Spam Detection in Twitter Data

Among all of an existing social media sites Twitter has grown to be the most admired by the internet users as it has changed the way of information exchange in recent years. But, as its popularity has spiked, spammers have emerged as one of twitter’s biggest limitations finding it easily accessible for attacking the trending topics to spoil useful content, generate traffic and revenue. They increase privacy concerns of the users by attacking their personal information. According to Twitter it mostly uses detection and blocking of spam creating accounts as a mechanism for dealing with spam issues. However, spammers always have the option of creating new accounts. So, there is a need of spam detection rather than detecting the accounts. Twitter is currently using Google Safe Browsing for the purpose of detecting and blocking of spam links but they are not proving to be much useful. In this paper we introduce bodywork that considers user based, tweet-based features and tweet text features. Then we use different machine learning algorithms. Our framework gives 95.695% accuracy with Naive Bayes algorithm.