R.
DOI: 10.1145/2629489
发表时间: 2014-10-01
影响因子: 22.7
作者:
Vrandecic, Denny;Kroetzsch, Markus
通讯作者: Kroetzsch, Markus
DOI: 10.18653/v1/d19-1005
发表时间: 2019-09
期刊: ArXiv
影响因子: --
作者:
Matthew E. Peters;Mark Neumann;IV RobertL.Logan;Roy Schwartz;Vidur Joshi;Sameer Singh;Noah A. Smith-
通讯作者: Matthew E. Peters;Mark Neumann;IV RobertL.Logan;Roy Schwartz;Vidur Joshi;Sameer Singh;Noah A. Smith-