Has anyone used this?
http://software.newsforge.com/print.pl?sid=06/03/24/1728247
Comments? Alternatives?
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
Frank Cox wrote:
Has anyone used this?
http://software.newsforge.com/print.pl?sid=06/03/24/1728247
Comments? Alternatives?
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
I use it with Outlook/Exchange at work, and I really like it. I've installed it on a couple of other people's machines and they say that it's a huge time saver.
I use Mozilla Thunderbird at home, which has built-in bayesian filtering similar to that of SpamBayes.
I find both to be very fast and have nearly perfect accuracy after a couple of weeks of training.
Frank Cox wrote:
Has anyone used this? [spambayes]
Yes, I'm using it. It's coping pretty well with around 2500 spams a month (that's not a request for more...)
I use procmail to pipe mail through it, and mutt to pipe mis-categorised e-mail through the relearning scripts.
Comments? Alternatives?
Personally, I find it works better for me than SpamAssassin. But the best review of antispam software for Linux that I know of is found at http://lwn.net/Articles/172491/ with a follow-up at http://lwn.net/Articles/173910/ .
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
I seem to have generated an RPM from source, without taking notes of how I did it. That was careless.
There is the option to unpack the source archive and run python setup.py bdist_rpm which is supposed to create an RPM. But it doesn't seem to do that on FC5, at least with 1.0.4.
James.
On Mon, 2006-05-08 at 16:58 -0600, Frank Cox wrote:
Has anyone used this?
http://software.newsforge.com/print.pl?sid=06/03/24/1728247
Comments? Alternatives?
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
-- MELVILLE THEATRE ~ Melville Sask ~ http://theatre.sasktelwebsite.net
I used it more that a year ago and at that time it was designed mostly for Windows use and the Linux part that implemented IMAP and POP did not work properly. Since someone says they use it under Linux maybe it improved.
But and no one asked it is based on a mistaken assumption that it is useful to have mail identified in addition to spam and ham as unknown. I don't think they call it unknown but that is the purpose. I can't go into the whole argument but to me this tri-classification is not only unnecessary but more trouble to deal with.
-- ======================================================================= Whenever someone tells you to take their advice, you can be pretty sure that they're not using it. ======================================================================= Aaron Konstam telephone: (210) 656-0355 e-mail: akonstam@sbcglobal.net
Aaron Konstam wrote:
But and no one asked it is based on a mistaken assumption that it is useful to have mail identified in addition to spam and ham as unknown. I don't think they call it unknown but that is the purpose. I can't go into the whole argument but to me this tri-classification is not only unnecessary but more trouble to deal with.
I, on the other hand, find it excellent. The program has the honesty to ask for help when it gets stuck.
What we'd all *like*, ideally, is an antispam program that could identify what we considered to be spam with 100% accuracy.
That turns out to be practically impossible. There will be e-mails that are border-line, e-mails that "look" like spam but are actually wanted (false positives), e-mails that "look" wanted but are really spam (false negatives), and ones that are pretty impossible to automatically classify.
The "unsure" category provides a place for the border-line and the Hard Cases, and massively reduces false positives and negatives (they usually end up in "unsure", instead of "good" or "spam").
So you get "good" folders that you can be pretty certain are good. You get "spam" folders that *very* *very* rarely have good e-mail in them. And you have a folder *marked* "dodgy". So you can quickly deal with it when you want, with the expectation that it's probably spam.
Of course, since the program is based on a modified Bayesian algorithm, you are expected to train on errors. You are expected to put a little bit of time into helping the program. "Unsure" is simply where e-mails go if the program needs to be trained on them.
James.
On Wednesday 10 May 2006 13:12, James Wilkinson wrote:
What we'd all *like*, ideally, is an antispam program that could identify what we considered to be spam with 100% accuracy.
That turns out to be practically impossible. There will be e-mails that are border-line, e-mails that "look" like spam but are actually wanted (false positives), e-mails that "look" wanted but are really spam (false negatives), and ones that are pretty impossible to automatically classify.
One should also bear in mind that 'one man's meat is another man's poison'. That alone makes accuracy impossible until any filtering program learns what is acceptable to you.
Anne
On Wed, May 10, 2006 at 01:25:57PM +0100, Anne Wilson wrote:
One should also bear in mind that 'one man's meat is another man's poison'.
