System and method for representing and retrieving knowledge in an adaptive cognitive network

• US 5,787,234 A
• Filed: 06/10/1997
• Issued: 07/28/1998
• Est. Priority Date: 06/22/1994
• Status: Expired due to Term

• Current Assignee: Aptean, Inc. (Aptean Superior Holdings, Inc.)
• Original Assignee: Transamerica Business Credit Corporation (Aegon Ltd.)
• Inventors: Molloy, Bruce G.
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US08/872,791
• Time in Patent Office: 413 Days
• Field of Search: 395/51
• US Class Current: 706/46
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06N 5/027 Frames

Abstract

An adaptive learning system in which knowledge is represented in the form of discrete concepts that are associated with one another. Each such association of concepts is called a case. In accordance with the invention, the degree of association of the concepts in the various cases can be specified by a mathematical score. As a result, if one concept such as the description of a problem is specified to this system, the system will identify associated concepts such as the likely solutions of the problem in the order of likelihood. The adaptive learning system is initialized by loading into the system cases which typically were derived from previous experience. The cases are stored by the is processor in a case table in long-term memory. Each of the concepts identified in the cases is stored in the concept table in the long-term memory. From these cases the processor can then generate in real time ordered lists of concepts which are most closely associated with free text or with one or more concepts that are provided as inputs to the system.

Claims

1. A method of storing and retrieving data in a computer system comprising the steps of:

abstracting the data into generic concepts,

storing in the memory of the computer system a set of cases, each case comprising one or more of said concepts,

retrieving from the computer system an ordered ranking of one or more concepts related to a query by;

comparing the query with the contents of the memory, and

evaluating a strength of association of the query with the concepts stored in the memory, wherein the step of evaluating a strength of association of the query with the concepts stored in memory comprises the steps of;

comparing the query with the concepts previously stored in the memory,

for each case where a match occurs between the query and a concept of the case, activating all the concepts in that case,

accumulating the number of activations of each concept and the number of queries for which the concept was activated, and

combining the number of activations and the number of queries for which the concept was activated to determine the strength of association between the query and the stored concepts.

2. The method of claim 1 wherein the query is free text.

3. The method of claim 1 wherein the query is a concept.

4. The method of claim 3 further comprising the steps of:

organizing the concepts into a limited number of types, and

identifying the query as relating to a particular type.

5. The method of claim 1 wherein the query is free text and/or a concept.

6. The method of claim 1 further comprising the step of storing in memory a lookup table which for each concept identifies the cases in which the concept is found.

7. The method of claim 6 wherein the step of comparing the query with the concepts is performed by comparing the query with the concepts in the lookup table.

8. The method of claim 7 wherein each concept has an identifier number and the step of storing a set of cases in memory comprises the step of forming in memory a table in which for each case the identifier numbers of the concepts in the case are grouped together.

9. The method of claim 8 wherein the step of activating all the concepts in a case includes the steps of:

identifying from the lookup table a case which includes a concept that matches with a query, and

reading from the case table the concept identifier numbers of each concept in the case identified from the lookup table.

10. The method of claim 1 wherein each concept has an identifier number and the step of storing a set of cases in memory comprises the step of forming in memory a table in which for each case the identifier numbers of the concepts in the case are grouped together.

11. A method of operating an adaptive learning system comprising the steps of:

generalizing specific events, actions, people, things and other entities so as to form a series of concepts,

storing the concepts in a computer memory,

retrieving from the computer memory associated concepts and providing them to a second computer memory, and

ordering the retrieved concepts in the second computer memory on the strength of their association, wherein the concepts are stored in computer memory in cases of associated concepts and the step of ordering the retrieved concepts on the strength of their association comprises the steps of;

comparing a query with the concepts previously stored in the memory,

for each case where a match occurs between the query and a concept of the case, activating all the concepts in that case,

accumulating the number of activations of each concept and the number of queries for which the concept was activated, and

combining the number of activations and the number of queries for which the concept was activated to determine the strength of association between the query and the stored concepts.

12. The method of claim 11 further comprising the step of storing in memory a lookup table which for each concept identifies the cases in which the concept is found.

13. The method of claim 12 wherein the step of comparing the query with the concepts is performed by comparing the query with the concepts in the lookup table.

14. The method of claim 13 wherein each concept has an identifier number and each case is stored in computer memory as a grouping of the identifier numbers of the associated concepts.

