Machine learning of context of data fields for various document types

Abstract

A method and system learns new forms to be incorporated into an electronic document preparation system. The method and system receive form data related to a new form having a plurality of data fields that expect data values based on specific functions. The method and system gather training set data including previously filled forms having completed data fields corresponding to the data fields of the new form. The method and system utilize machine learning in conjunction with the training set data to identify the correct function for each of the data fields of the new form.

What is claimed is: 
     
         1 . A computing system implemented method for learning and incorporating new forms in an electronic document preparation system, the method comprising:
 receiving form data related to a new form having a plurality of data fields;   gathering training set data related to previously filled forms, each previously filled form having completed data fields that each correspond to a respective data field of the new form;   generating, for a first selected data field of the plurality of data fields of the new form, dependency data indicating one or more possible dependencies for a correct function that provides a proper data value for the first selected data field;   generating, for the first selected data field, candidate function data including a plurality of candidate functions based on the dependency data and one or more operators selected from a set of operators;   generating, for each candidate function, test data by applying the candidate function to the training set data;   generating, for each candidate function, matching data by comparing the test data to the completed data fields corresponding to the first selected data field, the matching data indicating how closely the test data matches the corresponding completed data fields of the previously filled forms;   identifying, from the plurality of functions, the correct candidate function for the first selected data field of the new form by determining, for each candidate function, whether or not the candidate function is the correct function for the first selected data field of the new form based on the matching data;   generating, after identifying the correct function for the first data field, results data indicating the correct function for the first data field of the new form; and   outputting the results data.   
     
     
         2 . The method of  claim 1 , wherein outputting the results data includes outputting the results data for review by an expert. 
     
     
         3 . The method of  claim 2 , further comprising prompting an expert to approve the correct function for the first data field of the new form. 
     
     
         4 . The method of  claim 1 , further comprising, after identifying the correct function for the first selected data field of the new form, identifying a second correct function for a second selected data field from the plurality of data fields of the new form. 
     
     
         5 . The method of  claim 4 , wherein identifying the second correct function for the second selected data field of the new form includes:
 generating, for the second selected data field, second dependency data indicating one or more possible second dependencies for the second correct function;   generating, for the second selected data field, second candidate function data including a plurality of second candidate functions based on the second dependency data and one or more operators selected from a set of operators;   generating, for each second candidate function, second test data by applying the second candidate function to the second training set data;   generating, for each candidate function, second matching data by comparing the second test data to the completed data fields corresponding to the second selected data field, the second matching data indicating how closely the second test data matches the corresponding completed data fields of the previously filled forms;   identifying, from the plurality of second candidate functions, the second correct candidate function for the first selected data field of the new form by determining, for each second candidate function, whether or not the second candidate function is the second correct function for the second selected data field of the new form based on the second matching data;   generating, after identifying the second correct function for the second selected data field, second results data indicating the second correct function for the second data field of the new form; and   outputting the second results data.   
     
     
         6 . The method of  claim 1 , further comprising generating, for each candidate function, confidence score data based on the matching data and indicating a level of confidence that the candidate function is the correct function for the first data field. 
     
     
         7 . The method of  claim 1 , wherein the training set data includes historical financial data related to previously prepared financial documents, the historical financial data including the previously filled forms. 
     
     
         8 . The method of  claim 7 , wherein the historical financial data includes previously prepared financial documents that were previously filed with a government or financial institution. 
     
     
         9 . The method of  claim 1 , wherein the training set data includes fabricated financial data related to fabricated financial documents, the fabricated financial data including the previously filled forms. 
     
     
         10 . The method of  claim 9 , further comprising receiving the fabricated financial data from one or more third parties. 
     
     
         11 . The method of  claim 1 , further comprising generating historical document instructions data related to software instructions for completing historical forms. 
     
     
         12 . The method of  claim 11 , further comprising analyzing the historical document instructions data. 
     
