Constructing User Interaction Behaviors Net from System Log

Abstract

Traditional modeling system methods are mainly on system mechanism and structures. For many applications interaction between user and system is important. How to model and analyze user behaviors is becoming more and more important and it is applied to solve problems of user interest, information service recommendation, network security and so on. This paper introduces an algorithm to extract and model user interaction behaviors from system logs and represent them in the form of user interaction behaviors net. By combining user interaction behaviors net with user log datasets stored in system log, it is possible to describe user behaviors in a much more comprehensive and effective manner. Results show that the algorithm can successfully model user behaviors and supply a new description basis for user behavior feature extraction and identification.