【作者】 袁泉；

【导师】 刘群；

【作者基本信息】 哈尔滨工程大学 ， 计算机软件与理论， 2007， 硕士

【摘要】 商业经济的未来发展模式是以客户为中心的面向服务的运营模式，其核心是完善的服务保障体系和高效的经营管理策略。其中，Web日志分析和挖掘就是通过了解Web浏览用户的兴趣，来改善服务质量的一种商业智能的应用。本文将XML应用到OLAP中，并针对Web日志的分析和挖掘进行研究，主要做了如下几方面的工作。1．分析了XML在数据描述方面的优势，结合这些优势介绍XML在OLAP系统中的一些应用。2．提出一个基于XML的OLAP和数据挖掘过程。该过程考虑了异构数据源的集成，使用基于XML的数据立方体，OLAP系统与数据挖掘系统协同工作，并通过Internet发布分析和挖掘结果。3．具体分析了基于XML的OLAP和数据挖掘过程所涉及的关键技术，如基于XML的异构数据源集成的方法；引入了一种专门针对XML的多维表达式，用它构建数据立方体和进行多维查询，并介绍了使用XML描述Web日志立方体的具体方法。4．提出一种基于用户对网页分类的兴趣度来对用户聚类的方法，忽略用户对个别网页的点击量，并根据网页内容和深度设定网页的权重，可以理解为按用户宏观兴趣来聚类。最后通过实例，分析了这种方法与传统方法的区别。本文所做的研究工作，为基于XML的OLAP和数据挖掘系统的研究，和它在Web日志分析和挖掘中的应用，以及Web日志聚类的研究，提供了一定的参考价值。更多还原

【Abstract】 In the future, the development pattern of commerce economy is a business pattern which takes customer as center and is service oriented. This pattern’s core is perfect service ensures system and high-effect management control strategy. Web-log analysis and mining is an application of business intelligence whose purpose is to improve service quality through understanding web browsers’ interest. This dissertation applies XML in OLAP, aiming at analysis and mining web-log. There are several aspects this dissertation does as fallows:1. This dissertation has analyzed the advantages of data description with XML and introduced some applications of XML in OLAP system.2. A process of OLAP and data mining based on XML is proposed. This process takes isomerous data source integration into account and use data cube with XML. In this process, OLAP system and data mining system associate, and the result of analysis and mining is released with internet.3. This dissertation has analyzed concretely the technologies involved in the process of OLAP and data mining based on XML, such as isomerous data source integration based on XML; Has introduced a multi-dimension expression based on XML, which is used to construct data cube and multi-selection; Has introduced the method of describing Web-log with XML.4. A method which makes users clustering with the rate of interest of users toward web page classes is proposed. This method ignores the hits amounts, set web page weight with its content and depth. It can be considered as a method with users’ macro interest. At last, this dissertation analyzed the difference between this method and traditional ones.The research done by this dissertation, have provided certain reference value for the research of OLAP based on XML and its application in web log analysis and mining, and the research of web-log clustering.更多还原