# US 2019/0379628 A1 · Method and apparatus for detecting fake news in a social media network

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

- Number: US 2019/0379628 A1
- Document ID: US20190379628A1
- Status: Published application
- Kind: A1
- Application number: 16/435,064
- Filed: 2019-06-07
- Published: 2019-12-12
- Earliest claimed: 2018-06-07
- Assignees: Arizona Board of Regents on behalf of Arizona State University
- Applicants: Arizona Board of Regents on Behalf of Arizona State University
- Inventors: Liang Wu; Huan Liu
- Web page: https://patentgrep.com/patent/US20190379628A1

## CPC

- H04L 51/52: for supporting social networking services
- G06F 16/2365: {Ensuring data consistency and integrity}
- G06F 16/355: Creation or modification of classes or clusters
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- H04L 51/212: using filtering or selective blocking
- G06N 20/20: Ensemble learning
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Messages are transmitted in a social media network. Embeddings of social media network users in the social media network are inferred. Propagation pathways over which the plurality of messages are transmitted through the social media network are classified. Action is taken on one or more of the messages that are transmitted through the social media network, based on the classification of the propagation pathways over which the messages are transmitted through the social media network and the inferred embeddings of the social media network users.

## Claims

### Claim 1

A method applied to a plurality of messages in a social media network, the method comprising:
transmitting the plurality of messages in the social media network;
inferring embeddings of social media network users in the social media network;
classifying propagation pathways over which the plurality of messages are transmitted through the social media network; and
taking an action on one or more of the plurality of messages that are transmitted through the social media network based on the classification of the propagation pathways over which the plurality of messages are transmitted through the social media network and the inferred embeddings of the social media network users.

### Claim 2 (depends on 1)

The method of claim 1, wherein inferring embeddings of social media network users in the social media network comprises inferring embeddings of social media network users using Large-Scale Information Network Embedding (LINE) and incorporating community information.

### Claim 3 (depends on 1)

The method of claim 1, wherein classifying propagation pathways of the plurality of messages comprises classifying the propagation pathways of the plurality of messages utilizing a Long Short-Term Memory Recurrent Neural Network (LSTM-RNN).

### Claim 4 (depends on 1)

The method of claim 1, wherein classifying propagation pathways of the plurality of messages comprises classifying the propagation pathways of the plurality of messages as one of: a pathway for spreading fake news or spam, and a pathway for spreading real news.

### Claim 5 (depends on 1)

The method of claim 1, wherein taking an action on one or more of the plurality of messages that are transmitted over a propagation pathway based on the classification of the propagation pathway comprises filtering out the one or more of the plurality of messages that are transmitted over a propagation pathway that is classified as a fake news propagation pathway.

### Claim 6

A method of classifying a message posted and forwarded in a social media network, the method comprising:
receiving from the social media network information about a plurality of messages posted and forwarded in the social media network, including one or more sources of the plurality of messages, a plurality of social media network users that posted or forwarded the plurality of messages in the social media network, and a plurality of sequences in which the social media network users posted or forwarded the plurality of messages in the social media network;
receiving a new message posted and forwarded in the social media network;
receiving from the social media network information about the new message posted and forwarded in the social media network, including a source of the new message, a plurality of social media network users that posted or forwarded the new message in the social media network, and a sequence in which the plurality of social media network users posted or forwarded the new message in the social media network;
comparing the information about the new message with the information about the plurality of messages; and
classifying the new message responsive to the comparison.

### Claim 7 (depends on 6)

The method of claim 6, wherein comparing the information about the new message to the information about the plurality of messages comprises:
providing information about the new message as input to an LSTM-RNN model that is trained with the information about the plurality of messages;
producing an output from the LSTM-RNN model; and
wherein classifying the new message responsive to the comparison comprises classifying the new message responsive to the output of the LSTM-RNN model.

### Claim 8 (depends on 6)

The method of claim 6, wherein classifying the new message comprises classifying the new message as one of fake news, spam, and real news.

### Claim 9 (depends on 6)

The method of claim 6, further comprising taking an action on the new message based on its classification.

