Ensemble learning-based model for fake news detection

Technological advances in the 21st century have improved the life quality of humans worldwide. They have morphed the news-delivering sphere from a monopoly controlled by state-owned media to a free-speech-motivated playfield. News is no longer conveyed solely by journalists' articles and printed newspapers. Social media, and the Internet in general, allow anyone to share or deliver news to a large audience in real-time, very quickly and at a low cost. Unfortunately, fake news has spread widely along with true news online. Fake news can cause significant damage to companies, celebrities, and even ordinary individuals. Therefore, detecting fake news has become an important task. This paper presents an ensemble learning model that uses CNN, LSTM and C-LSTM to recognize fake news. The aim is to create a model that can effectively detect fake news in both short and long news statements. For this purpose, a combination of two datasets, namely the ISOT and LIAR datasets, were used and combined into a single corpus. Evaluation metrics were used to assess the models' performance: accuracy, recall, precision, and F1-score. When compared to the state-of-the-art, the proposed model achieved good results. Compared to the state-of-the-art, the proposed model achieved competitive results. An accuracy of 89.16% and an F1 score of 95.03% were obtained on the combined corpus, a precision of 89.47% on the ISOT dataset, and an accuracy of 53.23% and an F1-score of 71.80% on the LIAR dataset.