A Benchmark Dataset for Detecting and Explaining Implicit Hate Speech

• In Event: Advances in Machine Learning (II)

Abstract

Hate speech has grown significantly on social media, causing serious consequences for victims of all demographics. Despite much attention being paid to characterize and detect discriminatory speech, most work has focused on explicit or overt hate speech, failing to address the pervasive implicit hate speech that uses coded or indirect language to disparage a protected group or individual.
To fill this gap, this work introduces a theoretical taxonomy of implicit hate speech and a benchmark corpus with fine-grained labels for each message and its implication.
We present systematic analyses of our dataset and results achieved by state-of-the-art machine learning baselines in detecting and explaining implicit hate speech, as well as discuss certain challenges that existing approaches struggle with, suggesting that this dataset can serve as a useful research benchmark for understanding implicit hate speech online.

Author

• David Muchlinski, Georgia Institute of Technology