15. The method of claim 14 wherein each case is a linked list of concept identifier numbers.

16. The method of claim 14 wherein the step of activating all the concepts in a case includes the steps of:

identifying from the lookup table a case which includes a concept that matches a query; and

reading from the case table the concept identifier numbers of each concept in the case identified from the lookup table.

17. The method of claim 11 wherein the step of activating all the concepts in a case includes the steps of:

identifying from the lookup table a case which includes a concept that matches a query; and

reading from the case table the concept identifier numbers of each concept in the case identified from the lookup table.

18. The method of claim 11 wherein the query is free text.

19. The method of claim 11 wherein the query is a concept.

20. A method of operating an adaptive learning system stored in a computer comprising the steps of:

inputting into the computer a case comprising a plurality of associated concepts,

for each concept of the case that is inputted, comparing the concept with all the concepts in previously stored cases in the learning system,

for each concept in a case where a match occurs between a concept of the case that is being inputted and a concept of the case stored in the learning system, evaluating a strength of association of the inputted case with the concepts stored in the learning system, and

ordering the concepts stored in the learning system in accordance with said evaluated strength of association, wherein the strength of association is evaluated by accumulating at least first and second scores,

the first score being the number of times that there is a match between a concept of the case that is being inputted and one of the concepts in the previously stored cases, and

the second score being the number of concepts in the case being inputted which caused an increase in the first score.

21. The method of claim 20 further comprising the step of storing in memory a lookup table which for each concept identifies the cases in which the concept is found.

22. The method of claim 21 wherein the step of comparing the concept being inputted with the concepts previously stored is performed by comparing the concept being inputted with the concepts in the lookup table.

23. The method of claim 22 wherein each concept has an identifier number and each case is stored in computer memory as a grouping of the identifier numbers of the associated concepts.

24. The method of claim 23 wherein each case is a linked list of concept identifier numbers.

25. The method of claim 23 wherein the first score is accumulated by:

identifying from the lookup table each case which includes a concept that matches the concept being inputted, and

for all the concepts in all the cases that include a concept that matches the concept being inputted, counting the number of times each concept is found in such cases.

26. The method of claim 20 wherein the first score is accumulated by:

identifying from the lookup table each case which includes a concept that matches the concept being inputted, and

for all the concepts in all the cases that include a concept that matches the concept being inputted, counting the number of times each concept is found in such cases.

27. The method of claim 20 wherein the case that is inputted also comprises at least one instance of free text that is compared with all the concepts in previously stored cases in the learning system.

28. The method of claim 20 further comprising the step of displaying the ordered concepts for use in selecting the next concept to be inputted.

29. The method of claim 20 further comprising the step of adding the case being inputted to the previously stored cases in the learning system.

30. An adaptive learning system comprising:

storage means for storing information received from a user, the information received from a user being abstracted into generalized concepts and organized into sets of associated concepts, the storage means comprising means for storing a record of the concepts and means for storing a record of the sets of associated concepts,

a processor coupled to the storage means for calculating the degree of association between one or more concepts or free text specified at an input to the system and the concepts stored in the storage means and forming lists of concepts ordered in accordance with their degree of association,

means for activating all the concepts in a set when a match occurs between a concept or free text specified at an input to the system and one of the concepts in the set, and

means for counting the number of times each concept is activated.

31. The learning system of claim 30 further comprising a lookup table stored in memory which for each concept identifies the cases in which the concept is found.

32. The learning system of claim 30 wherein each concept has an identifier number and the record of sets of associated concepts is stored in memory in the form of a table which for each set groups together the identifier numbers of the concepts in that set.

33. The learning system of claim 32 wherein the processor includes means for comparing concepts or free text specified at an input to the system with concepts stored in the storage means.

34. The learning system of claim 30 configured to provide a help desk in which the concepts comprise identification of a problem, identification of a diagnosis and identification of action to be taken to correct the problem.

35. The learning system of claim 34 wherein the help desk advances through a case from problem, to diagnosis, to corrective action and the lists of concepts are provided to the help desk user to prompt his selection of a problem, a diagnosis and/or corrective action.

36. The learning system of claim 30 further comprising at least one additional storage means and processor, each storage means and processor constituting a user terminal, the user terminal being connected together by a network.

