I am getting dozens of spams a day that get past the spam filter. I don't know why; most of them are poker or viagra spams. Not only should spam.module easily recognize those words by now, but I've also manually added "usually spam" filters for them. Moreover, it does catch many more spams with these words in them. However, the sheer volume of uncaught spam is a huge inconvenience, and it seems obvious that they should be caught.

Any insight into what's going on here?

Comments

jeremy’s picture

Go to administer >> settings >> spam and check "advanced configuration". Save the configuration, now check "display spam rating" in the advanced section. Now look at the comments in question, and see what spam value they are being assigned (1-99).

jgoerzen@changelog.complete.org’s picture

In the past few days, I have been innundated by the mega-comments (as described in the other post). As I recall, though, most of these comments were being assigned scores in the 60-70 range. Trouble is, plenty of legitimate comments were as well. A lot of these comments were ones that had "real-looking" words and content in them, along with some sort of poker link or something.

jeremy’s picture

There were numerous bugs in the original spam module. I have written it solving many of these issues. Please read the release announcement here. Download the new module here.

jeremy’s picture

Status: Active » Fixed

The tokenizer logic was very broken in the earlier version of the spam module. The rewritten spam module solved these annoying problems.

osherl’s picture

Anonymous’s picture

jeremy’s picture

Status: Fixed » Closed (fixed)

Closing manually. (I think the project module is broken, it keeps marking them updated when they should be closed.)