System for determining degrees of similarity in email message information

• US 7,590,694 B2
• Filed: 01/16/2004
• Issued: 09/15/2009
• Est. Priority Date: 01/16/2004
• Status: Expired due to Fees

• Current Assignee: Gozoom.com, Inc.
• Original Assignee: Gozoom.com, Inc.
• Inventors: Yu, Tonny
• Primary Examiner(s): Flynn; Nathan J
• Assistant Examiner(s): Park; Jeong S
• Application Number:
  US10/760,140
• Publication Number:
  US 20050160148A1
• Time in Patent Office: 2,069 Days
• Field of Search: 709/206, 709/207
• 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

Similarity of email message characteristics is used to detect bulk and spam email. A determination of “sameness” for purposes of both bulk and spam classifications can use any number and type of evaluation modules. Each module can include one or more rules, tests, processes, algorithms, or other functionality. For example, one type of module may be a word count of email message text. Another module can use a weighting factor based on groups of multiple words and their perceived meanings. In general, any type of module that performs a similarity analysis can be used. A preferred embodiment of the invention uses statistical analysis, such as Bayesian analysis, to measure the performance of different modules against a known standard, such as human manual matching. Modules that are performing worse than other modules can be valued less than modules having better performance. In this manner, a high degree of reliability can be achieved. To improve performance, if a message is determined to be the same as a previous message, the previous computations and results for that previous message can be re-used. Users can be provided with options to customize or regulate bulk and spam classification and subsequent actions on how to handle the classified email messages.

Claims

1. A method for classifying email messages, the method comprising:

providing multiple independent modules each of which is configured to analyze email messages;

receiving an email message;

using a plurality of the independent modules to determine a level of sameness of the received email message with one or more prior email messages, wherein each module being used determines a level of sameness in a different manner than the other modules being used, and wherein at least some of the modules being used are each assigned a non-zero weight indicative of the module'"'"'s performance level;

determining an overall level of sameness for the received email message by combining results of at least two of the plurality of independent modules using the non-zero weights assigned to the modules;

evaluating the performance level for each of the independent modules that were used to determine the level of sameness for the received email message;

comparing the performance levels evaluated for the independent modules that were used to determine the level of sameness for the received email message;

adjusting the non-zero weights of at least two of the modules in response to comparing the performance levels, including increasing the non-zero weight of at least one of the modules and reducing the non-zero weight of at least another one of the modules; and

using the overall level of sameness determined for the received email message to classify the received email message into a category.

2. The method of claim 1, further comprising:

comparing the number of email messages classified in the category with a predetermined number; and

if the number of email messages is greater than the predetermined number then classifying the category as a first category type;

else

classifying the category as a second category type.

3. The method of claim 2, wherein the first category type is bulk email.

4. The method of claim 2, further comprising:

accepting a signal from a user input device to indicate processing of email messages in a category.

5. The method of claim 4, wherein the processing includes preventing the email messages in a category from being delivered to a user.

6. The method of claim 1, wherein the category is commercial email.

7. The method of claim 1, wherein Bayesian analysis is used.

8. The method of claim 1, further comprising:

accepting a signal from a user input device to set a parameter; and

using the parameter to adjust a weighting.

9. The method of claim 1, wherein one of the plurality of independent modules being used analyzes word count in an email message.

10. The method of claim 1, wherein one of the plurality of independent modules being used analyzes similarity of text in an email message.

11. The method of claim 1, wherein one of the plurality of independent modules being used analyzes a similarity of sender addresses.

12. The method of claim 1, wherein one of the plurality of independent modules being used analyzes a similarity of network routing.

13. The method of claim 1, wherein one of the plurality of independent modules being used analyzes a similarity of a hash of information in an email message.

14. The method of claim 1, wherein a message classification in a bulk category includes a determination of whether the number of email messages in a category exceed a predetermined number, the method further comprising:

submitting email messages in the bulk category to analysis to determine the level of commercial text.

15. The method of claim 14, further comprising:

preventing messages with a predetermined level of commercial text from being sent to an intended recipient.

16. The method of claim 14, further comprising:

intercepting the email messages from being sent to an intended recipient;

collecting the intercepted messages for a period of time; and

determining whether the collected messages are bulk messages, and if so, submitting the email messages in the bulk category to analysis to determine a level of commercial text.

17. The method of claim 16, further comprising:

preventing messages with a predetermined level of commercial text from being sent to an intended recipient.

18. The method of claim 1, further comprising:

assigning a lower rating to a module with a low performance level.

