Google Blogoscoped

Forum

Spam at GMail

Zim [PersonRank 10]

Monday, October 29, 2007
13 years ago2,991 views

Google is reporting from the Official Google Blog about how well the spam filter is working.
google.com/mail/help/fightspam ...
In this page, there's a easy-to-understand graph with the spam percentage growing, and the spam filter success percentage.
Also there's a youtube video explaining how the spam filter works. Very nice one:

+ Show video



For me, the spam filter works very well, and I don't get more than ten spam messages at the inbox. And I found personal messages in the spam folder not more than once per month.
What do you think about?

Philipp Lenssen [PersonRank 10]

13 years ago #

The most interesting bit in the video to me was that they're using OCR (same as used in Google Book search) to understand image attachments, to better filter for spam. Another interesting bit was the mention of "multiple authentication systems" to prevent non-spam landing in the spambox, though I don't know what he specifically means by that...

James Xuan [PersonRank 10]

13 years ago #

Im getting way more spam lately consisting only of images...

Roger Browne [PersonRank 10]

13 years ago #

I was having a big problem with Gmail. About twenty non-spam messages per day were being put in the Gmail spambox. I marked every one of them as "Not Spam", in the hope that this would improve the filtering.

About a month ago, I got a message from Gmail asking if it was OK for a human to read the emails that I was marking as "Not Spam", to help them better understand how to improve their spam filtering. I agreed to let them read my emails.

Now, since two weeks ago, my spam filtering has been astonishingly good. Thousands of good emails in my Inbox, and maybe only five spams per week in the Inbox. And not one single good email has appeared in the Spambox for two weeks. If this keeps up, I can stop checking the Spambox, which will be a huge relief.

It can't all be due to what they discovered in my emails – because there are many, many types of spam – but I'd like to think it helped them fine-tune their algorithms a bit.

James Xuan [PersonRank 10]

13 years ago #

I get around 5 a day in my inbox...

George R [PersonRank 10]

13 years ago #

Roger:

It might be interesting to see some samples of what you marked as "Not Spam" and authorized google to read. Do you have any that are not private?

Maybe you could work with Philipp on this. Addresses and personal information could be changed.

Kirby Witmer [PersonRank 10]

13 years ago #

I have a "extra" gmail address called thisisjunk[put at-character here]gmail.com. Its amazing, but I think some people must use it for a "junk" address, cause I get legitimate emails from monster.com or facebook.com, that I never asked for. :)

TOMHTML [PersonRank 10]

13 years ago #

Where is the "false positives" curve?

Mambo [PersonRank 10]

13 years ago #

<< Where is the "false positives" curve? >>

I believe that would be much more valuable data. In fact, it's completely critical (unlike inbox spam which is only a minor nuisance). I've acquired a bad habit of just deleting all my spam without checking it beforehand.

Tony Ruscoe [PersonRank 10]

13 years ago #

<< Where is the "false positives" curve? >>

That's a good question but it would be very skewed for the very reason Mambo gives. Even if the graph was there, it probably wouldn't even show 50% of the false positives as most people probably don't even check their spam.

Mysterius [PersonRank 10]

13 years ago #

Does any of this have to do with the Postini acquisition?

This thread is locked as it's old... but you can create a new thread in the forum. 

Forum home

Advertisement

 
Blog  |  Forum     more >> Archive | Feed | Google's blogs | About
Advertisement

 

This site unofficially covers Google™ and more with some rights reserved. Join our forum!