It’s All in the Embedding! Fake News Detection Using Document Embeddings

With the current shift in mass media landscape from journalistic rigor to social media, personalized is becoming new norm. Although digitalization progress of brings many advantages, it also increases risk spreading disinformation, misinformation, and malformation through use fake news. The emergence this harmful phenomenon has managed polarize society manipulate public opinion on particular topics, e.g., elections, vaccinations, etc. Such information propagated can distort perceptions generate unrest while lacking traditional journalism. Natural Language Processing Machine Learning techniques are essential for developing efficient tools that detect Models context textual data resolving news detection problem, as they manage encode linguistic features within vector representation words. In paper, we propose a approach uses document embeddings build multiple models accurately label articles reliable or fake. We present benchmark different architectures using binary multi-labeled classification. evaluated five large corpora accuracy, precision, recall. obtained better results than more complex state-of-the-art Deep Neural Network models. observe most important factor obtaining high accuracy encoding, not classification model's complexity.