Spammer Detection using Features based on User Relationships in Twitter

Chansik Lee, Juntae Kim

http://doi.org/

Twitter is one of the most famous SNS(Social Network Service) in the world. Twitter spammer accounts that are created easily by E mail authentication deliver harmful content to twitter users. This paper presents a spammer detection method that utilizes features based on the relationship between users in twitter. Relationship based features include friends relationship that represents user preferences and type relationship that represents similarity between users. We compared the performance of the proposed method and conventional spammer detection method on a dataset with 3% to 30% spammer ratio, and the experimental results show that proposed method outperformed conventional method in Naive Bayesian Classification and Decision Tree Learning.