Adaptive junk message filtering system

• US 7,640,313 B2
• Filed: 07/17/2007
• Issued: 12/29/2009
• Est. Priority Date: 02/25/2003
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Platt, John C., Rounthwaite, Robert L., Goodman, Joshua T., Heckerman, David E., Kadie, Carl M.
• Primary Examiner(s): Winder; Patrice
• Assistant Examiner(s): NGUYEN, MINH CHAU
• Application Number:
  US11/779,263
• Publication Number:
  US 20080010353A1
• Time in Patent Office: 896 Days
• Field of Search: 709223-224
• US Class Current: 709/206
• 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

The invention relates to a system for filtering messages—the system includes a seed filter having associated therewith a false positive rate and a false negative rate. A new filter is also provided for filtering the messages, the new filter is evaluated according to the false positive rate and the false negative rate of the seed filter, the data used to determine the false positive rate and the false negative rate of the seed filter are utilized to determine a new false positive rate and a new false negative rate of the new filter as a function of threshold. The new filter is employed in lieu of the seed filter if a threshold exists for the new filter such that the new false positive rate and new false negative rate are together considered better than the false positive and the false negative rate of the seed filter.

Claims

1. A system that facilitates adaptive data filtering, comprising:

a processor;

a memory communicatively coupled to the processor, the memory having stored therein computer-executable instructions configured to implement the data filtering system, including;

a first filter configured to label messages as junk based upon junk information associated with the messages, wherein the first filter is associated with a first accuracy rate;

a second filter configured to label the messages as junk based upon junk information associated with the messages, the second filter is initially associated with the first accuracy rate;

a filter output configured to receive labeled and unlabeled messages from the first filter and the second filter;

a user correction component configured to receive user actions overriding the initial labeling of the messages received at the filter output and calculate a first accuracy rate based upon the user actions; and

a filter control component configured to

train the second filter utilizing a threshold and the user actions, wherein if the probability that a message is junk exceeds the threshold, then the filter is trained to label the message as junk;

calculate a second accuracy rate for the second filter; and

route subsequently received messages to the second filter in lieu of the first filter if the second accuracy rate is better than the first accuracy rate.

2. The system of claim 1, wherein the second filter is employed in combination with the first filter.

3. The system of claim 1, wherein the junk information includes at least one of sender information, source IP address, sender name, sender e-mail address, sender domain name, unintelligible alphanumeric strings in identifier fields, terms and phrases in message text, features in message text, or embedded links to pop-up advertisements second.

4. The system of claim 1, wherein the first accuracy rate and the second accuracy rate comprises a false positive rate and a false negative rate.

5. The system of claim 1, wherein the second accuracy rate is a function of the threshold.

6. The system of claim 1, wherein the first filter is a seed filter configured to identify typical junk messages according to historical data.

7. The system of claim 1, wherein the user actions include at least one of explicitly or implicitly tagging the message as one of a junk message or a non-junk message.

8. The system of claim 1, wherein the user actions include at least one of tagging a message as a non-junk message, reading and deleting a message, forwarding a message, or replying to a message.

9. The system of claim 1, wherein the threshold is an optimal probability threshold (p*), wherein p*=N/(N+1), where N is a number of messages.

10. A method having stored computer-executable instructions that are executable on a processor that facilitates adaptive data filtering, the method comprising:

labeling messages by a first filter, as junk based upon junk information associated with the messages, wherein the first filter is associated with a first accuracy rate;

labeling the messages by a second filter, as junk based upon junk information associated with the messages, a second filter is initially associated with the first accuracy rate;

receiving by a filter output, labeled and unlabeled messages from the first filter and the second filter;

receiving by a user correction component, user actions overriding the initial labeling of the messages received at the filter output and calculating a first accuracy rate based upon the user actions; and

including a filter control component configured to;

training the second filter utilizing a threshold and the user actions, wherein if a probability that a message is junk exceeds the threshold, then the filter is trained to label the message as junk;

calculating a second accuracy rate for the second filter; and

routing subsequently received messages to the second filter in lieu of the first filter if the second accuracy rate is better than the first accuracy rate;

wherein the junk information includes at least one of sender information, source IP address, sender name, sender e-mail address, sender domain name, unintelligible alphanumeric strings in identifier fields, terms and phrases in message text, features in message text, or embedded links to pop-up advertisements.