| Summary: | 碩士 === 國立雲林科技大學 === 資訊管理系碩士班 === 91 === As the rapid development of the E-Commerce and Internet, it is possible that consumers can go shopping online conveniently. Most of the studies of web usage mining are using web server log as data source, and using the offline batch process to proceed data pre-processing and pattern mining, and apply the pattern to the online recommendation. However, using the web server log as mining data source, have to take many preprocessing task to eliminate irrelevant items, and using the offline batch process to mining patterns would cause the problem of couldn’t only deal with the newly inserted transactions.
In this study, we proposed a new online incremental web usage mining methodology, which utilized keeping track of user's browsing behavior in real time to solve the problem of using web server log, and utilized online incremental mining to solve the problem of traditional mining method which couldn't only deal with the newly inserted transactions. We also developed a supportive system based on the methodology to proceed experiment. For two weeks of the experiment, the evaluation result show that:
In the perspective of system executing performance, the browsing recommendation ratio was 97.9%, and the shopping cart recommendation ratio was 92.8%, and the average of overall browsing recommendation time was 4.16 seconds, and the average of overall shopping cart recommendation time was 0.52 second, and the average of overall browsing mining time was 12.83 minutes, and the average of overall shopping cart mining time was 1.41 minutes.In the perspective of system recommendation benefit, the overall recommendation precision was 37.4%,and the overall recommendation coverage was 80%,and the overall F1 was 51%.
|
|---|
