Answer Support System and Method

• US 20110161274A1
• Filed: 11/29/2010
• Published: 06/30/2011
• Est. Priority Date: 11/30/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Gao, Peng, Zou, Qing, Yang, Jie, Zhou, Wei
• Granted Patent: US 8,595,174 B2
• Time in Patent Office: Days
• US Class Current: 706/50
• CPC Class Codes: G06F 16/24573 using data annotations, e.g. user-defined metadata
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06Q 10/10 Office automation; Time management

Abstract

An answer support system generates a target cause to aid in solving a current problem. Upon receiving known problem symptoms of the current problem from the customer, a search of problem causes is conducted in a knowledge database to find potential causes corresponding to the known problem symptoms. If conventional techniques cannot determine the target cause of the current problem, an attribute is selected which is not among the attributes associated with the known problem symptoms in the database. The attribute selection is based on the attribute deemed most likely to identify the target cause of the current problem from among the potential causes found in the knowledge database.

Claims

1. A method of generating a target cause in an answer support system to aid in solving a current problem, comprising:

receiving known problem symptoms of the current problem, said known problem symptoms being included among stored problem symptoms associated with problem causes stored in a knowledge database;

searching the problem causes in said knowledge database to find a plurality of potential causes corresponding to the known problem symptoms;

attempting to determine a target cause of the current problem from among the plurality of potential causes based on attributes associated with the known problem symptoms;

in response to an inability to determine the target cause of the current problem, choosing as a selected attribute an attribute which is not among the attributes associated with the known problem symptoms, wherein an attribute value of the selected attribute is determined to be most likely to identify the target cause of the current problem from said plurality of potential causes found in the knowledge database;

determining the target cause based on the selected attribute; and

generating a solution based on the target cause.

2. The method of claim 1, further comprising:

maintaining the knowledge database to contain historical data mapping relationships between the problem causes and the stored problem symptoms;

wherein said attempting to determine the target cause is based on a difference between the historical data for the plurality of potential causes with respect to the known problem symptoms in the knowledge database.

3. The method of claim 1, wherein the selected attribute is a first selected attribute and the target cause is a first target cause, the method further comprising:

generating and sending a question associated with the selected attribute;

receiving as an answer to the question a new known problem symptom, and using the new known problem symptom as an attribute value for a second selected attribute; and

determining a second target cause based on the second selected attribute.

4. The method of claim 1, further comprising:

providing the solution to the current problem to a customer; and

in response to receiving a feedback that the current problem has been solved, storing the mapping relationship between the target cause of the current problem and the problem symptoms into the knowledge database.

5. The method of claim 1, wherein the structure of the knowledge database is a matrix, its rows represent the problem causes, its columns represent the attributes associated with the stored problem symptoms;

and

wherein elements in said matrix are number values representing respective ones of the attributes associated with the stored problem symptoms.

6. The method of claim 1, wherein the attempt to determine the target cause of the current problem is based on a known information ratio, the method further comprising:

calculating a known information ratio of each potential cause in the plurality of potential causes, wherein, the known information ratio of each potential cause is calculated as a sum of number values of the potential cause with respect to the known problem symptoms divided by the sum of the number values of the potential cause with respect to the known problem symptoms plus the sum of the number values of the potential cause with respect to all the problem symptoms of all the remaining attributes besides the attributes to which the known problem symptoms belong;

sorting the plurality of potential causes in descending order according to their known information ratios;

determining whether the difference between the known information ratios of the potential cause with the maximum known information effective ratio and the potential cause with the second maximum known information ratio is greater than a threshold; and

in response to the determination being yes, determining the potential cause with the maximum known information ratio as the target cause of the current problem.

7. The method of claim 6, wherein the threshold is the average value of the differences between the known potential information ratios of every two adjacent potential causes of the sorted plurality of potential causes.

8. The method of claim 6, further comprising:

for each remaining attribute in the remaining attributes, by using the number values of the potential cause with the maximum known information ratio in the plurality of potential causes with respect to the problem symptoms in the remaining attribute as the number of positive examples, and using the number values of all the other potential causes in the plurality of potential causes with respect to the problem symptoms in the remaining attribute as the number of negative examples, calculating the information gain of the remaining attribute; and

selecting the remaining attribute with the maximum information gain in the remaining attributes as the attribute to be determined.

9. The method of claim 6, wherein calculating dependencies of the plurality of causes is performed by using the Chi-square analysis.

