Challenge: lexical biases in hate speech detection are limited when applied to real-world data, exhibiting limited out-of-distribution robustness and perpetuating harmful social biase.
Approach: They propose to disentangle spurious and authentic artifacts and analyze their impact on out-of-distribution fairness and robustness.
Outcome: The proposed models show that spurious artifacts require different treatments to attain robustness and fairness in hate speech detection.

Similar Papers

Mitigating Biases in Hate Speech Detection from A Causal Perspective (2023.findings-emnlp)

Copied to clipboard

Challenge: Existing methods to detect hate speech are prone to spurious correlations between training data and labels, which could lead to biased treatment of vulnerable and minority groups.
Approach: They propose to use grammar induction to find grammar patterns for hate speech and analyze this phenomenon from a causal perspective.
Outcome: The proposed methods can detect hate speech from a causal perspective and are robust to different datasets.
Latent Hatred: A Benchmark for Understanding Implicit Hate Speech (2021.emnlp-main)

Copied to clipboard

Challenge: Existing studies on explicit or overt hate speech have failed to address a more pervasive form based on coded or indirect language.
Approach: They propose a theoretically-justified taxonomy of implicit hate speech and a benchmark corpus with fine-grained labels for each message and its implication.
Outcome: The proposed dataset will serve as a useful benchmark for understanding this multifaceted issue.
Robust Hate Speech Detection via Mitigating Spurious Correlations (2022.aacl-short)

Copied to clipboard

Challenge: a novel hate speech detection model can be used to detect word- and character-level adversarial attacks . existing adversarials assume that attackers replace the target words with other names to evade detection .
Approach: They propose a robust hate speech detection model that can defend against adversarial attacks . they describe the process of hate speech recognition by a causal graph and a regularized entropy loss function to quantify spurious correlation .
Outcome: The proposed model can defend against word- and character-level adversarial attacks.
Delving into Qualitative Implications of Synthetic Data for Hate Speech Detection (2024.emnlp-main)

Copied to clipboard

Challenge: Recent work on synthetic data for training models for NLP tasks reports mixed results on subjective tasks such as hate speech detection.
Approach: They propose to use synthetic data to train models for highly subjective tasks such as hate speech detection to investigate the potential and specific pitfalls of using it.
Outcome: The proposed model outperforms models trained with real data on hate speech detection tasks, but it fails to accurately reflect real-world data on linguistic dimensions and results in different class distributions.
Competency Problems: On Finding and Removing Artifacts in Language Data (2021.emnlp-main)

Copied to clipboard

Challenge: Recent work in NLP has documented dataset artifacts, bias, and spurious correlations . how to tell which features have spurious instead of legitimate correlations is typically left unspecified .
Approach: They propose a class of competency problems to formalize this notion into a classification . they show that realistic datasets will increasingly deviate from competency problems .
Outcome: The proposed model can be used to show that models are inappropriately affected by these less extreme biases.
HateDay: Insights from a Global Hate Speech Dataset Representative of a Day on Twitter (2025.acl-long)

Copied to clipboard

Challenge: Prior work on automated hate speech detection models has been limited due to systematic biases in evaluation datasets and poor performance across geographies.
Approach: They propose to construct a global hate speech dataset representative of social media settings from tweets posted on September 21, 2022.
Outcome: The proposed dataset covers eight languages and four English-speaking countries and covers eight countries where English is the main language on Twitter.
Multilingual Twitter Corpus and Baselines for Evaluating Demographic Bias in Hate Speech Recognition (2020.lrec-1)

Copied to clipboard

Challenge: Existing work on document classification models mainly uses synthetic monolingual data without ground truth for author demographic attributes.
Approach: They assemble and publish a multilingual Twitter corpus for the task of hate speech detection using inferred author demographic factors.
Outcome: The results show that the classifiers learn human biases and can be discriminatory towards certain demographic groups.
Data-Efficient Strategies for Expanding Hate Speech Detection into Under-Resourced Languages (2022.emnlp-main)

Copied to clipboard

Challenge: Hate speech datasets focus on English-language content, hindering effective models . annotating hateful content is expensive, time-consuming and potentially harmful to annotators.
Approach: They propose to use ISO 639-1 codes to fine-tune models on one source language and apply them to another language.
Outcome: The proposed approach performs well on some tasks, but fails on many others.
Data-Efficient Methods For Improving Hate Speech Detection (2023.findings-eacl)

Copied to clipboard

Challenge: Existing methods for hate speech detection are data-hungry and require large datasets.
Approach: They propose an input-level data augmentation technique EasyMix to improve hate speech detection in english and multilingual datasets.
Outcome: The proposed method improves the performance across english and multilingual datasets by 1% and 2-8%.
Exploring Cross-Cultural Differences in English Hate Speech Annotations: From Dataset Construction to Analysis (2024.naacl-long)

Copied to clipboard

Challenge: Existing datasets for hate speech detection neglect the cultural diversity within a single language.
Approach: They propose a CR**oss-cultural **E**nglish **Hate* speech dataset that uses culturally hateful keywords to identify posts from four countries plus the United States.
Outcome: The proposed dataset shows that only 56.2% of the posts in CREHate achieve consensus among all countries, with the highest pairwise label difference rate of 26%.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations