Mozilla/Thunderbird/Netscape spam filter

I believe all three use the same code for this? Anyway, I’ve been running Thunderbird for a few months now, training the spam filter by diligently marking every email as either Junk or Not Junk, including all the stuff I imported from Outlook Express. But still, it lets 15-20% of spam through, including obvious stuff like emails with subjects mentioning “v1ag.ra”, the like of which must have been marked as Junk thousands of times by now.
Has anyone had more success with it? 15% is still a hell of a lot of spam to manually mark and delete.

I don’t use the Mozilla mail program, but I do use a Bayesian filter on Outlook. Unless is was exactly “v1ag.ra” marked as junk, and not a subtle variation, it won’t be recognized as a Spam indicator.

Perhaps you need more spam for it to learn. I’ve got a couple years worth in my spam folder, but in MS outlook format. Do you think it might be possible for me to send it to you, you import into your spam folder, and re-train? I’m not too familiar with Mozilla mail.

Thanks Revtim. But I think you’d have to send each message individually rather than as an Outlook archive, because I don’t have Outlook so I couldn’t import it. Is that what you meant? If you just forwarded them to me I could obviously receive them in Thunderbird, but two years worth? That would take you a while, wouldn’t it?

As an experiment I exported them, but I couldn’t find a way to do it to anything but a .pst file, which I suppose is what you meant as an Outlook archive.

It turns out it’s only about 1 years worth, but as a .pst file it’s something like 13 megabytes. I was able to compress it to 7 meg with ZIP and 5 meg with RAR, but still it’s format you probably cannot use.

It might be easy for me to mark them in groups, and forward them as attachments. I’d try now but I’m running Linux right now and don’t have access to me Outlook.

I just remembered while typing this that I have a program called “Outport” which I mainly used to extract contact and calendar info for use with Ximian Evolution (a Linux Outlook-type prog), and I think it can export emails too. I’ll take a look at that too once I’m back on Windows.

Mozilla’s spam filtering works like hell, but you’ve really got to train it with enough messages before it works extremely reliably. Also, I’ve found that sometimes it mysteriously unmarks messages that had been flagged as junk, which I think causes it to consider those “not spam.” I never spent much time investigating this, though, as I eventually switched to SpamProbe, which is a server-side Bayesian spam filter. But give Google about 2000 spams to learn from and it should be awfully accurate.

OK, I played around with it, and maybe we have some options (I have too much time on my hands, being unemployed…)

Option 1) My copy of Thunderbird imported the spam from Outlook directly, so now I have spam and spam.msf files which represent the spam folder in Thunderbird format. Perhaps all it would take would be to copy these files into your directory where the folders are stored (it worked that way with Thunderbird, I found with experimentation). A RAR archive of these files is 1.9 megabytes, easily emailable.

Option 2) Outport exported them into about 17 hundred txt files. Perhaps there is some way to import these into your spam folder, or maybe inserting them into a single email is enough. A RAR archive of these files is only about 900K.

Let me know if you want me to email these to you. Perhaps you can post your email address, spelled out in a way such that email-searching spambots don’t find it, like
joe at shmoe dot com

Or, you can add your email to your SDMB profile, and remove it after I send the files.

Yes, please could you send them to me (my email’s in my profile now). I did play with Thunderbird’s import, but when I specified Outlook it complained that it couldn’t find the application. Anyway, let’s try it. I appreciate you going to all this trouble.

No sweat. Like I said, too much time on my hands!

I sent them in RAR format, use winrar to uncompress
http://www.rarlab.com/download.htm