基于D_S证据理论的Web事务识别方法

Web Transaction Identification Method Based on D_S Evidence Theory

基于D_S证据理论,提出一种新的Web事务识别方法,融合多个特征对Web事务进行综合评判,并选取计算代价较小的特征及高效的融合规则,保证算法性能的高效.实验结果表明,该方法与其他事务识别方法相比,具有更高的事务识别率.

Presents a novel Web transaction identification method based on Dempster-Shafer （D_S） evidence theory. This method fuses multiple features of Web session to decide whether user＇s interest is changed. Furthermore, light-computation features are used to develop an efficient fusion mechanism to guarantee high performance of the algorithm. Compared with other transaction identification methods, experimental results showed that our method achieves a high performance.