Deepbot: A Deep Neural Network based approach for Detecting Twitter Bots

Abstract

Social networks have played a very critical role in very aspect of our daily life. However, a wide variety of bots have been found which are designed for some malicious purposes such as spreading spam mes- sages and faking news. Although various techniques have been proposed, this task is still challenging if we want to judge whether the tweets are posted by a bot or not merely based on the textual information. For this challenge, the Deepbot is designed which adopts the Bi-LSTM model to analyze tweets and a Web interface is provided for public access which is developed using Web service. From our empirical studies, this system can achieve better classification accuracy.