> The caveat with this approach is that you would need to gather a > crapload of logs This is without doubt true >, and have people somehow denote whether each entry is > good, bad or just ugly. With email you can mark a whole message, but > with logs you'd have to do this line by line, unfortunately. This is not necessarily true. While I'm not an expert with genetic algorithm classifiers AND I think that these would need human interaction, neural algorithms can classify objects with very little need of guidance. In addition, you can train a neural network and then "monitor" its guesses. The more time and energy you can spend "correcting" its guesses, the better it develops. Stefano "Raistlin" Zanero System Administrator Gioco.Net public PGP key block at http://gioco.net/pgpkeys _______________________________________________ LogAnalysis mailing list LogAnalysisat_private https://lists.shmoo.com/mailman/listinfo/loganalysis
This archive was generated by hypermail 2b30 : Thu Aug 22 2002 - 11:46:41 PDT