A spam filter is software that automatically identifies and separates out unwanted bulk email, commonly by scoring each incoming message against rules or statistical models trained on known examples of spam. Programmer Paul Graham popularized a practical statistical approach using Bayesian probability in his 2002 essay A Plan for Spam, which influenced many of the spam filters built into email services afterward. Modern spam filters combine content analysis, sender reputation and machine learning to keep the great majority of unwanted email out of a user's inbox before it is ever seen.
Facts
Partially Attested
YearYear of the first scholarly publication applying a naive Bayes classifier to spam filtering; Paul Graham's essay A Plan for Spam, which popularized the approach, dates from August 2002. SignificancePaul Graham's 2002 essay A Plan for Spam showed that a Bayesian combination of individual word spam probabilities could filter spam acceptably well. 1 FAQs
How does a statistical spam filter learn what spam looks like?
It learns from a collection of spam and a collection of legitimate mail and combines the spam probability of each word in a message.
A statistical spam filter learns from examples rather than from a fixed list of banned words. In the approach Paul Graham described, the filter starts from two collections of mail, one of spam and one of legitimate mail, and anything a user deletes as spam is added to the spam collection while everything else goes into the legitimate collection. Each word then carries its own probability of appearing in spam, and the filter combines the probabilities of the words in a new message, using a Bayesian combination, to judge how likely the message is to be spam.
Connections
In Field
Source Email filtering (Wikipedia)
Sources
1. Paul Graham: A Plan for Spam
Essay text, opening sectionQuote, Essay text, opening section
In fact, I've found that you can filter present-day spam acceptably well using nothing more than a Bayesian combination of the spam probabilities of individual words.
2. Wikipedia: Naive Bayes classifier
Spam filtering sectionQuote, Spam filtering section
The first scholarly publication using the naive Bayes classifier for spam filtering was by Sahami et al. in 1998.
Email filtering (Wikipedia)
In Group: Web and Online Communication Software, Lead paragraph
Email filtering is the processing of email to organize it according to specified criteria.
In Field: Software Engineering, Lead section
The term can apply to the intervention of human intelligence, but most often refers to the automatic processing of messages at an SMTP server, possibly applying anti-spam techniques.
View the SourceReader Challenges (0)
No disputes yet. Spotted an error or a better source? Open the first one.
Sign in to dispute this or suggest a correction.