Hello Larry, Wednesday, January 21, 2004, 11:37:09 PM, you wrote:
LG> Along the same lines, I had the following: LG> describe MY_RBDY_INVTXTSZ1 MY: Invisible text size LG> rawbody MY_RBDY_INVTXTSZ1 /font\s+.*\bsize=.-\d\D/i LG> score MY_RBDY_INVTXTSZ1 0.5 LG> describe MY_RBDY_INVTXTSZ2 MY: Invisible text size with style LG> rawbody MY_RBDY_INVTXTSZ2 /size=.-\d\D style=.font-size: \dpx;\D/i LG> score MY_RBDY_INVTXTSZ2 0.5 LG> describe MY_RBDY_INVIMGSZE MY: Invisible image size LG> rawbody MY_RBDY_INVIMGSZE /width=.1\D height=.1\D/i LG> score MY_RBDY_INVIMGSZE 0.5 LG> They seem to hit legit opt-in advertisements though. I would be LG> curious to see the results against your corpus if you are inclined LG> and willing to spend the time. MY_RBDY_INVTXTSZ1 -- 5002s/876h of 91714 corpus (74113s/17601h) 01/22/04 Note that a <font size=-1> or -2 is not an invisible font size, but simply a reduced font size. MY_RBDY_INVTXTSZ2 -- 463s/0h of 91714 corpus (74113s/17601h) 01/22/04 I wouldn't think that any font-size in pixels would be "invisible", since one could do anything from tiny to gigantic using that specification, but in my corpus it hits only spam. MY_RBDY_INVIMGSZE -- 2190s/158h of 91714 corpus (74113s/17601h) 01/22/04 ham hit is mostly HTML newsletters from accepted sources and/or YahooGroups. Bob Menschel ------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
