Abstract :

The Rapid Growth Of Social Networking Platforms Such As Twitter Has Significantly Increased The Exposure Of Users And Servers To Complex Cyber-attacks, Including Botnet-driven Campaigns, Phishing, Distributed Denial-of-service (DDoS) Attacks, Vulnerability Exploitation, And Information Leakage. Traditional OSI-layer, Rule-based Security Mechanisms Suffer From High Time Complexity, Limited Adaptability, And An Inability To Detect Multiple Coordinated Attacks In Real Time. To Address These Limitations, This Research Proposes A Dual-Target Transformer-Driven Optimized Question Answers (OQA) Framework For Real-time Cyber Threat Detection And Prediction In Twitter Streams. The Proposed System Is Designed With Two Outputs: Primary Output Performs Accurate Cyber-attack Type Classification, While Secondary Output Determines The Relevance And Severity Of The Detected Threat For Intelligent Prioritization And Response. The Framework Utilizes A Large-scale Twitter Cyber-security Dataset And Integrates Advanced NLP Preprocessing, Including Contextual Normalization, Noise Removal, Semantic Enrichment, Along With Uniform Resource Locator (URL)-specific Feature Extraction And Destinationbased Preprocessing To Analyse Domain Reputation, Redirection Behaviour, And Landing Page Characteristics. A Transformer-based Learning Architecture Using Sentence Bidirectional Encoder Representations From Transformers (SBERT) Combined With The OQA Algorithm Is Employed For Deep Semantic Feature Learning, Supported By Synthetic Minority Over-sampling Technique (SMOTE)-based Data Balancing And Stochastic Gradient Descent (SGD) Optimization For Efficient Training Under Imbalanced Data Conditions. To Further Enhance Detection Performance, The System Is Complemented With Naïve Bayes, Deep Neural Networks (DNN), And Linear Discriminant Analysis (LDA) For Probabilistic Validation, Deep Behavioural Learning, And Topic-level Attack Interpretation. Experimental Results Demonstrate That The Proposed Dual-target Hybrid Framework Achieves High Accuracy, Low False Alarm Rates, And Robust Real-time Detection Capability For Multiple Cyber-attack Types On Twitter. The Outcomes Validate The Effectiveness Of The Proposed System As A Scalable, Intelligent, And Adaptive Cyber Defence Solution For Modern Social Network Environments.

Published:

24-4-2026

Issue:

Vol. 26 No. 4 (2026)

Page Nos:

3072-3085

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite