System

Abstract

A system is provided.SOLUTION: A system, comprising: means for detecting an offensive message to an account of a digital assistant; means for generating an advocacy message if the offensive message exceeds a certain threshold; means for distributing the generated advocacy message to a digital assistant of a collaborator; and means for posting the advocacy message to a social networking service by the collaborator via the digital assistant.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In social networking services (SNS), an increase in aggressive messages directed at specific users can cause mental stress and, in some cases, develop into a serious social problem. There is a need to provide an environment where users can use SNS safely without being affected by such aggressive messages. However, current SNS platforms do not provide sufficient, effective solutions to this problem. This is the problem that the present invention aims to solve. [Means for solving the problem]

[0005] The present invention provides a system for suppressing social networking site flame wars by detecting offensive messages and generating and distributing corresponding defense messages. Specifically, the system includes a means for detecting offensive messages directed at an information terminal account, a means for generating a defense message when the number of offensive messages exceeds a certain threshold, a means for distributing the generated defense message to collaborators' information terminals, and a means for collaborators to post the defense message on the social networking site via their information terminals. The generation of the defense message is automatically created using a generative model, and the detection of the offensive message is performed using a natural language processing algorithm to identify offensive content and negative sentiment. In this way, offensive messages can be dealt with quickly and effectively, providing an environment in which users can safely use the social networking site.

[0006] "Information terminal" refers to any electronic device that allows a user to connect to the Internet and send and receive information.

[0007] "Offensive Message" means a message that contains content that is offensive, insulting, or that induces negative feelings toward other users.

[0008] A "generative model" refers to an algorithm or system that uses artificial intelligence or machine learning to automatically generate new data or content.

[0009] "Advocacy messages" refer to messages that support, encourage, or defend a user who is under attack.

[0010] "Natural language processing algorithms" refers to technologies and methods that enable computers to understand, interpret, and generate human language.

[0011] "Contributors" refer to users who collaborate to post advocacy messages in the system.

[0012] "Social networking services (SNS)" refer to platforms and services that facilitate communication between people over the Internet.

[0013] "Posting means" refers to the functions and mechanisms for publishing advocacy messages on social media. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0015] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0016] First, the terms used in the following description will be explained.

[0017] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0018] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0019] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0020] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0021] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0022] [First embodiment]

[0023] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0024] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0025] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0026] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0027] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0028] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0029] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0030] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0032] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0033] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0034] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0035] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and a method for implementing the system will be described using a specific example.

[0036] Overall system overview

[0037] This system consists of a server, a terminal, and a user. The server monitors the SNS accounts of service subscribers and detects offensive messages. For detected offensive messages, the server uses a generative model to generate defensive messages and distributes them to the terminals of collaborators. The collaborators then post defensive messages to the SNS via their terminals, thereby mitigating the impact of the offensive messages.

[0038] Server Roles

[0039] 1. Starting the monitoring module

[0040] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[0041] 2. Detecting offensive messages

[0042] The server analyzes the data it receives using a natural language processing algorithm to identify offensive messages. The algorithm performs sentiment analysis of the language and detects content that is deemed offensive. Detected offensive messages are stored in a dedicated log.

[0043] 3. Generating Advocacy Messages

[0044] If the number of offensive messages exceeds a certain threshold, the server uses a generative model to generate defensive messages, which undergo quality checks and are added to a message queue.

[0045] 4. Send messages to collaborators

[0046] The server accesses the list of registered contributors and distributes the generated advocacy messages to the contributors' devices, quickly using APIs and notification systems.

[0047] Device Role

[0048] 1. Receiving a message

[0049] The collaborator's device receives the advocacy message sent from the server, which causes the message to be stored in the device's local data store.

[0050] 2. Submission Preparation and Notification

[0051] The device application notifies the user that a new message has been sent to the user interface, and after confirmation, the user is ready to post the message to the SNS.

[0052] User Roles

[0053] 1. Posting a message of support

[0054] The contributor user checks the advocacy message sent from the server and posts it to the SNS via the application on their device. By pressing the post button, the message is made public using the SNS API.

[0055] Specific examples

[0056] A concrete example is given below. For example, suppose a user receives a large number of offensive messages. In this case, the server detects these messages and generates a supportive message such as, "This user's behavior is admirable. Let's support them together." This message is sent to the collaborator's device, and the collaborator posts it on SNS, mitigating the impact of the offensive messages.

[0057] In this way, the SNS Shield system counters offensive messages in real time, providing a safe environment for users to use SNS.

[0058] The processing flow will be explained below.

[0059] Step 1: Starting the monitoring module

[0060] The server starts a module for monitoring the social media accounts of service subscribers, allowing the server to obtain subscribers' posts and comments in real time.

[0061] Step 2: Getting the Timeline Data

[0062] The server periodically collects subscribers' latest posts and comments via the SNS's API and stores them in a database.

[0063] Step 3: Detecting offensive messages

[0064] The server runs natural language processing algorithms on the collected data to detect offensive messages, specifically identifying messages that contain negative sentiment or offensive content.

