Method of selecting desired domains and for developing a seeding methodology for a knowledge base system

• US 6,493,697 B1
• Filed: 08/24/1999
• Issued: 12/10/2002
• Est. Priority Date: 08/24/1999
• Status: Expired due to Fees

• Current Assignee: Stream International Incorporated (TD SYNNEX Corp.)
• Original Assignee: Stream International Incorporated (TD SYNNEX Corp.)
• Inventors: Haughton, Debra Ann, Stier, Sharon
• Primary Examiner(s): Black, Thomas
• Assistant Examiner(s): Hirl, Joseph P.
• Application Number:
  US09/379,822
• Time in Patent Office: 1,204 Days
• Field of Search: 706/50, 706/45, 706/46, 706/59, 707/5, 707/3, 700/286, 700/97, 714/26
• US Class Current: 706/50
• CPC Class Codes: G06N 5/025 Extracting rules from data
• CPC Class Codes: Y10S 707/99936 Pattern matching access
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A method and a domain candidate benefit indicator for selecting knowledge domains for a knowledge base from at least one domain candidate involves developing a domain candidate benefit value for each domain candidate. Domain selection is based on a comparison of the domain candidate benefit values, which are numerical values developed by identifying for each domain candidate at least one forecasting attribute having an ability to forecast an extent of benefit of selecting the domain candidate to be the selected domain. Sub-domains may be selected by identifying sub-domain candidates in the selected domain and developing a sub-domain candidate benefit value for each sub-domain candidate. The selected sub-domain is selected based on a comparison of sub-domain candidate benefit values. The domain selection methodology may also be used to develop a knowledge base seeding methodology.

Claims

1. A method for selecting knowledge domains for knowledge bases, said method comprising:

identifying a domain candidate set having at least one domain candidate for a knowledge base;

developing a domain candidate benefit value for each said domain candidate, said domain candidate benefit value comprising a numerical value indicating a level of benefit of each said domain candidate associated therewith for becoming a selected knowledge domain for said knowledge base;

conducting a comparison of each said domain candidate benefit value; and

selecting said selected knowledge domain from said domain candidate set based on said comparison.

2. The method of claim 1, wherein developing said domain candidate benefit value further comprises:

identifying goals for said knowledge base;

identifying for said domain candidate set at least one forecasting attribute to forecast benefit extent, wherein said benefit extent comprises an extent of benefit, based on said goals, in selecting said domain candidate to be said selected knowledge domain;

for each said forecasting attribute, identifying selected forecasting instantiations of said forecasting attribute;

establishing said benefit extent for each of said selected forecasting instantiations; and

developing an attribute valuation system to demonstrate said benefit extent for each of said selected forecasting instantiations, said valuation system developed by assigning an attribute benefit value to each of said selected forecasting instantiations, each said attribute benefit value comprising a numerical value indicative of said benefit extent of said forecasting attribute associated thereto; and

for said domain candidate, developing an actual attribute value comprising numerical values and for each said forecasting attribute associated therewith, by developing an actual instantiation for said domain candidate;

assigning, for each said forecasting attribute associated therewith, an actual attribute score from said attribute valuation system of said forecasting attribute, said actual attribute score based on said actual instantiation; and

weighting said actual attribute score according to said goals to generate said actual attribute value; and

combining each said actual attribute value for said domain candidate to generate said domain candidate benefit value for said domain candidate.

3. The method of claim 2, wherein said combining comprises summing each said actual attribute value.

4. The method of claim 2, wherein said forecasting attribute is measurable, and said actual instantiation is identified by measuring said forecasting attribute.

5. The method of claim 2, wherein said actual instantiation is identified by identifying a characteristic of said forecasting attribute.

6. The method of claim 2,

wherein one of said goals comprises responding to requests for assistance;

wherein said forecasting attribute comprises a domain candidate problem frequency comprising a frequency of problems involving said domain candidate; and

wherein said actual attribute value comprises an actual domain candidate problem frequency value comprising a numerical value.

7. The method of claim 6, further comprising identifying said actual instantiation by measuring said domain candidate problem frequency with a percentage of queue metric comprising a percentage of requests for assistance involving said domain candidate compared to a total number of all requests for assistance.

8. The method of claim 2,

wherein one of said goals comprises responding to requests for assistance;

wherein said forecasting attribute comprises a first time fix frequency, which comprises a frequency that problems in said domain candidate are resolved in a first request for assistance; and

wherein said actual attribute value comprises an actual first time fix frequency value comprising a numerical value.

