A Comprehensive Study of Detection Methods for Deceptive Content across Social Media Platforms

Abstract

The rapid spread of deceptive content across social media platforms poses a significant threat to information authenticity and public discourse. This review critically examines 20 key studies from 2018 to 2024, focusing on the development of detection methods. Three prominent trends emerge: the rising use of deep learning techniques (40% of the studies), the creation of hybrid models combining multiple detection algorithms (25%), and a shift toward multimodal analysis, which addresses text, image, and video simultaneously. The highest accuracy recorded was 99%, achieved by systems utilizing BERT models, with the average accuracy being 93.65%. Despite these advancements, challenges remain. Notable issues include the high computational demands of advanced models, the absence of standardized evaluation metrics, and difficulties in compiling comprehensive and diverse training datasets. Future research should focus on addressing these challenges by developing cross-platform detection frameworks, improving real-time detection efficiency, and enhancing resistance to adversarial attacks. This review offers insights into the current state of detection technologies, highlighting both their strengths and limitations, and suggests directions for future exploration in tackling misinformation across multiple media formats.