Re: [Mimedefang] LZW, Gifs, and fingerprinting stock spams
Am Mittwoch November 1 2006 15:48 schrieb Dave Helton: > I have had very good success with this plugin for SA. > > http://wiki.apache.org/spamassassin/FuzzyOcrPlugin > > config file allows you to add/remove keywords, and the program > keeps a hash of known images so that they are not ocr'ed again. I am using it as well since some days. Works quite amazing, even with the provided short but reasonable wordlist. One fine thing also is, that you can tune it not to scan mails which already got a spamassassin score over a certain treshold. So you won't have to do an OCR on mails with images spamassassin already marked as spam :-) Jürgen ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
Re: [Mimedefang] LZW, Gifs, and fingerprinting stock spams
Jason Bertoch [Electronet] wrote: > Where might I find more information on the RPTN system? The searchable > archives don't seem to be working at the moment. We have a white paper (mostly marketing-oriented, alas): http://www.roaringpenguin.com/files/images/resources_files/White-Paper-RPTN.pdf (Sorry for the ridiculous URL; it predates our move to Drupal.) If you want a more technical white paper, please e-mail me off-list. RPTN is only available to CanIt customers, though we continue to mull over the possibility of making a SpamAssassin plugin and selling RPTN subscriptions. We haven't had enough people say "Yes, I'm willing to pay!" to prod us to do it yet, however. Regards, David. ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
RE: [Mimedefang] LZW, Gifs, and fingerprinting stock spams
I have had very good success with this plugin for SA. http://wiki.apache.org/spamassassin/FuzzyOcrPlugin config file allows you to add/remove keywords, and the program keeps a hash of known images so that they are not ocr'ed again. this plugin also understands animated gifs, something I've seen recently. I do not know how well it handles compressed images. needs testing. HTH -Dave Hughes Network Technologies -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Philip Prindeville Sent: Tuesday, October 31, 2006 10:26 PM To: mimedefang@lists.roaringpenguin.com Subject: [Mimedefang] LZW, Gifs, and fingerprinting stock spams I'm trying to do some stochastic analysis of stock spams and figure out if there's a common fingerprint that can be used to identify them... But first, I'm bumping up against some Perl issues. Seems that there aren't many modules out there that help deconstruct Gif formats. I'm using Image::Info::GIF, but need to decompress the compressed data portion. I tried to take the data and pass it to Compress::LZW directly, but most GIF's (at least for stocks, which don't use many colors) use 4, 6, or 8 bit codesizes. Unfortunately, Compress::LZW only handles 12 or 16 bits... Anyone familiar enough with either GIF formats or how to decompress the data to offer a leg up? Thanks, -Philip ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
RE: [Mimedefang] LZW, Gifs, and fingerprinting stock spams
> -Original Message- > Not to blow our own horn too much, but... > > We've had pretty good luck with our RPTN system. > > Regards, > > David. David, Where might I find more information on the RPTN system? The searchable archives don't seem to be working at the moment. Jason A. Bertoch Network Administrator [EMAIL PROTECTED] ElectroNet Intermedia Consulting 3411 Capital Medical Blvd. Tallahassee, FL 32308 (V) 850.222.0229 (F) 850.222.8771 ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
Re: [Mimedefang] LZW, Gifs, and fingerprinting stock spams
Kevin A. McGrail wrote: > I will also say that the stock image spams have been very effective at > thwarting traditional anti-spam techniques. Not to blow our own horn too much, but... We've had pretty good luck with our RPTN system. It's a shared Bayes database. A couple of hundred of our customer sites submit "votes" with word-counts to add to a large shared Bayes database, which we update and redistribute every night. The non-image junk in most of those spams usually scores very high in our Bayes implementation. Our current RPTN database contains words and word-pairs from 451,506 spams and 224,318 hams, for a total of just under 7 million tokens. Regards, David. ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
Re: [Mimedefang] LZW, Gifs, and fingerprinting stock spams
I'm trying to do some stochastic analysis of stock spams and figure out if there's a common fingerprint that can be used to identify them... Philip: Have you looked at Dallas' ImageInfo.pm? See http://www.rulesemporium.com/plugins.htm. It's a great place to start building image rules. However, I think you are barking up the wrong tree. The spams have been very effective at being randomized. I will also say that the stock image spams have been very effective at thwarting traditional anti-spam techniques. It's been an ebb and flow battle for weeks (months?) with them. But I am happy to say that if you use MIMEDefang, I've been VERY pleased with the results of the AOL-esque reverse DNS test that I wrote a few weeks ago. I'm continuing to tweak it but I just put the latest version up in http://www.peregrinehw.com/downloads/MIMEDefang/mimedefang-filter-KAM. I use this in conjunction with my ruleset which only SCORES the emails. I do NOT use this technique to block email like AOL. This may change. The rules are in http://www.peregrinehw.com/downloads/SpamAssassin/contrib/KAM.cf Good Luck! Regards, KAM ___ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
