| 摘要 |
The portion of email traffic that cannot be definitively determined to be spam or definitively determined to be ham (non-spam) is processed by sending a bounceback email to the sender, requiring the sender to reply using a sub-address created by the bounceback generator. The type of bounceback email depends on whether the domain of the received email indicates that the sender is or is not likely to be a spammer. When the sender is not likely to be a spammer, the bounceback email includes a sub-address in computer readable form. When the sender is likely to be a spammer, the bounceback email includes a test that must be solved to yield the sub-address; preferably, the test is very difficult for a computer to solve but reasonably easy for a human to solve. When a reply email to the bounceback email (“reply bounceback”) is received, the presence of the sub-address indicates that the sender is not a spammer, so the reply bounceback is determined to be ham and delivered to the destination mailbox. The bounceback can have the sub-address encoded in such a way that a human must be involved in decoding, which is appropriate for situations where the envelope sender domain of the original email is considered suspicious. The bounceback can have the sub-address presented in a computer readable format when the envelope sender domain of the original email is more trustworthy, if authenticated. Authentication information can be retained in a local private Sender Policy Framework (SPF) database, or shared in a centralized private SPF database.
|
