A Comprehensive Review of TriBERT-X: A Concatenated Transformer Approach for Explainable Cyberbullying Detection on Twitter

DOI

10.2991/978-94-6463-866-0_9How to use a DOI?

Keywords

Transformer-Based Models; Explainable AI (XAI); TriBERT-X; Ensemble Learning; Cyberbullying Detection

Abstract

Cyberbullying on social media comes with serious challenges that require advanced and interpretable detection mechanisms. We present TriBERT-X, a potent explainable model of transformers, uniting the strengths of ALBERT, RoBERTa, and DistilBERT through a unique concatenation-ensembling strategy. The TriBERT-X model is not like the conventional methods that are based on naturally handcrafted features, leveraging contextual embeddings from a mixture of transformers to have smoother jumps in performance. By averaging 92.3% accuracy and 91.8% F1-score on the cyberbullied detection task, it outperforms its constituents. TriBERT-X incorporates XAI mechanisms such as LIME and SHAP with high fidelity scores (LIME: 90.5% and SHAP: 93.2%) that give lucid justification for the prediction to allow transparency and trust. Fairness-aware training has also been applied to combat algorithmic bias and keep performance unbiased across demographic groups. In addition to the multilingual detection capability, which also includes the low-resource languages, TriBERT-X does the job in flexible online setting. The second scaling factor favors real-time content moderation through quantization and sparse attention mechanisms. TriBERT-X is built on privacy-preserving federated learning to ensure data privacy while continuously improving performance. As a cyberbullying detection system that is accurate, fair, and interpretable, TriBERT-X sets the stage for making digital spaces safer and more inclusive.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Vidhushavarshini Suresh Kumar
AU  - S. Ranjini
AU  - B. B. Sadhana
AU  - Tanvi Kalaskar
PY  - 2025
DA  - 2025/10/31
TI  - A Comprehensive Review of TriBERT-X: A Concatenated Transformer Approach for Explainable Cyberbullying Detection on Twitter
BT  - Proceedings of the International Conference on Intelligent Systems and Digital Transformation (ICISD 2025)
PB  - Atlantis Press
SP  - 77
EP  - 89
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6463-866-0_9
DO  - 10.2991/978-94-6463-866-0_9
ID  - Kumar2025
ER  -