Web-log Mining for Quantitative Temporal Event Prediction

摘要: The web log data embed much of users’ browsing behavior. From the logs, one can discover patterns that predict future requests based on their current These are very complex due to large size and sequential nature. In past, researchers have proposed different methods what pages will be visited next present visit patterns. this paper, we extend work when these page accesses occur. Our method is a novel extension association rule classification method. We traditional rules by including temporal information explicitly in each rule, reason about confidence prediction terms its region. compare two for event prediction, demonstrate effectiveness our empirically realistic explore tradeoff between accuracy mining time models.