Date of Publication : July 2026

Abstract: The study examines the effectiveness of Open Source Intelligence (OSINT) to produce action able intelligence for cyber threat detection and analysis, specifically investigating the method of collection and analysis of data from sources such as Telegram, Dark web and Twitter, for understanding of evolving cyber threats. Through OSINT research method, this study proposes new framework to identify, track and mitigate emerging cyber threats, such as a framework with integrated of BERT, Heterogeneous Graph Convolutional Network (HGCN) and Variational Autoencoders (VAE) model for enhanced cyber threat detection, BERT model to learn semantically related contexts, HGCN model to capture relationship across multiple platforms, and VAE model for detecting sophisticated anomaly in order to holistically study the heterogeneous data in OSINT. The results and analyses from this study have contributed to the effective use of OSINT based intelligence in identifying potential cyber threats and enhancing cyber threat detection and response process. The approach will use natural language processing (NLP) and machine learning to find relationships among threat indicators, and to produce timely actionable intelligence. This study has made contribution to development of secure and robust cyber threat intelligence system.

Reference :