ITYM mead. OTOH, ham, spam, hmm...
That alone makes accuracy impossible until any filtering program learns what is acceptable to you.
On Wed, 2006-05-10 at 13:12 +0100, James Wilkinson wrote:
Aaron Konstam wrote:
But and no one asked it is based on a mistaken assumption that it is useful to have mail identified in addition to spam and ham as unknown. I don't think they call it unknown but that is the purpose. I can't go into the whole argument but to me this tri-classification is not only unnecessary but more trouble to deal with.
I, on the other hand, find it excellent. The program has the honesty to ask for help when it gets stuck.
What we'd all *like*, ideally, is an antispam program that could identify what we considered to be spam with 100% accuracy.
That turns out to be practically impossible. There will be e-mails that are border-line, e-mails that "look" like spam but are actually wanted (false positives), e-mails that "look" wanted but are really spam (false negatives), and ones that are pretty impossible to automatically classify.
The "unsure" category provides a place for the border-line and the Hard Cases, and massively reduces false positives and negatives (they usually end up in "unsure", instead of "good" or "spam").
So you get "good" folders that you can be pretty certain are good. You get "spam" folders that *very* *very* rarely have good e-mail in them. And you have a folder *marked* "dodgy". So you can quickly deal with it when you want, with the expectation that it's probably spam.
Of course, since the program is based on a modified Bayesian algorithm, you are expected to train on errors. You are expected to put a little bit of time into helping the program. "Unsure" is simply where e-mails go if the program needs to be trained on them.
James.
All this is too technical a matter to deal with here. Training on the usure is not different then training on the ham which should be spam and the spam that should be ham. In any case you can't be over confident with spambayes. You still have to check for spam that is misclassified and ham which is misclassified. So now you have three streams to check rather than two. That to me is an extra pain.
On Wed May 10 2006 11:07 am, Aaron Konstam wrote:
You still have to check for spam that is misclassified and ham which is misclassified. So now you have three streams to check rather than two. That to me is an extra pain.
Sorry, but this is just not my experience. I happen to direct unsures to my inbox in Linux, and to a separate folder in Windows - either way is perfectly convenient. I never get false positives on any of the seven machines I run Spambayes on after a few days training. I just checked 400 spam in my mailbox to verify this before writing this - every once in a blue moon, I do that. There were no false positives. Training on spam that gets classified as unsure or as ham, is an ongoing process. Spammers constantly change their tactics and methods, so no method is ever going to catch it all. My worst trained machine, actually my Windows laptop running Outlook, lets the most crud through, up to 30 or so a day, that's out of over 400 spam messages that I get daily; I have one Linux box at work, my brand new FC5 machine, that only let three spams get through this same day - I download all my mail to multiple machines for various reasons, but it gives me a good metric to compare things...
On Mon May 8 2006 6:58 pm, Frank Cox wrote:
Has anyone used this?
http://software.newsforge.com/print.pl?sid=06/03/24/1728247
Comments? Alternatives?
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
I've used Spambayes on both Linux and Windows boxes for over a year now. It's a normal part of my installation of a new distro, these days. Spambayes can be detected by Kmail's anti-spam configuration wizard these days, and that's quite nice.
Installation is extremely simple. Unpack the tarball anywhere you prefer, and run 'python set_up.py' as root from the folder in which you unpacked the tarball - that installs everything (you can delete the folder after that). Then, you want to start it at boot-time. I use
'nohup python /usr/bin/sb_server.py &'
as a line in rc.local, or better, lately I've been using the above as a script file in my /.kde/startup folder - that invokes the spambayes server as user instead of root - it works just fine that way.
Then, either reboot, or start the spambayes server manually and open Firefox or any browser and put 'http://localhost:8880/' in your address bar and click on 'configure' - I usually just worry about the topmost two fields, leaving everything else in default.
In the topmost field, you enter the url of your pop mail servers, separated by commas. In the ports field, just pick random ports to proxy on, separated by commas ( I use 1110,1111,1112, for example)
After that, you have to configure your email client to use localhost for each pop server, and the corresponding port you entered in the second field, above.
Spambayes should start piping your mail through its filters after that, if you've got everything right. It will add a classification line to each mail header, labeling it as 'spam', 'unsure', or 'ham' (look at your header lines, and you'll this extra line as
'X-Spambayes-Classification: ham'
- use your email client filters to direct the mail to wherever you'd like it to end up according to these classifications. If you use kmail, as already mentioned, much of the email client header classification filtering gets done for you if you go through the anti-spam wizard...
