Source Identification of DeepFake Videos in Social Media-like Networks using Centrality Measures

The widespread occurrence of DeepFake videos on social media platforms makes it extremely difficult to pinpoint their source, which is essential for reducing their propagation and possible damage. This research uses centrality measurements in graph-based social networks to identify the primary accounts responsible for spread of misinformation. PageRank, Closeness, Betweenness, and Degree are some of the centrality metrics used to evaluate the influence and impact of nodes in the network. The approach combines a multi-criteria centrality evaluation, content similarity analysis, and graph pruning to determine the most likely source of detected DeepFake videos. The accuracy and resilience of the suggested framework in identifying sources and assessing information dispersion are shown by experimental findings obtained in a variety of network scenarios. Our results identify the deepfake content as well as determine the source account which released this content onto social media. Identifying and eliminating these accounts from a social media network would ensure safety and help users trust the information received online. Our research is an important first step in thwarting false information and boosting confidence in digital platforms.