Advanced Fake News and Hate Speech Detection Leveraging Spectral Graph Theory

Despite the proliferation of the Internet, prejudice and misinformation have persisted throughout history. False news and hate speech have long been a part of human nature, as has lying and slander. The proliferation of social media, however, has changed the timing, location, and nature of disinformation and hate speech. False news and hate speech are no longer confined to offline forums thanks to the Internet.This suggested method consists of preprocessing, feature extraction, and training the model. The impact of pre-processing was examined by analyzing various pre-processing procedures, which were subsequently followed by the suggestion of a combination of these strategies. We used Word2Vec for feature extraction; it looks at a word's context to try to figure out what it means and how it's related to other words. Throughout the training process, Spectral Graph Theory was regularly utilized. This cutting-edge method outperforms CNN and LSTM with an average accuracy of 94.95%.