Platform for document classification

• US 10,402,641 B1
• Filed: 03/19/2019
• Issued: 09/03/2019
• Est. Priority Date: 03/19/2019
• Status: Active Grant

• Current Assignee: Capital One Services LLC (Capital One Financial Corporation)
• Original Assignee: Capital One Services LLC (Capital One Financial Corporation)
• Inventors: Dang, Steven, Gould, Jason, Jiang, Jennifer, Akatsuka, Christopher, Slattery, Douglas, Pasam, Vijaya
• Primary Examiner(s): Thomas, Mia M
• Application Number:
  US16/358,046
• Time in Patent Office: 168 Days
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06F 18/2431 Multiple classes
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables

Abstract

A device obtains image data associated with a document. Using a first machine learning model, the device determines, for the document, a first classification of one of a plurality of document types and a first confidence score associated with the first classification, and a second classification of one of the plurality of document types and a second confidence score associated with the second classification based on the image data. The device determines a difference between the first confidence score and the second confidence score, compares the difference and a threshold value, and accept the first classification of the document when the difference satisfies the threshold value.

Claims

1. A method, comprising:

obtaining, by a processor, first information associated with a document,

wherein the first information includes image data;

determining, for the document, by the processor, and using a first machine learning model, a first classification of one of a plurality of document types and a first confidence score associated with the first classification based on the image data,

wherein the first confidence score indicates a first confidence level that the document corresponds to the first classification;

comparing, by the processor, the first confidence score and a first threshold value;

accepting, by the processor, the first classification of the document when the first confidence score satisfies the first threshold value;

obtaining, by the processor, second information associated with the document when the first confidence score fails to satisfy the first threshold value,

wherein the second information includes text data;

determining, for the document, by the processor, and using a second machine learning model, a second classification of one of the plurality of document types and a second confidence score associated with the second classification based on the text data,

wherein the second confidence score indicates a second confidence level that the document corresponds to the second classification;

comparing, by the processor, the second confidence score and a second threshold value; and

accepting, by the processor, the second classification of the document when the second confidence score satisfies the second threshold value.

2. The method of claim 1, wherein the image data includes an image of the document.

3. The method of claim 1, wherein:

the document contains a plurality of pages,

a first page of the document corresponds to a first document type,

a second page of the document corresponds to a second document type, and

the first document type is different than the second document type.

4. The method of claim 1, wherein the plurality of document types include at least one of:

a document containing personal identifying information,

a document containing tax information,

a document containing legal information, or

a document containing banking information.

5. The method of claim 1, wherein the first classification is one of a plurality of classifications, and the plurality of classifications includes a third classification for the document of one of the plurality of document types and a third confidence score associated with the third classification based on the image data,

wherein the third confidence score indicates a third confidence level that the document corresponds to the third classification,

wherein the third confidence score is less than the first confidence score,

wherein the method further comprises:

determining a difference between the first confidence score and the third confidence score when the first confidence score fails to satisfy the first threshold value;

comparing the difference to a third threshold value; and

accepting the first classification of the document when the difference satisfies the third threshold value.

6. The method of claim 5, wherein the third threshold value is associated with a confidence level spread.

7. The method of claim 1, further comprising:

assigning the document a label corresponding to the one of the plurality of document types associated with the first classification; and

storing the label for access by a third-party device.

8. The method of claim 7, wherein the document is a first document, the image data is first image data, and the label is a first label,

wherein the method further comprises:

obtaining second image data associated with a second document;

determining, for the second document and using the first machine learning model, a third classification of one of the plurality of document types and a third confidence score associated with the third classification based on the second image data,

wherein the third confidence score indicates a third confidence level that the second document corresponds to the third classification;

comparing the third confidence score and a third threshold value;

accepting the third classification of the second document when the third confidence score satisfies the third threshold value,

wherein the one of the plurality of document types associated with the third classification is different than the one of the plurality of document types associated with the first classification;

assigning the second document a second label corresponding to the one of the plurality of document types associated with the third classification; and

storing the second label for access by the third-party device.

9. A device, comprising:

one or more memories; and

one or more processors, communicatively coupled to the one or more memories, configured to;

obtain image data associated with a document;

determine, for the document and using a first machine learning model, a first classification of one of a plurality of document types and a first confidence score associated with the first classification, and a second classification of one of the plurality of document types and a second confidence score associated with the second classification based on the image data,

wherein the first confidence score indicates a first confidence level that the document corresponds to the first classification, and the second confidence score indicates a second confidence level that the document corresponds to the second classification,

wherein the first confidence score is greater than the second confidence score;

determine a difference between the first confidence score and the second confidence score;

compare the difference and a threshold value; and

accept the first classification of the document when the difference satisfies the threshold value.

