# US 2017/0286979 A1 · Architecture for predicting network access probability of data files accessible over a computer network

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

- Number: US 2017/0286979 A1
- Document ID: US20170286979A1
- Status: Published application
- Kind: A1
- Application number: 15/091,273
- Filed: 2016-04-05
- Published: 2017-10-05
- Earliest claimed: 2016-04-05
- Assignees: Shutterstock, Inc.
- Applicants: Shutterstock, Inc.
- Inventors: Alexander CHAVEZ; David CHESTER; Heath HOHWALD; Nathan HURST; Kevin Scott LESTER; Manor LEV-TOV
- Web page: https://patentgrep.com/patent/US20170286979A1

## CPC

- G06Q 30/0202: Market predictions or forecasting for commercial activities
- G06F 15/16: Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs {(coordinating program control therefor G06F9/52; in regulating and control system G05B)}
- G06F 16/288: {Entity relationship models}
- G06F 16/435: Filtering based on additional data, e.g. user or group profiles
- G06N 7/06: Simulation on general purpose computers
- G06Q 20/04: Payment circuits
- H04L 9/40: Network security protocols
- H04N 7/16: Analogue secrecy systems; Analogue subscription systems
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

Methods for predicting network access probability of data files accessible over a computer network are provided. In one aspect, a method includes generating a primary data vector for a media file based on a stored data representation of the file, and providing the data vector for the file to an algorithm that uses past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value. The method also includes receiving, as an output of the algorithm, a marketability score for the media file, the score indicative of a likelihood that a user will download the media file. Systems and machine-readable media are also provided.

## Claims

### Claim 1

A method comprising:
generating a primary data vector for a media file based on a stored data representation of the media file;
providing the data vector for the media file to an algorithm for predicting a marketability of the media file based on past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value; and
receiving, as an output of the algorithm, a marketability score for the media file, the marketability score indicative of a likelihood that a user will download the media file.

### Claim 2 (depends on 1)

The method of claim 1, wherein the at least one other media file comprises a plurality of other media files, and wherein the algorithm for predicting a marketability of the media file is configured to:
generate a plurality of additional data vectors for the plurality of other media files;
identify a position of the primary data vector and the plurality of additional data vectors in a vector space;
determine, based on the positions of the primary data vector and the plurality of additional data vectors in the vector space, that the primary data vector and the plurality of additional data vectors are within a distance threshold to one another;
analyze past user interaction with the plurality of other media files;
generate, for each of the plurality of other media files, a similarity score representing a similarity in the vector space between the media file and the plurality of other media files; and
generate the marketability score for the media file based on the similarity scores and the past user interaction with the at least one other media file.

### Claim 3 (depends on 2)

The method of claim 2, wherein generating the marketability score for the media file based on the similarity scores and the past user interaction with the at least one other media file comprises forming a weighted average by multiplying, for each of the plurality of other media files, the respective similarity score between the corresponding other media file and the media file times the average download rate of the corresponding other media file over a period of time.

### Claim 4 (depends on 3)

The method of claim 3, wherein the similarity score is generated using a Hamming distance between the corresponding other media file and the at least one media file, or using cosine similarity algorithm.

### Claim 5 (depends on 3)

The method of claim 3, wherein the weighted average is weighted at least in part based on a distance between the primary data vector for the media file in the vector space and the data vector in the vector space for the corresponding other media file.

### Claim 6 (depends on 1)

The method of claim 1, wherein the algorithm for predicting a marketability of the media file is configured to:
generate a plurality of other data vectors for the media files in the collection of media files;
cluster the plurality of other data vectors into a predetermined number of clusters according to a clustering algorithm;
generate a marketability value for each of the predetermined number of clusters based on past interaction information for the media files corresponding to the data vectors in the corresponding cluster;
assign the primary data vector to at least one of the predetermined number of clusters using the clustering algorithm; and
generate the marketability score for the media file based on the marketability value for the at least one cluster to which the primary data vector is assigned.

### Claim 7 (depends on 6)

The method of claim 6, wherein the primary data vector is assigned to a plurality of the predetermined number of clusters using the clustering algorithm, and the marketability score for the media file is generated based on the marketability value for the plurality of the predetermined number of clusters to which the primary data vector is assigned.

### Claim 8 (depends on 6)

The method of claim 6, wherein the marketability value for a cluster is based on the average download rate of all of the media files in the cluster over a period of time.

### Claim 9 (depends on 6)

The method of claim 6, wherein the primary data vector and each of the plurality of other data vectors is generated using a convolutional neural network.

