Balancing out-of-dictionary and in-dictionary recognition scores

• US 20070280537A1
• Filed: 06/05/2006
• Published: 12/06/2007
• Est. Priority Date: 06/05/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Slavik, Petr
• Granted Patent: US 7,899,251 B2
• Time in Patent Office: Days
• US Class Current: 382/187
• CPC Class Codes: G06V 30/2272 with lexical matching
• CPC Class Codes: G06V 30/36 Matching; Classification

Abstract

Various technologies and techniques are disclosed that improve handwriting recognition operations. A balancing factor is calculated that can be used in recognition mode to compare out-of-dictionary recognition scores with in-dictionary recognition scores. Correct ink samples of words are provided, with some in-dictionary and some out-of-dictionary words. One or more tuning sets are generated from the samples, the tuning set(s) having a percentage of out-of-dictionary words based upon what is expected for a typical user. A handwriting recognizer is run against the tuning set(s) to determine statistics. The statistics are based upon the recognized answer for each word compared to the correct answer. The statistics are used to calculate the balancing factor. During recognition mode, a user inputs handwriting to be recognized. The balancing factor is used to compare and combine an in-dictionary alternate list with an out-of-dictionary alternate list to reach a recognition decision.

Claims

1. A method for improving handwriting recognition comprising the steps of:

calculating a balancing factor for use in comparing at least one original out-of-dictionary recognition score with at least one in-dictionary recognition score; and

combining the balancing factor with the original out-of-dictionary recognition score to arrive at a revised out-of-dictionary recognition score.

2. The method of claim 1, further comprising:

comparing the in-dictionary recognition score to the revised out-of-dictionary recognition score.

3. The method of claim 1, wherein the revised out-of-dictionary recognition score is a product of the balancing factor and the original out-of-dictionary recognition score.

4. The method of claim 1, wherein the original out-of-dictionary recognition score is a product of a matching probability and a weak language model probability.

5. The method of claim 1, wherein the calculating step comprises:

providing a set of correct ink samples of words, the correct ink samples including at least some in-dictionary words and some out-of-dictionary words;

generating a tuning set from the correct ink samples, the tuning set having a number of out-of-dictionary words representing a particular percentage of out-of-dictionary words expected for a typical user;

running a handwriting recognizer against the tuning set to determine a set of statistics for a plurality of words in the tuning set, the statistics being based at least in part upon a recognized answer for each of the plurality of words compared to a correct answer; and

based on the set of statistics, calculating the balancing factor.

6. The method of claim 1, wherein the balancing factor can be adjusted at least in part by a user.

7. The method of claim 1, further comprising:

repeating the combining stage for a plurality of out-of-dictionary recognition scores to generate a plurality of revised out-of-dictionary recognition scores.

8. The method of claim 7, further comprising:

generating an out-of-dictionary alternate list from the plurality of revised out-of-dictionary recognition scores.

9. The method of claim 8, further comprising:

generating an in-dictionary alternate list with a plurality of in-dictionary recognition scores.

10. The method of claim 9, further comprising:

generating a combined alternate list from the in-dictionary alternate list and the out-of-dictionary alternate list.

11. The method of claim 10, further comprising:

using the combined alternate list to make a recognition decision.

12. A computer-readable medium having computer-executable instructions for causing a computer to perform the steps recited in claim 1.

13. A computer-readable medium having computer-executable instructions for causing a computer to perform steps comprising:

provide a set of correct ink samples of words, the correct ink samples including at least some in-dictionary words and some out-of-dictionary words;

generate a tuning set from the correct ink samples, the tuning set having a number of out-of-dictionary words representing a particular percentage of out-of-dictionary words expected for a typical user;

run a handwriting recognizer against the tuning set to determine a set of statistics for a plurality of words in the tuning set, the statistics being based at least in part upon a recognized answer for each of the plurality of words compared to a correct answer; and

based on the set of statistics, calculate a balancing factor that can be used to compare at least one in-dictionary recognition score with at least one out-of-dictionary recognition score in a recognition operation.

14. The computer-readable medium of claim 13, further having computer-executable instructions for causing a computer to generate the tuning set randomly.

15. The computer-readable medium of claim 13, further having computer-executable instructions for causing a computer to perform the step comprising:

programmatically determine the particular percentage of out-of-dictionary words expected for the typical user.

16. The computer-readable medium of claim 13, further having computer-executable instructions for causing a computer to perform the step comprising:

use the balancing factor to compare the at least one in-dictionary recognition score with the at least one out-of-dictionary recognition score in the recognition operation.

17. A method for improving handwriting recognition comprising the steps of:

receiving handwritten input from a user;

breaking the handwritten input into a plurality of segments;

determining a set of matching probabilities for at least a portion of the segments;

generating an in-dictionary alternate list with a plurality of in-dictionary recognition scores;

generating an out-of-dictionary alternate list with a plurality of out-of-dictionary recognition scores;

using a balancing factor to compare the in-dictionary alternate list with the out-of-dictionary alternate list;

generating a combined alternate list from the in-dictionary alternate list and the out-of-dictionary alternate list; and

using the combined alternate list to make a final recognition decision regarding the handwritten input.

18. The method of claim 17, further comprising:

displaying one or more characters representing the final recognition decision to the user.

19. The method of claim 17, wherein the balancing factor is multiplied by each out-of-dictionary recognition score to arrive at a revised corresponding out-of-dictionary recognition score.

20. A computer-readable medium having computer-executable instructions for causing a computer to perform the steps recited in claim 17.