| Summary: | 碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 92 === With the rapid growth of the World Wide Web and the development of E-commerce, mining and predicting user’s web browsing patterns have become a hot topic. The past researches on this field focus on mining users’ navigation patterns or clustering pageviews so as to model users’ behavior. However, none of them are concerned with the web log temporality, i.e., the start time of a user session in our definition. In this paper, we take into account the Web temporality for constructing the time-based user behavior model, based on which the user behavior can be predicted. In addition, we propose three methods to measure the changes of Web temporality in order to evaluate the applicability of a temporality model. Our experiments show that the precision of prediction can be improved more if there exist more distinct changes of temporality in the user’s browsing behaviors.
|
|---|