     
         13 . The method of  claim 12 , wherein generating the candidate function data includes generating the dependency data based on the historical document instructions data. 
     
     
         14 . The method of  claim 1 , further comprising generating natural language parsing data by performing natural language parsing analysis on the form data. 
     
     
         15 . The method of  claim 11 , wherein generating the dependency data includes generating the dependency data based on the natural language parsing data. 
     
     
         16 . The method of  claim 1 , wherein the set of operators includes one or more of:
 an addition operator;   a subtraction operator;   a division operator;   a multiplication operator;   an exponential operator;   logical operators;   a string comparison operator; and   existence condition operators.   
     
     
         17 . The method of  claim 1 , wherein the candidate functions generate numerical data values. 
     
     
         18 . The method of  claim 1 , wherein the possible dependencies include one or more of:
 a data field from the new form;   multiple data fields from the new form;   a data field from a form other than a new form;   multiple data fields from multiple forms other than the new form; and   a constant.   
     
     
         19 . The method of  claim 1 , wherein the new form is a new tax form. 
     
     
         20 . The method of  claim 19 , wherein the training set data includes previously prepared tax returns. 
     
     
         21 . The method of  claim 20 , wherein the training set data includes fabricated tax returns. 
     
     
         22 . A non-transitory computer-readable medium having a plurality of computer-executable instructions which, when executed by a processor, perform a method for learning and incorporating new forms in an electronic document preparation system, the instructions comprising:
 an interface module configured to receive form data representing to a new form having a plurality of data fields;   a data acquisition module configured to gather training set data related to previously filled forms, each previously filled form having completed data fields that each correspond to a respective data field of the new form; and   a machine learning module configured to identify a respective correct function for each of the data fields of the new form by generating candidate function data relating to a plurality of candidate functions based on dependency data indicating possible dependencies for each data field of the new form and including one or more operators from a set of operators, generating test data by applying the candidate functions to the training set data, and finding, for each of the data fields a respective correct function from the plurality of candidate functions based on a how closely the test data matches the candidate function data.   
     
     
         23 . The non-transitory computer-readable medium of  claim 22  wherein the machine learning module is configured to continue generating candidate functions for each data field of the new form until the corresponding correct function is found. 
     
     
         24 . The non-transitory computer-readable medium of  claim 22  wherein the machine learning module is configured to generate results data indicating the respective correct function for each data field of the new form. 
     
     
         25 . The non-transitory computer-readable medium of  claim 22  wherein the electronic document preparation system includes a financial document preparation system. 
     
     
         26 . The non-transitory computer-readable medium of  claim 22  wherein the financial document preparation system includes a tax return preparation system. 
     
     
         27 . A system for learning and incorporating new forms in an electronic document preparation system, the system comprising:
 at least one processor; and   at least one memory coupled to the at least one processor, the at least one memory having stored therein instructions which, when executed by any set of the one or more processors, perform a process including:   receiving, with an interface module of a computing system, form data related to a new form having a plurality of data fields;   gathering, with a data acquisition module of a computing system, training set data related to previously filled forms, each previously filled form having completed data fields that each correspond to a respective data field of the new form;   generating, with a machine learning module of a computing system, for a first selected data field of the plurality of data fields of the new form, dependency data indicating one or more possible dependencies for a correct function that provides a proper data value for the first selected data field;   generating, with the machine learning module, for the first selected data field, candidate function data including a plurality of candidate functions based on the dependency data and one or more operators selected from a set of operators;   generating, with the machine learning module, for each candidate function, test data by applying the candidate function to the training set data;   generating, with the machine learning module, for each candidate function, matching data by comparing the test data to the completed data fields corresponding to the first selected data field, the matching data indicating how closely the test data matches the corresponding completed data fields of the previously filled forms;   identifying, with the machine learning module, from the plurality of functions, the correct candidate function for the first selected data field of the new form by determining, for each candidate function, whether or not the candidate function is the correct function for the first selected data field of the new form based on the matching data;   generating, with the machine learning module and after identifying the correct function for the first data field, results data indicating the correct function for the first data field of the new form; and   outputting, with the interface module, the results data.   
     
