Case-based reasoning system and method with a search engine that compares the input tokens with view tokens for matching cases within view

• US 6,571,251 B1
• Filed: 12/30/1997
• Issued: 05/27/2003
• Est. Priority Date: 12/30/1997
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Shore, Thomas Alan, Koski, Dennis D., Moon, Charles Song Yop
• Primary Examiner(s): CHANNAVAJJALA, SRIRAMA T
• Application Number:
  US09/001,117
• Time in Patent Office: 1,974 Days
• Field of Search: 707/500, 707/514, 707/530-532, 707/1-10, 707/100-104, 707/200-203, 706/10, 706/12-13, 706/20, 706/45-50, 706/52, 706/60-61, 714/26, 714/37, 714/40
• US Class Current: 1/1
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• 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 case-based reasoning system includes a case database capable of storing a plurality of cases that each include one or more attributes. A view generator in the case-based reasoning system generates a view of the case database by representing each case within at least a selected subset of the plurality of cases within the case database with one or more uniform-length view tokens. An input parser in the case-based reasoning system provides a tokenized representation of an input incident that includes one or more input tokens. The case-based reasoning system further includes a search engine that compares the input tokens with the view tokens to identify one or more closely matching cases within the view. By searching the view rather than directly searching the case database, cases that closely match the input are efficiently identified.

Claims

1. A case-based reasoning system, comprising:

a case base capable of storing a plurality of cases that each include one or more attributes;

a view generator that generates a view of said case base by representing each case within at least a selected subset of said plurality of cases within said case base as one or more view tokens, all view tokens in said view having a uniform length;

an input parser that provides a tokenized representation of an input incident, said tokenized representation including one or more input tokens; and

a search engine that compares said input tokens with said view tokens to identify one or more closely matching cases within said view, wherein said one or more closely matching cases are efficiently located by searching said view rather than directly searching said case base.

2. The case-based reasoning system of claim 1, and further comprising a solution manager that presents said one or more closely matching cases to a user.

3. The case-based reasoning system of claim 1, wherein:

said view generator generates a plurality of views of said case base; and

said case-based reasoning system further comprises a case manager that selects one of said plurality of views as a selected view to be searched by said search engine prior to searching any other of said plurality of views.

4. The case-based reasoning system of claim 3, said plurality of views of said case base generated by said view generator comprising a first view including all attributes of said all cases in said at least said subset and a second view including fewer than all said attributes of all said cases in said at least said subset.

5. The case-based reasoning system of claim 1, wherein said view generator generates said view prior to receipt of said input incident by said input parser.

6. The case-based reasoning system of claim 1, each view token having associated therewith scoring information, wherein said search engine utilizes said scoring information to calculate a score for each case in said view and determines said one or more closely matching cases based upon said scores.

7. The case-based reasoning system of claim 6, wherein:

said view comprises a first view;

said view generator generates a second view of said case base; and

said search engine searches said second view if a highest ranking score of any case in said first view is below a selected threshold.

8. The case-based reasoning system of claim 1, said search engine performing only exact word matching between said input tokens and said view tokens.

9. The case-based reasoning system of claim 1, said input parser including a synonym table correlating textual inputs and keywords, wherein said input parser converts said textual inputs into keywords by reference to said synonym table.

10. The case-based reasoning system of claim 9, each of said keywords in said synonym table comprising a uniform length input token.

11. The case-based reasoning system of claim 9, each of said attributes comprising a property having one or more values associated therewith.

12. A method of case-based reasoning, said method comprising:

storing, in a case base, a plurality of cases that each include one or more attributes;

generating a view of said case base by representing each case within at least a selected subset of said plurality of cases within said case base as one or more view tokens, all view tokens in said view having a uniform length;

providing a tokenized representation of an input incident, said tokenized representation including one or more input tokens; and

comparing said input tokens with said view tokens to identify one or more closely matching cases within said view, wherein said one or more closely matching cases are efficiently located by searching said view rather than directly searching said case base.

13. The method of claim 12, and further comprising presenting said one or more closely matching cases to a user.

14. The method of claim 12, wherein said generating step comprises generating a plurality of views of said case base, said method further comprising:

selecting one of said plurality of views as a first view to process in response to said input tokens.

15. The method of claim 14, said step of generating a plurality of views of said case base comprising generating a first view including all properties of all cases in said at least said subset and generating a second view including fewer than all said properties of all said cases in said at least said subset.

16. The method of claim 12, said generating step being performed prior to said providing step.

17. The method of claim 12, each view token having associated therewith scoring information, said comparing step comprising:

calculating a score for each case in said view utilizing said scoring information; and

determining said one or more closely matching cases based upon scores of cases in said view.

18. The method of claim 17, said view comprising a first view, wherein said generating step includes generating a second view of said case base, said method further comprising:

comparing said input tokens with said view tokens in said second view to identify one or more closely matching cases within said second view, if a highest ranking score of any case in said first view is below a selected threshold.

19. The method of claim 12, said comparing step comprising performing only exact word matching between said input tokens and said view tokens.

20. The method of claim 12, and further comprising converting textual inputs into keywords by reference to a synonym table that correlates textual inputs and keywords.

21. The method of claim 20, and further comprising storing each of said keywords in said synonym table as a uniform length input token.

22. The method of claim 12, wherein said storing step comprises storing, in a case base, a plurality of cases that each include one or more attributes, wherein each of said attributes includes a property having one or more value associated therewith.

23. A program product that can be utilized by a data processing system to implement a case-based reasoning system, said data processing system including a case base capable of storing a plurality of cases that each include one or more attributes, said program product comprising:

a view generator that causes said data processing system to generate a view of said case base by representing each case within at least a selected subset of said plurality of cases within said case base as one or more view tokens, all view tokens in said view having a uniform length;

an input parser that causes said data processing system to provide a tokenized representation of an input incident, said tokenized representation including one or more input tokens; and

a search engine that causes said data processing system to compare said input tokens with said view tokens to identify one or more closely matching cases within said view, wherein said one or more closely matching cases are efficiently located by searching said view rather than directly searching said case base; and

a signal-bearing medium bearing said view generator, said input parser, and said search engine.

24. The program product of claim 23, and further comprising a solution manager within said signal-bearing medium that causes said data processing system to present said one or more closely matching cases to a user.

25. The program product of claim 23, wherein:

said view generator generates a plurality of views of said case base; and

said program product further comprises a case manager, within said signal-bearing medium, that causes said data processing system to select one of said plurality of views as a selected view to be searched by said search engine prior to searching any other of said plurality of views.

26. The program product of claim 25, said plurality of views of said case base generated by said view generator comprising a first view including all properties of said all cases in said at least said subset and a second view including fewer than all said properties of all said cases in said at least said subset.

27. The program product of claim 23, wherein said view generator causes said data processing system to generate said view prior to receipt of said input incident by said input parser.

28. The program product of claim 23, each view token having associated therewith scoring information, wherein said search engine causes said data processing system to utilize said scoring information to calculate a score for each case in said view and determines said one or more closely matching cases based upon said scores.

29. The program product of claim 28, wherein:

said view comprises a first view;

said view generator causes said data processing system to generate a second view of said case base; and

said search engine causes said data processing system to search said second view if a highest ranking score of any case in said first view is below a selected threshold.

30. The program product of claim 23, said search engine causing said data processing system to perform only exact word matching between said input tokens and said view tokens.

31. The program product of claim 23, said program product including a synonym table correlating textual inputs and keywords, wherein said input parser causes said data processing system to convert said textual inputs into keywords by reference to said synonym table.

32. The program product of claim 31, each of said keywords in said synonym table comprising a uniform length input token.