On Friday 27 October 2006 14:37, sophana wrote: > Frank Barknecht a écrit : > > I don't know if it is a coincidence. Spam started in the same time a > guy was complaining about sourceforge mailling lists, and proposed to > use google groups.
I wouldn't put this behind him considering what other people say about him and how he tried to manipulate the community and the projects. > It is reaching abnormal levels. > These new spams have no url inside. They are difficult to catch. Use a Bayesian spam filter. bogofilter is one for example, which I can recommend since I have used it for more than 2 years with great success. With it I never got a false positive and the number of spam messages that still escape and end up in my mailbox is less than 2 per week, but most of the time not even that, and the Bayesian spam filter can be trained to learn and reclassify such false negatives as spam and capture them the next time such similar messages arrive. For those who do not know, a Bayesian spam filter is one that can be trained on a per user basis and learns from corpus of spam and non-spam messages that are specific for that user. It will assign certain probabilities to each token it finds in the training emails and then will evaluate new email based on the probabilities of the tokens (words) that it contains into one of spam or non-spam. Even more new messages can not only be classified, but can also be used for further training the filter, which makes the Bayesian filter adapt to new conditions and evolve. In this way it becomes very efficient because it will be adapted to the user's internet behavioral patterns and will be much more efficient than a generic spam filter that uses traditional methods. As an ideea of how efficient such a filter it is, in the last 20 days, my filter has retained over 1300 spam messages with no false positives (in fact I never had a single false positive with it since installed) and missed less than 5-7 spams which ended up in my inbox and I had to retrain the filter to learn they were actually spam. In addition it is capable to catch spam messages that are sent not as text or html, but with the message put in an image attachment that contains the actual message. This capability is not present to my knowledge in any other spam filtering system which uses traditional methods. P.S. Nowadays many mail user agents already implement Bayesian filtering, even though they may call it something else. Their filters can be used instead of a standalone filter like bogofilter, however I have no experience with them to say how well they perform. -- Dan ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ sqlobject-discuss mailing list sqlobject-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sqlobject-discuss
