Acta Electronica Malaysia (AEM)

Social media usage is on the rise due to the proliferation of the internet. People share their interests and search for news of their interests on social media because news on social media can be accessible to anyone easily at the cheapest cost. This shift comes with several benefits as well as some drawbacks. Particularly, in an information-driven society, fake news has become an immense threat to humankind because of the continuous distribution of inauthentic content on social media. The primary difficulty is to distinguish false news from real news. This article proposes a two-step technique for detecting false news. Before everything, we use methods like preprocessing and text mining to turn unstructured material into a form that can be understood by a computer. The documents in the datasets have been converted into feature vectors using TF￾IDF vectorizer, as well as the weight of each word is provided with the document term matrix. Eight machine learning algorithms have been applied to the datasets transformed after preprocessing and text mining methods in the second step. At the end of this work, experimental evaluation of these machine learning algorithms is performed on public datasets, and the comparisons have been made, and the algorithms have been evaluated based on four evaluation metrics.