SYSTEM AND METHOD OF LEARNING-BASED MATCHING

• US 20120059853A1
• Filed: 08/30/2011
• Published: 03/08/2012
• Est. Priority Date: 01/18/2010
• Status: Active Grant

• Current Assignee: Salesforce.com, Inc.
• Original Assignee: Salesforce.com, Inc.
• Inventors: Jagota, Arun Kumar
• Granted Patent: US 9,026,552 B2
• Time in Patent Office: Days
• US Class Current: 707/780
• CPC Class Codes: G06F 16/29 Geographical information databases
• CPC Class Codes: G06F 16/95 Retrieval from the web
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: Y10S 707/944 Business related

Abstract

Some embodiments match address data in a database system having a corpus of address information stored therein. In some embodiments, an external address data for a company that is associated with a set of locations is received. Some embodiments determine whether the external address data matches existing address data that includes the set of locations using fuzzy matching techniques. In order to determine whether two addresses match, some embodiments perform fuzzy matching techniques that uses learned knowledge to identify aliases in the address pair. In situations where a contact is specified along with the external address data, the contact is tethered to one of the set of locations when the external address data is determined to match an address within the existing address data. Some embodiments create a new location for the company using the external address data when the external address data does not match the existing address data.

Claims

1. A computer-implemented method of matching address data in a database system having a corpus of address information stored therein, the method comprising:

receiving external address data for a company of a contact, the company associated with a set of locations in the database system;

determining whether the external address data matches an existing address data of one of the set of locations using one or more fuzzy matching techniques;

tethering, using a processor, the contact to the one of the set of locations when the external address data matches the existing address data; and

creating a new location for the company using the external address data when the external address data does not match the existing address data.

2. The method of claim 1 further comprising:

tethering the contact to the new location upon creation of the new location.

3. The method of claim 1, wherein using the one or more fuzzy matching techniques comprises comparing the external address data with each address data corresponding to each of the set of locations.

4. The method of claim 3, wherein each comparison yields a score corresponding to each address pair, each address pair comprising the external address data and address data of one of the set of locations.

5. The method of claim 3, wherein the comparing the external address data with each address data corresponding to each of the set of locations comprises:

comparing each address field of the external address data with each address field of each address data corresponding to each of the set of locations to obtain a first set of field match scores using a learned knowledge table;

comparing each address field of the external address data with each address field of each address data corresponding to each of the set of locations to obtain a second set of field match scores using a matching logic; and

determining a third set of field match scores by obtaining the higher score for each field between the first and second sets of field match scores, wherein the external address data is determined to match the existing address data of one of the set of locations when an overall address match score exceeds a threshold value, the overall address match score determined based at least in part on the third set of field match scores.

6. The method of claim 1, wherein the external address data comprises a plurality of address fields, each address field representing at least one of a street, a city, a state, a zip code, and a country.

7. The method of claim 1 further comprising:

determining whether the external address data is verified by a third party vendor that provides address validation services.

8. The method of claim 1 further comprising:

determining whether the external address data is validated, wherein the contact is capable of being tethered to the one of the set of locations when the external address data is validated.

9. The method of claim 1, wherein the external address data is either partial or complete.

10. The method of claim 1, wherein the database system is a multi-tenant database system integrating the received address data into the database system.

11. A system for matching address data in a database system having a corpus of address information stored therein, the system comprising:

a processor;

a network interface; and

a memory encoded with program code to be executed by the processor, comprising;

program code to receive external address data for a company of a contact, the company associated with a set of locations in the database system;

program code to determine whether the external address data matches an existing address data of one of the set of locations using one or more fuzzy matching algorithms;

program code to tether the contact to the one of the set of locations when the external address data matches the existing address data; and

program code to create a new location for the company using the external address data when the external address data does not match the existing address data.

12. The system of claim 11, the memory encoded with program code further comprising:

program code to determine whether the external address data is validated, wherein the contact is tethered to the one of the set of locations when the external address data is validated and matches the existing address data.

13. The system of claim 11, the memory encoded with program code further comprising:

program code to determine whether the external address data is validated;

program code to determine a nearest location in the set of locations to the external address data; and

program code to tether the contact to the one of the set of locations when the external address data matches the existing address data.

14. The system of claim 13, wherein the program code to determine the nearest location in the set of locations to the external address data comprises:

program code to determine a distance between the external address data and address data of each of the set of locations.

15. The system of claim 11, wherein each address data comprises a plurality of address fields, wherein using the one or more fuzzy matching algorithms comprises:

comparing each address field of the external address data with each address field of the existing address data to obtain a field match score for an address pair comprising the external address data and the existing address data.

16. A non-transitory computer-readable medium containing program code executable by a processor in a computer to match address data in a database system having a corpus of address information stored therein, the program code including instructions to:

receive external address data for a company of a contact, the company associated with a set of locations in the database system;

determine whether the external address data matches an existing address data of one of the set of locations using one or more fuzzy matching techniques;

tether the contact to the one of the set of locations when the external address data matches the existing address data; and

create a new location for the company using the external address data when the external address data does not match the existing address data.

17. The non-transitory computer-readable medium of claim 16, wherein each address data comprises a plurality of address fields, wherein the one or more fuzzy matching techniques comprises determining a weighted confidence score for at least one of a portion of the plurality of address fields, the weighted confidence scored determined based on a frequency that the portion of the plurality of address fields has appeared in training sets.

18. The non-transitory computer-readable medium of claim 16, further including instructions to:

determine whether the external address data is validated by a third party address validation service; and

receive validated address data from the third party address validation service when the external address data is validated, wherein whether the external address data matches the existing address data is determined using the validated address data.

19. The non-transitory computer-readable medium of claim 18, wherein the external address data comprises a plurality of external address fields and the validated address data comprises a plurality of normalized address fields, further including instructions to:

compare each external address field with a corresponding normalized address field;

determine that the external address and the validated address is a training set pair when at least a portion of an external address field and a corresponding portion of a normalized address field are not expressed in a same way;

determine a set of alias pairs from the training set pair, the set of alias pairs determined based on the portion of the external address field and its corresponding portion of the normalized address field; and

store the set of alias pairs in a learned knowledge table.

20. The non-transitory computer-readable medium of claim 16, wherein the external address data comprises a plurality of external address fields, wherein the instruction to determine comprises instructions to:

calculate a plurality of field match scores for the external address data and the existing address data using learned knowledge; and

calculate a record-level match score for the external address data using the calculated field match scores, wherein the record-level match score indicates a weighted match score between the two address data.