Hope that helps - there's more that can be configured, and I may have left something out, but that should be enough to get you going. You'll have to read the faq on the spambayes site to learn about training, but that's fairly straightforward. After several days, you'll find Spambayes removing well over 90% of your spam with almost no false positives. I get over 400 spams per day because of multiple published email addresses, but, Spambayes makes that completely manageable. One further comment, while it's true that Thunderbird uses bayesian filtering, and it's pretty good, it's not the same as Spambayes, which in my opinion is an even better implementation of bayesian principles - there are some pretty good docs on the net if you google for them, to explain the principles behind bayesian filtering - Spambayes is one of several implementations of the principles involved.
On Tue, 9 May 2006, Claude Jones wrote:
On Mon May 8 2006 6:58 pm, Frank Cox wrote:
Has anyone used this?
http://software.newsforge.com/print.pl?sid=06/03/24/1728247
Comments? Alternatives?
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
I've used Spambayes on both Linux and Windows boxes for over a year now. It's a normal part of my installation of a new distro, these days. Spambayes can be detected by Kmail's anti-spam configuration wizard these days, and that's quite nice.
Installation is extremely simple. Unpack the tarball anywhere you prefer, and run 'python set_up.py' as root from the folder in which you unpacked the tarball - that installs everything (you can delete the folder after that). Then, you want to start it at boot-time. I use
'nohup python /usr/bin/sb_server.py &'
as a line in rc.local, or better, lately I've been using the above as a script file in my /.kde/startup folder - that invokes the spambayes server as user instead of root - it works just fine that way.
Then, either reboot, or start the spambayes server manually and open Firefox or any browser and put 'http://localhost:8880/' in your address bar and click on 'configure' - I usually just worry about the topmost two fields, leaving everything else in default.
In the topmost field, you enter the url of your pop mail servers, separated by commas. In the ports field, just pick random ports to proxy on, separated by commas ( I use 1110,1111,1112, for example)
After that, you have to configure your email client to use localhost for each pop server, and the corresponding port you entered in the second field, above.
Spambayes should start piping your mail through its filters after that, if you've got everything right. It will add a classification line to each mail header, labeling it as 'spam', 'unsure', or 'ham' (look at your header lines, and you'll this extra line as
'X-Spambayes-Classification: ham'
- use your email client filters to direct the mail to wherever you'd like it
to end up according to these classifications. If you use kmail, as already mentioned, much of the email client header classification filtering gets done for you if you go through the anti-spam wizard...
Hope that helps - there's more that can be configured, and I may have left something out, but that should be enough to get you going. You'll have to read the faq on the spambayes site to learn about training, but that's fairly straightforward. After several days, you'll find Spambayes removing well over 90% of your spam with almost no false positives. I get over 400 spams per day because of multiple published email addresses, but, Spambayes makes that completely manageable. One further comment, while it's true that Thunderbird uses bayesian filtering, and it's pretty good, it's not the same as Spambayes, which in my opinion is an even better implementation of bayesian principles - there are some pretty good docs on the net if you google for them, to explain the principles behind bayesian filtering - Spambayes is one of several implementations of the principles involved.
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
On Tue May 9 2006 1:38 pm, Justin Zygmont wrote:
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
I've never taken the time to learn how to configure spamassassin properly, so I'm not a good judge. With default install settings, I did often note that when I had both spamassassin and spambayes as filters, that many messages were let through as low percentage by spamassassin, but were trapped by Spambayes. I don't know if that's much of a test, however.
On Tue, 9 May 2006, Claude Jones wrote:
On Tue May 9 2006 1:38 pm, Justin Zygmont wrote:
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
I've never taken the time to learn how to configure spamassassin properly, so I'm not a good judge. With default install settings, I did often note that when I had both spamassassin and spambayes as filters, that many messages were let through as low percentage by spamassassin, but were trapped by Spambayes. I don't know if that's much of a test, however.
spamassassin is easy, I just have procmail pipe it to spamd like this:
:0fw | spamc
:0 H * ^X-Spam-Status: Yes /dev/null
if you still get spam, just view the email headers and se what the score was. I was suprised to find some major spam messages coming through with a 0.0 score. Its a tough problem to beat sometimes..
