Dynamic determination of needed agent rules

• US 8,423,496 B1
• Filed: 12/22/2006
• Issued: 04/16/2013
• Est. Priority Date: 12/22/2006
• Status: Active Grant

• Current Assignee: Callahan Cellular LLC (Intellectual Ventures LLC)
• Original Assignee: Curen Software Enterprises LLC (Intellectual Ventures LLC)
• Inventors: Hauser, Robert R.
• Primary Examiner(s): Wong, Lut
• Application Number:
  US11/645,191
• Time in Patent Office: 2,307 Days
• Field of Search: 706/47
• US Class Current: 706/47
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/025 Extracting rules from data
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

A method for updating rule statistics in an execution environment by an agent that utilizes a rule engine, comprising collecting an activation statistic when a rule is activated, collecting a firing statistic when a rule is fired; and storing the activation statistic and the firing statistic in a rule history once processing by the rule engine is complete.

Claims

1. A tangible computer readable medium comprising instructions stored thereon, the instructions comprising:

instructions to request, from a remote rule repository, a first set of canonical rules associated with a first execution environment and a prior execution environment;

instructions to merge the first set of canonical rules with a second set of canonical rules associated with the first execution environment to form a total canonical rule set, wherein the first set of canonical rules and the second set of canonical rules are associated with a goal;

instructions to determine whether a threshold number of outcomes has been obtained;

instructions to remove, in response to the threshold number of outcomes being obtained, at least one rule from the total canonical rule set to form a pruned canonical rule set, wherein the pruned canonical rule set is configured to be moved from the first execution environment to a second execution environment in combination with moving an agent from the first execution environment to the second execution environment, and the removal of the at least one rule from the total canonical rule set is based at least in part on aggregate statistics associated with previously executed agent results and outcomes associated with the total canonical rule set, wherein the aggregate statistics indicate that the removed at least one rule was active in at least one negative outcome associated with the total canonical rule set;

instructions to encode the agent and the pruned canonical rule set to form an encoded agent and an encoded canonical pruned rule set; and

instructions to move the encoded agent and the encoded canonical pruned rule set from the first execution environment to the second execution environment.

2. The tangible computer readable medium of claim 1, further comprising instructions to store a previous agent outcome indexed by an agent identifier in the agent.

3. The tangible computer readable medium of claim 1, wherein the instructions to request comprise instructions to request the first set of canonical rules from a rule repository in the first execution environment.

4. The tangible computer readable medium of claim 3, further comprising instructions to request a rule history from the rule repository in the first execution environment.

5. The tangible computer readable medium of claim 1, further comprising instructions to determine the at least one rule by comparing a previous agent outcome against a rule history.

6. The tangible computer readable medium of claim 5, further comprising instructions to use rule activation or firing as a criterion for rule inclusion or exclusion.

7. The tangible computer readable medium of claim 5, further comprising instructions to use a rule left hand side condition being met as a criterion for rule inclusion or exclusion.

8. A computer-implemented method comprising:

requesting, from a remote rule repository, a first set of canonical rules associated with a first execution environment and a prior execution environment;

merging the first set of canonical rules with a second set of canonical rules associated with the first execution environment to form a total canonical rule set, wherein the first set of canonical rules and the second set of canonical rules are associated with a goal;

determining whether a threshold number of outcomes has been obtained;

removing, in response to the threshold number of outcomes being obtained, at least one rule from the total canonical rule set to form a pruned canonical rule set, wherein the pruned canonical rule set is configured to be moved from the first execution environment to a second execution environment in combination with moving an agent from the first execution environment to the second execution environment, and the removal of the at least one rule from the total canonical rule set is based at least in part on aggregate statistics associated with previously executed agent results and outcomes associated with the total canonical rule set, wherein the aggregate statistics indicate that the removed at least one rule was active in at least one negative outcome associated with the total canonical rule set;

encoding the agent and the pruned canonical rule set to form an encoded agent and an encoded canonical pruned rule set; and

moving the encoded agent and the encoded canonical pruned rule set from the first execution environment to the second execution environment.

9. The method of claim 8, further comprising storing a previous agent outcome indexed by an agent identifier in the agent.

10. The method of claim 8, wherein the requesting is performed by requesting the first set of canonical rules from a rule repository in the first execution environment.

11. The method of claim 10, further comprising requesting a rule history from the rule repository in the first execution environment.

12. The method of claim 8, further comprising determining the at least one rule by comparing a previous agent outcome against a rule history.

13. The method of claim 12, further comprising using rule activation or firing as a criterion for rule inclusion or exclusion.

14. The method of claim 12, further comprising using a rule left hand side condition being met as a criterion for rule inclusion or exclusion.

15. A system, comprising:

a memory comprising a working memory; and

a processor configured to be communicably coupled to the first memory, wherein the processor is configured to;

request, from a remote rule repository, a first set of canonical rules associated with a first execution environment and a prior execution environment;

merge the first set of canonical rules with a second set of canonical rules associated with the first execution environment to form a total canonical rule set, wherein the first set of canonical rules and the second set of canonical rules are associated with a goal;

determine whether a threshold number of outcomes has been obtained;

remove, if in response to the threshold number of outcomes being obtained, at least one rule from the total canonical rule set to form a pruned canonical rule set, wherein the pruned canonical rule set is configured to be moved from the first execution environment to a second execution environment in combination with moving an agent from the first execution environment to the second execution environment, and the removal of the at least one rule from the total canonical rule set is based at least in part on aggregate statistics associated with previously executed agent results and outcomes associated with the total canonical rule set, wherein the aggregate statistics indicate that the removed at least one rule was active in at least one negative outcome associated with the canonical rule set;

encode the agent and the pruned canonical rule set to form an encoded agent and an encoded canonical pruned rule set; and

move the encoded agent and the encoded canonical pruned rule set from the first execution environment to the second execution environment.

16. The system of claim 15, wherein the processor is further configured to store, in the agent, a previous agent outcome indexed by an agent identifier.

17. The system of claim 15, wherein the processor is further configured to request the first set of canonical rules from a rule repository in the first execution environment.

18. The system of claim 17, wherein the processor is further configured to request a rule history from the rule repository in the first execution environment.

19. The system of claim 15, wherein the processor is further configured to determine the at least one unnecessary rule by comparing a previous agent outcome against a rule history.

20. The system of claim 19, wherein the processor is further configured to use rule activation or firing as a criterion for rule inclusion or exclusion.

21. The system of claim 19, wherein the processor is further configured to use a rule left hand side condition being met as a criterion for rule inclusion or exclusion.