浏览行为数据在Web用法挖掘中的应用

The Application of Browsing Action Data in Web Usage Mining

网站用户的浏览行为的不确定性导致从现有的Web日志文件挖掘出的事务模式来预测用户行为越来越困难。引入一种新的数据类型浏览行为数据(BAD)用于提高Web用法挖掘的质量,BAD是一种特殊的浏览数据,如"复制","滚动","另存为",并且没有被日志文件所记录,并给出了BAD的定义。为了象Web日志文件一样记录BAD,介绍并使用一种现有的在线数据收集模块来捕获用户BAD。通过一个电子商务程序的实例说明BAD能增加现有Web事务挖掘算法的有效性。

The behavior of the users of a website may change so quickly that it becomes a real challenge to attempt to make predictions according to the trascation patterns coming from the analysis of a Web log file, This research explores a new data source called Browing Action data（BAD）to improve the effectiveness of WUM applications. BAD is a category of online browsing action, such as＂copy＂, ＂scroll＂, or ＂save as＂ and is not recorded in Web log files. This paper defines BAD. In order to make BAD recorded as the normal Web log, an online data collection module for capturing BAD is also proposed and discussed. The benefits of BAD which are illustrated by the implications of Web transaction mining algorithm for an EC application. In this paper the authors only adopt the occurrences of ＂scroll＂used in the mining process, there are different BAD that have varied attributes. How to combine varied BAD into the mining is considered to be immediate future work.