# US 2020/0005047 A1 · Standalone video classification

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

- Number: US 2020/0005047 A1
- Document ID: US20200005047A1
- Status: Published application
- Kind: A1
- Application number: 16/024,804
- Filed: 2018-06-30
- Published: 2020-01-02
- Earliest claimed: 2018-06-30
- Assignees: Microsoft Technology Licensing, LLC
- Applicants: Microsoft Technology Licensing, LLC
- Inventors: Gungor Polatkan; Mahesh S. Joshi; Fares Hedayati; Bonnie Bills
- Web page: https://patentgrep.com/patent/US20200005047A1

## CPC

- G06V 20/47: {Detecting features for summarising video content}
- G06F 18/241: relating to the classification model, e.g. parametric or non-parametric approaches
- G06F 18/24323: {Tree-organised classifiers}
- G06V 10/764: using classification, e.g. of video objects
- G06V 20/41: {Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items (segmenting video sequences G06V20/49)}
- G06V 2201/10: Recognition assisted with metadata
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Techniques for classifying videos as standalone or non-standalone are provided. Feature (or attribute) values associated with a particular video are identified. Feature values are extracted from metadata associated with the particular video and/or from within a transcript of the particular video. The extracted feature values of the particular video are input to a rule-based or a machine-learned model and the model scores the particular video. Once a determination pertaining to whether the particular video is standalone is made, information about the particular video being a standalone video is presented to one or more users within the network.

## Claims

### Claim 1

A method comprising:
identifying multiple co-viewing feature values between a particular video item and a plurality of other video items;
determining a time duration of the particular video item;
based on the multiple co-viewing feature values and the time duration, making a determination whether the particular video is a standalone video;
based on the determination, causing information about the particular video item being a standalone video to be presented to one or more users;
wherein the method is performed by one or more computing devices.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
storing training data that comprises a plurality of training instances, wherein each training instance in the plurality of training instances corresponds to a different video item, includes a plurality of feature values pertaining to the different video item, and includes a label that indicates whether the different video item is a standalone video;
using one or more machine learning techniques to train a classification model based on the training data;
wherein making the determination comprises inserting the multiple co-viewing feature values into the classification model that generates an output that is used to determine whether the particular video item is a standalone video.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
identifying text associated with the particular video;
analyzing the text for one or more textual features;
wherein making the determination is also based on the one or more textual features.

### Claim 4 (depends on 3)

The method of claim 3, wherein analyzing the text comprises determining whether the text includes one or more pre-defined phrases, wherein the one or more textual features includes whether the text includes the one or more pre-defined phrases.

### Claim 5 (depends on 3)

The method of claim 3, wherein analyzing the text comprises identifying a plurality of tokens in the text, determining a token embedding for each token in the plurality of tokens, and performing an aggregation of multiple token embeddings, wherein the one or more textual features is based on output of the aggregation.

### Claim 6 (depends on 3)

The method of claim 3, wherein analyzing the text comprises determining whether the text includes one or more time reference patterns, wherein the one or more textual features includes whether the text includes the one or more time reference patterns.

### Claim 7 (depends on 1)

The method of claim 1, wherein the one or more machine learning techniques include a gradient boosting technique, wherein the classification model comprises a plurality of decision trees.

### Claim 8 (depends on 1)

The method of claim 1, wherein the one or more machine learning techniques include a pruning technique, wherein the classification model comprises a plurality of decision trees.

### Claim 9

One or more storage media storing instructions which, when executed by one or more processors, cause:
identifying multiple co-viewing feature values between a particular video item and a plurality of other video items;
determining a time duration of the particular video item;
based on the multiple co-viewing feature values and the time duration, making a determination whether the particular video is a standalone video;
based on the determination, causing information about the particular video item being a standalone video to be presented to one or more users.

### Claim 10 (depends on 9)

The one or more storage media of claim 9, wherein the instructions, when executed by the one or more processors, further cause:
storing training data that comprises a plurality of training instances, wherein each training instance in the plurality of training instances corresponds to a different video item, includes a plurality of feature values pertaining to the different video item, and includes a label that indicates whether the different video item is a standalone video;
using one or more machine learning techniques to train a classification model based on the training data;
wherein making the determination comprises inserting the multiple co-viewing feature values into the classification model that generates an output that is used to determine whether the particular video item is a standalone video.

### Claim 11 (depends on 9)

The one or more storage media of claim 9, wherein the instructions, when executed by the one or more processors, further cause:
identifying text associated with the particular video;
analyzing the text for one or more textual features;
wherein making the determination is also based on the one or more textual features.

### Claim 12 (depends on 11)

The one or more storage media of claim 11, wherein analyzing the text comprises determining whether the text includes one or more pre-defined phrases, wherein the one or more textual features includes whether the text includes the one or more pre-defined phrases.

### Claim 13 (depends on 11)

The one or more storage media of claim 11, wherein analyzing the text comprises identifying a plurality of tokens in the text, determining a token embedding for each token in the plurality of tokens, and performing an aggregation of multiple token embeddings, wherein the one or more textual features is based on output of the aggregation.

### Claim 14 (depends on 11)

The one or more storage media of claim 11, wherein analyzing the text comprises determining whether the text includes one or more time reference patterns, wherein the one or more textual features includes whether the text includes the one or more time reference patterns.

### Claim 15 (depends on 9)

The one or more storage media of claim 9, wherein the one or more machine learning techniques include a gradient boosting technique, wherein the classification model comprises a plurality of decision trees.

### Claim 16 (depends on 9)

The one or more storage media of claim 9, wherein the one or more machine learning techniques include a pruning technique, wherein the classification model comprises a plurality of decision trees.

### Claim 17

A system comprising:
one or more processors;
one or more storage media storing instructions which, when executed by the one or more processors, cause:
storing course data that identifies a plurality of video courses, each of which contains a plurality of video items;
for each video course of the plurality of video courses:
for each video item in the plurality of video items that correspond to said each video course:
identifying a plurality of feature values of said each video item,
determining, based on the plurality of feature values, whether said each video item is a standalone video;
if it is determined that said each video item is a standalone video, then storing first classification data that classifies said each video item as a standalone video;
if it is determined that said each video item is a non-standalone video, then storing second classification data that classifies said each video item as a non-standalone video;
wherein a particular video course of the plurality of video courses includes at least one video item that is classified as a standalone video and at least one video item that is classified as a non-standalone video.

### Claim 18 (depends on 17)

The system of claim 17, wherein the instructions, when executed by the one or more processors, further cause:
identifying a set of standalone videos;
determining a relevance score for each standalone video in the set of standalone videos relative to a particular user;
based on the relevance score for each standalone video in the set of standalone videos, identifying a strict subset of the set of standalone videos;
receiving a content request that is initiated by the particular user;
in response to receiving the content request, causing the strict subset of the set of standalone videos to be presented on a computing device associated with the particular user.

### Claim 19 (depends on 18)

The system of claim 18, wherein the strict subset of the set of standalone videos is presented on a first portion of a web page, wherein the instructions, when executed by the one or more processors, further cause:
identifying a set of video courses of the plurality of video courses;
determining a second relevance score for each video course in the set of video courses relative to the particular user;
based on the second relevance score for each video course in the set of video courses, identifying a strict subset of the set of video courses;
in response to receiving the content request, causing the strict subset of the set of video courses to be presented on a second portion, of the web page, that does not overlap with the first portion of the web page.

### Claim 20 (depends on 17)

The system of claim 17, wherein a plurality of features of the plurality of feature values include two or more of a duration, an indication of a time reference pattern, and a co-viewing attribute.
