On 06/20/2011 03:19 AM, Mark Goodge wrote:
If the latter, then you may want to use a simple "bad word" filter to block emails containing egregiously bad language.
But be careful. Some years ago we had a system that had some overly-simplistic body checks for "bad" words. Then the CEO came to ask why an email he was trying to send kept being rejected. I kidded him that he shouldn't be trying to send porn and did some research. Sure enough, he was sending an attachment which, when base64 encoded, happened to contain a 4-letter sequence starting with the letter "f" that triggered the filter.
Similarly, don't expect anything close to perfection in filtering unwanted images short of prohibiting attachments altogether.
Finally, whatever method you choose to carry out the filtering, the action taken should always be to block rather than substitute. Apart from the fact that you have no right to edit the contents of someone else's message just because you dislike some of the words in it, there is a real danger that a misapplied filter will alter words it should not in a way which materially affects the meaning of the message. Google for 'clbuttic' for examples, and, while you're at it, look up the "Scunthorpe problem" for some other reasons why profanity filters are very hard to get right.
The classic example of this is the Christian news outlet OneNewsNow which had a filter to convert "objectionable" terms to something they deemed more acceptable. So when they displayed the AP news feed about Tyson Gay's record-breaking sprints at the Olympic games they reported instead that "Tyson Homosexual was a blur in blue, sprinting 100 meters faster than anyone ever has." Incorrect and embarrassing. (http://voices.washingtonpost.com/sleuth/2008/07/christian_sites_ban_on_g_word.html).
Cheers, Steve