37. The learning system of claim 36 further comprising a file server connected to the user terminals via the network, the file server maintaining a record of the concepts and a record of the sets of associated concepts.

38. An adaptive learning system comprising:

storage means for storing information received from a user, said information being abstracted into generalized concepts and organized into cases of associated concepts, the storage means comprising means for storing a concept table which for each concept associates a concept name with an identifier number and a case table which for each case associates a case identifier with the identifier numbers of its associated concepts,

a processor coupled to the storage means for calculating the degree of association between one or more concepts or free text specified at an input to the system and the concepts stored in the storage means,

means for activating all the concepts in a case when a match occurs between a concept or free text specified at an input to the system and one of the concepts in the case, and

means for counting the number of times each concept is activated.

39. The learning system of claim 38 wherein the storage means further comprises means for storing a lookup table which for each concept identifies the cases in which the concept is found.

40. The learning system of claim 38 wherein the processor includes means for comparing concepts or free text specified at an input to the system with concepts stored in the storage means.

41. The learning system of claim 38 configured to provide a help desk in which the concepts comprise identification of a problem, identification of a diagnosis and identification of action to be taken to correct the problem.

42. The learning system of claim 41 wherein the help desk advances through a case from problem, to diagnosis, to corrective action and the lists of concepts are provided to the help desk user to prompt his selection of a problem, a diagnosis and/or corrective action.

43. The learning system of claim 38 further comprising at least one additional storage means and processor, each storage means and processor constituting a user terminal, the user terminal being connected together by a network.

44. The learning system of claim 43 further comprising a file server connected to the user terminals via the network, the file server maintaining a record of the concepts and a record of the sets of associated concepts.

45. A method of storing and retrieving data in a computer system comprising the steps of:

abstracting the data into generic concepts,

storing in the memory of the computer system a set of cases, each case comprising one or more of said concepts,

retrieving from the computer system an ordered ranking of one or more concepts related to a query by;

comparing the query with the contents of the memory, and

evaluating a strength of association of the query with the concepts stored in the memory, wherein the step of evaluating a strength of association of the query with the concepts stored in memory comprises the steps of;

comparing the query with the concepts previously stored in th e memory, and

for each concept in a case where a match occurs between the query and the concept of the case stored in the computer system, accumulating at least first and second scores related to such match, and combining said scores.

46. The method of claim 45 the first score being the number of times that there is a match between the query and one of the concepts in the previously stored cases, and

the second score being the number of queries which caused an increase in the first score.

47. A method of storing and retrieving data in a computer system comprising the steps of:

abstracting the data into generic concepts,

storing in the memory of the computer system a set of cases, each case comprising one or more of said concepts,

retrieving from the computer system an ordered ranking of one or more concepts related to a query by;

comparing the query with the contents of the memory, and

evaluating a strength of association of the query with the concepts stored in the memory, wherein the step of evaluating a strength of association of the query with the concepts stored in memory comprises the steps of;

comparing the query with the concepts previously stored in the memory,

for each case where a match occurs between the query and a concept of the case, activating all the concepts in that case, and scoring the activated concepts.

48. A method of operating an adaptive learning system comprising the steps of:

generalizing specific events, actions, people, things and other entities so as to form a series of concepts,

storing the concepts in a computer memory,

retrieving from the computer memory associated concepts and providing them to a second computer memory, and

ordering the retrieved concepts in the second computer memory on the strength of their association, wherein the concepts are stored in computer memory in cases of associated concepts and the step of ordering the retrieved concepts on the strength of their association comprises the steps of;

comparing a query with the concepts previously stored in the memory,

for each case where a match occurs between the query and a concept of the case, activating all the concepts in that case, and scoring the activated concepts.

49. A method of operating an adaptive learning system comprising the steps of:

generalizing specific events, actions, people, things and other entities so as to form a series of concepts,

storing the concepts in a computer memory,

retrieving from the computer memory associated concepts and providing them to a second computer memory, and

ordering the retrieved concepts in the second computer memory on the strength of their association, wherein the step of ordering the retrieved concepts on the strength of their association comprises the steps of;

comparing a query with the concepts previously stored in the memory, and

for each concept where a match occurs between the query and a stored concept in the computer system, accumulating at least first and second scores related to such match, and combining said scores.

50. The method of claim 49, wherein the first score being the number of times that there is a match between the query and one of the previously stored concepts, and

the second score being the number of queries which caused an increase in the first score.

