Aggregate score matching system for transaction records

• US 6,636,850 B2
• Filed: 12/28/2000
• Issued: 10/21/2003
• Est. Priority Date: 12/28/2000
• Status: Active Grant

• Current Assignee: Fairisaac and Company Incorporated
• Original Assignee: Fairisaac and Company Incorporated
• Inventors: Lepien, Dana
• Primary Examiner(s): Mizrahi, Diane D.
• Application Number:
  US09/750,525
• Publication Number:
  US 20030041059A1
• Time in Patent Office: 1,027 Days
• Field of Search: 707/6, 707/3, 707/5, 707/10, 707/100, 705/10, 705/26, 709/224, 379/201.03, 370/52
• US Class Current: 707/780
• CPC Class Codes: G06F 16/24556 Aggregation; Duplicate elimination
• CPC Class Codes: G06F 16/90344 by using string matching techniques
• CPC Class Codes: Y10S 707/944 Business related
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

Method and apparatus that compares each field in two records to determine if the records match. The comparison function can be selected from among varied comparison function that enable close matches by examining all of the fields contained in the compared records.

Claims

1. A method for matching transactions against a known database of records comprising the steps of:

accepting a transaction record;

accepting a plurality of known records;

comparing each field as interest in the transaction record to the corresponding field in each of said known records;

creating a positive match score for each field in each of said known records that is based on the comparison of the transaction record to the known record;

summing the positive match scores for each of the fields for each of the known records;

comparing the sum of the positive match scores for each of the known records to a first threshold; and

declaring a match for the known record if said sum of the positive match scores for that known record exceeds the first threshold.

2. The method of claim 1 wherein the positive match score for each field of said known records is obtained from a configuration file.

3. The method of claim 1 wherein the first threshold is obtained from a configuration file.

4. The method of claim 1 further comprising the steps of:

creating a negative match score for each field in each of said known records that is based on the comparison of the transaction record to the known record;

summing the negative match scores for each of the fields for each of the known records;

comparing the sum of the negative match scores for each of the known records to a second threshold;

comparing the sum of the negative match scores for each of the known records to a second threshold; and

declaring a mismatch for the known record if said sum of the negative match scores for that known record falls below falls below the second threshold.

5. The method of claim 4 wherein the negative match score for each field of said known records is obtained from a configuration file.

6. The method of claim 4 wherein the second threshold is obtained from a configuration file.

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

subtracting the sum of the negative match scores from the sum of the positive match scores for each of the known records to yield a net match score for each of said known records;

comparing the net match score for each of said known records to a third threshold; and

declaring a match for the known record if said net match score for that known record exceeds the third threshold.

8. The method of claim 7 wherein the third threshold is obtained from a configuration file.

9. The method of claim 1 wherein the step of comparing each field against the corresponding field in a known record consists of one of the following methods:

comparing for equality if both fields are non-blank;

comparing for equality even if one or both fields are blank;

removing blanks and punctuation and then comparing for equality;

comparing gender code for equality;

comparing name and gender code for equality;

translating text to numeric equivalents and then comparing if both fields are non-blank;

compressing blanks and special characters and then comparing the shorter of the data lengths;

comparing in a close alpha manner where prior to performing an alpha comparison blanks, punctuation and numerics are removed and wherein the comparison allows one transposition and treats “

E” and

“

O”

as equal;

comparing for equality wherein the comparison allows blanks to compare to a preset value;

removing blanks and punctuation, then compressing the field and then comparing the left most characters;

comparing numerics only wherein the comparison allows one transposition;

isolating the right-most numerics and then comparing;

comparing numerics only, wherein the comparison allows one transposition and blanks compared to a value are assigned a negative score and blanks compared to blanks are assigned a positive score;

comparing for inequality if both fields are non-blank;

comparing first names for matches using nicknames, initials and full first names;

right justifying and then comparing;

reversing the field order and then comparing phonetically;

comparing fields phonetically; and

comparing in alpha/numeric manner but allowing one transposition.

10. An apparatus for matching a transaction against a known database of records comprising:

transaction reception unit that receives a transaction record;

legacy record reception unit that receives a plurality of known records; and

positive match field comparison unit that;

compares each field of the transaction record to a corresponding field in one record of said plurality of known records and increments a first accumulator whenever the field in the transaction record matches the field in the one known record; and

generates a record match indicator signal when the value stored in the first accumulator exceeds the value stored in a first threshold register.

11. The transaction matching apparatus of claim 10 wherein the first accumulator has a first programmable increment register and said first increment register is loaded from a configuration file.

12. The transaction matching apparatus of claim 10 wherein the value stored in said first threshold register is loaded from a configuration file.

13. The transaction matching apparatus of claim 10 further comprising:

negative match field comparison unit that;

compares each field of the transaction record to a corresponding field in one record of said plurality of known records and increments a second accumulator whenever the field in the transaction record does not matches the field in the one known record; and

generates a mis-match indicator signal when the value stored in the second accumulator exceeds the value stored in a second threshold register.

14. The transaction matching apparatus of claim 13 wherein the second accumulator has a second programmable increment register and said second increment register is loaded from a configuration file.

15. The transaction matching apparatus of claim 13 wherein the value stored in said second threshold register is loaded from a configuration file.

16. The transaction matching apparatus of claim 13 further comprising:

difference unit that receives the value stored in said first accumulator and the value stored in said second accumulator and stores the difference of these values in a difference register; and

difference match field comparison unit that generates a match indicator signal when the value stored in the difference register exceeds the value stored in a third threshold register.

