Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Promoting Security and Trust on Social Networks: Explainable Cyberbullying Detection Using Large Language Models in a Stream-Based Machine Learning Framework

Silvia García-Méndez, Francisco De Arriba-Pérez

TL;DR

This work addresses real-time cyberbullying detection on social networks by integrating a stream-based ML framework with Large Language Models for feature engineering and an explainability dashboard. It blends prompt-driven LLMS features with traditional NLP features, applying a two-stage feature selection and evaluating multiple streaming classifiers, with Adaptive Random Forest delivering the strongest performance. The approach achieves near 90% effectiveness on key metrics and provides natural-language explanations to enhance trust and accountability. The proposed system supports timely interventions in online communities and highlights future avenues for multi-label, sliding-window, and multimodal analyses to further improve robustness and fairness.

Abstract

Social media platforms enable instant and ubiquitous connectivity and are essential to social interaction and communication in our technological society. Apart from its advantages, these platforms have given rise to negative behaviors in the online community, the so-called cyberbullying. Despite the many works involving generative Artificial Intelligence (AI) in the literature lately, there remain opportunities to study its performance apart from zero/few-shot learning strategies. Accordingly, we propose an innovative and real-time solution for cyberbullying detection that leverages stream-based Machine Learning (ML) models able to process the incoming samples incrementally and Large Language Models (LLMS) for feature engineering to address the evolving nature of abusive and hate speech online. An explainability dashboard is provided to promote the system's trustworthiness, reliability, and accountability. Results on experimental data report promising performance close to 90 % in all evaluation metrics and surpassing those obtained by competing works in the literature. Ultimately, our proposal contributes to the safety of online communities by timely detecting abusive behavior to prevent long-lasting harassment and reduce the negative consequences in society.

Promoting Security and Trust on Social Networks: Explainable Cyberbullying Detection Using Large Language Models in a Stream-Based Machine Learning Framework

TL;DR

This work addresses real-time cyberbullying detection on social networks by integrating a stream-based ML framework with Large Language Models for feature engineering and an explainability dashboard. It blends prompt-driven LLMS features with traditional NLP features, applying a two-stage feature selection and evaluating multiple streaming classifiers, with Adaptive Random Forest delivering the strongest performance. The approach achieves near 90% effectiveness on key metrics and provides natural-language explanations to enhance trust and accountability. The proposed system supports timely interventions in online communities and highlights future avenues for multi-label, sliding-window, and multimodal analyses to further improve robustness and fairness.

Abstract

Social media platforms enable instant and ubiquitous connectivity and are essential to social interaction and communication in our technological society. Apart from its advantages, these platforms have given rise to negative behaviors in the online community, the so-called cyberbullying. Despite the many works involving generative Artificial Intelligence (AI) in the literature lately, there remain opportunities to study its performance apart from zero/few-shot learning strategies. Accordingly, we propose an innovative and real-time solution for cyberbullying detection that leverages stream-based Machine Learning (ML) models able to process the incoming samples incrementally and Large Language Models (LLMS) for feature engineering to address the evolving nature of abusive and hate speech online. An explainability dashboard is provided to promote the system's trustworthiness, reliability, and accountability. Results on experimental data report promising performance close to 90 % in all evaluation metrics and surpassing those obtained by competing works in the literature. Ultimately, our proposal contributes to the safety of online communities by timely detecting abusive behavior to prevent long-lasting harassment and reduce the negative consequences in society.
Paper Structure (17 sections, 2 figures, 4 tables)

This paper contains 17 sections, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Literature review
  3. Research contributions
  4. Methodology
  5. Preprocessing module
  6. Feature engineering module
  7. Feature analysis & selection module
  8. Classification module
  9. Explainability module
  10. Results and discussion
  11. Experimental data
  12. Preprocessing module
  13. Feature engineering module
  14. Feature analysis & selection module
  15. Classification module
  16. ...and 2 more sections

Figures (2)

  • Figure 1: System scheme.
  • Figure 2: Dashboard for explainability.