# US 11,443,648 B2 · Systems and methods to assist an instructor of a course

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 11,443,648 B2
- Document ID: US11443648B2
- Status: Granted
- Kind: B2
- Application number: 17/029,707
- Filed: 2020-09-23
- Published: 2021-02-18
- Granted: 2022-09-13
- Earliest claimed: 2014-08-26
- Assignees: Zoomi, Inc.
- Applicants: Zoomi, Inc.
- Inventors: Christopher Greg Brinton; Mung Chiang; Sangtae Ha; William D. Ju; Stefan Rudiger Rill; James Craig Walker; Elizabeth Tenorio
- Web page: https://patentgrep.com/patent/US11443648B2

## CPC

- G09B 7/00: Electrically-operated teaching apparatus or devices working with questions and answers (mechanically operated G09B3/00; computing arrangements G06F)
- G09B 5/02: with visual presentation of the material to be studied, e.g. using film strip
- G09B: EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS (devices for psychotechnics or for testing reaction times A61B5/16; games, sports, amusements A63; projectors, projector screens G03B)

## Abstract

A system and method for delivery of an online course, customized to a student based on captured student's actions in interacting with the course is disclosed. The actions are compared to actions of others, where the actions of others are correlated with known learning results. In part, such comparison involves tagged content and the tagging process is also a part of the present invention.

## Claims

### Claim 1

A method for a processor-based server in communication with at least one data store and a multimedia interactive student graphical user interface (GUI) operating at least in part on a student's workstation; said GUI configured to capture student actions and an associated time factor for each action for delivery to said server; to select and deliver interactive student-customized on-line course modules to said GUI, module delivery customized based on a student's determined understanding of delivered content, said content absent assessment modules, said customization occurring during course delivery, comprising the steps of:
with an engine for processing sound, image, and natural language, identifying learning concepts in a module by tagging content words and topics based on their frequencies in each said module, storing said tags in association with a corresponding module and associating said modules with said frequencies of terms in a course syllabus;
formulating motifs from behavioral data of prior students, each said motif including a sequence of student actions together with associated time, said motifs formed using a probabilistic mixture model wherein each position in a motif is described using at least a mixture of (1) a position-dependent motif model where each position is described as a multinomial distribution specifying one or more probabilities and (2) a position-independent background model where each position is described as a multinomial distribution specifying character occurrence independent of position, and said formulating further including a maximum likelihood estimation with an expectation-maximization algorithm to maximize the expectation of the joint likelihood of the mixture model;
associating said motifs with said learning concepts using said tags;
further associating said motifs with student understanding by algorithmically extracting similarities between prior user behavioral data and assessments by using collaborative filtering, thereby forming motif vectors;
storing said motifs and motif vectors in said at least one data store;
relative to a particular student, delivering a first course module in a sequence of modules, said sequence reflecting a full course, to said GUI, said module retrieved from said at least one data store and including at least one of text, video, and images;
during on-going delivery of said first course module, said GUI capturing all student actions and associated times, said server receiving said captured student actions relative to said delivered portions of said module and associated times;
formulating an individual student vector for a particular student, said individual student vector initially formed as a user model for said particular student, said individual student vector modified based on the particular student's behavioral data on-goingly as the student progresses in a module, said student behavioral data including actions and associated times;
comparing said individual student vector for said particular student with said motif vectors in said data store to form a predictive understanding of the particular student by forming a file-to-file similarity matrix and using said matrix in a supervised learning algorithm to mathematically determine a best match between said individual student vector and said motif vectors;
based on said best match, selecting alternate content as appropriate for said particular student; and
during on-going module delivery, replacing content in a not yet delivered portion of the course with said alternate content in consistency with said best match;
wherein said module is delivered on demand.

### Claim 2 (depends on 1)

The method of claim 1, wherein said tagging is achieved using at least one of natural language processing, text to speech, and optical character recognition.

### Claim 3 (depends on 1)

The method of claim 1, wherein delivery of said first course module halts based on an identified student knowledge gap, and said processor selects and delivers a second course module aligned to said identified knowledge gap.

### Claim 4 (depends on 1)

The method of claim 1, wherein each said sequence of captured actions includes any of reflecting, reviewing, skimming, or speeding.

### Claim 5 (depends on 1)

The method of claim 1, wherein said alternate content as appropriate for said particular student is further selected based on historical or utility scores of alternate next modules provided to students with similarities in their user models.

### Claim 6 (depends on 1)

The method of claim 1, wherein said student understanding is determined by algorithmic association with correct on first attempt in prior testing.

