Web usage mining is an active research area for its uses in web site maintenance and for the potential economical impact. In the past, re- search has focused on off-line statistical analysis, learning the user behavior and on identifying most frequently visited structures. We propose and study on-line monitoring of web usage. We devise efficient real-time algorithms for identifying most visited sites and site-paths. We further provide advance warning when there is a potential denial of service attack. In our system named W3Live, we have implemented algorithms and live event warnings using LOGML and the graph library of the WWWPal suite.
W3Live is free software licensed under the GPL!
We have written a paper about W3Live for the Worldwide Web Computing Conference, 2004. Its acceptance/rejection is pending.
For testing W3Live, a tool for generating web log files (named genlog) has been built. Go here to download and learn about genlog.