Deep Learning Approach For Rumour Detection In Twitter: A Comparative Analysis

5 Pages Posted: 15 Aug 2019

See all articles by Pavithra


Government Engineering College, Sreekrishnapuram


Government Engineering College, Sreekrishnapuram

Date Written: August 15, 2019


Social media has become popular as a platform allowing users to follow events and break news. The rapid emer- gence and spread of new information is one of the features that characterizes social media. This leads to rumour being circulated. It is technically very challenging to automatically detect rumours. Rumour detection can be formally defined as, Given a claim Ci, convert it into a propagation tree to determine whether the source tweet is Rumour/ Non-rumour using different neural networks, where Ci is a claim consisting of ri, xi1, xi2, xim where ri is the source tweet and (xi1, xi2, xim) corresponds to responsive tweets. Most of the previous studies used supervised models based on feature engineering which focus on text mining from sequential microblog streams. This work primarily focuses on a comparative study for social media rumour detection using different neural network models such as Convolutional Neural Network(CNN), Recurrent Neural Network(RNN) and Recursive Neural Network(RvNN). Detecting rumours from Twitter mes- sages is of primary concern. The model of CNN and RNN depends only on text. I.e. tweets for source and response. And the RvNN model depends on the text and the structure of the propagation. The RvNN model outperforms the other models as it capture the propagation structure of the tweet and it’s response.

Keywords: Rumour/Non-rumour, Twitter, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Recursive Neural Network (RvNN)

Suggested Citation

C P, Pavithra and Joseph, Shibily, Deep Learning Approach For Rumour Detection In Twitter: A Comparative Analysis (August 15, 2019). In proceedings of the International Conference on Systems, Energy & Environment (ICSEE) 2019, GCE Kannur, Kerala, July 2019. Available at SSRN: or

Pavithra C P (Contact Author)

Government Engineering College, Sreekrishnapuram ( email )


Shibily Joseph

Government Engineering College, Sreekrishnapuram ( email )


Register to save articles to
your library


Paper statistics

Abstract Views
PlumX Metrics