Dynamic message filtering

• US 7,257,564 B2
• Filed: 10/03/2003
• Issued: 08/14/2007
• Est. Priority Date: 10/03/2003
• Status: Active Grant

• Current Assignee: Mailgate, LLC
• Original Assignee: Tumbleweed Communications Corporation (Axway Software SA)
• Inventors: Loughmiller, Scott, Olson, Mike, Ready, Jeff, Maedge, Ehren, White, Phil, Collier, Jason
• Primary Examiner(s): Vincent, David
• Assistant Examiner(s): Fernandez Rivas, Omar F.
• Application Number:
  US10/678,602
• Publication Number:
  US 20050076084A1
• Time in Patent Office: 1,411 Days
• Field of Search: 706 10- 16, 706/18, 706/20, 706/909, 709/202, 709/206, 709/207, 726/1, 726/2, 714/1, 714/4, 714/26, 714/100, 714/699
• US Class Current: 706/16
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]
• CPC Class Codes: H04L 51/212 using filtering or selective blocking

Abstract

Dynamically filtering and classifying messages, as good messages, bulk periodicals, or spam. A regular expression recognizer, and pre-trained neural networks. The neural networks distinguish “likely good” from “likely spam,” and also operate at a more discriminating level to distinguish among the three categories above. A dynamic whitelist and blacklist; sending addresses are collected when the number of their messages indicates the sender is good or a spammer. A dynamically selected set of regular expressions input to the neural networks.

Claims

1. A method for blocking delivery of unwanted spam messages, comprising the steps of:

recognizing patterns including words and groups of words in a messages;

applying a plurality of machine learning techniques including a two-level neural network responsive to the recognized patterns in order to classify the message, the two levels of neural networks include;

a primary neural network level that determines if the message is likely a non-spam message or likely a spam message; and

a secondary neural network level that includes a pair of neural networks including;

a first secondary level neural network that determines if a likely non-spam message from the first neural network level is actually a non-spam message or a bulk message, and

a second secondary level neural network, different from said first secondary level neural network, that determines if a likely spam message from the first neural network level is a spam message or a bulk message;

for messages classified as bulk, providing user access to at least a listing of message subject field data corresponding to said bulk messages; and

for messages classified as spam, blocking delivery of the messages to at least one intended recipient.

2. A method as in claim 1, wherein for at least one of the classifications the neural networks classify the message in one of three classifications, wherein more than one path through the neural networks exists for the message to arrive at that classification.

3. A method as in claim 1, further comprising the step of dynamically maintaining the neural networks responsive to classification of the message.

4. A method as in claim 1, further comprising the step of applying rules to the message to help classify the message.

5. A method as in claim 4, wherein if the message is classified by the rules, the step of applying the neural networks is skipped.

6. A method as in claim 4, wherein the rules utilize a whitelist, a blacklist, or both the whitelist and the blacklist.

7. A method as in claim 6, further comprising the step of dynamically maintaining the whitelist, the blacklist, or both the whitelist and the blacklist responsive to classification of the message.

8. A method as in claim 6, wherein the step of recognizing expressions further includes the step of applying a genetic algorithm to select a set of regular expressions to be recognized.