PREDICTING SYMPTOMS OF RUN-TIME PROBLEMS BASED ON ANALYSIS OF EXPERT DECISION MAKING

• US 20140068330A1
• Filed: 09/06/2012
• Published: 03/06/2014
• Est. Priority Date: 09/06/2012
• Status: Active Grant

• Current Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Original Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Inventors: Hecox, Andrew, Romig-Koch, Gavin
• Granted Patent: US 9,026,856 B2
• Time in Patent Office: Days
• US Class Current: 714/26
• CPC Class Codes: G06F 11/0706 the processing taking place on a specific hardware platform or in a specific software environment
• CPC Class Codes: G06F 11/079 Root cause analysis, i.e. error or fault diagnosis in a hardware test environment G06F11/22; in a software test environment G06F11/36
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: H04L 41/5061 characterised by the interaction between service providers and their network customers, e.g. customer relationship management

Abstract

Systems, methods, and non-transitory computer-readable storage media for receiving historical data describing behavior of human subject-matter experts, wherein the historical data links customer problems with solutions, receiving a plurality of human-generated algorithms describing patterns for linking customer problems with solutions based on problem-specific diagnostic data, comparing each algorithm of the plurality of algorithms with the historical data to determine respective predictive scores for linking a customer problem type with a particular solution, and ranking at least part of the plurality of algorithms based on the respective similarity scores.

Claims

1. A method comprising:

receiving historical data describing behavior of human subject-matter experts, wherein the historical data links customer cases with solutions;

identifying, from the historical data, a problem signature associated with a problem type;

applying an algorithm to the historical data to determine a predictive accuracy score for the algorithm; and

assigning the predictive accuracy score to the algorithm for the problem type.

2. The method of claim 1, wherein the historical data is structured data.

3. The method of claim 1, wherein the historical data comprises one of a stack trace, a core file, or a memory dump.

4. The method of claim 1, further comprising:

receiving additional historical data; and

upon determining that the additional historical data diverges at least a threshold distance from the historical data, updating the predictive accuracy score based on the additional historical data.

5. The method of claim 1, wherein identifying the problem signature further comprises:

comparing each algorithm of a plurality of algorithms with the historical data;

forming groups within the historical data based on a common problem signature; and

comparing the historical data based on the groups.

6. The method of claim 5, wherein the solutions are associated with a software product, the method further comprising:

flagging, based on ranking at least part of the plurality of algorithms and a reporting threshold, a portion of the software product as problematic.

7. A system comprising:

a processing device;

a memory having stored therein instructions which, when executed by the processing device, cause the processing device to;

receive log data associated with a problem;

identify a problem type of the problem;

select an algorithm having a predictive accuracy score associated with the problem type, wherein the predictive accuracy score is based on historical problem data; and

based on the algorithm, identify a likely resolution for the problem.

8. The system of claim 7, wherein the instructions, when executed by the processing device, further cause the processing device to:

bias the predictive accuracy score based on similarities of the historical data to an environment of the customer problem.

9. The system of claim 7, wherein the instructions, when executed by the processing device, further cause the processing device to:

identify a top ranked resolution as the likely resolution; and

perform an action to resolve the customer problem based on the top ranked resolution.

10. The system of claim 7, wherein the likely resolution is selected from a data store of knowledge base articles.

11. The system of claim 7, wherein the instructions, when executed by the processing device, further cause the processing device to:

determine that the historical data has insufficient information to identify the problem type of the problem;

identify a second customer problem representing a second problem type that is similar to the problem type; and

select the algorithm based on the second problem type.

12. The system of claim 7, wherein the likely resolution conforms with a solution format defined by an external organization.

13. The system of claim 7, wherein the instructions, when executed by the processing device, further cause the processing device to:

identify likely algorithms, from which the algorithm is selected, based on linking behavior of subject-matter experts in resolving problems.

14. The system of claim 7, wherein the instructions, when executed by the processing device, further cause the processing device to:

implement at least part of the likely resolution for the problem.

15. A non-transitory computer-readable storage medium having stored therein instructions which, when executed by a processing device, cause the processing device to perform a method comprising:

identifying, from historical data describing behavior of human subject-matter experts, a problem signature associated with a problem type, wherein the historical data links customer cases with solutions;

applying an algorithm to the historical data to determine a predictive accuracy score for the algorithm;

assigning the predictive accuracy score to the algorithm for the problem type;

receiving log data associated with a customer problem having a customer problem type;

identifying, based on the algorithm, the predictive accuracy score, and the customer problem type, a resolution for the problem;

applying the resolution to the customer problem; and

updating at least one of the algorithm or the predictive accuracy score based on results of applying the resolution to the customer problem.

16. The non-transitory computer-readable storage medium of claim 15, the instructions, when executed by the processing device, further causing the processing device to perform a method comprising:

tracking a success rate of the resolution; and

updating a data store of resolutions based on the success rate.

17. The non-transitory computer-readable storage medium of claim 15, wherein applying the resolution to the customer problem comprises at least one of presenting the resolution to a human problem resolution agent, presenting the resolution to a customer, and implementing at least part of the resolution via a customer computing device.

18. The non-transitory computer-readable storage medium of claim 15, wherein the log data is received via a network from an automatic reporting mechanism operating on a customer computing device.

19. The non-transitory computer-readable storage medium of claim 15, the instructions, when executed by the processing device, further causing the processing device to perform a method comprising:

filtering the historical data based on the customer problem type.

20. The non-transitory computer-readable storage medium of claim 15, wherein identifying the resolution is further based on a predictive accuracy threshold.