Rumor Detection Using Various Deep Learning Approaches

Abstract

As the endless development in web 2.0 and ease of access methods, devices upcoming new technologies like Social Media, Mobile, Analytics and Cloud-generates infinite stream of data. The misinformation can spread widely and rapidly in online social network. Due to potential harm this circulate may bring to public, so false rumor detection is demanding and important. Previous studies are mainly based on various machine learning algorithms and deep learning techniques. In this paper, various rumor detection techniques using Deep Learning Models like Long Short Term Memory (LSTM), Convolutional Neural Network (CNN), CNN-LSTM, Bidirectional LSTM (BiLSTM) and CuDNNLSTM( layer with LSTM)  on textual data are performed and  analysis has been done. These models perform binary classification of tweets into rumors and non-rumors. Comparative Analysis has been done with results on same dataset by existing machine learning algorithms and our deep learning models.  Our deep learning models outperforms the baseline machine learning algorithms.