### Claim 7 (depends on 1)

The method of claim 1, wherein said frequency is further calculated using an automated thesaurus so as to aggregate like-meaning terms.

### Claim 8 (depends on 1)

The method of claim 1, wherein in the steps of formulating motifs, motif vectors and individual student vectors, intervals greater than a selected length are discounted.

### Claim 9 (depends on 1)

The method of claim 1, wherein said particular student's understanding is further based on weighting factors associated with frequency or quantity of student rewinds.

### Claim 10 (depends on 1)

The method of claim 1, wherein said step of comparing further includes using at least Multiple EM for Motif Elicitation (MEME) to find similar behaviors between said individual student vector and motif vectors.

### Claim 11 (depends on 1)

The method of claim 1, wherein said step of comparing further includes using at least one of cosine similarity and mathematical correlations between student actions and associated times with stored motifs.

### Claim 12 (depends on 1)

The method of claim 1, wherein said alternate content is further determined based on at least one instructor directed change to at least one module.

### Claim 13 (depends on 1)

The method of claim 1, wherein said processor further comprises a sensor for sensing at least one of movement or sound of a student and any sensing is included as an action.

### Claim 14

A system for customizing on-line modular delivery of a course comprising;
a processor-based server;
at least one data store; and
an interactive student GUI for collecting and forwarding to said server student behaviors together with associated time factors, where said student behaviors comprise at least mouse and keyboard actions;
wherein said at least one data store is used for storing a collection of course module with the frequency of its text terms, a syllabus with the frequency of its text terms, at least one user model, and a collection of known behaviors with associated topical understanding; and
said processor-based server provides course modules to a student selected based on determined student topical material understanding, said understanding based on said user model and a comparison of said student events to said known behaviors, implementing a method of:
with an engine for processing sound, image, and natural language, identifying learning concepts in a module by tagging content words and topics based on word and topic frequencies in each said module, storing said tags in association with a corresponding module and associating said modules with said frequencies and with terms in a course syllabus;
formulating motifs from behavioral data of prior students, each said motif including a sequence of student actions together with associated time, said motifs formed using a probabilistic mixture model wherein each position in a motif is described using a mixture of (1) a position-dependent motif model where each position is described as a multinomial distribution specifying one or more probabilities and (2) a position-independent background model where each position is described as a multinomial distribution specifying character occurrence independent of position, and said formulating further including a maximum likelihood estimation with an expectation-maximization algorithm to maximize the expectation of the joint likelihood of the mixture model;
associating said motifs with said learning concepts using said tags;
further associating said motifs with student understanding by algorithmically extracting similarities between prior user behavioral data and assessments by using collaborative filtering, thereby forming motif vectors;
storing said motifs and motif vectors in said at least one data store;
relative to a particular student delivering a first course module in a sequence of modules, said sequence representing a full course, to said GUI, said first course module retrieved from said at least one data store and including at least one of text, video, and images;
during on-going delivery of said module, said GUI capturing all student actions and associated times, said server receiving said captured student actions relative to said delivered portions of said module and associated times;
formulating an individual student vector for a particular student, said individual student vector initially formed as a user model for said particular student, said individual student vector modified based on the particular student's behavioral data on-goingly as the student progresses in a module, said student behavioral data including actions and associated times;
comparing said individual student vector for said particular student with said motif vectors in said data store to form a predictive understanding of the particular student by forming a file-to-file similarity matrix and using said matrix in a supervised learning algorithm to mathematically determine a best match between said individual student vector and said motif vectors;
based on said best match, selecting alternate content as appropriate for said particular student; and
during on-going module delivery replacing content in a not yet delivered portion of the course with said alternate content in consistency with said best match.

### Claim 15 (depends on 14)

The system of claim 14, further comprising a sensor for sensing at least one of movement or sound of a student.

### Claim 16 (depends on 14)

The system of claim 14, wherein said step of comparing further includes using at least Multiple EM for Motif Elicitation (MEME) to find similar behaviors between said individual student vector and motif vectors.

### Claim 17 (depends on 14)

The system of claim 14, wherein said step of comparing further includes using at least one of cosine similarity and mathematical correlations between student actions and associated times with stored motifs.

### Claim 18 (depends on 14)

The system of claim 14, wherein said alternate content is further determined based on at least one instructor directed change to at least one module.

### Claim 19 (depends on 14)

The system of claim 14, wherein said particular student's understanding is further based on weighting factors associated with frequency or quantity of student rewinds.

### Claim 20 (depends on 14)

The system of claim 14, wherein in the steps of formulating motifs, motif vectors and individual student vectors, intervals greater than a selected length are discounted.