Research Of Web Data Mining

The resources on WWW are increasing day by day. One important problem is how to get the required information via effective approaches. Web mining is a technology to find users' browsing model or relative web pages. Web mining will give reasonable advices for web masters, investors and advertisers, etc. It can also provide powerllil intelligent searching engine and customized services to end-users. Since web is such an information system that is unstructured, dynamic and distributed, it is difficult to mine it directly. However, the log of a web server has an integrated structure. We would mine web logs to implement web data usage mining.The thesis analyzes the current research situation of web mining and proposes the problem. Data mining and its technologies are introduced. The relationship between XML and Web mining is presented.The method of applying mining systems to web usage is described in detail, which includes three phases:Firstly, the data are prepared. The preparation of web information is important in web usage mining. It is also a heavy work. The quality of data preparation would affect the results of data mining directly. The thesis discusses the following issues during the preparation: data collection, data cleaning, user identification, session identification, transaction identification and path completion.Secondly, the patterns are discoveried, which are important in the research. The thesis presents some common technologies of path analysis, association rules, classification and clustering. Based on the description of the questions, a time-sequence mining algorithm with high efficiency is presented, which guarantees the integrality of time-sequence mining and higher efficiency.Finally, pattern analysis and applications are investigated. Pattern analysis is...