Resolving similar entities from a transaction database

• US 10,140,664 B2
• Filed: 03/14/2013
• Issued: 11/27/2018
• Est. Priority Date: 03/14/2013
• Status: Active Grant

• Current Assignee: Palantir Technologies Incorporated
• Original Assignee: Palantir Technologies Incorporated
• Inventors: Erenrich, Daniel
• Primary Examiner(s): Iwarere, Oluseye
• Application Number:
  US13/827,491
• Publication Number:
  US 20140279299A1
• Time in Patent Office: 2,084 Days
• Field of Search: 705 30, 705 35
• CPC Class Codes: G06F 16/22 Indexing; Data structures therefor; Storage structures
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/10 Tax strategies

Abstract

A technique for identifying related transaction records from a database storing transaction records for multiple entities includes grouping transaction records with a common attribute value into transaction record sets, receiving a selection of an exemplar record set and determining the probability the transaction record set stores transaction records associated with a first entity. Other operations include resolving the transaction record set as storing transaction records associated with the first entity. This improves the process of identifying related transaction records because related transaction records missed by string comparisons transaction record attributes are detected.

Claims

1. A method for identifying related transaction records from a database storing transaction records for multiple entities performed by one or more processors of a computer system, the method comprising:

acquiring a plurality of transaction record sets, wherein each transaction record set includes one or more of the transaction records sharing a common attribute value and wherein the transaction records are credit or debit transactions processed by a financial institution for a merchant;

receiving a selection of or selecting an exemplar record set of the plurality of transaction record sets, wherein the exemplar record set comprises a plurality of the transaction records associated with a first entity of the multiple entities;

for at least one of the acquired plurality of transaction record sets;

determining a probability that the transaction record set stores transaction records associated with the first entity based at least in part on a machine learning classifier, the machine learning classifier being trained using one or more first pairs of transaction record sets and one or more second pairs of transaction record sets, wherein a first pair of transaction record sets represents a common entity and a second pair of transaction record sets represents unrelated entities, and

upon determining the probability exceeds a threshold, resolving the transaction record set as storing transaction records associated with the first entity, the resolving including merging the transaction records of the transaction record set into the exemplar record set that comprises the plurality of the transaction records associated with the first entity.

2. The method of claim 1, wherein attributes of the transaction records include one or more of the following:

an identification of the merchant from which the transaction originates;

an identification of the credit or debit account owner;

an amount of the transaction;

a date of the transaction;

a time of the transaction; and

a location of where the transaction originated.

3. The method of claim 2, further comprising determining, for at least one of the transaction record sets, aggregate values for the attributes of the transaction record set;

and

determining aggregate values for attributes of the exemplar record set.

4. The method of claim 1, wherein determining a probability comprises passing a transaction record set and the exemplar record set to a classifier, wherein the classifier is configured to determine the probability that the transaction record set stores transaction records associated with the first entity.

5. The method of claim 4, wherein the classifier is a random forest classifier.

6. The method of claim 4, wherein the classifier evaluates features of at least one of the acquired transaction records, including at least one of a word overlap count, word frequency, a word-based or character based cosine similarity, merchant category codes, and numeric city codes associated with the at least one of the acquired transaction record.

7. The method of claim 4 wherein the classifier evaluates features of at least one of the acquired transaction records including at least one of a fractional difference in size of an average ticket-size in the transaction record, a standard deviation between the average ticket-sizes in the transaction records, and a fractional difference in a magnitude of ticket-size variances.

8. The method of claim 1, further comprising, performing an analysis on a set of the transaction records, wherein the set includes the transaction records of the exemplar record set and the transaction records resolved as associated with the first entity.

9. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform an operation for identifying related transaction records from a database storing transaction records for multiple entities, the method comprising:

acquiring a plurality of transaction record sets, wherein each transaction record set includes one or more of the transaction records sharing a common attribute value and wherein the transaction records are credit or debit transactions processed by a financial institution for a merchant;

receiving a selection of or selecting an exemplar record set of the plurality of transaction record sets, wherein the exemplar record set comprises a plurality of the transaction records associated with a first entity of the multiple entities;

for at least one of the acquired plurality of transaction record sets;

determining a probability that the transaction record set stores transaction records associated with the first entity based at least in part on a machine learning classifier, the machine learning classifier being trained using one or more first pairs of transaction record sets and one or more second pairs of transaction record sets, wherein a first pair of transaction record sets represents a common entity and a second pair of transaction record sets represents unrelated entities, and

upon determining the probability exceeds a threshold, resolving the transaction record set as storing transaction records associated with the first entity, the resolving including merging the transaction records of the transaction record set into the exemplar record set that comprises the plurality of the transaction records associated with the first entity.

10. The computer readable medium of claim 9, wherein attributes of the transaction records include one or more of the following:

an identification of the merchant from which the transaction originates;

an identification of the credit or debit account owner;

an amount of the transaction;

a date of the transaction;

a time of the transaction; and

a location of where the transaction originated.

11. The computer readable medium of claim 10, further comprising determining, for at least one of the received transaction record set, aggregate values for the attributes of the transaction record set;

and

determining aggregate values for attributes of the exemplar record set.

12. The computer readable medium of claim 9, wherein determining a probability comprises passing a transaction record set and the exemplar record set to a classifier, wherein the classifier is configured to determine the probability that the transaction record set stores transaction records associated with the first entity.

13. The computer readable medium of claim 12, wherein the classifier is a random forest classifier.

14. The computer readable medium of claim 9, further comprising, performing an analysis on a set of the transaction records, wherein the set includes the transaction records of the exemplar record set and the transaction records resolved as associated with the first entity.

20. The system of claim 9, further comprising, performing an analysis on a set of the transaction records, wherein the set includes the transaction records of the exemplar record set and the transaction records resolved as associated with the first entity.

15. A computer system, comprising:

a memory; and

a processor storing one or more programs configured to perform an operation for identifying related transaction records from a database storing transaction records for multiple entities, the method comprising;

acquiring a plurality of transaction record sets, wherein each transaction record set includes one or more of the transaction records sharing a common attribute value and wherein the transaction records are credit or debit transactions processed by a financial institution for a merchant;

receiving a selection of or selecting an exemplar record set of the plurality of transaction record sets, wherein the exemplar record set comprises a plurality of the transaction records associated with a first entity of the multiple entities;

for at least one of the acquired plurality of transaction record sets;

determining a probability that the transaction record set stores transaction records associated with the first entity based at least in part on a machine learning classifier, the machine learning classifier being trained using one or more first pairs of transaction record sets and one or more second pairs of transaction record sets, wherein a first pair of transaction record sets represents a common entity and a second pair of transaction record sets represents unrelated entities, and

upon determining the probability exceeds a threshold, resolving the transaction record set as storing transaction records associated with the first entity, the resolving including merging the transaction records of the transaction record set into the exemplar record set that comprises the plurality of the transaction records associated with the first entity.

16. The system of claim 15, wherein attributes of the transaction records include one or more of the following:

an identification of the merchant from which the transaction originates;

an identification of the credit or debit account owner;

an amount of the transaction;

a date of the transaction;

a time of the transaction; and

a location of where the transaction originated.

17. The system of claim 16, further comprising determining, for at least one of the received transaction record sets, aggregate values for the attributes of the transaction record set;

and

determining aggregate values for attributes of the exemplar record set.

18. The system of claim 15, wherein determining a probability comprises passing a transaction record set and the exemplar record set to a classifier, wherein the classifier is configured to determine the probability that the transaction record set stores transaction records associated with the first entity.

19. The system of claim 18, wherein the classifier is a random forest classifier.