Re: [SAtalk] RD Message body/subject filter help

[Fred I-IS.COM]

Title: RE: [SAtalk] RD Message body/subject filter help

I created a program in VB to do it for me, it was 2 nested loops, checking AA,AB...ZY,ZZ and if not found in the dictionary, it was a candidate.   Some others that were removed:  VB, MX, XM  (they could appear in normal e-mail). CJ failed for Discjockey   I tried to find pairs that should never show up in normal e-mail.   The program went through all possible combo's and then I tried the same test with 3 seperate dictionaries and compared the results.   A few others only showed up in 1 or 2 words but I as few false positives as possible.   ii can be used in roman numerals, also in sur-names. (ASCII) (antiimperialism) (antiinflammatory) are those words? Frederic Tarasevicius Internet Information Services, Inc.   ----- Original Message ----- From: Chris Santerre To: 'Fred I-IS.COM' ; Spamassassin-Talk (E-mail) Sent: Thursday, August 07, 2003 3:35 PM Subject: RE: [SAtalk] RD Message body/subject filter help SWEET! How about checking on double letters? like "ii" ?   Did you reverse some of them? I see 'jc' but not 'cj'. 'cj' shows up in a word maybe??   It gets even better with 3 letter combos. -----Original Message----- From: Fred I-IS.COM [mailto:[EMAIL PROTECTED] Sent: Thursday, August 07, 2003 3:18 PM To: Spamassassin-Talk (E-mail) Subject: Re: [SAtalk] RD Message body/subject filter help I did a quick check and came up with this list which does not appear to be in any English words I can find.   cx fv,fz gv,gx jb,jc,jf,jg,jq,jw,jx,jz kq,kx,kz pq,pz qa,qf,qh,qj,qk,qm,qn,qs,qv,qx,qy,qz sx tq vj,vk,vq,vx wq,wv xj,xk,xz yy zf,zj,zq,zx   I created the following rules based on this:   body ODD_OTHER /(vj|vk|xj|xk|yy|zf|zj)/i describe ODD_OTHER Contains odd letter combinations. score ODD_OTHER 0.1   body ODD_Q0 /(j|k|p|q|t|v|w|z)q/i describe ODD_Q0 Contains odd letter combination with Q. score ODD_Q0 0.1   body ODD_Q1 /q(a|f|h|j|k|m|n|s|y)/i describe ODD_Q1 Contains odd letter combination with Q (2). score ODD_Q1 0.1   body ODD_V /(f|g|q|w)v/i describe ODD_V Contains odd letter combination with V. score ODD_V 0.1   body ODD_X /(c|g|j|k|q|s|v|z)x/i describe ODD_X Contains odd letter combination with X. score ODD_X 0.1   body ODD_Z /(f|j|k|p|q|x)z/i describe ODD_Z Contains odd letter combination with Z. score ODD_Z 0.1   body ODD_J /j(b|c|f|g|w)/i describe ODD_J Contains odd letter combination with J. score ODD_J 0.1   Can anyone see anything wrong with these tests or think of a word which might conflict?   I checked this against 3 dictionaries and my ham corpus and it turned up clean.   Frederic Tarasevicius Internet Information Services, Inc.     ----- Original Message ----- From: Mike Kuentz (2) To: [EMAIL PROTECTED] Sent: Thursday, August 07, 2003 12:15 PM Subject: RE: [SAtalk] RD Message body/subject filter help That's not all it matches, from the standard dictionary off of my Redhat box.  grep 'yb' /usr/share/dict/linux.words anybody bodybuilder bodybuilders bodybuilding capybara Charybdis cybernetic cybernetics cyberspace daybreak donnybrook everybody honeybee hybrid keyboard keyboards lazybones maybe piggyback piggybacked piggybacking piggybacks playback playboy psilocybin Rayburn soybean storyboard Sybil Tyburn volleyball Volleyballs Granted, there some funky words in that list.  'yj' is a lot more promising as it only matches skyjack Mike -----Original Message----- From: Fred I-IS.COM [mailto:[EMAIL PROTECTED]] Sent: Thursday, August 07, 2003 11:42 AM To: [EMAIL PROTECTED] Subject: Re: [SAtalk] RD Message body/subject filter help rawbody MY_OBFUY /y(b|j)/i describe MY_OBFUY Y with unusual chars score MYU_OBFUY .45 This one catches MAYBE Frederic Tarasevicius Internet Information Services, Inc.