17. The transaction matching apparatus of claim 16 wherein the third threshold register is loaded from a configuration file.

18. The transaction matching apparatus of claim 10 wherein the positive match field comparison unit is programmable to operate in one of the following modes:

equality mode that recognizes a field match condition if each character in each field position in both records is identical and one or both records contains some non-blank characters;

equality with blanks mode that recognizes a field match condition if each character in each field position in both records is identical;

compressed alphanumeric mode that recognizes a match condition if each character in each filed position in both records is identical after having first removes all blank and punctuation characters and compressing both fields;

gender compare mode that recognizes a field match condition when the both fields contains a gender code that is identical;

gender and name compare mode that recognizes a field match condition when both fields contains a gender code and name that are identical;

numeric equivalents mode that recognizes a field match condition when each field in each record is the same value after having converted a textual string into a numeric equivalent;

compressed minimum length mode that recognizes a field match condition when each character position in each field is identical in as many character positions that are non-blank in the shortest of the two fields after having removed all blank and punctuation characters and then compressing both fields;

close alpha match mode that recognizes a field match condition when all characters in each character position of each fields are identical after having removed blank characters, punctuation characters, numeric characters and allowing two character positions between the two fields to be transposed and treating the characters “

E” and

“

O”

as equal;

blank equivalent mode that recognizes a field match condition when all characters in each character position of each fields are identical or one of the fields contains all blank characters;

compressed mode that recognizes a field match condition when all characters in the left most character positions of each field are identical after first compressing each field;

numeric mode that recognizes a field match condition when each of the character positions in the right most portion of each field are identical after having first isolated the numeric characters in each field;

transposed numeric mode that recognizes a field match condition when the numeric values in each of the compared fields is identical except for one transposition or when a both fields contain only blank characters;

not equivalent mode that recognizes a field match condition when in fact the two fields from two records are not identical;

nickname mode that recognizes a field match condition when the name specified in the field in the first record is the nickname of the name specified in the field in the second record or when both records contain a single initial that is identical in each record or when the field in one record has an initial and the field in the other record contains a name beginning with that initial or a nickname corresponding to a name beginning with the initial;

right justification mode that recognizes a field match condition when each character position in each field from the two record being compared contains the identical character after having right justified both fields;

reverse phonetic mode that recognizes a field match condition when a phonetic code for each field is identical after having first reversed to order of the characters in each field being compared and before creating said phonetic code;

phonetic mode that recognizes a field match condition when a phonetic code for each field is identical; and

single transposition mode that recognizes a field match condition when each character position in each field being compared is identical except for the transposition of two character position in any of the two fields being compared.

19. The transaction matching apparatus of claim 13 wherein the negative match field comparison unit is programmable to operate in one of the following modes:

equality mode that recognizes a field mis-match condition if each character in each field position in both records is not identical and one or both records contains some non-blank characters;

equality with blanks mode that recognizes a field mis-match condition if each character in each field position in both records is not identical;

compressed alphanumeric mode that recognizes a mis-match condition if each character in each filed position in both records is not identical after having first removes all blank and punctuation characters and compressing both fields;

gender compare mode that recognizes a field mis-match condition when the both fields contain a gender code that is not identical;

gender and name compare mode that recognizes a field mis-match condition when both fields contains a gender code and name that are not identical;

numeric equivalents mode that recognizes a field mis-match condition when each field in each record is not the same value after having converted a textual string into a numeric equivalent;

compressed minimum length mode that recognizes a field mis-match condition when each character position in each field is not identical in as many character positions that are non-blank in the shortest of the two fields after having removed all blank and punctuation characters and then compressing both fields;

close alpha match mode that recognizes a field mis-match condition when all characters in each character position of each fields are not identical after having removed blank characters, punctuation characters, numeric characters and allowing two character positions between the two fields to be transposed and treating the characters “

E” and

“

O”

as equal;

blank equivalent mode that recognizes a field mis-match condition when all characters in each character position of each fields are not identical or one of the fields contains all blank characters;

compressed mode that recognizes a field mis-match condition when all characters in the left most character positions of each field are not identical after first compressing each field;

numeric mode that recognizes a field mis-match condition when each of the character positions in the right most portion of each field are not identical after having first isolated the numeric characters in each field;

transposed numeric mode that recognizes a field mis-match condition when the numeric values in each of the compared fields is not identical except for one transposition or when a both fields contain only blank characters;

not equivalent mode that recognizes a field mis-match condition when in fact the two fields from two records are identical;

nickname mode that recognizes a field mis-match condition when the name specified in the field in the first record is not the nickname of the name specified in the field in the second record and when both records contain a single initial that is not identical in each record or when the field in one record has an initial or the field in the other record does not contain a name beginning with that initial and a nickname not corresponding to a name beginning with the initial;

right justification mode that recognizes a field mis-match condition when each character position in each field from the two record being compared does not contain the identical character after having right justified both fields;

reverse phonetic mode that recognizes a field mis-match condition when a phonetic code for each field is not identical after having first reversed to order of the characters in each field being compared and before creating said phonetic code;

phonetic mode that recognizes a field mis-match condition when a phonetic code for each field is not identical; and

single transposition mode that recognizes a field mis-match condition when each character position in each field being compared is not identical except for the transposition of two character position in any of the two fields being compared.