[0065] Step 4: Threshold check

[0066] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, proceeds to the next step.

[0067] Step 5: Generating an advocacy message

[0068] The server uses the generative model to generate advocacy messages, such as "This user is doing great work, let's support them."

[0069] Step 6: Check message quality

[0070] The server quality checks the generated advocacy message for correctness, and if incorrect, it is regenerated.

[0071] Step 7: Prepare the message for distribution

[0072] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list, and adds them to the message queue for distribution.

[0073] Step 8: Distribute messages to collaborators

[0074] The server distributes messages to the collaborators' devices. Messages are sent to the collaborators' devices through APIs or notification systems.

[0075] Step 9: Receiving a message

[0076] The terminal receives the advocacy message sent from the server and stores it in a local data store.

[0077] Step 10: User Interface Notification

[0078] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[0079] Step 11: Review your advocacy message

[0080] The user checks the notification on the terminal and confirms the content of the received advocacy message.

[0081] Step 12: Post an Advocacy Message

[0082] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[0083] Step 13: Feedback on submission results

[0084] The terminal notifies the server that the message has been successfully posted, which causes a posting log to be recorded on the server.

[0085] Step 14: Evaluate the effectiveness of the system

[0086] The server evaluates the effectiveness of the system based on the feedback and improves the generative model and algorithms as needed.

[0087] Example 1

[0088] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0089] On modern online platforms, online flame wars caused by offensive messages are becoming more frequent, placing a greater mental burden on users. This problem significantly undermines the environment in which users can use the platform safely. Conventional systems have not established a method for effectively suppressing these offensive messages, and it is particularly difficult to respond in real time. Therefore, there is a need to provide a safe and secure environment for users by quickly detecting offensive messages on online platforms and responding appropriately.

[0090] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0091] In this invention, the server includes a means for monitoring abusive messages to the account of the information processing device, a means for generating a defense message when the abusive messages exceed a certain threshold, and a means for distributing the generated defense message to the information processing device of the collaborator. This makes it possible to detect abusive messages in real time and respond quickly to prevent online flaming and provide a safe and secure environment for users.

[0092] "Information processing device" generally refers to electronic devices capable of data processing, such as computers and smartphones.

[0093] An "account" is data including identification information and authentication information that allows a user to be individually managed on an online platform.

[0094] "Offensive messages" are messages that are intended to hurt, insult, or cause anxiety to others.

[0095] "Generative AI models" refer to algorithms or frameworks for generating text or data using artificial intelligence techniques.

[0096] "Natural language processing algorithms" refer to computational techniques for understanding, generating, and manipulating human language.

[0097] A "threshold" is a value that meets a particular condition or criterion, and when exceeded, triggers a particular action.

[0098] "Advocacy messages" refer to positive messages that are generated to mitigate the impact of aggressive messages and protect the target audience.

[0099] "Online Platform" refers to a website or application that enables users to share information, exchange, or interact over the Internet.

[0100] "Distribution" refers to the sharing of specific information or data with multiple recipients.

[0101] The present invention relates to a system for suppressing flame wars caused by offensive messages on online platforms. The system is composed of an information processing device, a server, and a user. Specific embodiments for carrying out the invention are described below.

[0102] Configuration of information processing device

[0103] The information processing device is an electronic device capable of data processing, such as a computer or smartphone, and functions as a terminal used by collaborators. This information processing device includes a network connection function for receiving messages and a local data store for saving and displaying received messages. The specific implementation uses an SQLite database.

[0104] Server Configuration

[0105] The server consists of the following main modules:

[0106] Monitoring module: Implemented using Python language and Django framework, it monitors the accounts of service subscribers.

[0107] Natural language processing module: Analyzes and detects offensive messages using TensorFlow and PyTorch.

[0108] Generative module: Creates advocacy messages using a generative AI model (e.g., OpenAI GPT-3).

[0109] Notification module: Distributes generated advocacy messages to collaborators' information processing devices using RESTful APIs and the Firebase notification system.

[0110] User Roles

[0111] The collaborating user checks the advocacy message received from the server and posts it to the online platform via an information processing device. The user uses the application on their device to press the post button, which publishes the message through the SNS API (e.g., Twitter API).

[0112] Overview of the invention

[0113] The server monitors the subscriber's account on the online platform and detects offensive messages using a natural language processing module. A generation module generates a defensive message for the detected offensive message. The defensive message is distributed to the collaborator's information processing device, and after the user has reviewed it, the user can post it on the online platform to mitigate the impact of the offensive message.

[0114] Specific examples

[0115] For example, if a user receives an offensive message such as "I completely disagree with your opinion. I think it's wrong," the server detects this message and uses a generative AI model (e.g., GPT-3) to generate a defensive message such as "I think it's okay to have different opinions. Let's respect this user's opinion." This message is then sent to the collaborator's information processing device, and the collaborator posts it on an online platform to mitigate the impact of the offensive message.

[0116] Examples of prompt statements

[0117] Offensive messages:

[0118] "I don't agree with you at all. I think you're wrong."

[0119] Advocacy message:

[0120] "It's good to have different opinions. Let's respect this user's opinion."