9. The method of claim 8, further comprising identifying said actual instantiation by measuring said first time fix frequency with a first time fix rate metric.

10. The method of claim 2, further comprising organizing said forecasting attribute into sub-attributes, developing an attribute valuation system for said sub-attributes, and treating said sub-attributes as forecasting attributes when evaluating said domain candidate.

11. The method of claim 2,

wherein one of said goals comprises responding to requests for assistance;

wherein one of said forecasting attribute comprises an escalation frequency, which comprises a frequency that problems involving said domain candidate are forwarded for assistance in resolution; and

wherein said actual attribute value comprises an actual escalation frequency value comprising a numerical value.

12. The method of claim 11, further comprising identifying said actual instantiation by measuring said escalation frequency with a first metric comprising a rate of escalation to mentor metric, with a mentor escalation attribute score being assigned based on said rate of escalation to mentor metric.

13. The method of claim 12, further comprising:

measuring said escalation frequency with a second metric comprising a rate of escalation to customer metric, with a customer escalation attribute score being assigned based on said rate of escalation to customer mentor metric;

assigning an escalation to mentor attribute weight; and

assigning an escalation to customer attribute weight that is greater than said escalation to mentor attribute weight.

14. The method of claim 12, further comprising identifying said actual instantiation by measuring said escalation frequency with a rate of escalation to customer metric.

15. The method of claim 2,

wherein one of said goals comprises problem resolution by agents;

wherein one of said forecasting attributes comprises an extent of differences between a new agent and an experienced agent in said problem resolution involving said domain candidate; and

wherein said actual attribute value comprises an actual extent of differences value comprising a numerical value.

16. The method of claim 15, further comprising identifying said actual instantiation by measuring said extent of differences with an average handle time gap metric.

17. The method of claim 2, wherein one of said forecasting attributes comprises measurability, comprising an extent that said domain candidate can be measured, and wherein said actual attribute value comprises an actual measurability value comprising a numerical value indicative of said extent that said domain candidate can be measured.

18. The method of claim 17, further comprising identifying said actual instantiation by evaluating said measurability of said domain candidate to be measured by a series of measurability metrics.

19. The method of claim 18, wherein one of said goals comprises responding to requests for assistance by at least one agent and wherein said series of measurability metrics comprises:

a request measurement to measure a number of requests for assistance that are received about said domain candidate;

response satisfaction measurements to measure satisfaction of requesters with responses to said requests for assistance; and

quality measurements to measure an extent of quality of said responses; and

wherein said series of measurability metrics are used to develop said actual measurability value comprising said numerical value indicative of said extent that said domain candidate can be measured.

20. The method of claim 2,

wherein one of said goals comprises responding to requests for assistance;

wherein one of said forecasting attributes comprises training that a team receives for said domain candidate; and

wherein said actual attribute value comprises an actual training value comprising a numerical value.

21. The method of claim 20, further comprising identifying said actual instantiation by measuring said training with a training metric to measure an amount of time spent training for a domain candidate, compared to a total amount of time spent training.

22. The method of claim 2,

wherein one of said goals comprises responding to requests for assistance;

wherein said forecasting attribute comprises a repetition frequency comprising a frequency of repetition of a problem in said requests for assistance involving said domain candidate; and

wherein said actual attribute value comprises an actual repetition value comprising a numerical value.

23. The method of claim 22, further comprising identifying said actual instantiation by measuring said repetition frequency with a repetition metric comprising a percent of times that a problem is repeated in said requests for assistance involving said domain candidate, compared to a total number of said requests for assistance involving said domain candidate.

24. The method of claim 2,

wherein one of said goals comprises responding to requests for assistance;

wherein said forecasting attribute comprises leverageability which comprises a leverageability extent, comprising an extent that knowledge about said domain candidate can provide benefit outside of said domain candidate; and

wherein said actual attribute value comprises an actual leverageability value comprising a numerical value.

25. The method of claim 24, further comprising identifying said actual instantiation by measuring said leverageability with a leverageability metric, wherein said leverageability metric is used to develop said actual leverageability value comprising said numerical value indicative of said leverageability extent.

