human spam filter claiming good posts as bad, trigger words

  • Resolved LynnW

    (@lcwoodward)


    8 years, 3 months ago

    Hi Paul,
    Have a client install of the simple security firewall that has started throwing good comments into the spam filter. It’s the “human spam filter” portion. It has marked some comments as “pending” (and I had it set to toss those into spam; I have since changed that to toss them into the comment approval area. This takes care of the problem on the blog owner’s end, but will the commenters see their comment as “pending approval” or will the comment appear to be lost to them?).

    This blog gets a lot of comments, and it seems this behavior just started Dec 29. The human spam filter mentions these words or parts of words as affronts:
    port
    -and
    )-
    ry
    Problem is “port” is part of the word “important” “ry” shows up in “story” “very” and “history”. The other examples were just folks making a — for an m-dash and forgetting a space. I would give you more examples, but the client has already cleaned the comments and approved them! (he doesn’t waste time!)

    Thanks for all your hard work on your GREAT WP plugins!

    https://wordpress.org/plugins/wp-simple-firewall/

Viewing 2 replies - 1 through 2 (of 2 total)
  • Plugin Author Paul

    (@paultgoodchild)

    8 years, 3 months ago

    Hi Lynn,

    The issue with the human anti-spam is that it’s bound to catch false positives… and with a site that gets a lot of comments, then proportionally they’ll get more than average false positives.

    The best thing here is perhaps to disable the human spam component if it’s too much hassle to mitigate for the site owner. The spam bot system will block all the robots of course.

    So the decision is:
    1) Disable the human spam filter and deal with the human spam manually; or
    2) Keep the human spam filter on and deal with the false positives manually.

    It’s a question then of which is the least work or troublesome.

    Hope that helps! 🙂

    Thread Starter LynnW

    (@lcwoodward)

    8 years, 3 months ago

    Thanks, Paul, I just saw your response, thank you.
    I think I wasn’t clear. The human spam feature in the plugin just started making these false positives around the end of Dec. That’s why I sent you a note. And it’s making them on PARTS of words or lack of spacing, which is a common typo.
    (those listed above), and, among others:
    oy
    !!!
    I was able to search through 1400 spam comments with the keyword “pending” to find them without checking each one. We’re happy with option 2) that you listed, just having the filter put them in the approval area instead of the spam area.

    But I thought you’d want to know as this is a recent change in the filter’s behavior and may not be ideal.
    Thank you!
    Lynn

Viewing 2 replies - 1 through 2 (of 2 total)
  • The topic ‘human spam filter claiming good posts as bad, trigger words’ is closed to new replies.