Menu
▾
▴
#69 Do not move emails when marking as spam or ham
The Thunderbird extension always seems to move messages that get marked as spam to the spamato folder and messages that get marked as ham to the Inbox. This is very bad, since it makes sense after installation to tell Spamato about my own Ham/Spam emails. These emails, especially ham are usually already organized in apropriate folders and moving them will destroy that.
At least there should be an option whether to move or not. There is an option in the web-configuration interface, but that only seems to apply to messages that get classified as spam when they arrive, not those that are manually marked. However that configuration option adds to the confusion because the move when doing a manual classification is not expected.
Logged In: YES
user_id=1217053
Originator: NO
That is a very good point (although rather a Feature Request than a bug), thanks for posting.
Do you think that two additional buttons that you would put to your toolbar only at the beginning when you want to "train" the systme would do the job? Or do you have any other idea of how a simple interface could look like to realize your proposal?
Logged In: YES
user_id=1472154
Originator: YES
OK, lets separate the issues here.
First, I think spamato should have the functionality to train and re-train on pre-existing ham and spam collections. Especially for the Bayesian filter (and other learning/statistics based future plugins) this is absolutely essential. One difficulty here is how to decide which of the plugins should "see" those existing messages in a training step, especially when a user re-trains several times.
The other issue is how to implement all this in the GUI. I guess the simplest first approach would be to use the existing "Spam" and "Ham" buttons for this: if it is possible to easily switch off the automatic moving in the preferences, one can simply mark whatever messages one wants and mark them as spam or ham. If I have a whole folder of ham messages, I'd simply mark the whole folder.
I am suggesting this because at least for the Bayesian learning, there is no difference between "report this as spam" and "retrain this as spam" -- so would the existing "ham" and "spam" buttons not be usable for this?
However, it is hard to figure out how to find those buttons initially when they are not automatically placed in the toolbar. So maybe there could be a button in the prefs to do this or at least a hint.
Also, many people like to have important functions also present in the context menu, so maybe there could be a "Spamato" option in the context menu, maybe only activated by a pref.
Of course there is also the problem of the disappearing toolbar buttons (and in my case, the disappearing Tool->Spamato menu), so maybe the context menu entry would be at least a workaround for this too.
Logged In: YES
user_id=1217053
Originator: NO
I think the plugins/filters should decide on their own whether they want to use data for training purpose or not. There might be a switch in every filter or whatever is send as "training data" is always handled. This shouldn't be a big problem, at least not for the user.
So ok, we keep the Spam/Not Spam buttons or maybe transform it to a "menu button" such as the one for "Get Mail". Than the user can easily report spam/ham either as training data or as "real" data. However, this has to be documented... some users might not be able to distinguish between these two types or in some cases it might not be clear what to use.
(I have refiled this as a Feature Request.)
Logged In: YES
user_id=862508
Originator: NO
I agree!
wouldn't it be easier to just tag the messages at spam that are spam only. Like Thunderbirds own Junk Mail controls... It's adds a small icon to the mail when it is marked as junk..
Since when is it logical to move ham mail.... (except previously spam marked mails)
You should have the option to move spam e-mail ONLY, there is no reason to move ham mail.
Logged In: YES
user_id=1217053
Originator: NO
Yes, we will consider this FR for release 1.0. Thanks for the additional opinion.