Method and apparatus for storing corrected words with previous user-corrected recognition results to improve recognition
First Claim
Patent Images
1. A method, comprising the steps of:
- comparing information representing user input with a first set of recognition data;
generating a list of candidate recognition results based on the step of comparing information representing user input with the first set of recognition data, wherein each candidate recognition result in the list has an associated value representing a likelihood of the candidate recognition result correctly replicating the user input as typewritten text;
comparing each candidate recognition result in the list with a second set of recognition data, wherein the second set of recognition data represents previous user-corrected recognition results explicitly corrected by a user;
enhancing the associated value for a candidate recognition result when the candidate recognition result is found in the second set of recognition data to reflect a higher likelihood of the candidate recognition result correctly replicating the user input;
presenting to the user the candidate recognition result that has a highest likelihood of correctly replicating the user input;
processing the candidate recognition result with user instructions to modify it least a portion of the candidate recognition result that has the highest likelihood of correctly replicating the user input to generate a user-corrected recognition result; and
adding the user-corrected recognition result to the second set of recognition data, wherein the user-corrected recognition result is used for later comparison with subsequent lists of recognition results.
4 Assignments
0 Petitions
Accused Products
Abstract
A cache is used to provide additional information to a recognition system about what words are likely to be used in a current context. Words are added to the cache only if they are explicitly corrected by the user. If the user forgets to correct (or postpones correcting) a recognition error, nothing is added to the cache, and therefore, there is no negative impact on the recognition system.
36 Citations
14 Claims
-
1. A method, comprising the steps of:
-
comparing information representing user input with a first set of recognition data; generating a list of candidate recognition results based on the step of comparing information representing user input with the first set of recognition data, wherein each candidate recognition result in the list has an associated value representing a likelihood of the candidate recognition result correctly replicating the user input as typewritten text; comparing each candidate recognition result in the list with a second set of recognition data, wherein the second set of recognition data represents previous user-corrected recognition results explicitly corrected by a user; enhancing the associated value for a candidate recognition result when the candidate recognition result is found in the second set of recognition data to reflect a higher likelihood of the candidate recognition result correctly replicating the user input; presenting to the user the candidate recognition result that has a highest likelihood of correctly replicating the user input; processing the candidate recognition result with user instructions to modify it least a portion of the candidate recognition result that has the highest likelihood of correctly replicating the user input to generate a user-corrected recognition result; and adding the user-corrected recognition result to the second set of recognition data, wherein the user-corrected recognition result is used for later comparison with subsequent lists of recognition results. - View Dependent Claims (2, 3, 4, 5, 6, 7)
-
-
8. A device, comprising:
-
a digitizing input device for receiving user input and for providing information representing the user input; a display for displaying the user input; a memory having an operating system and application programs stored therein, including recognition instructions, a first set of recognition data and a second set of recognition data; and digital processing circuitry including a programmable digital processing device capable of processing the recognition instructions in accordance with the operating system to; compare the information representing the user input with the first set of recognition data; generate a list of candidate recognition results based on comparing information representing the user input with the first set of recognition data, wherein each candidate recognition result in the list has an associated having a value representing a likelihood of the candidate recognition result correctly replicating the user input as typewritten text; compare each candidate recognition result in the list with the second set of recognition data, wherein the second set of recognition data represents previous user-corrected recognition results; enhance the associated value for a candidate recognition result when the candidate recognition result is found in the second set of recognition data to reflect a higher likelihood of correctly replicating the user input; present to a user the candidate recognition result that has a highest likelihood of correctly replicating the user input; implement user instructions to modify at least a portion of the candidate recognition result that has the highest likelihood of correctly replicating the user input to generate a user-corrected recognition result; and add the user-corrected recognition result to the second set of recognition data, wherein the user-corrected recognition result is used for later comparison with subsequent lists of recognition results. - View Dependent Claims (9, 10, 11, 12, 13, 14)
-
Specification