Join the AMA (Ask Me Anything) with the Firefox leadership team to celebrate Firefox 20th anniversary and discuss Firefox’s future on Mozilla Connect. Mark your calendar on Thursday, November 14, 18:00 - 20:00 UTC!

Ovo će web mjesto raditi na ograničen način, dok obavljamo održavanje stranice. Ako neki članak ne riješi tvoj problem i ako želiš postaviti pitanje, naša zajednica za podršku spremna je pomoći na Twitteru @FirefoxSupport i na Redditu /r/firefox.

Pretraži podršku

Izbjegni prevare podrške. Nikad te nećemo tražiti da nas nazoveš, da nam pošalješ telefonski broj ili da podijeliš osobne podatke. Prijavi sumnjive radnje pomoću opcije „Prijavi zlouporabu”.

Saznaj više

Can I use wild cards when defining message filter criteria?

  • 6 odgovora
  • 2 imaju ovaj problem
  • 1 prikaz
  • Posljednji odgovor od Toad-Hall

more options

Basically I want to sort my incoming mail by using wild cards in either the to field of the from field.

Basically I want to sort my incoming mail by using wild cards in either the to field of the from field.

Svi odgovori (6)

more options

see info here: http://kb.mozillazine.org/Filters_%28Thunderbird%29

basically, no.

If you are trying to create a filter to sort Junk then do not bother, Set up Junk Controls and train them to recognise Junk.

Info on Junk Controls: http://kb.mozillazine.org/Junk_Mail_Controls

Perhaps you could give an example of what it is you are trying to do?

more options

Message filters provide a kind of wildcard option with "contains" or "begins with".

more options

"contains" and "begins with" is better than a poke in the eye with a sharp stick, but completely inadequate for filtering out spammers who have a nearly infinite capacity to intersperse other characters in between the letters of the word "PENIS". Let's please catch up to the 1990s and get the real deal regexps, Thunderbird.

more options

lilbrownbat: re: "contains" and "begins with" is better than a poke in the eye with a sharp stick, but completely inadequate for filtering out spammers who have a nearly infinite capacity to intersperse other characters

You are correct. That's why you should not be wasting time trying to create your own zillion versions of filters when there is already an in built system using Bayesian filtering. The advice is to set up the Thunderbird Junk controls and then train them.

More info:

more options

Toad-Hall:

Re: "That's why you should not be wasting time trying to create your own zillion versions of filters when there is already an in built system using Bayesian filtering. The advice is to set up the Thunderbird Junk controls and then train them."

...except that it's not "training". All it's doing is saying, "Yes, this is junk", "No, this isn't junk", and leaving it entirely up to the black box to decide what (if anything) to do with that information. "Training" would involve an element of explaining WHY it's junk. That's the useful purpose served by wildcards in filters. I realize that it's nontrivial, but imagine if there were a mechanism to describe what characterizes the spam: "Yes, this is junk (as far as I'm concerned), and you can tell because it has the following characteristics."

Look at it this way: Thunderbird users have two types of recourse to unwanted email, one being this black box mechanism that eventually (perhaps) filters out some spam, but lets other, very similar spam through. After saying "Yes this is spam" enough times, I've filtered out ONE spammer. Since I can't really "train" the algorithm, it's a painfully slow and woefully inadequate process. As a result, I end up having to create junk filters without wildcards, and they're also inadequate. There's a huge gap between the two tools. Either allowing real "training", or real wildcards, would close that gap.

more options

Did you read the info at the links supplied because your response implies that you didn't? Bayesian filtering information: http://en.wikipedia.org/wiki/Bayesian_spam_filtering

This method is used by many spam detecting software, which you could or may already use eg: spamassassin You do not need to explain why it is junk because the algorithms are designed to detect why it is Junk; it just needs to know whether you think it is junk or not. Most people would accept the same certain type of emails as junk, but there is a wide variation on personal needs of what is junk; hence why it needs 'training'.

When I initially started using thunderbird, I set up Junk controls and immediately trained it following the instructions exactly, stating what was Junk and what was not junk. yes, it was a bit of a bore at first, but soon started to work very well. I've used this for years and rarely get false positives.


http://kb.mozillazine.org/Junk_Mail_Controls 'Training the Junk Mail Controls this heading gives very clear instructions on how to go about creating a good junk control filter. I followed them to the letter and have no issues.

I have found that many people use the 'mark as Junk' on junk emails in the Inbox and 'mark as not junk' any good emails in the Junk folder, but when initially training rarely select good emails in the Inbox and 'mark as not junk'.

Bayesian filtering requires at least 100 bad messages be marked as spam and 100 good messages marked as not junk to function. To work best, it needs a few hundred of each marked.

If you mark a thousand spam messages but do not mark legitimate messages, or very few, it won't work well.

It's best to mark different types of messages good and bad - marking 500 messages from the same source is not as good as marking 500 messages from different people.

Tweaking the Junk Mail Controls You may want to try this first before retraining the Junk Controls.

mail.adaptivefilters.junk_threshold is the preference that determines at what "level" messages are classified as junk. Read this section if you want to modify this theshold to see if this improves the situation.

If you feel that you want to retrain your Junk controls.

  • Tools > Options > Security > Junk tab
  • click on 'Reset Training Data' button
  • click on OK