Extracting features and writing styles from short text messages for compromised tweet detection is always a challenge. Short messages, such as tweets, do not have enough data to perform statistical authorship attribution. ...