Web Pre-fetching Of Data Mining Research

Although the speed of the network has been improved considerably in recent years,the rapid expansion of using the Internet and the inherited character of delay in thenetwork makes the network very traffic. The network makes no guarantee on thequality of service for the user. It is not very practical that we solve this problem onlyby enhancing the hardware of the network. This paper studies the Web predictionmodels and presents a new Web prediction method. It is very useful to enhance thespeed when the user browses the Web pages.This paper firstly analyzes the current Web prediction method, and set up the Webprediction model with the data mining technology. Secondly, on the base of the modelthe paper define the binary and the multi prediction strategy. Binary predictionstrategy is very simple, but it is inefficient because there are too much analysis andprediction. Because the interest of the user is an interest chain, he usually visits thenext pages along the links in the current page. So we can mine the interest chains ofthe user and get his interest models. With these models, we can request a series ofWeb pages when the user is visiting a Web page and do not predict again. This is multiprediction strategy. Obviously, multi prediction strategy can reduce the times ofanalysis.Finally, this paper develops a platform with VB.Net and analyzes the performanceof the binary and multi prediction strategy. The test manifests that these strategies arevery effective to enhance the speed when the user visits the pages.