On Tue May 9 2006 4:52 pm, Justin Zygmont wrote:
spamassassin is easy, I just have procmail pipe it to spamd
like this:
:0fw : | spamc | :0 H
- ^X-Spam-Status: Yes
/dev/null
if you still get spam, just view the email headers and se what the score was. I was suprised to find some major spam messages coming through with a 0.0 score. Its a tough problem to beat sometimes.
Let me rephrase this, since it seems to have raised Aaron's hackles. I'm not condeming spamassassin. I'm saying I am not competent to judge spamassassin, because I haven't taken the time to learn how it works, or how to adjust the parameters by which if judges what is/isn't spam.
The simple point I made in my first message was that the default configuration set up by kmail to pipe mail through spamassassin was letting more spam through then filtering through spambayes was - that's it - just a comparison of a default, unaltered, install. No judgement intended about spamassassin - if my post's wording implied otherwise, then, my words didn't convey what I was trying to say ---
On Tuesday 09 May 2006 23:09, Claude Jones wrote:
Let me rephrase this, since it seems to have raised Aaron's hackles. I'm not condeming spamassassin. I'm saying I am not competent to judge spamassassin, because I haven't taken the time to learn how it works, or how to adjust the parameters by which if judges what is/isn't spam.
The simple point I made in my first message was that the default configuration set up by kmail to pipe mail through spamassassin was letting more spam through then filtering through spambayes was - that's it - just a comparison of a default, unaltered, install.
I haven't been following this thread, so aplogies now if I tread on sore toes.
I, too, didn't have much time to spare for configuration, though I did look at it briefly. I did find, though, that one simple change to kmail made a huge difference. In the filter configuration for classifying as spam, I changed it to the following: sa-learn -L --spam --sync
The difference was immediately obvious. I understand that this is not the default setting because it is slower than the default, but it hasn't caused me any problems. Of course, manually applying the filter to missed spam improves performance also.
Anne
On Wed May 10 2006 4:41 am, Anne Wilson wrote:
In the filter configuration for classifying as spam, I changed it to the following: sa-learn -L --spam --sync
You're not treading on my toes. Where did you put that entry, above?
The difference was immediately obvious. I understand that this is not the default setting because it is slower than the default, but it hasn't caused me any problems. Of course, manually applying the filter to missed spam improves performance also.
I did manually train spamassassin during one period when I tried to use it because I was having configuration problems with Spambayes. There was not a dramatic effect - it seemed like I'd classify something as spam, and the same message would keep getting through as ham - I made a mini-attempt to learn about configuring spamassassin, but was in a hurry and didn't readily find answers. Shortly after, I sorted out my Spambayes issue, so I stopped using spamassassin. I've read repeatedly that spamassassin works well, so I chalked up my experience to inexperience, and poor configuration. Someone on this list whose views I respect once said that the best anti-spam strategy was a combination of spamassassin and spambayes. Now, having got my curiosity up, I'm discovering some man pages I didn't find previously, and see that spamassassin has a bayes-ian component to its filtering. Looks like its time to revisit the subject, for me.
On Wednesday 10 May 2006 14:19, Claude Jones wrote:
On Wed May 10 2006 4:41 am, Anne Wilson wrote:
In the filter configuration for classifying as spam, I changed it to the following: sa-learn -L --spam --sync
You're not treading on my toes. Where did you put that entry, above?
Settings > Configure Filters > Classify as Spam > Execute Command
You'll find that you only need to change the ending of the command. I read on the kde-pim list that it was considered too slow for accounts with huge amounts of mail, but it's not caused me any problems.
Anne
On Wed May 10 2006 11:49 am, Anne Wilson wrote:
Settings > Configure Filters > Classify as Spam > Execute Command
You'll find that you only need to change the ending of the command. I read on the kde-pim list that it was considered too slow for accounts with huge amounts of mail, but it's not caused me any problems.
Thanks, found it - so now I have both Spamassassin and Spambayes running on one of my machines. Now, I just have to figure out how to train Spamassassin...
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Claude Jones wrote:
Thanks, found it - so now I have both Spamassassin and Spambayes running on one of my machines. Now, I just have to figure out how to train Spamassassin...