10. The device of claim 9, wherein the difference is a first difference and the threshold value is a first threshold value,

wherein the one or more processors are further configured to:

obtain text data associated with the document when the first difference fails to satisfy the first threshold value;

determine, for the document and using a second machine learning model, a third classification of one of the plurality of document types and a third confidence score associated with the third classification, and a fourth classification of one of the plurality of document types and a fourth confidence score associated with the fourth classification based on the text data,

wherein the third confidence score indicates a third confidence level that the document corresponds to the third classification, and the fourth confidence score indicates a fourth confidence level that the document corresponds to the fourth classification,

wherein the third confidence score is greater than the fourth confidence score;

determine a second difference between the third confidence score and the fourth confidence score;

compare the second difference and a second threshold value; and

accept the third classification of the document when the second difference satisfies the second threshold value.

11. The device of claim 9, wherein the image data includes an image of the document.

12. The device of claim 9, wherein the plurality of document types include at least one of:

a document containing personal identifying information,

a document containing tax information,

a document containing legal information, or

a document containing banking information.

13. The device of claim 9, wherein the threshold value is associated with a confidence level spread.

14. The device of claim 9, wherein the one or more processors are further configured to:

assign the document a label corresponding to the one of the plurality of document types associated with the first classification; and

store the label for access by a third-party device.

15. A non-transitory computer-readable medium storing instructions, the instructions comprising:

one or more instructions that, when executed by one or more processors, cause the one or more processors to;

receive first image data associated with a first page of a document;

receive second image data associated with a second page of the document;

determine, for the first page of the document and using a first machine learning model, a first classification of one of a plurality of document types and a first confidence score associated with the first classification based on the first image data,

wherein the first confidence score indicates a first confidence level that the first page of the document corresponds to the first classification;

determine, for the second page of the document and using the first machine learning model, a second classification of one of the plurality of document types and a second confidence score associated with the second classification based on the second image data,

wherein the second confidence score indicates a second confidence level that the second page of the document corresponds to the second classification;

compare the first confidence score and a first threshold value;

compare the second confidence score and a second threshold value;

accept the first classification of the first page of the document when the first confidence score satisfies the first threshold value;

accept the second classification of the second page of the document when the second confidence score satisfies the second threshold value,

wherein the one of the plurality of document types associated with the second classification is different than the one of the plurality of document types associated with the first classification;

assign the first page of the document a first label corresponding to the first classification;

assign the second page of the document a second label corresponding to the second classification;

store the first label for access by a third-party device; and

store the second label for access by the third-party device.

16. The non-transitory computer-readable medium of claim 15, wherein the one or more instructions, when executed by the one or more processors, further cause the one or more processors to:

obtain text data associated with the first page of the document when the first confidence score fails to satisfy the first threshold value;

determine for the first page of the document, using a second machine learning model, a third classification of one of the plurality of document types and a third confidence score associated with the third classification based on the text data,

wherein the third confidence score indicates a third confidence level that the first page of the document corresponds to the third classification;

compare the third confidence score and a third threshold value; and

accept the third classification of the document when the third confidence score satisfies the third threshold value.

17. The non-transitory computer-readable medium of claim 15, wherein:

the first image data includes a first image of the first page of the document, and

the second image data includes a second image of the second page of the document.

18. The non-transitory computer-readable medium of claim 15, wherein the plurality of document types include at least one of:

a document containing personal identifying information,

a document containing tax information,

a document containing legal information, or

a document containing banking information.

19. The non-transitory computer-readable medium of claim 15, wherein the one or more instructions, when executed by the one or more processors, further cause the one or more processors to:

receive third image data associated with a third page of the document;

determine, for the third page of the document and using the first machine learning model, a third classification of one of the plurality of document types and a third confidence score associated with the third classification based on the third image data,

wherein the third confidence score indicates a third confidence level that the first page of the document corresponds to the third classification;

compare the third confidence score and a third threshold value; and

accept the third classification of the third page of the document when the third confidence score satisfies the third threshold value,

wherein the one of the plurality of document types associated with the third classification is different than the one of the plurality of document types associated with the first classification and the one of the plurality of document types associated with the second classification.

20. The non-transitory computer-readable medium of claim 19, wherein the one or more instructions, when executed by the one or more processors, further cause the one or more processors to:

assign the third page of the document a third label corresponding to the third classification; and

store the third label for access by the third-party device.