Registered Member
|
Hi there! This is my first post here, have mercy on me
I recently switched from Thunderbird to Kmail (Kontact) and really like it. Kontact is really cool! Well anyway i got a question concerning the spamfilters: I installed spamassassin and bogofilter and they work quite well, but I'm wondering if one of them would be sufficient. as far as i understood bogofilter uses Bayesian filtering while spamassassin uses Bayesian filtering and analyses the mail headers. bogofilter seems to be a lot faster than spamassassin. please correct me if i got something wrong. now for my question: does it make any sense to run both of them? if not, which one should I choose? now I'm running both of them at the same time, sometimes bogo will find the spam sometimes assassin will. thank you for you answers edit: oups! i posted into the wrong subforum could someone please move this to the Office & Productivity section. thanks! |
Manager
|
There's no need to run both, in fact I'd advise against it. I changed to bogofilter some time back because it was much faster than SA. Collect at least 100 samples of good mail - some list and some private, as well as collecting spam, so that you can train bogofilter. As spammers keep changing tactics you may need to run additional training from time to time. Every couple of weeks or so I run the command
bash /usr/share/bogofilter/contrib/contrib/trainbogo.sh -c -H /home/anne/Maildir/.INBOX.bogotrain_ham/cur/ -S /home/anne/Maildir/.INBOX.bogotrain_spam/cur/ (all on one line). Note that bogotrain_ham is the folder-name where I move 'good' messages if they turn up in the unsure folder.
annew, proud to be a member of KDE forums since 2008-Oct and a KDE user since 2002.
Join us on http://userbase.kde.org |
Registered Member
|
thank you for you hints annew! now i only got to decide which one to use...
so the command you wrote tells bogofilter that all the messages i kept in my inbox folder are ham, right? nice, i didn't think of that before - that already makes a lot of training! i read somewhere that you should keep your spam messages in order to be able to train your spamfilter later, but i think that's only valid if you switch spam filters, right? isn't spamassassin better at catching spam? from what i read it is "much more sophisticated". or is it just better in the beginning, and after training it doesn't matter any more? thanks for your help |
Manager
|
I guess you could use your inbox, but if you sort mail, as I do, it's worth copying samples of all your normal mail. Besides, setting up bogofilter should create some holding folders for you. (Currently I'm using it in server mode, rather than on my laptop, but I'm fairly sure that it does.)
Messages that bogofilter puts into the spam folder can be deleted immediately, but it will put some into the Unsure folder - fewer as time goes on, but with spammers changing tactics regularly there will always be some new types. You move any of those that are spam into the spam folder, so that the training learns about them. Similarly you copy or move the good messages from the Unsure folder into the ham training folder.
I've never seen anything convincing that either is 'better at catching spam'. I found bogofilter less obtrusive and easier to train than SA. Others are perfectly happy with SA. In my opinion the important thing is to make a decision then train the one you choose. I believe either will do a good job.
You're welcome
annew, proud to be a member of KDE forums since 2008-Oct and a KDE user since 2002.
Join us on http://userbase.kde.org |
Registered Member
|
thanks a lot annew!
in kmail (kontact) i had to set up all the folders manually. there was no spam folder and no unsure folder. i activated bogofilter and spamassassin through the anti-spam wizard which created filters and asked me where to put spam. i was never asked where to put unsure mails. did i do something wrong? should i have done it another way? training through the spam and ham buttons in kmail works just as good as through the command line commands, right? i can't decide on which filter to use - most of the articles comparing the two are so old that SA didn't have bayesian filtering at that time. the way i see it: bogofilter: +fast | -more training necessary spamassassin: +less training necessary (rules) | -slow of course this is a simplification, but for a normal user who just wants to sort out his mail with the spam and ham buttons it should be true. can anybody give me some advice on this, please. is that correct? did i miss something? |
Registered Member
|
may i ask another question:
is there a way to whitelist the emailaddresses in my kaddressbook? or is there a way to automatically whitelist all email addresses that i send emails to? and another: shouldn't the spamassassin command in the automatically created filter use "sa-learn -L --spam --sync" instead of the --no-sync switch? i thought a sync is needed in order to make SA take the training into account when scanning new messages. |
Manager
|
Probably not. It's a long time since I used bogofilter at client level, and my memory is not reliable.
Probably - again, memory problems
Bayesian filtering depends on learning what you consider to be ham or spam. It's as simple as that. If you don't train, you will not get good results. I have friends that welcome mail that I immediately delete. For them it's ham, for me it's spam. If you want to know how it works, read http://www.bgl.nu/bogofilter/ . There's a useful FAQ at http://bogofilter.sourceforge.net/faq.shtml , and if you decide to go with bogofilter I would recommend joining the mailing list - it's low traffic and very helpful. http://www.bogofilter.org/mailman/listinfo/bogofilter
annew, proud to be a member of KDE forums since 2008-Oct and a KDE user since 2002.
Join us on http://userbase.kde.org |
Registered users: abc72656, Bing [Bot], Google [Bot], lockheed, Yahoo [Bot]