Word Embeddings via Causal Inference: Gender Bias Reducing and Semantic Information Preserving-猫眼课题宝

DOI：

10.1609/aaai.v36i11.21443

发表时间：

2021-12

期刊：

影响因子：

通讯作者：

中科院分区：

其他

文献类型：

作者：

Lei Ding;Dengdeng Yu;Jinhan Xie;Wenxing Guo;Shenggang Hu;Meichen Liu;Linglong Kong;Hongsheng Dai;Yanchun Bao;Bei Jiang

随着自然语言处理（NLP）在日常生活中的广泛应用，NLP模型中继承的社会偏见变得更加严重和成问题。先前的研究表明，在人类生成的语料库上训练的词嵌入具有强烈的性别偏见，可以在下游任务中产生歧视性结果。以往的去偏方法主要集中在对偏置进行建模，只隐含地考虑语义信息，而完全忽略了偏置和语义成分之间复杂的潜在因果结构。为了解决这些问题，我们提出了一种新的方法，利用因果推理框架，以有效地消除性别偏见。所提出的方法使我们能够构建和分析复杂的因果机制，促进性别信息流，同时保留甲骨文的语义信息的词嵌入。我们的综合实验表明，该方法在性别去偏见任务中取得了最先进的结果。此外，我们的方法在单词相似性评估和各种外部下游NLP任务中具有更好的性能。

With widening deployments of natural language processing (NLP) in daily life, inherited social biases from NLP models have become more severe and problematic. Previous studies have shown that word embeddings trained on human-generated corpora have strong gender biases that can produce discriminative results in downstream tasks. Previous debiasing methods focus mainly on modeling bias and only implicitly consider semantic information while completely overlooking the complex underlying causal structure among bias and semantic components. To address these issues, we propose a novel methodology that leverages a causal inference framework to effectively remove gender bias. The proposed method allows us to construct and analyze the complex causal mechanisms facilitating gender information flow while retaining oracle semantic information within word embeddings. Our comprehensive experiments show that the proposed method achieves state-of-the-art results in gender-debiasing tasks. In addition, our methods yield better performance in word similarity evaluation and various extrinsic downstream NLP tasks.

Word Embeddings via Causal Inference: Gender Bias Reducing and Semantic Information Preserving

文献摘要

被引文献

相似文献