Heuristic Filtering

Heuristic simply means speculative. A message containing the phrase “mortgage” by itself does not necessarily set off a red flag. It could be a reminder email from the bank that the mortgage payment is overdue. But the word “mortgage”, combined with “bad credit”, “no problem” and “apply online” might tell a different story.

Heuristic filtering works by subjecting email messages through thousands of pre-defined rules against the message envelope, header and content. Each rule assigns a numerical score to the probability of the message being spam. The result of the final equation is known as the Spam Score.

The spam score is then measured against the user’s desired level of spam sensitivity - whether low, medium or high sensitivity. Setting a higher level of sensitivity leads to more spam being captured, but has the adverse effect of filtering legitimate emails as spam. This is known as a false-positive.

Based on the spam score and sensitivity threshold, the messages are organized into spam, not spam, and unsure folders. There as exceptions to this rule however :-

• Messages from senders on the Approved List is never considered spam
• Messages from senders on the Blocked List is always considered spam
• Messages triggering custom rules defined by the user is never considered spam

Certain words add to the spam score while others subtract from it. Words such as “free”, “click here”, “apply online”, capitalized words, excessive use of punctuations or disguised words such as “p0rn” or “C1ALIS” will add to the probability of the message being spam. It has been statistically shown that messages with the word “FREE” in red bold capital letters, has a probability of exceeding 99 percent to be spam. Some of the factors which reduces the spam score are using words such as “Newsletter”, “Issue Number”, “List”, or placing a date on the message.

Heuristic engines however suffer from a main drawback. As spammers learn and adopt according to the filtering rules employed, so too must the heuristic database. This has kept a cat and mouse game of playing catch up and is therefore important that the spam filter software supports updating its heuristic database as part of its feature. Additionally, heuristic engines tend to be very error prone and a good spam filter software should not only have a low false-positive rate, but should also allow custom rules to be defined.

Many spam filter software uses several methods to stop spam. MailWasherPro is one such example which uses both a Heuristic and Bayesian spam filter solution which boasts over 98 percent accuracy and is also the lowest priced anti-spam software tested. It works with MS Outlook and Outlook Express.

Click for a free trial download :-

Download your copy now