Automated processing of receipts and invoices

1. A computer implemented method for processing a receipt, the method comprising:

receiving an image of a receipt;

executing an optical character recognition (OCR) of the receipt content embodied in the image, wherein executing the OCR of the receipt content results in machine-encoded text content;

generating tokens from the machine-encoded text content;

aggregating the generated tokens into a plurality of data groups, wherein at least one data group of the plurality of data groups comprises a plurality of generated tokens; and

for each data group of the plurality of data groups:

identifying likely interpretations of the tokens of the data group;

identifying a plurality of likely products for the data group according to the likely interpretations of the tokens of the data group, wherein each likely product of the plurality of likely products is associated with a confidence value indicative of a likelihood that the likely product is a correct interpretation of the corresponding information of the receipt image;

determining whether the confidence scores associated with the plurality of likely products fall below a confidence threshold, and if the confidence scores fall below the confidence threshold:

providing the plurality of likely products to an evaluator;

obtaining feedback from the evaluator regarding the plurality of likely products; and

updating the confidence values associated with the plurality of likely products according to the obtained feedback; and

storing the plurality of likely products of the data group in association with the image of the receipt in a data store.

2. The computer implemented method of claim 1 , further comprising, for each generated token:

classifying the token with a token classification as to what the token likely represents;

wherein identifying the plurality of likely products for the data group comprises identifying the plurality of likely products for the data group according to the token classifications of the aggregated tokens of the data group.

3. The computer implemented method of claim 2 , wherein generating tokens from the machine-encoded text content comprises:

identifying a vendor corresponding to the receipt; and

generating tokens from the machine encoded text content according to known information corresponding to the identified vendor.

4. The computer implemented method of claim 3 , further comprising identifying a receipt image, the receipt image corresponding to a sub-image of the received image of the receipt that includes only content of the receipt.

5. The computer implemented method of claim 4 , wherein executing the OCR of the receipt content embodied in the image comprises executing the OCR of content in the receipt image.

6. The computer implemented method of claim 5 , wherein at least one data group of the plurality of data groups comprises tokens aggregated from a plurality of horizontal lines of the machine-encoded text content.

7. The computer implemented method of claim 6 , wherein classifying the token with a token classification as to what the token likely represents comprises classifying the token with a token classification in view of known information of the identified vendor.

8. The computer implemented method of claim 7 , wherein the steps of generating tokens, aggregating the generated tokens into a plurality of data groups, identifying likely interpretations of the tokens, identifying the plurality of likely products for each data group, and storing the plurality of likely products of the data group in association with the image of the receipt in a data store are conducted according to known information of the identified vendor.

9. The computer implemented method of claim 7 , further comprising, upon identifying a vendor of the receipt, re-executing the steps of executing the OCR of the receipt content, generating the tokens, aggregating the generated tokens into a plurality of data groups, identifying likely interpretations of the tokens, and identifying the plurality of likely products for each data group according to known information of the identified vendor.

10. A non-transitory computer readable medium bearing computer executable instructions which, when executed on a computing system comprising at least a processor retrieved from the medium, carry out a method for processing a receipt, the method comprising:

receiving an image of a receipt;

executing an optical character recognition (OCR) of the receipt content embodied in the image, wherein executing the OCR of the receipt content results in machine-encoded text content;

generating tokens from the machine-encoded text content;

aggregating the generated tokens into a plurality of data groups; and

for each data group of the plurality of data groups:

identifying likely interpretations of the tokens of the data group;

identifying a plurality of likely products for the data group according to the likely interpretations of the tokens of the data group, wherein each likely product of the plurality of likely products is associated with a confidence value indicative of a likelihood that the likely product is a correct interpretation of the corresponding information of the receipt image;

determining whether the confidence scores associated with the plurality of likely products fall below a confidence threshold, and if the confidence scores fall below the confidence threshold:

providing the plurality of likely products to an evaluator;

obtaining feedback from the evaluator regarding the plurality of likely products; and

updating the confidence values associated with the plurality of likely products according to the obtained feedback; and

storing the plurality of likely products of the data group in association with the image of the receipt in a data store.

11. The non-transitory computer readable medium of claim 10 , wherein generating tokens from the machine-encoded text content comprises generating tokens from the machine encoded text content according to known information corresponding to the identified vendor.

12. The non-transitory computer readable medium of claim 11 , wherein the method further comprises further comprising identifying a receipt image, the receipt image corresponding a sub-image of the received image of the receipt that includes only content of the receipt.

13. The non-transitory computer readable medium of claim 12 , wherein executing the OCR of the receipt content embodied in the image comprises executing the OCR of content in the receipt image.

14. The non-transitory computer readable medium of claim 13 , wherein the method further comprises, for each generated token:

classifying the generated token with a token classification;

wherein identifying likely interpretations of the tokens of the data group comprises identifying likely interpretations of the tokens of the data group according to the token classifications of the aggregated tokens of the data group.

15. The non-transitory computer readable medium of claim 14 , wherein classifying each generated token with a token classification comprises classifying each generated token with a token classification in view of known information of the identified vendor.

16. The non-transitory computer readable medium of claim 15 , wherein the steps of generating tokens, aggregating the generated tokens into a plurality of data groups, identifying likely interpretations of the tokens, identifying the plurality of likely products for each data group, and storing the plurality of likely products of each data group in association with the image of the receipt in a data store are conducted according to known information of the identified vendor.

17. The non-transitory computer readable medium of claim 15 , wherein the method further comprises, upon identifying a vendor of the receipt, re-executing the steps of executing the OCR of the receipt content, generating the tokens, aggregating the generated tokens into a plurality of data groups, identifying likely interpretations of the tokens, and identifying the plurality of likely products for each data group according to known information of the identified vendor.

18. A computer device for processing a receipt, the computer device comprising a processor and a memory, wherein the processor executes instructions stored in the memory as part of or in conjunction with additional components to process the receipt, the additional components comprising:

a receipt processor that, in execution and in conjunction with additional components:

receives an image of a receipt;

captures receipt data from the receipt; and

stores the captured receipt data in a data store in conjunction with the receipt;

an Optical Character Recognition (OCR) component, wherein in execution the OCR component transcribes the receipt content embodied in the image of the receipt into machine-encoded text content;

a token generator that, in execution, generates tokens from the machine-encoded text content such that each item of the machine-encoded text content is encapsulated in a generated token of the generated tokens;

a data group generator that, in execution, constructs data groups according aggregations of horizontal lines of tokens of the receipt; and

a matching component that, in execution and for each data group:

identifies likely interpretations of the tokens of the data group;

identifies a plurality of likely products for the data group according to the likely interpretations of the tokens of the data group; and

stores the plurality of likely products for the data group in association with the image of the receipt in a data store.

19. The computer device of claim 18 , wherein the matching component, in execution and for each data group, is further configured to associate a confidence value to each likely product of the plurality of likely products, each confidence value indicative of a likelihood that the likely product is a correct interpretation of the corresponding information of the receipt image.

20. The computer device of claim 19 , wherein the matching component, in execution and for each data group, is further configured to:

determine whether the confidence scores associated with the plurality of likely products fall below a confidence threshold, and if the confidence scores fall below the confidence threshold:

providing the plurality of likely products to an evaluator;

obtaining feedback from the evaluator regarding the plurality of likely products; and

updating the confidence values associated with the plurality of likely products according to the obtained feedback.