Social Media Manipulation Deep Learning based Disinformation Detection

Abstract

The rapid grow and use of different social platforms enhanced communication between different entities and their audiences plus the transformation through digitalization of existing, e.g., ideas and businesses, or the creation of new ones fully existing or depending on this digital environment. Nevertheless, next to these promising aspects, social media is a vulnerable digital environment where a diverse plethora of cyber incidents are planned and executed engaging a diverse range of targets. Among these, social media manipulation through threats like disinformation and misinformation produce a broad span of effects that cross digital borders into the human realm by influencing and altering human believes, behaviour, and attitudes towards specific ideas, institutions, or people. To tackle these issues, existing academic, social platforms, dedicated organizations, and institutions efforts exist for building specific advanced and intelligent solutions for detecting and preventing them. Regardless, these efforts embed defender’s perspective and are focused locally, at target level, without being designed to fit a broader agenda of producing and/or strengthening social media security awareness. On this behalf, this research proposes a deep learning-based disinformation detection solution for facilitating and/or enhancing social media security awareness in respect to offender’s perspective. To achieve this objective, a Data Science approach is taken based on the Design Science Research methodology, and the results obtained are discussed with a keen on further field developments regarding intelligent, transparent, and responsible solutions countering social manipulation through realistic participation and contribution of different stakeholders from different disciplines.