Speeding up web access using weighted association rules
PReMI'05 Proceedings of the First international conference on Pattern Recognition and Machine Intelligence
| Hi-index | 0.00 |
In this paper we study data mining techniques as toolsfor reducing the time needed to access web pages in theenvironment of a corporation where users connect to theInternet through a proxy server. We add a data miningserver to the traditional web architecture. This servercomputes, using sequential patterns, the pages likely tobe requested by the users, taking into account their webaccess history. Then, the server loads these pages intothe proxy's cache in order to have them available whenthey are actually asked for. We describe our implementationand the results obtained, showing that the averageaccess time to a web page can be dramatically reducedusing this technique. Further, we also implemented ourproposal using simple association rules, showing that sequentialpatterns result in a smaller and more accurate setof rules, as a consequence of taking into account the orderof the requests. Finally, we discuss the differenceswith other prefetching proposals.