Sequential Patterns Mining Based On Web Logs And Its Application In E-Commerce

Electronic commerce has become popular with the rapid development of Internet.E-Commerce system has been accumulated a mass of access data, which is significantresource for the enterprise. Data Mining is devoted to data analysis and understanding,finding potential knowledge in the data. So a large number of useful potentialinformation can be taken out from the data which has been explosively growing. WebMining used in web data analysis become an important research filed.Web log mining is a kind of Web data mining. It uses the technology of DataMining to analyze the server log, and mining user access patterns such as associationrules, sequential patterns, etc. from the access log of the web.Sequential patterns mining based on web logs and its application in E-Commerceare researched in this thesis. Sequential patterns mining were first used in basketanalysis. One of the most common algorithms is PrefixSpan. The algorithm used inthis thesis is based on PrefixSpan and make some improvement on it, such as addingmaximal gap constraint. Then the algorithm can not only do basket analysis but alsomining Web access sequence. Data Preprocessing is veryimportant in Web log mining,A task including Data Cleaning, User Identification, Session Identification and PathCompletion is carried out in a preprocessing for web log file. Afterwards, the methodof mining Web access sequential patterns is presented, which identify the maximalforward path firstly, and then mining frequent sequential patterns with sequentialpatterns mining algorithm.Web access sequential patterns mining are very meaningful in practice. Thedevelopment of E-Commerce provides a new commercial pattern for the enterprise.Web access sequential patterns is useful in E-Commerce, it can be used in improvingWeb structure and also providing personal service.