An integrated approach for malicious tweets detection using NLP
Sagar Gharge, Manik K. Chavan · 2017
Many previous works have focused on detection of malicious user accounts. Detecting spams or spammers on Twitter has become a recent area of research in social network. However, we present a method based on two new aspects: the identification of spam-tweets without knowing previous background of the user; and the other based on analysis of language for detecting spam on twitter in such topics that are in trending at that time. Trending topics are the topics of discussion that are popular at that time. This growing micro blogging phenomenon therefore benefits spammers. Our work tries to detect spam tweets in based on language tools. We first collected the tweets related to many trending topics, labelling them on the basis of their content which is either malicious or safe. After a labelling process we extracted a many features based on the language models using language as a tool. We also evaluate the performance and classify tweets as spam or not spam. Thus our system can be applied for detecting spam on Twitter, focusing mainly on analysing of tweets instead of the user accounts.