[Spambayes] Feature improvement idea - Grammar and spell check rating!

skip at pobox.com skip at pobox.com
Tue Jul 18 21:55:49 CEST 2006


    Robert> At the risk of bloating this awesome piece of software, I submit
    Robert> that Grammar and Spell checking (at least localized English)
    Robert> would be a good way to easily identify illegitimate email. 

Any ideas for how grammar checking might be done?  Note that the current
architecture treats each message as a set of tokens.  No token ordering
exists at the scoring phase.

    Robert> A two phased approach would leave the existing code base
    Robert> untouched, and a second (optional) feature added on.  Perhaps
    Robert> even leveraging the existing host (Outlook) for the bulk of the
    Robert> work, and just look at the total number of errors.  This idea
    Robert> comes from a quick look at what consistently ends up in the Junk
    Robert> Suspects folder.

Note that Outlook isn't the only SpamBayes platform.  Unless multiple
platform-dependent plugins were developed, the solution would have to be
platform-independent.

Skip


More information about the SpamBayes mailing list