179518 - Automatically mark own (sent) emails as non-junk

Status: NEW

(MailNews Core :: Filters, enhancement)

Description

[Christian :Biesinger (don't email me, ping me on IRC)]

Currently, my own emails appeare as "junk status unknown" in my "Sent"-Folder.

however, I can guarantee that they are not spam :)
would be nice if they would be marked non-spam automatically.

Comment 1

[Jo Hermans]

*** Bug 179401 has been marked as a duplicate of this bug. ***

Comment 2

[Alfonso Martinez]

This should apply only to sent mails (imho).
Viruses fake the sender so if those mails are processed automatically as
non-junk, it could degrade the quality of the process. Yes, I know that due to
the bayesian system one mail won't ruin the system, but I just think that it's
better to avoid the use of bad data in order to get better results.

Comment 3

[Christian :Biesinger (don't email me, ping me on IRC)]

>This should apply only to sent mails (imho).

er, yeah, that's what I meant when I filed this bug.

Comment 4

[WADA:World Anti-bad-Duping Agency]

Sent mail became "Junk Status Unknown" on Win-Me too.
Please change OS=All. I'm not aothorized to change this field.

By the way, posted message to a news group also became "Junk Status Unknown"
when saved in my local Sent folder.
I can garantee my news posting is NOT SPAM too :-)

Comment 5

[Dan Mosedale (:dmosedale, :dmose)]

Some comments from Seth, from bug 179530: 

[junk mail] auto tokenize outgoing messages?

would it be worth auto tokenizing the messages the users sends, as "good" mail?

similar to how we collect outgoing email addresses?

on each mail I send, we run the message through the tokenizer.
we have the body and headers (locally, so no network delay)

possible upsides:
1) requires the user to do less manual training before they see good results
2) things I write about and people I send to aren't junk.

possible downsides:
1) people who respond to spam
2) too much data for training.dat?  (would we stop doing this at a certain point?)

Comment 6

[Dan Mosedale (:dmosedale, :dmose)]

*** Bug 179530 has been marked as a duplicate of this bug. ***

Comment 7

[WADA:World Anti-bad-Duping Agency]

Tested on 2002111404-trunc/Win-Me.
My message saved in Drafts Folder has been also set "Junk Status Unknown".
I'm glad if this is marked as "Not Junk" automatically.
Regards.

Comment 8

[(not reading, please use [e-mail] instead)]

we should be careful with adding to the good group on training.dat on every
message send.

that might group the group too fast.

maybe we should be smart and only add (automatically) if the good group is
small, to help auto-populate it?

Comment 9

[(not reading, please use [e-mail] instead)]

note, there is no more "unknown" icon, so this might be a moot point.

Comment 10

[Joe McCabe]

Third party themes still display an "unknown" icon.

Comment 11

[Vaclav Dvorak]

I don't think this is a "moot point", even if there is no "unknown" icon. The
training is the point, not the icon. Might be solved by a combination of bug
11039 and bug 181631, though.

Comment 12

[Matthew Kershaw]

What's the latest on this bug? I'm getting kind of sick of manually flagging
every email I send as 'Not Junk.' I save every email I send, and file them
according to topic. And the weird thing is, my own email address is in my
address book -- and I have the option set to not flag as junk any email in my
address book!

Comment 13

[Vaclav Dvorak]

Matthew: do you mean to imply that Mozilla marks some emails that you sent
yourself as spam? (Because you didn't say that.) If not, what's the reason that
you manually flag every email you send as not junk?

Comment 14

[RDL]

When I send mail in Thunderbird, I use the 'OutBox' extension to send later. I
use the Mnenhy extensions and the mail appears in the 'Unsent Messages' folder
with the 'unknown' junk status marker visible. I gather the same would also be
true for those using certain themes. After actually sending, I choose to have a
copy kept in the 'Sent' folder. The status on this copy is still set as
'unknown' and that stays with it wherever I move it, either by manual move, or
by manually running filters on the folder, until I manually change the status.
Whereas my setting the status of such sent copies to 'not junk' might add to the
Junk killer training, it apparently has no effect on the marking of sent copies
which are all (without exception) still marked as 'unknown' even though this
hardly ever happens now to my incoming mail. You will see, further on, why I
have very good reason, quite apart from accuracy and decent presentation, to not
want this misleading marker to be applied to my own output.

If 'sent copies' are not subject to acquired rules (training) is there any
point in ever marking outgoing mail or its copies, as unknown?

If this has some protective intent, say against parasitic mailing, couldn't the
training knowledge at least be applied to output? Currently I have the tedious
and apparently fruitless task of marking all my own outgoing copies as 'not junk'. 

I could include such setting in my output distribution filters but (a) this
carries the risk of automatically concealing something of which I really should
be made aware (eg 'alien' mail in the 'Sent' folder)(b) doesn't really address
the fact that I would hardly regard mail which I have just sent, by pressing a
button, as anything other than 'not junk'!

Mnenhy provides a feature called "folder storage" which allows for custom column
layout in the thread pane and it has been suggested that I could use that to
turn off the offending column in the 'Sent' folder display. However, the junk
status travels with the sent copies wherever they go and I use manually
activated filters to distribute them into appropriate other folders (in fact,
with the input from that person/body etc but labelled to distinguish
from input in the same conversations). I certainly don't want to switch off the
status display in those folders, since that would conceal information about any
incoming mail which still has 'undecided' status in the same folder(s).

The handling of filtering on output copies seems to have moved on somewhat since
19 months ago when this matter last received post and is perhaps only one step
away from the option of automatic filtering of output copies on transmission as
for input mail on receipt. However, as far as I can see, this particular problem
has stood still for about 19 months and this may be because the reason why
anyone would need (let alone want) to have their output mail copies' junk status
processed instead of defaulted to 'unknown' needed to be elaborated.

RDL

Comment 15

[Wayne Mery (:wsmwk)]

*** Bug 333346 has been marked as a duplicate of this bug. ***

Comment 17

[Dan Mosedale (:dmosedale, :dmose)]

Assigning bugs that I'm not actively working on back to nobody; use
SearchForThis as a search term if you want to delete all related bugmail at
once.

Comment 18

[Wayne Mery (:wsmwk)]

A collection of ideas like this could be used to quickly populate the training data - say perhaps for the first N messages.  Would it be more effective than attempting to develop a single starter corpus (that might work only modestly well?). Comment 5 seems to be a fair summary.