Electronic handwriting analysis through adaptive machine-learning

Patent metadata

Field	Value
Publication number	US-10740601-B2
Application number	US-201815905595-A
Country	US
Kind code	B2
Filing date	Feb 26, 2018
Priority date	Apr 10, 2017
Publication date	Aug 11, 2020
Grant date	Aug 11, 2020

Abstract

An improved machine learning system is provided. For example, a content management server may provide a digital assessment of a user's handwriting to assess the user's knowledge of a language. The assessment may comprise adaptive technology to help determine initial questions to provide to the user as well as follow-up questions to clarify appropriate remediation content in a particular context. The content management server may also provide real-time analysis, including assessing multiple users at the same time in adjusting the assessment based on the digital input from each of these users. In some examples, the content management server may incorporate handwriting analysis methods to perform object detection and score handwriting input.

First claim

What is claimed is:

1. A content management system for improving distribution and conversion of academic data, the content management system comprising:
one or more processors; and
one or memories coupled with the one or more processors, wherein the one or more processors and the one or more memories are configured to:
receive, by the one or more processors, a training data set comprising at least a model test item and a model test response, wherein:
the model test item being associated with a plurality of letters to spell a word associated with the model test item, and
the model test response comprising a first digital representation of handwritten user input associated with the model test item;

determine a model response score based on an analysis of the model test item;
generate a model ground truth scoring method that correlates the model response score with the training data set;
receive a second response, the second response comprising a second digital representation of handwritten user input associated with a second test item and being associated with a first interaction with the second test item;
receive a third response, the third response comprising a third digital representation of handwritten user input associated with the second test item and being associated with a second interaction with the second test item;
analyze, without user input, the third response using an automated handwriting assessment method, the model ground truth scoring method, and a time delta between a first duration of the first interaction with the second test item and a second duration of the second interaction with the second test item by executing a machine learning engine; and
determine, without user input, a user response score based on the analysis of the third response.

2. The content management system of claim 1 , wherein the one or more memories are further configured to:
analyze individual letters in the third digital representation of handwritten user input;
analyze spelling in the third digital representation of handwritten user input; and
correlate a combination of a likelihood of a word as the third response to a test item presented at a user interface of a user device.

3. The content management system of claim 1 , wherein the one or more memories are further configured to:
provide a plurality of test items to a plurality of user devices, at least one of the plurality of test items being associated with a second plurality of letters to spell a second word.

4. The content management system of claim 3 , wherein the plurality of test items are provided at a same time to the plurality of user devices.

5. The content management system of claim 1 , wherein the analysis of the third response using the automated handwriting assessment method and the model ground truth scoring method is performed in real-time and the one or more memories are further configured to:
alter the second test item based at least in part on the real-time analysis.

6. The content management system of claim 1 , wherein the model ground truth scoring method includes averaging scores to determine the model response score.

7. A computer-implemented method for rendering a markup language document, the method comprising:
receiving a training data set comprising at least a model test item and a model test response, wherein:
the model test item being associated with a plurality of letters to spell a word associated with the model test item, and
the model test response comprising a first digital representation of handwritten user input associated with the model test item;

determining a model response score based on an analysis of the model test item;
generating a model ground truth scoring method that correlates the model response score with the training data set;
receiving a second response, the second response comprising a second digital representation of handwritten user input associated with a second test item and being associated with a first interaction with the second test item;
receiving a third response, the third response comprising a third digital representation of handwritten user input associated with the second test item and being associated with a second interaction with the second test item;
analyzing, without user input, the third response using an automated handwriting assessment method, the model ground truth scoring method, and a time delta between a first duration of the first interaction with the second test item and a second duration of the second interaction with the second test item by executing a machine learning engine; and
determining, without user input, a user response score based on the analysis of the third response.

8. The computer-implemented method of claim 7 , further comprising:
analyzing individual letters in the third digital representation of handwritten user input;
analyzing spelling in the third digital representation of handwritten user input; and
correlating a combination of a likelihood of a word as the third response to a test item presented at a user interface of a user device.

9. The computer-implemented method of claim 7 , further comprising:
providing a plurality of test items to a plurality of user devices, at least one of the plurality of test items being associated with a second plurality of letters to spell a second word.

10. The computer-implemented method of claim 9 , wherein the plurality of test items are provided at a same time to the plurality of user devices.

11. The computer-implemented method of claim 7 , wherein the analysis of the third response using the automated handwriting assessment method and the model ground truth scoring method is performed in real-time and the method further comprising:
altering the second test item based at least in part on the real-time analysis.

12. The computer-implemented method of claim 7 , wherein the model ground truth scoring method includes averaging scores to determine the model response score.

13. One or more non-transitory computer-readable storage media collectively storing computer-executable instructions that, when executed by one or more computer systems, configure the one or more computer systems to collectively perform operations comprising:
receiving a training data set comprising at least a model test item and a model test response, wherein:
the model test item being associated with a plurality of letters to spell a word associated with the model test item, and
the model test response comprising a first digital representation of handwritten user input associated with the model test item;

determining a model response score based on an analysis of the model test item;
generating a model ground truth scoring method that correlates the model response score with the training data set;
receiving a second response, the second response comprising a second digital representation of handwritten user input associated with a second test item and being associated with a first interaction with the second test item;
receiving a third response, the third response comprising a third digital representation of handwritten user input associated with the second test item and being associated with a second interaction with the second test item;
analyzing, without user input, the third response using an automated handwriting assessment method, the model ground truth scoring method, and a time delta between a first duration of the first interaction with the second test item and a second duration of the second interaction with the second test item by executing a machine learning engine; and
determining, without user input, a user response score based on the anal

Assignees

• Pearson Education Inc

Inventors

• Wang Zhaodong

Classifications

• G09B5/02 Primary

  with visual presentation of the material to be studied, e.g. using film strip · CPC title

• G06V30/18057

  Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN] · CPC title

• G06V30/166

  Normalisation of pattern dimensions · CPC title

• G06V30/347

  Sampling; Contour coding; Stroke extraction · CPC title

• G06N3/045

  Combinations of networks · CPC title