A gated recurrent unit-based model for fake news detection in online social networks

Fake news and online rumors can mislead information, disrupt order, and destabilize society. This misleading information is propagated through online social media and other digital platforms, which results in loss of public confidence, stimulation of social unrest, and a threat to national security. In this work, we present a Gated Recurrent Unit-based fake news recognition approach to reduce the impact of misinformation. Since we want to build an effective model, we pre-process a fake-news-detection dataset having real and fake news by performing effective natural language processing (NLP) tasks such as news text cleaning, tokenization, word embedding, etc. The Gated Recurrent Unit (GRU) architecture is used based on its ability to execute sequential data, which allows the model to find patterns and the semantic relations within news content. The model proposed attained an accuracy of 99.55% and proved to be effective for detecting fake information. Our recommendation is to continue the fight against misinformation and other such cybercrimes.