Detecting Minority Arguments for Mutual Understanding: A Moderation Tool for the Online Climate Change Debate

Proceedings of the 29th International Conference on Computational Linguistics, pages 6715–6725 October 12–17, 2022. 6715 Detecting Minority Arguments for Mutual Understanding: A Moderation Tool for the Online Climate Change Debate Cedric Waterschoot and Ernst van den Hemel KNAW Meertens Instituut Oudezijds Achterburgwal 185 1012DK Amsterdam, The Netherlands cedric.waterschoot@meertens.knaw.nl ernst.van.den.hemel2@meertens.knaw.nl Antal van den Bosch Utrecht Institute of Linguistics OTS Utrecht University The Netherlands a.p.j.vandenbosch@uu.nl Abstract Moderating user comments and promoting healthy understanding is a challenging task, especially in the context of polarized topics such as climate change. We propose a mod- eration tool to assist moderators in promoting mutual understanding in regard to this topic. The approach is twofold. First, we train classi- fiers to label incoming posts for the arguments they entail, with a specific focus on minority arguments. We apply active learning to further supplement the training data with rare argu- ments. Second, we dive deeper into singular arguments and extract the lexical patterns that distinguish each argument from the others. Our findings indicate that climate change arguments form clearly separable clusters in the embed- ding space. These classes are characterized by their own unique lexical patterns that provide a quick insight in an argument’s key concepts. Additionally, supplementing our training data was necessary for our classifiers to be able to adequately recognize rare arguments. We argue that this detailed rundown of each argument provides insight into where others are coming from. These computational approaches can be part of the toolkit for content moderators and researchers struggling with other polarized de- bates. 1 Introduction Even though a consensus has existed within the scientific community on the topic of human-caused climate change for some time, the online debate remains very polarized. Online comment spaces are typically overwhelmed with a large quantity of contributions. This information flood hinders the promotion of mutual understanding and inclusiv- ity in debate spaces. Additionally, climate change presents a splintered debate with niche opinions and many viewpoints. The recognition of these niche arguments are vital to support the moderator in adhering to the heterogeneous discussion that climate change presents. This setting presents op- portunities for mutual understanding by improving issue awareness and the quality of deliberation. In this paper, we construct a twofold approach to support mutual understanding in the online cli- mate change discussion. First, we aim to classify posts for the argument they present. Second, we dive deeper into singular arguments to create an overview of the lexical patterns in each argument- specific sub-corpus. We conclude the paper by discussing the limitations of modeling nuanced ar- gumentation by a computational method and link our approach to fields struggling with content mod- eration and polarized debates. 2 Background 2.1 Argument Mining & Stance Detection Our application falls under the umbrella of ’ar- gument mining’ and ’stance detection’. Within Natural Language Processing, argument mining is defined as the automated identification and extrac- tion of argumentation found in natural language (Lawrence and Reed, 2019). Following the stark increase in the availability of textual data found on online …

Content truncated to 500 words