Dynamic rule reordering for message classification

• US 9,116,879 B2
• Filed: 05/25/2011
• Issued: 08/25/2015
• Est. Priority Date: 05/25/2011
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Gandhi, Mauktik H., Kavishwar, Shashank, Lamanna, Charles W.
• Primary Examiner(s): AL HASHEMI, SANA A
• Application Number:
  US13/115,804
• Publication Number:
  US 20120303624A1
• Time in Patent Office: 1,553 Days
• Field of Search: 707/687, 707/706, 707/736, 707/740, 707/758, 707/813
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24532 of parallel queries
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: H04L 51/212 using filtering or selective blocking
• CPC Class Codes: H04L 51/42 Mailbox-related aspects, e.g. synchronisation of mailboxes

Abstract

Embodiments are directed to generating a customized classification rule execution order and to identifying optimal ordering rules for previously processed data. In an embodiment, a computer system fingerprints a message received via a computer network. The fingerprinting identifies specific characteristics of the message. The computer system compares the message'"'"'s fingerprint to various stored message fingerprints generated from previously received messages. The comparison determines that the fingerprint does not match the stored fingerprints. The computer system applies classification rules to the message according to a predetermined rule execution order to determine a classification for the message. The computer system then generates a customized classification rule execution order to order those classification rules that optimally identified the message'"'"'s class at the top of the customized classification rule execution order.

Claims

1. At a computer system including at least one processor and a memory, in a computer networking environment including a plurality of computing systems, a computer-implemented method for generating a customized classification rule execution order, the method comprising:

identifying one or more specific characteristics of a message received via a computer network;

fingerprinting the message based at least in part on the identified one or more specific characteristics of the message;

comparing the message'"'"'s fingerprint to one or more stored message fingerprints generated from previously received messages, the comparison determining that the fingerprint does not match the stored fingerprints;

applying a plurality of classification rules to the message to determine a classification for the message, the rules applied sequentially according to a predetermined rule execution order; and

generating a customized classification rule execution order, based on the results of the sequential application in the predetermined order of the plurality of classification rules to the message, the customized classification rule execution order defining a new sequential order of the classification rules that optimally identifies the message'"'"'s class by applying the fewest number of rules of the customized classification rule execution order when applied sequentially to the message.

2. The method of claim 1, further comprising associating the generated customized classification rule execution order with the fingerprint for the message.

3. The method of claim 2, further comprising:

an act of fingerprinting a subsequently received message;

an act of determining that fingerprint of the subsequently received message matches the fingerprint generated for the original message; and

an act of using the generated customized classification rule execution order to process the subsequently received message.

4. The method of claim 3, further comprising refining the generated customized classification rule execution order based on the execution of the subsequently received message.

5. The method of claim 1, wherein the customized classification rule execution order is changed dynamically based on one or more portions of aggregated data.

6. The method of claim 1, wherein the classification rules are placed at the beginning of the rule execution order.

7. The method of claim 6, wherein the generated customized classification rule execution order comprises the least number of classification rules used to properly classify a message.

8. The method of claim 1, further comprising refining the customized classification rule execution order for fingerprints that match messages on a sufficiently frequent basis.

9. The method of claim 8, wherein fingerprints that do not sufficiently frequently match messages are removed.

10. The method of claim 1, wherein the classification rules include positive score rules and negative score rules, and wherein the customized classification rule execution order is determined solely for the positive score rules.

11. The method of claim 1, wherein the classification rules include positive score rules and negative score rules, and wherein the customized classification rule execution order is determined solely for the negative score rules.

12. The method of claim 1, further comprising:

an act of receiving a first subsequent message;

an act of saving the first subsequent message and a corresponding determined customized classification rule execution order for the first subsequent message;

an act of receiving a second subsequent message; and

an act of determining that the second subsequent message matches both the originally received message and the first subsequent message.

13. The method of claim 12, further comprising applying a combination of the customized classification rule execution order of the original message and the customized classification rule execution order of the first subsequent message.

14. The method of claim 12, further comprising applying at least one of the customized classification rule execution order of the original message and the customized classification rule execution order of the first subsequent message, according to which is the most similar to the second subsequent message'"'"'s customized classification rule execution order.

15. The method of claim 12, further comprising an act of applying at least one of the customized classification rule execution order of the original message and the customized classification rule execution order of the first subsequent message, according to which message is the most similar to the second subsequent message.

16. A computer program product for implementing a method for identifying optimal ordering rules for previously processed data, the computer program product comprising one or more computer-readable storage media having stored thereon computer-executable instructions that, when executed by one or more processors of the computing system, cause the computing system to perform the method, the method comprising:

identifying one or more specific characteristics of a message received via a computer network;

fingerprinting the message based at least in part on the identified one or more specific characteristics of the message;

comparing the message'"'"'s fingerprint to one or more stored message fingerprints generated from previously received messages, wherein the comparison determines that the fingerprint matches one or more of the stored fingerprints;

accessing a previously generated customized classification rule execution order corresponding to the matched fingerprint, the customized classification rule execution order having been determined by executing a plurality of classification rules upon the previously received messages;

sequentially applying one or more of the plurality of classification rules according to the accessed customized classification rule execution order to the message; and

determining a classification for the message based at least in part on the sequential application of the one or more of the plurality of classification rules.

17. The computer program product of claim 16, further comprising refining the customized classification rule execution order based on the execution of the customized classification rule execution order on the received message.

18. The computer program product of claim 16, further comprising an act of storing a new rule execution order based on the message classification.

19. The computer program product of claim 16, further comprising:

an act of receiving a subsequent message;

an act of determining that the original message is included in the subsequent message; and

an act of applying the customized classification rule execution order of the original message.

20. A computer system comprising the following:

one or more processors;

system memory;

one or more computer-readable storage media having stored thereon computer-executable instructions that, when executed by the one or more processors, causes the computing system to perform a method for generating a customized classification rule execution order, the method comprising the following;

identifying one or more specific characteristics of a message received via a computer network;

fingerprinting the message based at least in part on the identified one or more specific characteristics of the message;

comparing the message'"'"'s fingerprint to one or more stored message fingerprints generated from previously received messages, wherein the comparison determines that the fingerprint does not match the stored fingerprints;

applying a plurality of classification rules to the message to determine a classification for the message, the rules applied sequentially according to a predetermined rule execution order;

generating a customized classification rule execution order based on the results of the sequential application in the predetermined order of the plurality of classification rules to the message, the customized classification rule execution order defining a new sequential order of the classification rules that optimally identifies the message'"'"'s class by applying the fewest number of rules of the customized classification rule execution order when applied sequentially to the message;

associating the generated customized classification order with the fingerprint for the message;

fingerprinting a subsequently received message;

determining that fingerprint of the subsequently received message matches the fingerprint generated for the original message; and

using the generated customized classification rule execution order to process the subsequently received message.