A Social Bots Detection Based on Behavior Enhanced with Machine Learning Technique
Abstract
Social bots spread more and more on social platforms through the development of the Internet. Therefore, the detection of these social bot computers that threaten social networks is called for an effective detection algorithm. They may produce fake messages, rumor and even manipulate opinions of the public. Massive social booms have recently been created and spread widely on the social platform and have a negative effect on the public and safety. The aim of bot detection is to differentiate bots from humans, and over recent years it has received more and more attention. We propose behavior enhanced with machine learning detection model in this paper. The model proposed regards user content as time text data rather than plain text in order to extract latent time patterns. In addition, this method combines content and behavior information with a machine learning method. This is the first trial to apply the machine learning techniques on neural network to bot detection, to our best knowledge. Real-world experiments collected from Twitter also demonstrate that our proposed model is effective.