Semantic networks of fake news: German-speaking, spanish-speaking and russian-speaking data

Abstract

The spread of fake news is a growing concern, raising significant ethical and security issues in virtual communication (Iskanderova, 2024). Recent approaches integrate psychology, computational linguistics, and machine learning. Researchers have developed classifiers for detecting deceptive opinion spam with high accuracy, have combined verbal, physiological, and gestural cues. Linguistic markers also have been explored to distinguish truthful from deceptive language (Vogler & Pearl, 2020), aiding in detecting fraudulent online content (Jakupov et al., 2023). An analysis of text and visual content manipulation technologies is presented in Arnaudo et al. (2021). Advancements in machine learning have introduced benchmark datasets for detecting false information (Barsever et al., 2023), support vector machines (Nguyen et al., 2020), and transformer-based models like BERT for fake news detection (Barsever et al., 2020). The ExoFIA model integrates multimodal features and attention mechanisms to analyze fake news on social media (Li et al., 2023). Fact-checking and source reliability assessment also play a key role in misinformation detection. The rise of social media has exacerbated fake news dissemination, facilitated by the absence of temporal and geographic constraints. The COVID-19 pandemic intensified the issue, leading to an "infodemic" of misleading content (WHO, 2021). Governments and tech companies, including Facebook and Google, took measures to counter misinformation. Studies examined public trust in information sources (WHO, 2021) and patterns of misinformation spread (Tasnim et al., 2020). Furthermore, misinformation about vaccines had significant societal impacts (Carrieri et al., 2019).