STATISTICAL RECORD LINKAGE CALIBRATION FOR REFLEXIVE AND SYMMETRIC DISTANCE MEASURES AT THE FIELD AND FIELD VALUE LEVELS WITHOUT THE NEED FOR HUMAN INTERACTION

• US 20090271359A1
• Filed: 04/24/2009
• Published: 10/29/2009
• Est. Priority Date: 04/24/2008
• Status: Active Grant

• Current Assignee: LexisNexis Risk Solutions, Inc. (RELX PLC)
• Original Assignee: LexisNexis Risk & Information Analytics Group, Inc. (RELX PLC)
• Inventors: BAYLISS, David Alan
• Granted Patent: US 8,046,362 B2
• Time in Patent Office: Days
• US Class Current: 706/54
• CPC Class Codes: G06F 16/215 Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
• CPC Class Codes: G06F 16/2246 Trees, e.g. B+trees
• CPC Class Codes: G06F 16/24 Querying
• CPC Class Codes: G06F 16/245 Query processing
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/282 Hierarchical databases, e.g. IMS, LDAP data stores or Lotus Notes
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Disclosed is a system for, and method of, calculating parameters used to determine whether records and entity representations should be linked. The system and method use a symmetric and reflexive function to allow for linking records and entity representations whose field values differ. The system and method apply iterative techniques such that parameters from each linking iteration are used in the next linking iteration. The system and method need no human interaction in order to calibrate and utilize record matching formulas used for the linking decisions.

Claims

1. A computer implemented iterative process for generating entity representations in a computer implemented database using a record matching formula and for generating parameters for the record matching formula, the database comprising a plurality of records, each record comprising a plurality of fields, each field capable of containing a field value, wherein at least a portion of parameters for the record matching formula are generated using a symmetric and reflexive function and configured for a particular field value appearing in a selected field of at least one record, and wherein the process provides for linking records or entity representations with non-identical field values, the process comprising:

selecting a distance;

calculating a field value weight, the field value weight reflecting a likelihood that an arbitrary record in the database comprises a field value in the selected field of the arbitrary record that lies within the distance of the particular field value as determined by the symmetric and reflexive function;

forming a plurality of entity representations in the database, each entity representation comprising at least two records linked using a first instance of the record matching formula, at least one entity representation comprising a first record comprising the particular field value, the first record linked to a second record using a first instance of the record matching formula comprising the field value weight;

calculating a revised field value weight, the revised field value weight reflecting a likelihood that an arbitrary entity representation in the database comprises a field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function;

linking at least two entity representations in the database based on a second instance of the record matching formula, wherein the second instance of the record matching formula comprises the revised field value weight, whereby a number of entity representations in the database is reduced by the linking at least two entity representations relative to a number of entity representations in the database prior to the linking at least two entity representations; and

retrieving information from at least one record in the database.

2. The method of claim 1 further comprising iterating the calculating a revised field value weight and the linking at least two entity representations at least once prior to the retrieving.

3. The method of claim 1, wherein the record matching formula comprises a weighted sum of probabilities that two records match.

4. The method of claim 1, wherein the field value weight comprises a logarithm of a probability that an arbitrary record in the database comprises a field value in the selected field of the arbitrary record that lies within the distance of the particular field value as determined by the symmetric and reflexive function, and wherein the revised field value weight comprises a logarithm of a probability that an arbitrary entity representation in the database comprises a field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function.

5. The method of claim 1, wherein the symmetric, reflexive function is not an edit distance function.

14. The system of claim 1, wherein the field value weight comprises a logarithm of a probability that an arbitrary record in the database comprises a field value in the selected field of the arbitrary record that lies within the distance of the particular field value as determined by the symmetric and reflexive function, and wherein the revised field value weight comprises a logarithm of a probability that an arbitrary entity representation in the database comprises a field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function.

15. The system of claim 1l, wherein the symmetric, reflexive function is not an edit distance function.

6. A computer implemented iterative process for generating entity representations in a computer implemented database using a record matching formula and for generating parameters for the record matching formula, the database comprising a plurality of records, each record comprising a plurality of fields, each field capable of containing a field value, wherein at least a portion of parameters for the record matching formula are generated using a symmetric and reflexive function, configured for a particular field, and independent of any field value in the selected field, and wherein the process provides for linking records or entity representations with non-identical field values, the process comprising:

selecting a distance;

calculating a plurality of field value weights, each field value weight reflecting a likelihood that an arbitrary record in the database comprises a different field value, in the selected field of the arbitrary record, that lies within the distance of the field value as determined by the symmetric and reflexive function;

calculating a field weight for the selected field, the field weight for the selected field derived from each of the plurality of field value weights;

forming a plurality of entity representations in the database, each entity representation comprising at least two records linked using a first instance of the record matching formula, at least one entity representation comprising a first record linked to a second record using a first instance of the record matching formula comprising the field weight for the selected field;

calculating a plurality of revised field value weights, each revised field value weight reflecting a likelihood that an arbitrary entity representation in the database comprises a different field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function;

calculating a revised field weight for the selected field, the revised field weight for the selected field derived from each of the plurality of revised field value weights;

linking at least two entity representations in the database based on a second instance of the record matching formula, wherein the second instance of the record matching formula comprises the revised field value weight, whereby a number of entity representations in the database is reduced by the linking at least two entity representations relative to a number of entity representations in the database prior to the linking at least two entity representations; and

retrieving information from at least one record in the database.

