摘要
在W eb会话识别中,传统的路径补充算法仅考虑了点击"后退"按钮造成日志记录的不连续性。针对此问题,提出基于多窗口的路径补充思想,并结合确定的有限自动机(DFA)来实现该算法。对于用户从新窗口打开页面,异步并发访问网站时造成日志记录的不连续,该算法能避免按"后退"方式进行路径补充,而是采用多窗口方式。通过实例验证了该方法比传统路径补充算法更能准确地反映用户实际的访问路径。
In web session recognition,traditional algorithm of path supplement just considers the incontinuity of records in logs caused by clicking the" back" button. Aiming at the problem, in this paper it proposes an idea of path supplement based on multi-window, and implements the algorithm in combination with Deterministic Finite State Automata (DFA). For the incontinuity of records in logs caused by opening source pages from the new window and browsing the website with asynchronous concurrent mode,this algorithm does path supplement with the mode of multi-window rather than with the" backward" mode. In the end of this paper, it shows through examples that this algorithm reflects the users 'actual browsing paths more accurately than traditional algorithm.
出处
《计算机应用与软件》
CSCD
2009年第7期46-47,51,共3页
Computer Applications and Software
基金
高等学校博士学科点专项科研基金(20060285008)
关键词
WEB使用挖掘
会话构造
路径补充
多窗口
Web usage mining
Session reconstruction
Path supplement
Multi window