Agency notice processing system

1. A method for training an artificial intelligence system to process agency notices, the method comprising:

identifying, by a computer system, historical data that includes historical text generated from optical character recognition performed on historical images of the agency notices and historical metadata for items in the historical images of the agency notices;

generating, by the computer system, annotations for the historical text, wherein the annotations identify the items in the historical text and locations of the items in the historical text; and

training, by the computer system, the artificial intelligence system using the historical data and the annotations.

2. The method of claim 1 further comprising:

processing, by the artificial intelligence system, new agency notices received after the artificial intelligence system has been trained using the historical data and the annotations.

3. The method of claim 1 , wherein training the artificial intelligence system using the historical data and the annotations comprises:

training, by the computer system, the artificial intelligence system using the historical metadata in the historical data and the annotations.

4. The method of claim 1 further comprising:

identifying, by the computer system, validation data for additional agency notices, wherein the validation data includes validation text generated from optical character recognition performed on validation images of additional agency notices and validation metadata for the items in the validation images of the additional agency notices;

sending, by the computer system, the validation text to the artificial intelligence system;

receiving, by the computer system, results from the artificial intelligence system processing the validation text; and

determining, by the computer system, whether the artificial intelligence system identifies the items in the validation text within a desired level of accuracy.

5. The method of claim 4 further comprising:

performing, by the computer system, additional training of the artificial intelligence system using additional historical data that includes additional historical text generated from the optical character recognition performed on additional historical images of additional agency notices and additional historical metadata for the items in the additional historical images of the additional agency notices and using additional annotations identifying the items and locations of the items in the additional historical text in response to the desired level of accuracy being absent in the artificial intelligence system.

6. The method of claim 4 , wherein the artificial intelligence system is trained using a first machine learning algorithm and further comprising:

performing, by the computer system, additional training of the artificial intelligence system using the historical data and the annotations and a second machine learning algorithm in response to a desired level of accuracy being absent.

7. The method of claim 4 , wherein the historical data is organized using a first type of clustering and further comprising:

performing, by the computer system, additional training of the artificial intelligence system using the historical data organized using a second type of clustering and the annotations in response to the desired level of accuracy being absent.

8. The method of claim 1 further comprising:

checking a quality of the annotations using the metadata to determine whether an annotation has an error that is selected from a group comprising a missing annotation and an incorrect annotation.

9. The method of claim 1 , wherein the items comprise at least one of an agency, an organization name, an employer identifier, a notice type, an amount, a notice date, or a notice period.

10. The method of claim 1 , wherein the artificial intelligence system is a neural network with natural language processing.

11. An artificial intelligence training system comprising:

a computer system that identifies historical data that includes historical text generated from optical character recognition performed on historical images of agency notices and historical metadata for items in the historical images of the agency notices; generates annotations for the historical text, wherein the annotations identify items in the historical text and locations of the items in the historical text; and trains the artificial intelligence system using the historical data and the annotations.

12. The artificial intelligence training system of claim 11 , wherein the computer system processes new agency notices using the artificial intelligence system after the artificial intelligence system has been trained using the historical data and the annotations.

13. The artificial intelligence training system of claim 11 , wherein in training the artificial intelligence system using the historical data and the annotations, the computer system trains the artificial intelligence system using the historical metadata in the historical data and the annotations.

14. The artificial intelligence training system of claim 11 , wherein the computer system identifies validation data, wherein the validation data includes validation text generated from the optical character recognition performed on validation images of additional agency notices and validation metadata for the items in the validation images of the additional agency notices; sends the validation text to the artificial intelligence system; receives results from the artificial intelligence system processing the validation text; and determines whether the artificial intelligence system identifies the items in the validation text within a desired level of accuracy.

15. The artificial intelligence training system of claim 14 , wherein the computer system performs additional training of the artificial intelligence system using additional historical data that includes additional historical text generated from the optical character recognition performed on additional historical images of additional agency notices and additional historical metadata for the items in the additional historical images of the additional agency notices and using additional annotations identifying the items and locations of the items in the additional historical text in response to the desired level of accuracy being absent in the artificial intelligence system.

16. The artificial intelligence training system of claim 14 , wherein the artificial intelligence system is trained using a first machine learning algorithm and wherein the computer system performs additional training of the artificial intelligence system using the historical data and the annotations and a second machine learning algorithm in response to the desired level of accuracy being absent.

