Classifying Twitter Spammer based on User's Behavior using Decision Tree

Abstract

Twitter is one of microblogging service that widely used by people. Its popularity invites spammers to disturb other users with a large number of spam tweets. Spammers send untrusted news, unwanted tweets to another twitter accounts to introduce a product and service, a job with high salary, promote a new website, spread advertise to generate sales that could harm other users. This paper collects a hundred accounts from non-spammer and spammer. After that, manually classified as a non-spammer and spammer. User's behavior characteristics, which could give many clues to classify spammer. This paper applies profile users as features for the machine learning to classify users as a non-spammer or spammer. This paper applies seven attributes such as the statuses count, followers count, friends count, the age of account, average tweets per day, average limits between tweets, verified user or not. Using a Decision Tree method, we could classify non-spammer and spammer. The accuracy of the classification of non-spammer and spammer is 88,235%