【作者】 王书舟；

【导师】 高中文；

【作者基本信息】 哈尔滨理工大学 ， 控制理论与控制工程， 2003， 硕士

【摘要】 Web使用挖掘就是从服务器日志文件和客户交易数据中挖掘有意义的用户访问模式和潜在的客户群，使企业能够提供个性化信息服务和开展有针对性的电子商务活动。随着越来越多的业务在互联网上开展，用户使用Web的规律成了各企业共同关注的一大热点。因此，采用Web挖掘智能地、自动地提取出有价值的知识，构建自适应网站，提高WWW的效率，具有十分重要的现实意义和广阔的应用前景。基于国内外最新研究成果，论述了Web使用挖掘的内容、特点和挖掘过程。定义了自适应网站，提出了基于Web挖掘实现自适应网站的一种方案:在已存在的站点上进行用户访问模式挖掘，预测用户感兴趣的页面，并以增加链接的方式把指向这些页面的链接推荐给用户，动态地改变网站结构。给出了系统体系结构，阐述了系统实现的目标、设计所遵循的原则和源数据的收集等主要问题描述了系统挖掘用户访问模式的过程和算法。提出了一种预处理功能模型，采用基于cookie技术和扩充日志属性的用户识别方法，有效地识别通过同一代理服务器访问网站的不同用户。采用了一种扩展性良好、高效多能的聚类分析算法对用户访问模式进行挖掘，即直接对网站的拓扑结构和用户浏览信息进行处理的关联矩阵方法，避免了复杂的会话识别和事务识别。提出了自适应网站进行自动页面调整的方法，根据挖掘推荐结果，通过在网页中嵌入ASP代码的方式，用程序实现自动在页面中增加动态链接，对不同用户展现不同的网站视图，并给出了源程序。对系统进行了实际运行测试，得到了可行性验证。更多还原

【Abstract】 It is important for the modern enterprises to have the ability of discovering useful user access pattern and corresponding potential customers from large volume of use access logs, so that they can provide personal information service and make their electronic commerce strategies. With more and more business spread in Internet, enterprise consentrate on the principle of access pattern. It become practical significance and a vastitude foreground to abstract useful knowledge, construct adaptive web site,improve efficiency of WWW with web mining intellectively and automaticallyBased on the latest researched results, this paper discuss the content , characteristics and processes of web usage mining. In this paper we define adaptive web sites, puts forward the scheme of it based on web usage mining .Mining the user access pattern on the exist web station, forecast web page interested by user, recommand them to user through adding hyperlink of them. so as to change the constrature of the web site. We expatiate such primary question as the goal , rules to design adaptive web site and collection of sourse of data for mining.we describe the algorithm and processes of user access pattern , and presents a data preprocessing model. In the process of preprocessing,a user identification method based on cookie technology and extending Web Log attributes are adopted. the method can distinguish effectively the multiple users using the same one proxy server .The expansible and efficient URL-UserID relevant matrix clustering algorithm is used for user access pattern mining, used for dealing with information of user’s browsing patterns according to web site’s directed graph defined and avoid complicated session recognition or transaction recognition.We present the method of adjusting web page automaticly in adaptive web site. according to result of web usage mining, the program add dynamic hyperlink automaticly by inserting ASP code in web page to present different view to unlike user. And the source program are given.The feasibility of adaptive web sites has been tested by experiments更多还原