It can be as simple as:
$ sa-learn --spam /path/to/spam/folder $ sa-learn --ham /path/to/ham/folder
Obviously, you'll want to automate and integrate that into your MUA(s), but that shouldn't be too tough. And you already have a nice body of ham and spam to train on so it should be interesting to see which performs better.
One tweak I made was to bump the scores for messages that matched various levels of Bayes scores. Once trained it is highly accurate so I don't really need much more than a BAYES_99 hit from Spamassassin to comfortably drop mail to my spam folder. The defaults are a little more conservative (as they should be).
- -- Todd OpenPGP -> KeyID: 0xD654075A | URL: www.pobox.com/~tmz/pgp ====================================================================== I believe in the noble, aristocratic art of doing absolutely nothing. And someday, I hope to be in a position where I can do even less.
On Tue, 2006-05-09 at 13:52 -0700, Justin Zygmont wrote:
On Tue, 9 May 2006, Claude Jones wrote:
On Tue May 9 2006 1:38 pm, Justin Zygmont wrote:
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
I've never taken the time to learn how to configure spamassassin properly, so I'm not a good judge. With default install settings, I did often note that when I had both spamassassin and spambayes as filters, that many messages were let through as low percentage by spamassassin, but were trapped by Spambayes. I don't know if that's much of a test, however.
spamassassin is easy, I just have procmail pipe it to spamd like this:
:0fw | spamc
:0 H
- ^X-Spam-Status: Yes
/dev/null
if you still get spam, just view the email headers and se what the score was. I was suprised to find some major spam messages coming through with a 0.0 score. Its a tough problem to beat sometimes..
That is where training come in, What you describe should not be happening. Do you train you classifier?
On Wed, 10 May 2006, Aaron Konstam wrote:
On Tue, 2006-05-09 at 13:52 -0700, Justin Zygmont wrote:
On Tue, 9 May 2006, Claude Jones wrote:
On Tue May 9 2006 1:38 pm, Justin Zygmont wrote:
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
I've never taken the time to learn how to configure spamassassin properly, so I'm not a good judge. With default install settings, I did often note that when I had both spamassassin and spambayes as filters, that many messages were let through as low percentage by spamassassin, but were trapped by Spambayes. I don't know if that's much of a test, however.
spamassassin is easy, I just have procmail pipe it to spamd like this:
:0fw | spamc
:0 H
- ^X-Spam-Status: Yes
/dev/null
if you still get spam, just view the email headers and se what the score was. I was suprised to find some major spam messages coming through with a 0.0 score. Its a tough problem to beat sometimes..
That is where training come in, What you describe should not be happening. Do you train you classifier?
that just crossed my mind as you mentioned it, please let me know what you do about this.
On Tue, 2006-05-09 at 15:09 -0400, Claude Jones wrote:
On Tue May 9 2006 1:38 pm, Justin Zygmont wrote:
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
I've never taken the time to learn how to configure spamassassin properly, so I'm not a good judge. With default install settings, I did often note that when I had both spamassassin and spambayes as filters, that many messages were let through as low percentage by spamassassin, but were trapped by Spambayes. I don't know if that's much of a test, however. -- Claude Jones Bluemont, VA, USA
I try to be pleasant and relate positively to what people say on the list. But when someone says that he have never taken the trouble to learn how to configure a program and then says it does not work well for him, I am speechless.
On Tue, 2006-05-09 at 15:09 -0400, Claude Jones wrote:
On Tue May 9 2006 1:38 pm, Justin Zygmont wrote:
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
I've never taken the time to learn how to configure spamassassin properly, so I'm not a good judge. With default install settings, I did often note that when I had both spamassassin and spambayes as filters, that many messages were let through as low percentage by spamassassin, but were trapped by Spambayes. I don't know if that's much of a test, however. -- Claude Jones Bluemont, VA, USA
I try to be pleasant and relate positively to what people say on the list. But when someone says that he have never taken the trouble to learn how to configure a program and then says it does not work well for him, I am speechless.
-- Aaron Konstam akonstam@sbcglobal.net
That's not what I read there at all. He said, in effect, that the default setting seemed to let spam through spamassassin but not for Spambayes. He also indicated that this was probably not a good test.
I saw no indication of a value judgement in Claude's response, only a hint that, in his experience, you probably ought not to trust the default settings. settings.