51. A method of operating an adaptive learning system stored in a computer comprising the steps of:

inputting into the computer a case comprising a plurality of associated concepts,

for each concept of the case that is inputted, comparing the concept with all the concepts in previously stored cases in the learning system,

for each concept in a case where a match occurs between a concept of the case that is being inputted and a concept of the case stored in the learning system, evaluating a strength of association of the inputted case with the concepts stored in the learning system, and

ordering the concepts stored in the learning system in accordance with said evaluated strength of association,

wherein the step of evaluating the strength of the association comprises the step of;

activating all the concepts in a stored case where a match occurs between a concept being inputted and a concept in the stored case.

52. A method of operating an adaptive learning system stored in a computer comprising the steps of:

inputting into the computer a case comprising a plurality of associated concepts,

for each concept of the case that is inputted, comparing the concept with all the concepts in previously stored cases in the learning system,

for each concept in a case where a match occurs between a concept, of the case that is being inputted and a concept of the case stored in the learning system, evaluating a strength of association of the inputted case with the concepts stored in the learning system, and

ordering the concepts stored in the learning system in accordance with said evaluated strength of association, wherein the step of evaluating the strength of the association comprises the step of;

activating all the concepts in a stored case where a match occurs between a concept being inputted and a concept in the stored case;

wherein the step of evaluating the strength of the association further comprises the steps of;

accumulating the number of activations of each concept and the number of cases being inputted for which the concept in the stored case was activated, and

combining the number of activations and the number of cases being inputted for which the concept in the stored case was activated to determine the strength of association between the concept being inputted and the stored concepts.

53. The method of claim 52 further comprising the step of storing in memory a lookup table which for each concept identifies the cases in which the concept is found.

54. The method of claim 53 wherein the step of comparing the concept being inputted with the concepts previously stored is performed by comparing the concept being inputted with the concepts in the lookup table.

55. The method of claim 54 wherein each concept has an identifier number and each case is stored in computer memory as a grouping of the identifier numbers of the associated concepts.

56. The method of claim 55 wherein each case is a linked list of concept identifier numbers.

57. The method of claim 56 wherein the number of activations is accumulated by:

identifying from the lookup table each case which includes a concept that matches the concept being inputted, and

for all the concepts in all the cases that include a concept that matches the concept being inputted, counting the number of times each concept is found in such cases.

58. A method of operating an adaptive learning system stored in a computer comprising the steps of:

inputting into the computer a case comprising a plurality of associated concepts,

for each concept of the case that is inputted, comparing the concept with all the concepts in previously stored cases in the learning system,

for each concept in a case where a match occurs between a concept of the case that is being inputted and a concept of the case stored in the learning system, evaluating a strength of association of the inputted case with the concepts stored in the learning system, and

ordering the concepts stored in the learning system in accordance with said evaluated strength of association, wherein the strength of association is evaluated by accumulating at least first and second scores, and combining said scores.

59. An adaptive learning system comprising:

storage means for storing information received from a user, the information received from a user being abstracted into generalized concepts and organized into sets of associated concepts, the storage means comprising means for storing a record of the concepts and means for storing a record of the sets of associated concepts,

a processor coupled to the storage means for calculating the degree of association between one or more concepts or free text specified at an input to the system and the concepts stored in the storage means and forming lists of concepts ordered in accordance with their degree of association, and

a lookup table stored in memory which for each concept identifies the cases in which the concept is found.

60. An adaptive learning system comprising:

storage means for storing information received from a user, said information being abstracted into generalized concepts and organized into cases of associated concepts, the storage means comprising means for storing a concept table which for each concept associates a concept name with an identifier number and a case table which for each case associates a case identifier with the identifier numbers of its associated concepts,

a processor coupled to the storage means for calculating the degree of association between one or more concepts or free text specified at an input to the system and the concepts stored in the storage means, wherein the storage means further comprises means for storing a lookup table which for each concept identifies the cases in which the concept is found.

61. A method of storing and retrieving data in a computer system comprising the steps of:

abstracting the data into generic concepts,

storing in the memory of the computer system a set of cases, each case comprising one or more of said concepts,

retrieving from the computer system an ordered ranking of one or more concepts related to a query by;

comparing the query with the contents of the memory, and

evaluating a strength of association of the query with the concepts stored in the memory.