Method for efficiently detecting covered rules in a knowledge base

• US 5,592,590 A
• Filed: 07/01/1994
• Issued: 01/07/1997
• Est. Priority Date: 07/01/1994
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Jolly, Christopher H.
• Primary Examiner(s): HAFIZ, TARIQ R
• Application Number:
  US08/272,972
• Time in Patent Office: 921 Days
• Field of Search: 395/50-51, 395/60-67, 395/75-77, 395/600
• US Class Current: 706/59
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition

Abstract

A method for verifying a plurality of rules stored in a knowledge base for consistency. The method identifies rules in the knowledge base that are covered or made useless by at least one rule, two rules, or at three rules. Each rule in the database has at least one option and a corresponding range for satisfying the option. Each of the plurality of rules are compiled into a term. The terms are then arranged into a hierarchical listing with high priority terms being placed before low priority terms. Each term is removed from the hierarchical listing with the high priority terms being removed before the low priority terms. Each removed term is then added to a sum of previously removed terms. Then it is determined if the removed term is covered by the terms in the sum of previously removed terms. Coverage is determined by using an interval consensus procedure which uses an absorption rule, a partial absorption rule, a continuation rule, a simplification rule, and consensus rule.

Claims

1. A method for verifying a plurality of prioritized rules stored in a knowledge base for consistency, each prioritized rule having at least one option and a corresponding range for satisfying the option, the method comprising the steps of:

compiling each of the plurality of prioritized rules into a prioritized term, each prioritized term having at least one variable corresponding to each option and range in a rule;

arranging the prioritized terms into a hierarchical listing with high priority terms being placed before low priority terms;

for each prioritized term in the hierarchical listing, removing a highest priority term remaining in the hierarchical listing;

determining if the removed prioritized term is covered by any combination of prioritized terms in a sum of previously removed prioritized terms, the rules corresponding to the covered prioritized terms being covered; and

adding each removed prioritized term to the sum of previously removed prioritized terms.

2. The method according to claim 1, wherein the step of compiling includes assigning a value to each variable, the value representing the range of each option.

3. The method according to claim 2, wherein the step of assigning includes forming a union of the ranges for each term, dividing the union of ranges into non-overlapping intervals, numbering the intervals consecutively with the smallest number beginning with zero, and assigning a bit vector to the numbered intervals.

4. The method according to claim 1, wherein the step of determining includes applying an interval consensus procedure.

5. The method according to claim 4, wherein the interval consensus procedure includes using an absorption rule to guarantee detection of rules covered by one rule.

6. The method according to claim 4, wherein the interval consensus procedure includes using a partial absorption rule, a continuation rule, a simplification rule, and a consensus rule to guarantee detection of rules covered by one and two rules.

7. The method according to claim 4, wherein the interval consensus procedure includes using an absorption rule, a partial absorption rule, a continuation rule, a simplification rule, and a consensus rule to guarantee detection of rules covered by one, two, and three rules.

8. A method for verifying a plurality of prioritized rules stored in a knowledge base for consistency, each prioritized rule having at least one option and a corresponding range for satisfying the option, the knowledge base being used in a collection system including a personal computer for editing the prioritized rules in the knowledge base, a database containing delinquent account information, and a mainframe for processing the account information with the prioritized rules, the method comprising the steps of:

compiling each of the plurality of prioritized rules into a prioritized term, each prioritized term having at least one variable corresponding to the options and ranges in the rules;

arranging the prioritized terms into a hierarchical listing with high priority terms being placed before low priority terms;

for each prioritized term in the hierarchical listing, removing a highest priority term remaining in the hierarchical listing;

determining if the removed prioritized term is covered by any combination of prioritized terms in a sum of previously removed prioritized terms, the rules corresponding to the covered prioritized terms being covered; and

adding each removed prioritized term to the sum of previously removed prioritized terms.

9. The method according to claim 8, wherein the step of compiling includes assigning a value to each variable, the value representing the range of each option.

10. The method according to claim 9, wherein the step of assigning includes forming a union of the ranges for each term, dividing the union of ranges into non-overlapping intervals, numbering the intervals consecutively with the smallest number beginning with zero, and assigning a bit vector to the numbered intervals.

11. The method according to claim 8, wherein the step of determining includes applying an interval consensus procedure.

12. The method according to claim 11, wherein the interval consensus procedure includes using an absorption rule to guarantee detection of rules covered by one rule.

13. The method according to claim 11, wherein the interval consensus procedure includes using a partial absorption rule, a continuation rule, a simplification rule, and a consensus rule to guarantee detection of rules covered by one and two rules.

14. The method according to claim 11, wherein the interval consensus procedure includes using an absorption rule, a partial absorption rule, a continuation rule, a simplification rule, and a consensus rule to guarantee detection of rules covered by one, two, and three rules.

15. A method for verifying a plurality of prioritized rules stored in a knowledge base for consistency, each prioritized rule having at least one option and a corresponding range for satisfying the option, the knowledge base being used in a collection system including a personal computer for editing the prioritized rules in the knowledge base, a database containing delinquent account information, and a mainframe for processing the account information with the prioritized rules, the method comprising the steps of:

compiling each of the plurality of prioritized rules into a prioritized term in the mainframe, each prioritized term having at least one variable corresponding to the options and ranges in the rules;

arranging the prioritized terms into a hierarchical listing in the mainframe with high priority terms being placed before low priority terms;

for each prioritized term in the hierarchical listing, removing a highest priority term remaining in the hierarchical listing;

determining if the removed prioritized term is covered by any combination of prioritized terms in a sum of previously removed prioritized terms, the rules corresponding to the covered prioritized terms being covered;

adding each removed prioritized term to the sum of previously removed prioritized terms;

applying uncovered prioritized terms to the accounts in the account database; and

initiating collection of delinquent accounts after the uncovered prioritized terms have been applied to the account database.