Method for rating a match for a given entity found in a list of entities

• US 5,659,731 A
• Filed: 06/19/1995
• Issued: 08/19/1997
• Est. Priority Date: 06/19/1995
• Status: Expired due to Term

• Current Assignee: The New Dun & Bradstreet Corporation
• Original Assignee: DUN & Bradstreet Incorporated (Dun & Bradstreet Corp.)
• Inventors: Gustafson, Eric S.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): ROBINSON, GRETA LEE
• Application Number:
  US08/491,743
• Time in Patent Office: 792 Days
• Field of Search: 395/600, 395/604, 395/238, 395/54, 364/253.1
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/2462 Approximate or statistical queries
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

In a system that accepts a given search entity from a user and utilizes a database to identify a possible matching entity from a large list of entries, a method is provided for evaluating the reliability of the matching entity. Preferably, the method is carried out with minimal human intervention. A user inputs a plurality of attributes to identify a given entity, the system identifies a possible matching entity, and assigns a numerical grade to reflect the match quality of each attribute. Thereafter, the method assigns a grade to each attribute score, assembles the grades into a key, uses the key to address a memory, and retrieves a confidence code or quality indicator from the memory. The confidence codes are based on empirical information and reflect the overall quality of the match for the particular entity.

Claims

1. A method for utilizing and evaluating information, automatically and without human intervention, derived from a matching system, the matching system being of the type which searches an extensive database containing information on a plurality of entities, each entity being identified by a plurality of attributes, and the system being of the type which matches the attributes of an given entity with the attributes of entities stored within the database to indicate the identity of closely matching entities along with numerical scores for each attribute indicating the quality of the match for each of the attributes, the method comprising the steps of:

assigning a grade to the score of a plurality n of the attributes, with the grade being selected from a small number of possible grades distinguishing between at least a clear match, a clear mismatch, and a possible match condition;

assembling the grades for each of the n attributes to produce a key for a particular closely matching entity; and

addressing a memory with the key to retrieve a match indicator that reflects the overall quality of the match for the particular entity, the memory containing matching indicators based on empirical information for the same or similar grade keys.

2. The method according to claim 1, wherein the matching indicators stored in the memory are also based on statistical formulations.

3. The method according to claim 1, wherein the grade key is an n digit key.

4. The method according to claim 1, wherein each attribute grade is based upon the numerical score associated with that attribute.

5. The method according to claim 1, wherein the small number of possible grades further includes a grade for a no-entry condition.

6. The method according to claim 5, wherein a no-entry grade is assigned to a no-entry condition in an attribute of the given entity.

7. The method according to claim 5, wherein a no-entry grade is assigned to a no-entry condition in an attribute of an entry stored in the database.

8. The method according to claim 1, wherein the small number of possible grades further includes a first no-entry grade assigned to a no-entry condition in an attribute of the given entity and a second no-entry grade assigned to a no-entry condition in an attribute of an entry stored in the database.

9. In a computerized system for storing information on a large group of business entities and automatically selecting a member of the group as a likely match with an given entity, the system including a database for storing a compilation of the large group of entities which are identified by a plurality of attributes, and a processor for accepting information specifying the attributes of a given entity and searching the compilation to identify possible matches with the listed entities, the processor being programmed to score the quality of a possible match for a plurality n of the attributes of each identified entity, a method for automatically, and without human intervention, determining the quality of the match produced by the processor and comprising the steps of:

assigning a grade to the score of each of the n attributes, the grade being selected from a limited grade set including distinctive grades for a match, a mismatch, a possible match, and a no-entry condition;

composing a key from the grades for the respective attributes;

addressing a table with the key; and

extracting information from the table at an address specified by the key, the information reflecting the reliability of the match based on stored information statistically obtained from similar keys for other matches.

10. In a computerized system for storing information on a large group of business entities and selecting a member of the group as a likely match with an given entity, the system including a database for storing a compilation of the entities which are specified by information on a plurality of attributes, and a processor for accepting information specifying attributes of a given entity and searching the compilation to identify possible matches with the stored entities, the processor being programmed to score the quality of a possible match for each attribute of an identified entity, a method for automatically and without human intervention determining the quality of the match produced by the processor comprising the steps of:

selecting a number n of the attributes which will be used to grade the reliability of the match;

assigning a grade to the score associated with each of the n attributes, the grades being selected to at least include distinct grades for a match, a mismatch, a possible match and a no-entry condition;

assembling a key from the grades assigned to the n attributes;

providing a memory table addressable by all possible keys, the table having match reliability data stored therein; and

addressing the table to obtain the match reliability data specified by the key.

11. The method according to claim 10, further including the step of analyzing the obtained reliability data to provide a confidence level for the key.

12. A method of processing commercial transactions comprising the steps of:

accessing a commercial database including a list of a large number of business entities with associated business data, the business entities being identified by a plurality of attributes;

automatically searching within that database to identify entities that possibly match a given entity specified by a plurality of attributes;

obtaining numerical scores reflecting the quality of the match of each attribute;

transforming the numerical score for each attribute to a grade selected from a limited subset of grades, the subset of grades reflecting a likely match, no match, and possible match conditions;

using the grades for the possible match to form a key, for each identified entity;

utilizing the key to address a memory table to retrieve reliability information stored in the memory, the reliability information being derived from empirical information; and

using the reliability information to select a matching entity from the list of identified entities.

13. The method according to claim 12, wherein the subset of grades further includes a grade reflecting a no-entry condition.

14. The method according to claim 12, further including the step of using the reliability information in connection with the business data associated with the selected entity to automatically reach a decision for processing the commercial transaction.

15. The method according to claim 14, wherein the commercial transaction is a credit granting/denying transaction and the associated business data is a credit rating.