Methods, Systems, and Products for Estimating Answers to Questions

• US 20110153537A1
• Filed: 12/19/2009
• Published: 06/23/2011
• Est. Priority Date: 12/19/2009
• Status: Active Grant

• Current Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: Berthier, Robin, Vesonder, Gregory T., Cukier, Michel, Hiltunen, Matti, Kormann, David, Sheleheda, Daniel
• Granted Patent: US 8,626,691 B2
• Time in Patent Office: Days
• US Class Current: 706/47
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound

Abstract

Methods, systems, and products answer questions using heuristics. A question and an input are received. A set of heuristic rules is retrieved, and the question is answered by evaluating an accuracy of each heuristic rule. Multiple accuracies from the set of heuristic rules are combined to produce a confidence level for the answer to the question.

Claims

1. A method of answering questions, comprising:

receiving a question;

receiving an input;

retrieving a set of heuristic rules from memory;

answering the question by evaluating an accuracy of each heuristic rule;

combining accuracies from the set of heuristic rules to produce a confidence level for the answer to the question.

2. The method according to claim 1, further comprising combining the accuracies using a Bayesian inference.

3. The method according to claim 1, further comprising receiving Internet Protocol traffic information as the input.

4. The method according to claim 1, further comprising receiving NetFlow information as the input.

5. The method according to claim 1, further comprising merging NetFlow information having same source and destination end points.

6. The method according to claim 5, further comprising generating bidirectional flows by merging flows collected during a time window that have opposite source and destination end points.

7. The method according to claim 6, further comprising separating valid from invalid bidirectional flows by i) considering all UDP flows as valid and ii) considering TCP flows valid only if a request and a reply carry at least two packets and a TCP acknowledgement flag.

8. A system, comprising:

a processor executing code stored in memory, the code causing the processor to;

receiving a question;

receiving an input;

retrieving a set of heuristic rules from memory;

answering the question by evaluating an accuracy of each heuristic rule;

combining accuracies from the set of heuristic rules to produce a confidence level for the answer to the question.

9. The system according to claim 8, further comprising code that causes the processor to combine the accuracies using a Bayesian inference.

10. The system according to claim 8, further comprising code that causes the processor to receive Internet Protocol traffic information as the input.

11. The system according to claim 8, further comprising code that causes the processor to receive NetFlow information as the input.

12. The system according to claim 8, further comprising code that causes the processor to merge NetFlow information having same source and destination end points.

13. The system according to claim 8, further comprising code that causes the processor to generate bidirectional flows by merging flows collected during a time window that have opposite source and destination end points.

14. The system according to claim 13, further comprising code that causes the processor to separate valid from invalid bidirectional flows by i) considering all UDP flows as valid and ii) considering TCP flows valid only if a request and a reply carry at least two packets and a TCP acknowledgement flag.

15. A computer readable medium storing processor executable instructions for performing a method, the method comprising:

receiving a question;

receiving an input;

retrieving a set of heuristic rules from memory;

answering the question by evaluating an accuracy of each heuristic rule;

combining accuracies from the set of heuristic rules to produce a confidence level for the answer to the question.

17. The computer readable medium according to claim 17, further comprising instructions for receiving Internet Protocol traffic information as the input.

16. The computer readable medium according to claim 17, further comprising instructions for combining the accuracies using a Bayesian inference.

18. The computer readable medium according to claim 17, further comprising instructions for receiving NetFlow information as the input.

19. The computer readable medium according to claim 17, further comprising instructions for merging NetFlow information having same source and destination end points.

20. The computer readable medium according to claim 17, further comprising instructions for generating bidirectional flows by merging flows collected during a time window that have opposite source and destination end points.