中文摘要：

线上购物相对线下购物的一个特点是可以将用户的购物过程记录到网站的Web日志中,为研究用户网上行为特征提供必要的数据支持。本文详细介绍了对Web日志数据进行预处理的一系列步骤和方法,并以某出版社网站18天的日志数据为实证,提取出用户的访问路径（访问页面的先后次序）,分析路径信息得到用户在不同页面之间访问的转换概率,从而定量地衡量购物流程中各步骤之间的转换率和贡献率,提炼出用户最有可能的前向转换路径和后向转换路径,以期对网站流程优化和商品促销设置提供决策支持。

英文摘要：

Recording users＇ shopping process in Web log is a feature of shopping online relative to shopping offline,and provides the necessary data for studying users＇ characteristic behaviors online.This paper introduces a series of steps and methods of Web log data preprocessing,and empirically analyzes the 18 days of log data of one publishing house,extracts the users＇ visit path（the sequence of visit pages）,calculates the transit probability which users visit among the different pages,so as to quantitatively measure the transit rate and contribution rate among the steps of shopping process,and find out users＇ most likely forward transit path and backward transit path.We hope this paper could help making decision for process optimization and promotion online.