A recurrent neural network model for detecting fake news on social media

초록

The growing popularity of social media platforms for sharing news and videos has made it easier for users to access and share information instantly. However, verifying the credibility of such content remains a significant challenge. These platforms enable the rapid spread of fake news, which often leads to the distribution of inaccurate information. Since social media content is largely unrestricted, users frequently share news without verifying its source or accuracy, causing fake news to spread quickly and sometimes go viral. This can have harmful effects on society. Therefore, it is essential to ensure that the news shared on social media is accurate to prevent users from being misinformed, which is crucial for positive social development. This study proposes a recurrent neural network model using artificial intelligence and machine learning to detect and verify fake news on social media. The framework includes steps such as defining the problem, using datasets labeled as "fake" and "true," and applying natural language processing techniques. The authors conducted data cleaning, feature engineering, and visualization of real and fake news before converting text into tokens. The proposed model achieved a high accuracy of 99.96% with a minimal loss of 0.0083 after processing over 14 million tokens using 128 layers. © 2025 The Authors.

키워드