17. The artificial intelligence training system of claim 14 , wherein the historical data is organized using a first type of clustering and wherein the computer system performs additional training of the artificial intelligence system using the historical data organized using a second type of clustering and the annotations in response to the desired level of accuracy being absent.

18. The artificial intelligence training system of claim 11 , wherein the computer system checks a quality of the annotations using the metadata to determine whether an annotation has an error that is selected from a group comprising a missing annotation and an incorrect annotation.

19. The artificial intelligence training system of claim 11 , wherein the items comprise at least one of an agency, a client organization, a notice type, an amount, or a notice period.

20. The artificial intelligence training system of claim 11 , wherein the artificial intelligence system is a neural network with natural language processing.

21. An artificial intelligence training system comprising:

a computer system;

a data loader in the computer system, wherein the data loader retrieves historical data that includes historical text generated from optical character recognition performed on historical images of agency notices and historical metadata for items in the historical images of the agency notices and retrieves validation data, wherein the validation data includes validation text generated from optical character recognition performed on validation images of additional agency notices and validation metadata for the items in the validation images of the additional agency notices;

an annotator in the computer system, wherein the annotator generates annotations for the historical text in which the annotations identify items and locations of the items in the historical text;

a trainer in the computer system, wherein trainer trains the artificial intelligence system using the historical data and the annotations; and

a validator in the computer system, wherein the validator sends validation text to the artificial intelligence system, receives results from the artificial intelligence system processing the validation text, and determines whether the artificial intelligence system identifies the items in the validation images within a desired level of accuracy.

22. A computer program product for training an artificial intelligence system to process agency notices, the computer program product comprising:

a computer-readable-storage media with first program code, second program code, and third computer-readable storage media;

first program code, stored on the computer-readable-storage media, for identifying historical data that includes historical text generated from optical character recognition performed on historical images of the agency notices and historical metadata for items in the historical images of the agency notices;

second program code, stored on the computer-readable-storage media, for generating annotations for the historical text, wherein the annotations identify items in the historical text and locations of the items in the historical text; and

third program code, stored on the computer-readable-storage media, for training the artificial intelligence system using the historical data and the annotations.

23. The computer program product of claim 22 further comprising:

fourth program code, stored on the computer-readable-storage media for processing, by the artificial intelligence system, new agency notices received after the artificial intelligence system has been trained using the historical data and the annotations.

24. The computer program product of claim 22 , wherein the third program code comprises:

program code, stored on the computer-readable-storage media, for training the artificial intelligence system using the historical metadata in the historical data and the annotations.

25. The computer program product of claim 22 , wherein the annotations are first annotations and further comprising:

fourth program code, stored on the computer-readable-storage media, for identifying validation data for additional agency notices, wherein the validation data includes validation text generated from optical character recognition performed on validation images of additional agency notices and validation metadata for the items in the validation images of the additional agency notices;

fifth program code, stored on the computer-readable-storage media, for sending the validation text to the artificial intelligence system;

sixth program code, stored on the computer-readable-storage media, for receiving results from the artificial intelligence system processing the validation text; and

seventh program code, stored on the computer-readable-storage media, for determining whether the artificial intelligence system identifies the items in the validation text within a desired level of accuracy.

26. The computer program product of claim 25 further comprising:

fourth program code, stored on the computer-readable-storage media, for performing additional training of the artificial intelligence system using additional historical data that includes additional historical text generated from the optical character recognition performed on additional historical images of additional agency notices and additional historical metadata for the items in the additional historical images of the additional agency notices and using additional annotations identifying the items and locations of the items in the additional historical text in response to the desired level of accuracy being absent in the artificial intelligence system.

27. The computer program product of claim 25 , wherein the artificial intelligence system is trained using a first machine learning algorithm and further comprising:

fourth program code, stored on the computer-readable-storage media, for performing additional training of the artificial intelligence system using the historical data and the annotations and a second machine learning algorithm in response to a desired level of accuracy being absent.

28. The computer program product of claim 25 , wherein the historical data is organized using a first type of clustering and further comprising:

fourth program code, stored on the computer-readable-storage media, for performing additional training of the artificial intelligence system using the historical data organized using a second type of clustering and the annotations in response to the desired level of accuracy being absent.