Detecting Hoax Content on Social Media Using Bi-LSTM and RNN

Online media, such as websites and applications, have become a communication tool available on the internet. Social media is part of online that can be used to spread news, opinions, or even hoaxes, through Twitter. Although hoaxes are difficult eliminate, several systems been built using deep learning approaches process text images detect truthfulness news. In this study, four were methods, namely Bi-directional Long Short-Term Memory (Bi-LSTM), Recurrent Neural Network (RNN), hybrid RNN-Bi-LSTM, Bi-LSTM-RNN. Feature extraction was performed Term Frequency - Inverse Document (TF-IDF) feature expansion Global Vectors (GloVe). The data has adjusted according keyword fake news mainstream portals. This study attempted scenarios compare various methods built, with aim finding best method provides highest accuracy. results showed Bi-LSTM had accuracy 96.48%, while Bi-LSTM-RNN ranked second an 96.36%, followed by RNN 95.49%, RNN-Bi-LSTM 95.34%.