You too Brutus! Trapping Hateful Users in Social Media: Challenges, Solutions & Insights

TL;DR

This paper explores the detection of hateful users on social media by combining textual and social network features using Graph Neural Networks, achieving high accuracy and generalization across platforms.

Contribution

It introduces a semi-supervised GNN approach that leverages both text and social connections, outperforming traditional models in hate speech user detection.

Findings

AGNN achieves 0.791 macro F1 on Gab dataset

Model generalizes well across platforms in zero-shot setting

Hateful users have distinctive network neighborhood signatures

Abstract

Hate speech is regarded as one of the crucial issues plaguing the online social media. The current literature on hate speech detection leverages primarily the textual content to find hateful posts and subsequently identify hateful users. However, this methodology disregards the social connections between users. In this paper, we run a detailed exploration of the problem space and investigate an array of models ranging from purely textual to graph based to finally semi-supervised techniques using Graph Neural Networks (GNN) that utilize both textual and graph-based features. We run exhaustive experiments on two datasets -- Gab, which is loosely moderated and Twitter, which is strictly moderated. Overall the AGNN model achieves 0.791 macro F1-score on the Gab dataset and 0.780 macro F1-score on the Twitter dataset using only 5% of the labeled instances, considerably outperforming all the other models including the fully supervised ones. We perform detailed error analysis on the best performing text and graph based models and observe that hateful users have unique network neighborhood signatures and the AGNN model benefits by paying attention to these signatures. This property, as we observe, also allows the model to generalize well across platforms in a zero-shot setting. Lastly, we utilize the best performing GNN model to analyze the evolution of hateful users and their targets over time in Gab.