摘要
[目的/意义]鉴于已有基于点击流的用户模型大多简单地采用页面类型序列代替行为序列,提出一种根据点击流访问页面序列到用户行为的映射方案,解决用户行为建模的问题。[方法/过程]本文在分析网页URL参数、页面内容等特征的基础上,以81 759个电商用户会话为测试样本,提出并实现从页面到用户行为的映射方法,给出一种依据原始日志建立用户行为序列来描述会话的方案。[结果/结论]分析反映出在会话层面上已有研究不易得到的行为特征,得到6类具备不同行为模式的会话:功能探索会话、卖家管理会话、营销推动会话、资料管理会话、商品浏览会话、检索依赖会话。基于点击流对用户会话建模,可以得出用户会话中行为序列特征,对实现准确营销与推荐具有重要价值。
[ Purpose/significance] Most user session models based on click -through sequences take sequences of the page types, but not users＇ behaviors. This paper aims to construct a user behavior typology and model user session be- haviors using the typology. I Method/processl By analyzing features of URL parameter and pages contents, this paper takes 81 759 e-commerce user session behaviors for examples and proposes a novel approach to model user sessions with E - commerce click - through data by mapping movements from URL to URL to a typology of user behaviors. I Result/con- clusionl This approach is tested with a sample of 81 759 user sessions. It recognizes 6 different types of sessions by their behavior sequence patterns. The behavior typology is useful in modeling session behavior and the recognized behavior pat- terns may be sued for marketing and recommendation.
出处
《图书情报工作》
CSSCI
北大核心
2015年第1期119-126,共8页
Library and Information Service
基金
国家自然科学基金项目"面向电子商务生态平衡的目录导购机制研究"(项目编号:71373015)研究成果之一
关键词
电商网站
会话模型
用户行为
商品搜索
商品浏览
e-commerce sites session modeling user behavior product information seeking product informationbrowsing