19. The method of claim 1, further comprising:

assigning a higher weighting to a module with a high performance level.

20. The method of claim 1, further comprising:

preventing a module with a low performance level from being used in a subsequent determination of a level of sameness.

21. The method of claim 1, wherein the step of evaluating comprises using non-manual statistical analysis to compare the level of sameness determined by each of the used independent modules to a benchmark.

22. The method of claim 21, wherein the step of adjusting comprises:

increasing the non-zero weight of one or more of the modules that determine a substantially similar level of sameness as the benchmark; and

decreasing the non-zero weight of one or more of the modules that determine a substantially different level of sameness than the benchmark.

23. The method of claim 1, wherein each determined level of sameness is an indication of a degree of certainty that the received email message is the same as one or more prior email messages.

24. The method of claim 1, wherein at least one of the used independent modules determines a level of sameness of the entire received email message with the entire one or more prior email messages.

25. The method of claim 1, wherein:

a first one of the provided independent modules is configured to analyze similarity of word counts of email messages;

a second one of the independent modules is configured to analyze similarity of text of email messages;

a third one of the independent modules is configured to analyze similarity of sender addresses of email messages; and

a fourth one of the independent modules is configured to analyze similarity of network routings of email message.

26. An apparatus for classifying email messages, the apparatus comprising

a processor for executing instructions included in a machine-readable medium, the machine-readable medium including:

one or more instructions for providing multiple independent modules each of which is configured to analyze email messages;

one or more instructions for receiving an email message;

one or more instructions for using a plurality of the independent modules to determine a level of sameness of the received email message with one or more prior email messages, wherein each module being used determines a level of sameness in a different manner than the other modules being used, and wherein at least some of the modules being used are each assigned a non-zero weight indicative of the module'"'"'s performance level;

one or more instructions for determining an overall level of sameness for the received email message by combining results of at least two of the plurality of independent modules using the non-zero weights assigned to the modules;

one or more instructions for evaluating the performance level for each of the independent modules that were used to determine the level of sameness for the received email message;

one or more instructions for comparing performance levels evaluated for the independent modules that were used to determine the level of sameness for the received email message;

one or more instructions for adjusting the non-zero weights of at least in response to comparing the performance levels, including increasing the non-zero weight of at least one of the modules and reducing the non-zero weight of at least another one of the modules; and

one or more instructions for using the overall level of sameness determined for the received email message to classify the received email message into a category.

27. A machine-readable storage medium including instructions executable by a processor for classifying email messages, the machine-readable storage medium including:

one or more instructions for providing multiple independent modules each of which is configured to analyze email messages;

one or more instructions for receiving an email message;

one or more instructions for using a plurality of the independent modules to determine a level of sameness of the received email message with one or more prior email messages, wherein each module being used determines a level of sameness in a different manner than the other modules being used, and wherein at least some of the modules being used are each assigned a non-zero weight indicative of the module'"'"'s performance level;

one or more instructions for determining an overall level of sameness for the received email message by combining results of at least two of the plurality of independent modules using the non-zero weights assigned to the modules;

one or more instructions for evaluating the performance level for each of the independent modules that were used to determine the level of sameness for the received email message;

one or more instructions for comparing performance levels evaluated for the independent modules that were used to determine the level of sameness for the received email message;

one or more instructions for adjusting the non-zero weights of at least in response to comparing the performance levels, including increasing the non-zero weight of at least one of the modules and reducing the non-zero weight of at least another one of the modules; and

one or more instructions for using the overall level of sameness determined for the received email message to classify the received email message into a category.

28. An apparatus for classifying email messages, the apparatus comprising:

means for providing multiple independent modules each of which is configured to analyze email messages;

means for receiving an email message;

means for using a plurality of the independent modules to determine a level of sameness of the received email message with one or more prior email messages, wherein each module being used determines a level of sameness in a different manner than the other modules being used, and wherein at least some of the modules being used are each assigned a non-zero weight indicative of the module'"'"'s performance level;

means for determining an overall level of sameness for the received email message by combining results of at least two of the plurality of independent modules using the non-zero weights assigned to the modules;

means for evaluating the performance level for each of the independent modules that were used to determine the level of sameness for the received email message;

means for comparing performance levels evaluated for the independent modules that were used to determine the level of sameness for the received email message;

means for adjusting the non-zero weights of at least two of the modules in response to comparing the performance levels, including increasing the non-zero weight of at least one of the modules and reducing the non-zero weight of at least another one of the modules; and

means for using the overall level of sameness determined for the received email message to classify the received email message into a category.