Gustafson, Tim wrote: >> 1) FPs on highly technical mail due to words not known to the spell >> checker. >> > > I hadn't thought of that, but people who are dealing with highly > technical e-mails would probably also be able to customize their > local.cf file to effectively turn off the rule. > Well, if you customers are "professional agencies" as cited below, a lot of their email will fall in this category.
Note that by "highly technical" I'm not just referring to scientific papers. I'm using technical here to mean "Used in or peculiar to a specific field or profession; specialized". Most professional outfits deal extensively with the "lingo" of their profession. Contracting, Accounting, Lawyers, Human Resources... Also consider proper names of companies, products, etc. All of which will not be known to the checker. Of course you could train your spell checker to your companies local mail words.. however, at that point you've implemented a low-quality version of a bayes checker. > >> 2) FPs on email sent by folks of the text-message generation. (OMG did >> > u > >> c he 8 it all!) >> > > All my clients are professional agencies, and I don't think this really > applies to them. > Really? Are you sure? I personally doubt this frees you from such things. Perhaps not in such a teenage form, but plenty of "professional agencies" are full of crackberry addicts. > >> 3) FPs on email sent by lazy/stupid folks that can't spell. >> (Translation: management material) >> > > I don't mind these getting blocked. In fact, I'd love it if every time > someone sent me a very poorly written e-mail they got a bounce message > back telling them to turn on the spell check option in their e-mail > program. > Until it happens to an email sent by the person who creates or signs your paycheck.... > >> 4) relatively quick and easy for spammers to adapt to. >> > > True. I hadn't thought of your book text example. > > >> 5) Relatively high CPU usage, given the above caveats in accurate. >> > > For me, this is less of a concern...my e-mail is a quad 3.2ghz machine > with 4GB of RAM, and it runs at about 2% processor usage most of the > time. :) > > As for whether or not a lot of SPAM has mispelled words in it or not: > every single e-mail that came to my inbox this morning that was SPAM but > wasn't tagged as SPAM were pure gibberish, and they would have been > caught by my proposed filter. :) > Wow.. none of the mis-spelled gibberish spams make it to my inbox.. ever. Bayes tears them to shreds and DRUGS_* finishes them off. Occasionally new variants of the "word-salad with image" spams get by when SARE hasn't come up with a new rule for the variant, but those are all full of properly spelled words. > So, while I agree that this filter wouldn't suit everyone, it would > certainly be a nice option to have.