10. A computer program product comprising a machine readable non-transitory medium including a program of instructions for generating a target cause in an answer support system to aid in solving a current problem, wherein the program of instructions upon being executed on a computer causes the computer to perform activities comprising:

receiving known problem symptoms of the current problem, said known problem symptoms being included among stored problem symptoms associated with problem causes stored in a knowledge database;

searching the problem causes in said knowledge database to find a plurality of potential causes corresponding to the known problem symptoms;

attempting to determine a target cause of the current problem from among the plurality of potential causes based on attributes associated with the known problem symptoms;

in response to an inability to determine the target cause of the current problem, choosing as a selected attribute an attribute which is not among the attributes associated with the known problem symptoms, wherein an attribute value of the selected attribute is determined to be most likely to identify the target cause of the current problem from said plurality of potential causes found in the knowledge database;

determining the target cause based on the selected attribute; and

generating a solution based on the target cause.

11. The computer program product of claim 10, further comprising:

maintaining the knowledge database to contain historical data mapping relationships between the problem causes and the stored problem symptoms;

wherein said attempting to determine the target cause is based on a difference between the historical data for the plurality of potential causes with respect to the known problem symptoms in the knowledge database.

12. The computer program product of claim 10, wherein the selected attribute is a first selected attribute and the target cause is a first target cause, the method further comprising:

generating and sending a question associated with the selected attribute;

receiving as an answer to the question a new known problem symptom, and using the new known problem symptom as an attribute value for a second selected attribute; and

determining a second target cause based on the second selected attribute.

13. The computer program product of claim 10, further comprising:

providing the solution to the current problem to a customer; and

in response to receiving a feedback that the current problem has been solved, storing the mapping relationship between the target cause of the current problem and the problem symptoms into the knowledge database.

14. The computer program product of claim 10, wherein the structure of the knowledge database is a matrix, its rows represent the problem causes, its columns represent the attributes associated with the stored problem symptoms;

and

wherein elements in said matrix are number values representing respective ones of the attributes associated with the stored problem symptoms.

15. The computer program product of claim 10, wherein the attempt to determine the target cause of the current problem is based on a known information ratio, the method further comprising:

calculating a known information ratio of each potential cause in the plurality of potential causes, wherein, the known information ratio of each potential cause is calculated as a sum of number values of the potential cause with respect to the known problem symptoms divided by the sum of the number values of the potential cause with respect to the known problem symptoms plus the sum of the number values of the potential cause with respect to all the problem symptoms of all the remaining attributes besides the attributes to which the known problem symptoms belong;

sorting the plurality of potential causes in descending order according to their known information ratios;

determining whether the difference between the known information ratios of the potential cause with the maximum known information effective ratio and the potential cause with the second maximum known information ratio is greater than a threshold; and

in response to the determination being yes, determining the potential cause with the maximum known information ratio as the target cause of the current problem.

16. The computer program product of claim 15, wherein the threshold is the average value of the differences between the known potential information ratios of every two adjacent potential causes of the sorted plurality of potential causes.

17. The computer program product of claim 15, further comprising:

for each remaining attribute in the remaining attributes, by using the number values of the potential cause with the maximum known information ratio in the plurality of potential causes with respect to the problem symptoms in the remaining attribute as the number of positive examples, and using the number values of all the other potential causes in the plurality of potential causes with respect to the problem symptoms in the remaining attribute as the number of negative examples, calculating the information gain of the remaining attribute; and

selecting the remaining attribute with the maximum information gain in the remaining attributes as the attribute to be determined.

18. The computer program product of claim 15, wherein calculating dependencies of the plurality of causes is performed by using the Chi-square analysis.

19. An answer support system configured to generate a target cause to aid in solving a current problem, comprising:

a network interface configured to receive known problem symptoms of the current problem;

a storage memory configured to store a knowledge database comprising said known problem symptoms included among stored problem symptoms associated with problem causes, wherein said knowledge database is configured for searching the problem causes to find a plurality of potential causes corresponding to the known problem symptoms; and

a microprocessor configured to attempt to determine a target cause of the current problem from among the plurality of potential causes based on attributes associated with the known problem symptoms;

wherein, in response to an inability to determine the target cause of the current problem, the answer support system is configured to choose as a selected attribute an attribute which is not among the attributes associated with the known problem symptoms, wherein an attribute value of the selected attribute is determined to be most likely to identify the target cause of the current problem from said plurality of potential causes found in the knowledge database, determine the target cause based on the selected attribute, and generate a solution based on the target cause.

20. The answer support system of claim 19, further comprising:

wherein the knowledge database is maintained to contain historical data mapping relationships between the problem causes and the stored problem symptoms;

wherein said attempting to determine the target cause is based on a difference between the historical data for the plurality of potential causes with respect to the known problem symptoms in the knowledge database; and

wherein the selected attribute is a first selected attribute and the target cause is a first target cause, the method further comprising;

the microprocessor further being configured to generate and send a question associated with the selected attribute; and

the network interface further being configured to receive as an answer to the question a new known problem symptom, and using the new known problem symptom as an attribute value for a second selected attribute, wherein a second target cause is based on the second selected attribute.