COT: A Generative Approach for Hate Speech Counter-Narratives via Contrastive Optimal Transport

TL;DR

This paper introduces a contrastive optimal transport framework for generating targeted, diverse, and relevant counter-narratives to hate speech, addressing limitations of previous methods focused on fluency.

Contribution

The novel framework combines optimal transport, self-contrastive learning, and target-oriented decoding to improve relevance, diversity, and target interaction in hate speech counter-narrative generation.

Findings

Significantly outperforms existing methods on benchmark datasets.

Effectively incorporates hatred target information into generation.

Enhances diversity and relevance of generated counter-narratives.

Abstract

Counter-narratives, which are direct responses consisting of non-aggressive fact-based arguments, have emerged as a highly effective approach to combat the proliferation of hate speech. Previous methodologies have primarily focused on fine-tuning and post-editing techniques to ensure the fluency of generated contents, while overlooking the critical aspects of individualization and relevance concerning the specific hatred targets, such as LGBT groups, immigrants, etc. This research paper introduces a novel framework based on contrastive optimal transport, which effectively addresses the challenges of maintaining target interaction and promoting diversification in generating counter-narratives. Firstly, an Optimal Transport Kernel (OTK) module is leveraged to incorporate hatred target information in the token representations, in which the comparison pairs are extracted between original and transported features. Secondly, a self-contrastive learning module is employed to address the issue of model degeneration. This module achieves this by generating an anisotropic distribution of token representations. Finally, a target-oriented search method is integrated as an improved decoding strategy to explicitly promote domain relevance and diversification in the inference process. This strategy modifies the model's confidence score by considering both token similarity and target relevance. Quantitative and qualitative experiments have been evaluated on two benchmark datasets, which demonstrate that our proposed model significantly outperforms current methods evaluated by metrics from multiple aspects.