Cross-lingual Zero- and Few-shot Hate Speech Detection Utilising Frozen Transformer Language Models and AXEL

TL;DR

This paper presents a novel approach using frozen Transformer models and the AXEL classification block for cross-lingual zero- and few-shot hate speech detection, achieving competitive results on multilingual datasets.

Contribution

Introduces a new architecture combining frozen Transformers with AXEL for improved cross-lingual hate speech detection in low-resource languages.

Findings

Achieved competitive results on English and Spanish hate speech datasets.

Demonstrated effectiveness of the AXEL classification block.

Enabled meaningful future comparisons through data re-sampling.

Abstract

Detecting hate speech, especially in low-resource languages, is a non-trivial challenge. To tackle this, we developed a tailored architecture based on frozen, pre-trained Transformers to examine cross-lingual zero-shot and few-shot learning, in addition to uni-lingual learning, on the HatEval challenge data set. With our novel attention-based classification block AXEL, we demonstrate highly competitive results on the English and Spanish subsets. We also re-sample the English subset, enabling additional, meaningful comparisons in the future.