     
         28 . The system of  claim 27 , wherein outputting the results data includes outputting the results data for review by an expert. 
     
     
         29 . The system of  claim 28 , further comprising prompting an expert to approve the correct function for the first data field of the new form. 
     
     
         30 . The system of  claim 27 , wherein the process further includes, after identifying the correct function for the first selected data field of the new form, identifying a second correct function for a second selected data field from the plurality of data fields of the new form. 
     
     
         31 . The system of  claim 30 , wherein identifying the second correct function for the second selected data field of the new form includes:
 generating, for the second selected data field, second dependency data indicating one or more possible second dependencies for the second correct function;   generating, for the second selected data field, second candidate function data including a plurality of second candidate functions based on the second dependency data and one or more operators selected from a set of operators;   generating, for each second candidate function, second test data by applying the second candidate function to the second training set data;   generating, for each candidate function, second matching data by comparing the second test data to the completed data fields corresponding to the second selected data field, the second matching data indicating how closely the second test data matches the corresponding completed data fields of the previously filled forms;   identifying, from the plurality of second candidate functions, the second correct candidate function for the first selected data field of the new form by determining, for each second candidate function, whether or not the second candidate function is the second correct function for the second selected data field of the new form based on the second matching data;   generating, after identifying the second correct function for the second selected data field, second results data indicating the second correct function for the second data field of the new form; and   outputting the second results data.   
     
     
         32 . The system of  claim 27 , wherein the process includes generating, for each candidate function, confidence score data based on the matching data and indicating a level of confidence that the candidate function is the correct function for the first data field. 
     
     
         33 . The system of  claim 27 , wherein the training set data includes historical financial data related to previously prepared financial documents, the historical financial data including the previously filled forms. 
     
     
         34 . The system of  claim 33 , wherein receiving the form data includes receiving the form data automatically from a third party. 
     
     
         35 . The system of  claim 27 , wherein the training set data includes fabricated financial data related to fabricated financial documents, the fabricated financial data including the previously filled forms. 
     
     
         36 . The system of  claim 35 , wherein the process includes receiving the fabricated financial data from one or more third parties. 
     
     
         37 . The system of  claim 27 , wherein the process includes generating historical document instructions data related to software instructions for completing historical forms. 
     
     
         38 . The system of  claim 37 , wherein the process includes generating the dependency data by analyzing the historical document instructions data. 
     
     
         39 . The system of  claim 38 , wherein the process includes generating the candidate function data based on the historical document instructions data. 
     
     
         40 . The system of  claim 27 , wherein the process includes generating natural language parsing data by performing natural language parsing analysis on the form data. 
     
     
         41 . The system of  claim 37 , wherein the process includes generating the dependency data based on the natural language parsing data. 
     
     
         42 . The system of  claim 27 , wherein the set of operators includes one or more of:
 an addition operator;   a subtraction operator;   a division operator;   a multiplication operator;   an exponential operator;   logical operators;   a string comparison operator; and   existence condition operators.   
     
     
         43 . The system of  claim 27 , wherein the candidate functions generate numerical data values. 
     
     
         44 . The system of  claim 27 , wherein the possible dependencies include one or more of:
 a data field from the new form;   multiple data fields from the new form;   a data field from a form other than a new form;   multiple data fields from multiple forms other than the new form; and   a constant.   
     
     
         45 . The system of  claim 27 , wherein the new form is a new tax form. 
     
     
         46 . The system of  claim 45 , wherein the training set data includes previously prepared tax return. 
     
     
         47 . The system of  claim 45 , wherein the process includes updating a tax return preparation system based on the correct function. 
     
     
         48 . The system of  claim 47 , wherein the process further includes updating a tax return preparation interview script to include one or more prompts, notifications, or explanations to a user based on the correct function.