A Benchmark for Cross-Domain Argumentative Stance Classification on Social Media

TL;DR

This paper introduces a large, multi-domain benchmark for argumentative stance classification on social media, utilizing platform rules and language models to reduce manual annotation efforts and evaluate various learning settings.

Contribution

It presents a novel multidomain dataset with over 4,500 claims and nearly 31,000 arguments, and benchmarks multiple learning paradigms for stance classification.

Findings

Benchmark covers 21 domains, enhancing diversity.

Language models can effectively classify stance with limited supervision.

The dataset facilitates evaluation of supervised, zero-shot, and few-shot methods.

Abstract

Argumentative stance classification plays a key role in identifying authors' viewpoints on specific topics. However, generating diverse pairs of argumentative sentences across various domains is challenging. Existing benchmarks often come from a single domain or focus on a limited set of topics. Additionally, manual annotation for accurate labeling is time-consuming and labor-intensive. To address these challenges, we propose leveraging platform rules, readily available expert-curated content, and large language models to bypass the need for human annotation. Our approach produces a multidomain benchmark comprising 4,498 topical claims and 30,961 arguments from three sources, spanning 21 domains. We benchmark the dataset in fully supervised, zero-shot, and few-shot settings, shedding light on the strengths and limitations of different methodologies. We release the dataset and code in this study at hidden for anonymity.