Knowledge driven pre-trained form key mapping
The disclosure herein describes generating input key-standard key mappings for a form. A set of input key-value pairs are received, and a subset of candidate form types are determined from a set of form types using the input key-value pairs. A set of standard keys associated with the determined subset of candidate form types are obtained. A set of input key-standard key pairs are generated using the set of input key-value pairs and the obtained set of standard keys and the set of input key-standard key pairs are narrowed using a narrowing rule. Ranking scores for each input key-standard key pair of the narrowed set of input key-standard key pairs are generated. Each input key of the set of input key-vale pairs is mapped to a standard key of the set of standard keys using at least the generated ranking scores of the narrowed set of input key-standard key pairs.
1. A computerized method comprising:
receiving a set of input key-value pairs associated with a form;
determining a subset of candidate form types from a set of form types using the set of input key-value pairs;
obtaining a set of standard keys associated with the determined subset of candidate form types;
generating a set of input key-standard key pairs using the set of input key-value pairs and the obtained set of standard keys;
narrowing the set of input key-standard key pairs using a narrowing rule;
generating ranking scores for each input key-standard key pair of the narrowed set of input key-standard key pairs using a trained model; and
mapping each input key of the set of input key-value pairs to a standard key of the set of standard keys using at least the generated ranking scores of the narrowed set of input key-standard key pairs.
2. The computerized method of claim 1 , wherein determining the subset of candidate form types from the set of form types using the set of input key-value pairs includes:
generating, using a trained form type model, key vectors of input keys of the set of input key-value pairs;
aggregating the generated key vectors into an aggregate vector of the form;
comparing the aggregate vector of the form to aggregate vectors of the set of form types; and
selecting the subset of candidate form types from the set of form types based on the comparing of the aggregate vector of the form to the aggregate vectors of the set of form types.
3. The computerized method of claim 1 , wherein narrowing the set of input key-standard key pairs using a narrowing rule includes:
modifying an input key of the set of input key-standard key pairs using the narrowing rule;
identifying a standard key of the set of input key-standard key pairs that matches the modified input key; and
removing a subset of input key-standard key pairs that include the input key modified by the narrowing rule from the set of input key-standard key pairs based on identifying the standard key that matches the modified input key.
4. The computerized method of claim 3 , wherein modifying the input key using the narrowing rule includes at least one of the following: removing punctuation from the input key, changing whitespace in the input key, and removing whitespace from the input key.
5. The computerized method of claim 1 , wherein the set of form types includes at least one of the following: a standard set of form types and a customer-specific set of form types, wherein the form with which the received set of input key-value pairs is associated is associated with a customer of the customer-specific set of form types.
6. The computerized method of claim 1 , wherein mapping each input key of the set of input key-value pairs to a standard key of the set of standard keys using at least the generated ranking scores of the narrowed set of input key-standard key pairs further includes:
identifying standard keys of the narrowed set of input key-standard key pairs associated with highest ranking scores for each input key of the narrowed set of input key-standard key pairs;
determining a subset of form types associated with the identified standard keys;
selecting a form type of the determined subset of form types using the identified standard keys; and
mapping the input keys to standard keys of the selected form type for which the associated ranking scores are the highest values.
7. The computerized method of claim 6 , wherein selecting the form type of the determined subset of form types using the identified standard keys includes selecting the form type with which a largest quantity of identified standard keys are associated.
8. One or more non-transitory computer storage media having computer-executable instructions that, upon execution by a processor, cause the processor to at least:
receive a set of input key-value pairs associated with a form;
determine a subset of candidate form types from a set of form types using the set of input key-value pairs;
obtain a set of standard keys associated with the determined subset of candidate form types;
generate a set of input key-standard key pairs using the set of input key-value pairs and the obtained set of standard keys;
narrow the set of input key-standard key pairs using a narrowing rule;
generate ranking scores for each input key-standard key pair of the narrowed set of input key-standard key pairs using a trained model; and
map each input key of the set of input key-value pairs to a standard key of the set of standard keys using at least the generated ranking scores of the narrowed set of input key-standard key pairs.
9. The one or more non-transitory computer storage media of claim 8 , wherein determining the subset of candidate form types from the set of form types using the set of input key-value pairs includes:
generating, using a trained form type model, key vectors of input keys of the set of input key-value pairs;
aggregating the generated key vectors into an aggregate vector of the form;
comparing the aggregate vector of the form to aggregate vectors of the set of form types; and
selecting the subset of candidate form types from the set of form types based on the comparing of the aggregate vector of the form to the aggregate vectors of the set of form types.
10. The one or more non-transitory computer storage media of claim 8 , wherein narrowing the set of input key-standard key pairs using a narrowing rule includes:
modifying an input key of the set of input key-standard key pairs using the narrowing rule;
identifying a standard key of the set of input key-standard key pairs that matches the modified input key; and
removing a subset of input key-standard key pairs that include the input key modified by the narrowing rule from the set of input key-standard key pairs based on identifying the standard key that matches the modified input key.
11. The one or more non-transitory computer storage media of claim 10 , wherein modifying the input key using the narrowing rule includes at least one of the following: removing punctuation from the input key, changing whitespace in the input key, and removing whitespace from the input key.
12. The one or more non-transitory computer storage media of claim 8 , wherein the set of form types includes at least one of the following: a standard set of form types and a customer-specific set of form types, wherein the form with which the received set of input key-value pairs is associated is associated with a customer of the customer-specific set of form types.
13. The one or more non-transitory computer storage media of claim 8 , wherein mapping each input key of the set of input key-value pairs to a standard key of the set of standard keys using at least the generated ranking scores of the narrowed set of input key-standard key pairs further includes:
identifying standard keys of the narrowed set of input key-standard key pairs associated with highest ranking scores for each input key of the narrowed set of input key-standard key pairs;
determining a subset of form types associated with the identified standard keys;
selecting a form type of the determined subset of form types using the identified standard keys; and
mapping the input keys to standard keys of the selected form type for which the associated ranking scores are the highest values.
14. The one or more non-transitory computer storage media of claim 13 , wherein selecting the form type of the determined subset of form types using the identified standard keys includes selecting the form type with which a largest quantity of identified standard keys are associated.