# Addressing Antisocial Behavior in Multi-Party Dialogs Through Multimodal Representation Learning

arXiv: https://arxiv.org/abs/2510.17289 · HTML: https://mlsanity.com/paper/2510.17289/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 43 references that are in the map.
- First posted on arXiv 20 Oct 2025.

## Builds on (most cited first, 13 of 13)

- [CamemBERT: a Tasty French Language Model](https://mlsanity.com/paper/1911.03894/) (2019)
- [FlauBERT: Unsupervised Language Model Pre-training for French](https://mlsanity.com/paper/1912.05372/) (2019)
- [Toward a Perspectivist Turn in Ground Truthing for Predictive Computing](https://mlsanity.com/paper/2109.04270/) (2021)
- [Signed Graph Attention Networks](https://mlsanity.com/paper/1906.10958/) (2019)
- [Anyone Can Become a Troll: Causes of Trolling Behavior in Online Discussions](https://mlsanity.com/paper/1702.01119/) (2017)
- [Echo Chambers: Emotional Contagion and Group Polarization on Facebook](https://mlsanity.com/paper/1607.01032/) (2016)
- [The Structure of Toxic Conversations on Twitter](https://mlsanity.com/paper/2105.11596/) (2021)
- [CamemBERT 2.0: A Smarter French Language Model Aged to Perfection](https://mlsanity.com/paper/2411.08868/) (2024)
- [Graph embeddings for Abusive Language Detection](https://mlsanity.com/paper/2101.02988/) (2021)
- [WAC: A Corpus of Wikipedia Conversations for Online Abuse Detection](https://mlsanity.com/paper/2003.06190/) (2020)
- [Characterizing Engagement Dynamics across Topics on Facebook](https://mlsanity.com/paper/2211.15988/) (2022)
- [Whole-Graph Representation Learning For the Classification of Signed Networks](https://mlsanity.com/paper/2409.20073/) (2024)
- [Conversation-Based Multimodal Abuse Detection Through Text and Graph Embeddings](https://mlsanity.com/paper/2503.12994/) (2025)

## Abstract (arXiv)

Antisocial behavior (ASB) on social media -- including hate speech, harassment, and cyberbullying -- poses growing risks to platform safety and societal well-being. Prior research has focused largely on networks such as X and Reddit, while textitmulti-party conversational settings remain underexplored due to limited data. To address this gap, we use textitCyberAgressionAdo-Large, a French open-access dataset simulating ASB in multi-party conversations, and evaluate three tasks: textitabuse detection, textitbullying behavior analysis, and textitbullying peer-group identification. We benchmark six text-based and eight graph-based textitrepresentation-learning methods, analyzing lexical cues, interactional dynamics, and their multimodal fusion. Results show that multimodal models outperform unimodal baselines. The late fusion model textttmBERT + WD-SGCN achieves the best overall results, with top performance on abuse detection (0.718) and competitive scores on peer-group identification (0.286) and bullying analysis (0.606). Error analysis highlights its effectiveness in handling nuanced ASB phenomena such as implicit aggression, role transitions, and context-dependent hostility.