On Tue May 9 2006 5:48 pm, Aaron Konstam wrote:
I try to be pleasant and relate positively to what people say on the list. But when someone says that he have never taken the trouble to learn how to configure a program and then says it does not work well for him, I am speechless.
Aaron, slow down. You're letting your fingers race in front of you. Read again what I said - I qualified my statement twice using the phrase "I am no judge" in one instance, and "I don't know if that's much of a test, however" in another. I think it should have been clear that I was saying that I wasn't competent to judge the issue - what else could I have meant?
On Tue, 2006-05-09 at 10:38 -0700, Justin Zygmont wrote:
On Tue, 9 May 2006, Claude Jones wrote:
On Mon May 8 2006 6:58 pm, Frank Cox wrote:
Has anyone used this?
http://software.newsforge.com/print.pl?sid=06/03/24/1728247
Comments? Alternatives?
It looks very cool. Are there any up-to-date rpm packages around anywhere? The only ones that I could find are substantially out-of-date.
I've used Spambayes on both Linux and Windows boxes for over a year now. It's a normal part of my installation of a new distro, these days. Spambayes can be detected by Kmail's anti-spam configuration wizard these days, and that's quite nice.
Installation is extremely simple. Unpack the tarball anywhere you prefer, and run 'python set_up.py' as root from the folder in which you unpacked the tarball - that installs everything (you can delete the folder after that). Then, you want to start it at boot-time. I use
'nohup python /usr/bin/sb_server.py &'
as a line in rc.local, or better, lately I've been using the above as a script file in my /.kde/startup folder - that invokes the spambayes server as user instead of root - it works just fine that way.
Then, either reboot, or start the spambayes server manually and open Firefox or any browser and put 'http://localhost:8880/' in your address bar and click on 'configure' - I usually just worry about the topmost two fields, leaving everything else in default.
In the topmost field, you enter the url of your pop mail servers, separated by commas. In the ports field, just pick random ports to proxy on, separated by commas ( I use 1110,1111,1112, for example)
After that, you have to configure your email client to use localhost for each pop server, and the corresponding port you entered in the second field, above.
Spambayes should start piping your mail through its filters after that, if you've got everything right. It will add a classification line to each mail header, labeling it as 'spam', 'unsure', or 'ham' (look at your header lines, and you'll this extra line as
'X-Spambayes-Classification: ham'
- use your email client filters to direct the mail to wherever you'd like it
to end up according to these classifications. If you use kmail, as already mentioned, much of the email client header classification filtering gets done for you if you go through the anti-spam wizard...
Hope that helps - there's more that can be configured, and I may have left something out, but that should be enough to get you going. You'll have to read the faq on the spambayes site to learn about training, but that's fairly straightforward. After several days, you'll find Spambayes removing well over 90% of your spam with almost no false positives. I get over 400 spams per day because of multiple published email addresses, but, Spambayes makes that completely manageable. One further comment, while it's true that Thunderbird uses bayesian filtering, and it's pretty good, it's not the same as Spambayes, which in my opinion is an even better implementation of bayesian principles - there are some pretty good docs on the net if you google for them, to explain the principles behind bayesian filtering - Spambayes is one of several implementations of the principles involved.
Are you able to tell if its much more effective than spamassassin? I get spam coming through spamassassin with a 0.0 score!
Are you saying that mail that is assigned a 0 spam score and should be spam are evaluated incorrectly? Then something is wrong with your configuration. My objection to spambayes is theit assertion that classifying mail into three catagories spam, ham and unknown does not make for better spam detection. In my opinion the papers they have defending this position are just plain wrong. But spambayes does work to cull out spam but in my experience it is no better than spamasssasin.
On Tue May 9 2006 5:44 pm, Aaron Konstam wrote:
My objection to spambayes is theit assertion that classifying mail into three catagories spam, ham and unknown does not make for better spam detection.
Could you explain? I've read their site many times, and have never come across this assertion. I just looked at the faq, and I see no such claim. They talk quite matter-of-factly about dealing with "unsure" classified messages, but there's nothing said about using such classification "does not make for better spam detection". Maybe I've missed something...
-
Aaron Konstam -
Anne Wilson -
Claude Jones -
Eugen Leitl -
Frank Cox -
James Wilkinson -
Justin Zygmont -
Peter Gordon -
Richard England -
Todd Zullinger