7. The method of claim 6 further comprising iterating the calculating a plurality of revised field value weights, the calculating a revised field weight, and the linking at least two entity representations at least once prior to the retrieving.

8. The method of claim 6, wherein the record matching formula comprises a weighted sum of probabilities that two records match.

9. The method of claim 6, wherein each field value weight comprises a logarithm of a probability that an arbitrary record in the database comprises a field value in the selected field of the arbitrary record that lies within the distance of the particular field value as determined by the symmetric and reflexive function, and wherein each revised field value weight comprises a logarithm of a probability that an arbitrary entity representation in the database comprises a field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function.

10. The method of claim 6, wherein the symmetric, reflexive function is not an edit distance function.

11. A computer system for iteratively generating entity representations in a computer implemented database using a record matching formula and for generating parameters for the record matching formula, the database comprising a plurality of records, each record comprising a plurality of fields, each field capable of containing a field value, wherein at least a portion of parameters for the record matching formula are generated using a symmetric and reflexive function and configured for a particular field value appearing in a selected field of at least one record, and wherein the system provides for linking records or entity representations with non-identical field values, the system comprising:

a computer implemented database comprising a plurality of records, each record comprising a plurality of fields, each field capable of containing a field value;

an electronic memory storing a selected distance;

a processor programmed to calculate a field value weight, the field value weight reflecting a likelihood that an arbitrary record in the database comprises a field value in the selected field of the arbitrary record that lies within the distance of the particular field value as determined by the symmetric and reflexive function;

a processor programmed to form and store a plurality of entity representations in the database, each entity representation comprising at least two records linked using a first instance of the record matching formula, at least one entity representation comprising a first record comprising the particular field value, the first record linked to a second record using a first instance of the record matching formula comprising the field value weight;

a processor programmed to calculate a revised field value weight, the revised field value weight reflecting a likelihood that an arbitrary entity representation in the database comprises a field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function, and

a processor programmed to link and store at least two entity representations in the database based on a second instance of the record matching formula, wherein the second instance of the record matching formula comprises the revised field value weight, whereby a number of entity representations in the database is reduced by the linking at least two entity representations relative to a number of entity representations in the database prior to the linking at least two entity representations.

12. The system of claim 11 further comprising program logic configured to repeat calculating a revised field value weight and linking at least two entity representations at least once.

13. The system of claim 11, wherein the record matching formula comprises a weighted sum of probabilities that two records match.

16. A computer system for iteratively generating entity representations in a computer implemented database using a record matching formula and for generating parameters for the record matching formula, the database comprising a plurality of records, each record comprising a plurality of fields, each field capable of containing a field value, wherein at least a portion of parameters for the record matching formula are generated using a symmetric and reflexive function, configured for a particular field, and independent of any field value in the selected field, and wherein the process provides for linking records or entity representations with non-identical field values, the process comprising:

a computer implemented database comprising a plurality of records, each record comprising a plurality of fields, each field capable of containing a field value;

an electronic memory storing a selected distance;

a processor programmed to calculate a plurality of field value weights, each field value weight reflecting a likelihood that an arbitrary record in the database comprises a different field value, in the selected field of the arbitrary record, that lies within the distance of the field value as determined by the symmetric and reflexive function;

a processor programmed to calculate a field weight for the selected field, the field weight for the selected field derived from each of the plurality of field value weights;

a processor programmed to form and store a plurality of entity representations in the database, each entity representation comprising at least two records linked using a first instance of the record matching formula, at least one entity representation comprising a first record linked to a second record using a first instance of the record matching formula comprising the field weight for the selected field;

a processor programmed to calculate a plurality of revised field value weights, each revised field value weight reflecting a likelihood that an arbitrary entity representation in the database comprises a different field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function;

a processor programmed to calculate a revised field weight for the selected field, the revised field weight for the selected field derived from each of the plurality of revised field value weights; and

a processor programmed to link and store at least two entity representations in the database based on a second instance of the record matching formula, wherein the second instance of the record matching formula comprises the revised field value weight, whereby a number of entity representations in the database is reduced by the linking at least two entity representations relative to a number of entity representations in the database prior to the linking at least two entity representations.

17. The system of claim 16 further comprising program logic configured to iterate calculating the plurality of revised field value probabilities, calculating a revised field weight, and linking and storing at least two entity representations at least once.

18. The system of claim 16, wherein the record matching formula comprises a weighted sum of probabilities that two records match.

19. The system of claim 16, wherein each field value weight comprises a logarithm of a probability that an arbitrary record in the database comprises a field value in the selected field of the arbitrary record that lies within the distance of the particular field value as determined by the symmetric and reflexive function, and wherein each revised field value weight comprises a logarithm of a probability that an arbitrary entity representation in the database comprises a field value, in the selected field of a record in the arbitrary entity representation, that lies within the distance of the particular field value as determined by the symmetric and reflexive function.

20. The system of claim 16, wherein the symmetric, reflexive function is not an edit distance function.