摘要： 提出一种基于数据挖掘的网络主题用户数量计算模型。从网络服务器访问日志中挖掘网络用户使用记录，应用窗口函数识别多个IP相同的用户，通过分析用户行为的时间特征实现对虚假点击的过滤，构建能够表达用户主观兴趣取向的用户向量，从而自动计算各个网络主题在给定时间片内的用户数量。实验结果证明，该模型具有较高的计算准确性，能为管理人员决策提供技术支持。

关键词: 用户行为分析, 窗口函数, 用户兴趣向量, 时间片, 用户数量计算模型

Abstract: This paper presents a model for computing user quantity of Web topic which is based on Web data mining. It mines users usage record from Web server access log, applies the method of window function to identify the multi-users of a terminal owning identical IP, and analyzes the timing character of Web users’ behavior to filter the fraud data. The user interesting vector is constructed which can reflect the reading preference of a user in some extent, the Web users quantity of a Web topic in given time span can be calculated. Experimental results show that the model has the high accuracy, it can provide targeted decision support for administrators.

Key words: user behavior analysis, window function, user interest vector, time slice, user quantity computation model

中图分类号: 

• TP39