Early Detection of Sinhala Fake News in Social Media

With the advancement of technology, people started to rely on social media to get news updates. The use of social media is easy, cost-effective, and saves time. People can access any news whenever they want without worrying about the time or where they are. Even though social media makes everything simple and easy, it also accelerates the propagation of fake news. The manual fake news identification process is time-consuming and requires expert human resources. Therefore, social media fake news detection has become a hot topic in the research world. However, only limited studies have addressed Sinhala fake news detection problem. This study proposes a new approach to identify Sinhala fake news in the early propagation stages before reaching a broader audience. The proposed method consists of content-based techniques to extract features and a deep learning approach to classify news items. Experimental results show that the model gained an 82% accuracy with FastText word embedding and the LSTM model.