Case-based reasoning system and method for scoring cases in a case database

• US 6,021,411 A
• Filed: 12/30/1997
• Issued: 02/01/2000
• 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, Moon, Charles Song Yop, Brophy, Gary R.
• Primary Examiner(s): Amsbury, Wayne
• Assistant Examiner(s): PARDO, THUY N
• Application Number:
  US09/001,083
• Time in Patent Office: 763 Days
• Field of Search: 707/3, 707/5, 707/103, 707/104, 707/6, 707/10, 707/102, 707/1, 395/750.05
• 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 707/99936 Pattern matching access
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99944 Object-oriented database structure

Abstract

A case-based reasoning system is disclosed that includes a case database and a search engine. The case database is capable of storing a plurality of cases that each include one or more attributes that each have an associated match weight. Match weights of attributes in different cases are separately specified. In response to receipt of an incident including one or more input terms, the search engine scores the relative closeness of a selected case to the incident utilizing the match weights of attributes in the selected case that match input terms in the incident.

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, each of said attributes comprising a property having a value associated therewith, wherein each of said properties of said attributes has a match weight associated therewith and each value has a mismatch weight associated therewith, wherein match weights of attributes in different cases are separately specified, said plurality of cases including a selected case; and

a search engine that, in response to receipt of an incident including one or more input terms that each have an absent weight associated therewith, scores the relative closeness of said selected case to said incident utilizing (1) the match weights of attributes in said selected case that match input terms in said incident, (2) the mismatch weight of each value for which the associated property has a matching input term and the value itself has no matching input term, and (3) the absent weight of each input term having no matching attribute in said selected case.

2. The case-based reasoning system of claim 1, wherein said absent weights are user-specified.

3. The case-based reasoning system of claim 1, each of said match weights and mismatch weights having a contribution factor associated therewith, wherein said search engine scales match weights and mismatch weights utilized in scoring said selected case by multiplying said match weights and mismatch weights by their respective contribution factors.

4. The case-based reasoning system of claim 1, each of said attributes containing one or more terms that each have an absent weight associated therewith, wherein said search engine scores the relative closeness of said selected case to said incident utilizing the absent weight of each term in said selected case having no matching input term in said incident.

5. The case-based reasoning system of claim 1, wherein said search engine scores a plurality of cases including said selected case and outputs a best scoring case as a best match for said incident.

6. The case-based reasoning system of claim 1, wherein said match weights are user-specified.

7. The case-based reasoning system of claim 1, wherein said match weights are individually specified for each attribute in said selected case.

8. A method of case-based reasoning, comprising:

storing a plurality of cases including a selected case in a case base, wherein said plurality of cases each include one or more attributes that each comprises a property having a value associated therewith, wherein each value has a mismatch weight associated therewith;

associating a match weight with the property of each attribute, wherein match weights of attributes in different cases are separately specified; and

in response to receipt of an incident including one or more input terms that each have an absent weight associated therewith, scoring the relative closeness of said selected case to said incident utilizing (1) the match weights of attributes in said selected case that match input terms in said incident, (2) the mismatch weight of each value for which the associated property has a matching input term and the value itself has no matching input term, and (3) the absent weight of each input term having no matching attribute in said selected case.

9. The method of claim 8, said scoring step comprising scoring the relative closeness of said selected case using user-specified absent weights.

10. The method of claim 8, each of said match weights and mismatch weights having a contribution factor associated therewith, wherein said scoring step includes scaling match weights and mismatch weights utilized in scoring said selected case by multiplying said match weights and mismatch weights by their respective contribution factors.

11. The method of claim 8, each of said attributes containing one or more terms that each have an absent weight associated therewith, wherein said scoring step further comprises scoring the relative closeness of said selected case to said incident utilizing the absent weight of each term in said selected case having no matching input term in said incident.

12. The method of claim 8, wherein said scoring step comprises scoring said plurality of cases and outputting a best scoring case as a best match for said incident.

13. The method of claim 8, wherein said associating step is performed in response to entry of user-specified match weights.

14. The method of claim 8, wherein said associating step comprises associating individually specified match weights with attributes in said selected case.

15. A program product for causing a computer system to implement a method of case-based reasoning, said program product being embodied in a signal-bearing medium and comprising:

program code means for causing said computer system to store a plurality of cases including a selected case in a case base, wherein said plurality of cases each include one or more attributes, wherein each of said attributes comprises a property having a value associated therewith and each value has a mismatch weight associated therewith;

program code means for causing said computer system to associate a match weight with the property of each attribute, wherein match weights of attributes in different cases are separately specified; and

program code means, responsive to receipt of an incident including one or more input terms that each have an absent weight associated therewith, for causing said computer system to score the relative closeness of said selected case to said incident utilizing (1) the match weights of attributes in said selected case that match input terms in said incident, (2) the mismatch weight of each value for which the associated property has a matching input term and the value itself has no matching input term, and (3) the absent weight of each input term having no matching attribute in said selected case.

16. The program product of claim 15, said program code means for causing said computer system to score comprising program code means for causing said computer system to score the relative closeness of said selected case using user-specified absent weights.

17. The program product of claim 15, each of said match weights and mismatch weights having a contribution factor associated therewith, wherein said program code means for causing said computer system to score includes program code means for causing said computer system to scale match weights and mismatch weights utilized in scoring said selected case by multiplying said match weights and mismatch weights by their respective contribution factors.

18. The program product of claim 15, each of said attributes containing one or more terms that each have an absent weight associated therewith, wherein said program code means for causing said computer system to score further comprises program code means for causing said computer system to score the relative closeness of said selected case to said incident utilizing the absent weight of each term in said selected case having no matching input term in said incident.

19. The program product of claim 15, wherein said program code means for causing said computer system to score comprises program code means for causing said computer system to score said plurality of cases and output a best scoring case as a best match for said incident.

20. The program product of claim 15, wherein said program code means for causing said computer system to associate performs said association in response to entry of user-specified match weights.

21. The program product of claim 15, wherein said program code means for causing said computer system to associate comprises program code means for causing said computer system to associate individually specified match weights with attributes in said selected case.