26. The method of claim 1, further comprising, once said selected knowledge domain is selected, selecting a selected knowledge sub-domain for said knowledge base by:

identifying a sub-domain candidate set from said selected knowledge domain, said sub-domain candidate set having at least one sub-domain candidate;

developing a sub-domain candidate benefit value for each said sub-domain candidate, said sub-domain candidate benefit value comprising a numerical value and indicating benefit provided by said sub-domain candidate associated therewith in becoming said selected knowledge sub-domain;

conducting a comparison of said sub-domain candidate benefit values; and

selecting said sub-selected knowledge domain from said sub-domain candidates based on said comparison.

27. The method of claim 26, wherein said sub-domain candidate further comprises a functionality within said selected knowledge domain.

28. The method of claim 1, wherein said domain candidate is identified through a review of reports and surveys.

29. The method of claim 1, wherein said domain candidate set comprises a plurality of subjects.

30. The method of claim 1, wherein said domain candidate set comprises a plurality of functionalities within a subject area.

31. The method of claim 1, wherein said domain candidate, when said knowledge base is developed in support of a product, comprises troubleshooting, usage of said product, interoperability of said product with external applications, installation of said product and configuration of said product.

32. The method of claim 1, further comprising seeding said domain candidate set into said knowledge base,

wherein said domain candidate benefit value further comprises a seeding priority value comprising a numerical value and indicating a level of importance of seeding said domain candidate into said knowledge base;

and

wherein said conducting a comparison step comprises conducting a first comparison of each said domain candidate benefit value; and

further comprising, after said selecting step, providing a seeding order assignment to each said domain candidate, wherein said seeding order assignment identifies a numerical order in which each said domain candidate is to be seeded into said knowledge base; and

wherein said providing step comprises;

assigning a first seeding order assignment to said selected knowledge domain;

removing said selected knowledge domain from said domain candidate set;

conducting a next comparison of each said domain candidate benefit value for each said domain candidate remaining in said domain candidate set;

selecting said next selected knowledge domain from said domain candidate set based on said next comparison;

assigning a next seeding order assignment to said next selected knowledge domain, wherein said next seeding order assignment indicates that said next selected knowledge domain is to be seeded next after said domain candidate that was just previously removed from said domain candidate set;

removing said next selected knowledge domain from said domain candidate set; and

for each said domain candidate remaining in said domain candidate set, repeating said steps of conducting said next comparison, selecting said next selected knowledge domain, assigning said next seeding order assignment, and removing said next selected knowledge domain, until each said domain candidate has had assigned thereto said seeding order assignment.

33. The method of claim 32, wherein each said domain candidate comprises a sub-domain of a previously selected knowledge domain for said knowledge base.

34. The method of claim 1, further comprising:

recording each said domain candidate and said domain candidate benefit value associated therewith in a domain candidate benefit indicator in order to facilitate said comparison;

reviewing said domain candidate benefit indicator to identify a selected domain candidate comprising said domain candidate having said domain candidate benefit value associated therewith with a value greater than each other said domain candidate benefit value; and

selecting said selected domain candidate as said selected knowledge domain.

35. The method of claim 15, further comprising identifying said actual instantiation by measuring said extent of differences with experience metrics to measure said extent of differences, wherein said experience metrics comprise a first time fix rate, an escalation rate, and a measurement of customer satisfaction with said problem resolution;

and wherein said experience metrics are used to develop said actual extent of differences value comprising said numerical value indicative of said extent of differences.

49. The method of claim 19,

wherein said quality measurements comprise a rate of escalation of problems to mentors, a rate of escalation of problems to customers, and a first time fix rate of resolving problems in a domain candidate on a first request for assistance;

and

wherein said quality measurements are used to develop said actual measurability value comprising said numerical value indicative of said extent that said domain candidate can be measured.

50. The method of claim 19, wherein said response satisfaction measurements comprise a measure of a total average handle time and a measure of the average handle time by said agent.

36. A domain candidate benefit indicator for use in selecting knowledge domain candidates as knowledge domains for knowledge bases, said domain candidate benefit indicator comprising:

a domain candidate benefit value comprising a combination of actual attribute values for said domain candidate, each of said actual attribute values comprising numerical values and developed from an actual attribute score;

said actual attribute score developed by conducting an evaluation of a forecasting attribute of said domain candidate;

said actual attribute score indicative of an extent of benefit to a knowledge base of selecting said domain candidate as a selected knowledge domain, as indicated by said evaluation of said forecasting attribute; and

weighting said actual attribute score according to selected goals to generate said each of said actual attribute values.