### Claim 10 (depends on 9)

The method of claim 9, wherein taking an action on the new message comprises one of: allowing the message to be forwarded by a social media network user in the social media network, and removing the new message from the social media network.

### Claim 11

A system applied to a plurality of messages in a social media network, the system comprising:
a processor to execute software instructions;
a storage device in which to store the social media data;
software instructions that when executed by the processor cause the system to:
transmit the plurality of messages in the social media network;
infer embeddings of social media network users in the social media network;
classify propagation pathways over which the plurality of messages are transmitted through the social media network; and
take an action on one or more of the plurality of messages that are transmitted through the social media network based on the classification of the propagation pathways over which the plurality of messages are transmitted through the social media network and the inferred embeddings of the social media network users.

### Claim 12 (depends on 11)

The system of claim 11, wherein the software instructions that when executed by the processor cause the system to infer embeddings of social media network users in the social media network comprise software instructions that when executed by the processor cause the system to infer embeddings of social media network users using Large-Scale Information Network Embedding (LINE) and incorporating community information.

### Claim 13 (depends on 11)

The system of claim 11, wherein the software instructions that when executed by the processor cause the system to classify propagation pathways of the plurality of messages comprises software instructions that when executed by the processor cause the system to classify the propagation pathways of the plurality of messages utilizing a Long Short-Term Memory Recurrent Neural Network (LSTM-RNN).

### Claim 14 (depends on 11)

The system of claim 11, wherein the software instructions that when executed by the processor cause the system to classify propagation pathways of the plurality of messages comprises software instructions that when executed by the processor cause the system to classify the propagation pathways of the plurality of messages as one of: a pathway for spreading fake news or spam, and a pathway for spreading real news.

### Claim 15 (depends on 11)

The system of claim 11, wherein the software instructions that when executed by the processor cause the system to take an action on one or more of the plurality of messages that are transmitted over a propagation pathway based on the classification of the propagation pathway comprises software instructions that when executed by the processor cause the system to filter the one or more of the plurality of messages that are transmitted over a propagation pathway that is classified as a fake news propagation pathway.

### Claim 16

A system of classifying a message posted and forwarded in a social media network, the system comprising:
a processor to execute software instructions;
a storage device in which to store the social media data;
software instructions that when executed by the processor cause the system to:
receive from the social media network information about a plurality of messages posted and forwarded in the social media network, including one or more sources of the plurality of messages, a plurality of social media network users that posted or forwarded the plurality of messages in the social media network, and a plurality of sequences in which the social media network users posted or forwarded the plurality of messages in the social media network;
receive a new message posted and forwarded in the social media network;
receive from the social media network information about the new message posted and forwarded in the social media network, including a source of the new message, a plurality of social media network users that posted or forwarded the new message in the social media network, and a sequence in which the plurality of social media network users posted or forwarded the new message in the social media network;
compare the information about the new message with the information about the plurality of messages; and
classify the new message responsive to the comparison.

### Claim 17 (depends on 16)

The system of claim 16, wherein the software instructions that when executed by the processor cause the system to compare the information about the new message to the information about the plurality of messages comprises software instructions that when executed by the processor cause the system to:
provide information about the new message as input to an LSTM-RNN model that is trained with the information about the plurality of messages;
produce an output from the LSTM-RNN model; and
wherein the software instructions that when executed by the processor cause the system to classify the new message responsive to the comparison comprises software instructions that when executed by the processor cause the system to classify the new message responsive to the output of the LSTM-RNN model.

### Claim 18 (depends on 16)

The system of claim 16, wherein the software instructions that when executed by the processor cause the system to classify the new message comprises software instructions that when executed by the processor cause the system to classify the new message as one of fake news, spam, and real news.

### Claim 19 (depends on 16)

The system of claim 16, further comprising software instructions that when executed by the processor cause the system to take an action on the new message based on its classification.

### Claim 20 (depends on 19)

The system of claim 19, wherein the software instructions that when executed by the processor cause the system to take an action on the new message comprises one of software instructions that when executed by the processor cause the system to:
allow the message to be forwarded by a social media network user in the social media network; and
remove the new message from the social media network.
