Research On Key Technologies For E-Commerce Oriented Web Usage Mining

With the quick development of internet,World Wide Web has permeated every aspect of human life.Challenged by the huge body of information on the Web which has characteristic:dynamic, heterogeneous and semi-structured.How to make an adequate use of the data resource and how to extract the interesting,implicit and valuable model and knowledge have become an enormously important topic of research in the field of data mining. Web usage mining is the application of established data mining techniques to analyze web sites usage.For an E-commerce company,this means detecting future customers likely to make a large number of purchases,or predicting which online users will click on what advertisements based on observation of prior users who have behaved both positively and negatively to the advertisement banners. The research of this dissertation mainly focuses on the key techniques such as classification,clustering,association rules and sequential patterns used to find out user access patterns from Web usage data as well as the principle of main algorithms by use of information entropy,concept lattice,D-S evidence,fuzzy sets,F-tree,algebra lattice. The paper illustrates the approach to electronic commerce oriented Web usage mining in terms of Web pages classification,Web pages clustering,Web usage association rules mining and Web usage sequential patterns discovery and provides a broad prospect for the development of electronic commerce oriented Web usage mining.