|
I am interested in the problem of analyzing voluminous logs and correlating that with user-submitted errors. There was a good article on that not long ago -- maybe in the CACM? Can anyone help? P.S. I'd add the tag "Bayesian" but my rep isn't high enough -- if someone appropriately worthy were to edit it in, it might be helpful... |
|||
|
|
|
Interesting... Was it "Symptom-based problem determination using log data abstraction" |
|||
|
|
|
Try using either Bayesian Belief Networks or Naive Bayesian Classification as search terms. Google Scholar may turn up the specific article you are looking for. The SpamBayes spam detector uses NBC, and a writeup can be found at: http://www.paulgraham.com/spam.html Wikipedia also has a decent write-up on Bayes theorem and Bayesian filters. |
|||
|
|
