Utilizing document classification for grooming attack recognition

Grooming attack recognition is a complex issue that is difficult to address using simple word matching in order to identify potential hazard for minor users. In this paper, the utilization of document classification to create patterns from real dialogs is proposed. Furthermore, a decision making method that results in generating proper warning signals based on the classification results is introduced. The decision making method is then applied using the best ranked algorithm with a comparative evaluation which conducted on seven document classification algorithms.