37. A method for developing seeding methodologies for seeding sub-domains into knowledge bases, said methodologies having preferred orders for seeding said knowledge bases with said sub-domains, said method comprising:

identifying a sub-domain set comprising at least one sub-domain in a selected knowledge domain for a knowledge base;

developing, for each said sub-domain, a seeding priority value that indicates a level of importance of seeding said sub-domain into said knowledge base, said seeding priority value comprising a numerical value indicative of said level of importance;

conducting a comparison of each said seeding priority value; and

developing a preferred order for seeding each said sub-domain in said sub-domain set into said knowledge base based on said comparison.

38. The method of claim 37, wherein

conducting said comparison of each said seeding priority value further comprises ranking each said sub-domain in order by numerical value of said sub-domain seeding priority value associated therewith, said ranking comprising:

assigning a first ranking to said sub-domain having said seeding priority value greater than each other said seeding priority value; and

assigning a last ranking to said sub-domain having said seeding priority value lower than each other said seeding priority value; and

developing said preferred order for seeding further comprises assigning a seeding order assignment for each said sub-domain in ascending numerical order based on said ranking, with said sub-domain having said first ranking assigned a first seeding position, and with said sub-domain having said last ranking assigned a last seeding position.

39. The method of claim 37, further comprising recording an identification of said sub-domains and their associated seeding priority values into a seeding methodology indicator in order to facilitate said ranking.

40. The method of claim 37, further comprising identifying a sub-domain seeding volume to indicate an extent of said seeding of each of said sub-domains prior to activation of said knowledge base, said sub-domain seeding volume comprising an estimate of how many records need to be entered into said knowledge base in order to capture knowledge about each said sub-domain.

41. The method of claim 40, further comprising identifying a domain seeding volume for said selected knowledge domain, said domain seeding volume calculated by summing each said sub-domain seeding volume for each said sub-domain.

42. The method of claim 40, wherein developing said methodology further comprises representing said methodology in a domain matrix detailing said preferred order for seeding and seeding information for each said sub-domain, said seeding information comprising said sub-domain seeding volume.

43. The method of claim 42, wherein said seeding information further comprises an identification of an extent of knowledge available about each said sub-domain and an identification of knowledge reservoirs for each said sub-domain.

44. The method of claim 37, wherein said identifying step comprises identifying said sub-domain by type of query to be used in accessing said knowledge about said knowledge domain.

45. A method of determining seed case priority for sub-domains of a domain for a knowledge base, said method comprising:

identifying a sub-domain set comprising at least one sub-domain in a selected knowledge domain for a knowledge base; and

developing, for each said sub-domain, a seeding priority value that indicates a level of importance of seeding said sub-domain into said knowledge base;

said seeding priority value comprising a numerical value indicative of said level of importance; and

said seeding priority value determined by evaluating selected characteristics of said sub-domain.

46. The method of claim 45, wherein said characteristics comprise at least one of the following:

an extent that users of said knowledge base will seed themselves, an extent that users of said knowledge base can seed themselves, and a diversity of questions that will be presented to said users.

47. The method of claim 45, wherein said sub-domain is a product, and said characteristics comprise at least one the following:

product history, product complexity, and expected life cycle of said product.

48. A method for creating a knowledge base, comprising:

developing a domain candidate benefit indicator to identify a selected domain for said knowledge base, further comprising;

identifying a domain candidate set having at least one domain candidate for a knowledge base;

developing a domain candidate benefit value for each said domain candidate, said domain candidate benefit value comprising a numerical value indicating a level of benefit of each said domain candidate associated therewith for becoming a selected knowledge domain for said knowledge base;

recording each said domain candidate and said domain candidate benefit value associated therewith in a domain candidate benefit indicator;

reviewing said domain candidate benefit indicator to identify a selected domain candidate comprising said domain candidate having said domain candidate benefit value associated therewith with a value greater than each other said domain candidate benefit value; and

selecting said selected domain candidate as said selected knowledge domain;

identifying sub-domains of said selected knowledge domain; and

developing a seeding methodology for seeding said knowledge base prior to activation of said knowledge base, said methodology comprising a preferred order for seeding said sub-domains and, for each of said sub-domains, an estimate of how many records need to be entered into said knowledge base in order to capture knowledge about each of said sub-domains.