So,

the antifeature is more or less ready. There are two points left.

1. the upgrade script (a new setting and a new column in the posting table)
2. the handling of the approval with simultaneous marking the entry as ham

First one is a no brainer but the second point is a bit tricky. We can assume, that a forum team member, who is allowed to approve postings will approve a posting only if it is no spam. From the point of preventing unnecessary work, an approval should also mark a posting as ham.

Until now we provide the feature to mark a posting as ham or spam. If you mark a posting as a ham, you have the choice between two functions. There is the option of marking a posting as a ham and training the Bayesian filter or omitting the training, even if the Bayesian filter is activated in the forum settings. This is done with two buttons, each for one of the choices.

In the approval form there is only the one button for approval.

I am not sure, if it makes sense to add further buttons to the form. If further buttons should be added, which should be added?

Option A: Add all buttons, a ham-classification form would provide. This does not look right to me. It's complex and confusing.

[OK - Release posting][Report and flag as ham][Flag as ham only]

Option B: Combine the function of the buttons. Also this does not look simple enough to me even it is simpler and therefore much better than option A.

[OK - Release posting, report and flag as ham][OK - Release posting amd flag as ham only]

Option C: Leave it at the one button and always run the Bayesian filter training, even for very short posts, if the filter is activated. This looks right to me but might be not clear enough to others. Furthermore it needs a complex decision making in the code and it is possible to train the filter with crap (one word or one sentence postings).

[OK - Release posting]

Opinions?

Tschö, Auge