A novel model for user clicks identification based on hidden semi-Markov

摘要: User click pattern is a critical part of cyber behavior, which important for web operators and designers. Usually, massive mixed HTTP requests, we observed, are caused by the clicks from numerous browsers. Moreover, ever growing complexity makes it more difficult to identify user accurately. In this paper, propose novel identification method based on hidden semi-Markov model. develop parameter estimation state algorithms our order initialize model value improve applicability real websites, selection algorithm K-means clustering reveal in-line objects pages different websites. We evaluate with data set, collected at backbone Telecom. The experiment results demonstrated that works quite well.