### Claim 10 (depends on 1)

The method of claim 1, wherein the at least one other media file comprises a plurality of other media files, and wherein the algorithm for predicting a marketability of the media file is configured to:
generate a plurality of other data vectors for the media files in the collection of media files and a download indicator identifying of whether the media file corresponding to the data vector was previously downloaded;
provide the plurality of other data vectors and their corresponding download indicators to a supervised learning algorithm to generate a mapping function that maps a vector representation of an input media file to a probability that the input media file will be downloaded;
apply the mapping function generated by the supervised learning algorithm to the primary data vector for the media file to generate a probability that the media file will be downloaded; and
generate the marketability score for the media file based on the probability that the media file will be downloaded.

### Claim 11 (depends on 10)

The method of claim 10, wherein the other data vector corresponding to a media file in the collection and its corresponding download indicator is provided to the supervised learning algorithm according to the number of times the corresponding media file has been downloaded by a user.

### Claim 12

A system comprising:
a memory comprising a media file; and
a processor configured to execute instructions to:
generate a primary data vector for a media file based on a stored data representation of the media file;
provide the data vector for the media file to an algorithm for predicting a marketability of the media file based on past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value; and
receive, as an output of the algorithm, a marketability score for the media file, the marketability score indicative of a likelihood that a user will download the media file.

### Claim 13 (depends on 12)

The system of claim 12, wherein the at least one other media file comprises a plurality of other media files, and wherein the algorithm for predicting a marketability of the media file is configured to:
generate a plurality of additional data vectors for the plurality of other media files;
identify a position of the primary data vector and the plurality of additional data vectors in a vector space;
determine, based on the positions of the primary data vector and the plurality of additional data vectors in the vector space, that the primary data vector and the plurality of additional data vectors are within a distance threshold to one another;
analyze past user interaction with the plurality of other media files;
generate, for each of the plurality of other media files, a similarity score representing a similarity in the vector space between the media file and the plurality of other media files, the similarity score generated using a Hamming distance between the corresponding other media file and the at least one media file, or using cosine similarity algorithm; and
generate the marketability score for the media file based on the similarity scores and the past user interaction with the at least one other media file by forming a weighted average comprising multiplying, for each of the plurality of other media files, the respective similarity score between the corresponding other media file and the media file times the average download rate of the corresponding other media file over a period of time.

### Claim 14 (depends on 13)

The system of claim 13, wherein the weighted average is weighted at least in part based on a distance between the primary data vector for the media file in the vector space and the data vector in the vector space for the corresponding other media file.

### Claim 15 (depends on 12)

The system of claim 12, wherein the algorithm for predicting a marketability of the media file is configured to:
generate a plurality of other data vectors for the media files in the collection of media files;
cluster the plurality of other data vectors into a predetermined number of clusters according to a clustering algorithm;
generate a marketability value for each of the predetermined number of clusters based on past interaction information for the media files corresponding to the data vectors in the corresponding cluster;
assign the primary data vector to at least one of the predetermined number of clusters using the clustering algorithm; and
generate the marketability score for the media file based on the marketability value for the at least one cluster to which the primary data vector is assigned.

### Claim 16 (depends on 15)

The system of claim 15, wherein the primary data vector is assigned to a plurality of the predetermined number of clusters using the clustering algorithm, and the marketability score for the media file is generated based on the marketability value for the plurality of the predetermined number of clusters to which the primary data vector is assigned.

### Claim 17 (depends on 15)

The system of claim 15, wherein the marketability value for a cluster is based on the average download rate of all of the media files in the cluster over a period of time, and wherein the primary data vector and each of the plurality of other data vectors is generated using a convolutional neural network.

### Claim 18 (depends on 12)

The system of claim 12, wherein the at least one other media file comprises a plurality of other media files, and wherein the algorithm for predicting a marketability of the media file is configured to:
generate a plurality of other data vectors for the media files in the collection of media files and a download indicator identifying of whether the media file corresponding to the data vector was previously downloaded;
provide the plurality of other data vectors and their corresponding download indicators to a supervised learning algorithm to generate a mapping function that maps a vector representation of an input media file to a probability that the input media file will be downloaded;
apply the mapping function generated by the supervised learning algorithm to the primary data vector for the media file to generate a probability that the media file will be downloaded; and
generate the marketability score for the media file based on the probability that the media file will be downloaded.

### Claim 19 (depends on 10)

The system of claim 10, wherein the other data vector corresponding to a media file in the collection and its corresponding download indicator is provided to the supervised learning algorithm according to the number of times the corresponding media file has been downloaded by a user.

### Claim 20

A non-transitory machine-readable storage medium comprising machine-readable instructions for causing a processor to execute a method comprising:
generating a primary data vector for a media file based on a stored data representation of the media file;
providing the data vector for the media file to an algorithm for predicting a marketability of the media file based on past interaction information for at least one other media file from a collection of media files having a degree of similarity with the media file above a threshold similarity value; and
receiving, as an output of the algorithm, a marketability score for the media file, the marketability score indicative of a likelihood that a user will download the media file.