[0121] In this way, the system responds to offensive messages in real time, providing a safe and secure environment for users to use online platforms.

[0122] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0123] Step 1:

[0124] The server activates a monitoring module to monitor the service subscriber's account.

[0125] Input: Service subscriber account information

[0126] Specific operation: Uses the Django framework and executes the management command python manage.py runserver.

[0127] Data processing: Send a request to the SNS API to retrieve timeline data.

[0128] Output: Timeline data is saved to a database.

[0129] Step 2:

[0130] The server analyzes the acquired timeline data using a natural language processing algorithm to detect offensive messages.

[0131] Input: Saved timeline data

[0132] Specific operation: Loads a TensorFlow sentiment analysis model and performs analysis on the data.

[0133] Data processing: Processing data with sentiment analysis algorithms to identify offensive messages.

[0134] Output: Offensive messages are saved to a dedicated log file.

[0135] Step 3:

[0136] The server uses a generative AI model to generate a defensive message when an offensive message exceeds a certain threshold.

[0137] Input: offensive message

[0138] Specific behavior: Send an offensive message as a prompt to the GPT-3 API.

[0139] Data processing: The generated advocacy messages are reviewed in a quality check module.

[0140] Output: Advocacy messages that meet the criteria are added to the message queue.

[0141] Step 4:

[0142] The server distributes the advocacy message generated based on the collaborator list to the information processing device.

[0143] Input: Deferred message in message queue

[0144] Specific behavior: Sends notifications using RESTful APIs and the Firebase notification system.

[0145] Data processing: Extract the message from the message queue.

[0146] Output: The message is distributed to the collaborators' information processing devices.

[0147] Step 5:

[0148] The collaborator's terminal receives the advocacy message sent from the server.

[0149] Input: Notification from the server

[0150] Specific operation: Set up a listener to receive notifications from Firebase.

[0151] Data processing: Received messages are saved in a SQLite database.

[0152] Output: Saved advocacy messages are logged to a local data store.

[0153] Step 6:

[0154] The terminal notifies the user of the arrival of a new message and displays it on the user interface, allowing the user to check the message and prepare it for posting.

[0155] Input: Saved advocacy message

[0156] Specific operation: Notifies using the notification manager (Notification Manager for Android, UNUserNotificationCenter for iOS).

[0157] Data processing: Display the message content within the app.

[0158] Output: The user has reviewed the message and is ready to hit the post button.

[0159] Step 7:

[0160] A user posts an advocacy message to an online platform via an application on the terminal.

[0161] Input: Saved advocacy messages and user actions

[0162] Specific operation: By pressing the post button, a POST request is sent to the SNS API (e.g. Twitter API).

[0163] Data processing: Generates request data to be sent to the SNS API.

[0164] Output: The advocacy message is published on an online platform.

[0165] (Application example 1)

[0166] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0167] In recent years, there has been an increase in offensive and abusive messages on social networking services (SNS). These messages can have a negative impact on users' mental health and can lead to a phenomenon known as "flaming." While methods to prevent and mitigate this are needed, effective countermeasures are currently lacking. Furthermore, because manual responses are time-consuming and labor-intensive, real-time automated countermeasures are needed.

[0168] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0169] In this invention, the server includes means for detecting offensive messages to the account of the information terminal, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the information terminal of the collaborator, means for the collaborator to post the defense message to the social networking service via the information terminal, means for monitoring the SNS account in real time, means for notifying the generated defense message, and means for the notified user to post the message with one click after confirmation. This makes it possible to respond quickly and automatically to offensive messages and minimize the impact of flame wars on the SNS.

[0170] An "information terminal" is an electronic device that a user uses to access the Internet.

[0171] An "offensive message" is a message that contains negative or offensive content and may cause psychological harm to a user.

[0172] The "threshold" is a criterion for determining whether an offensive message exceeds a certain standard value.

[0173] "Advocacy messages" are messages that positively evaluate and support users and are generated to counter aggressive messages.

[0174] A "collaborator" is a user whose role is to receive advocacy messages and post them on social media from their own account.

[0175] "Real time" is a time concept in which information is acquired and processed immediately, meaning a nearly simultaneous response.

[0176] "Notifications" are alerts or messages sent to collaborators' information terminals when a specific event occurs.

[0177] "One-click posting" is a feature that allows users to post messages on social media with minimal operations, i.e., a single click or tap.

[0178] A "natural language processing algorithm" is a technology that analyzes human language and understands meaning and emotions.

[0179] A "generative AI model" is an artificial intelligence technology that automatically generates new text based on existing data.

[0180] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and is implemented as follows.

[0181] Overall system configuration

[0182] This system consists of a server, the collaborator's information terminal, and the user who operates it. The server monitors SNS accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators. The information terminal receives defensive messages, and posts them to the SNS after user confirmation.

[0183] Hardware and software configuration

[0184] Hardware:

[0185] Server: a high-performance computer

[0186] Information device: Smartphone (iPhone or Android device)

[0187] software:

[0188] Server-side program:

[0189] Natural Language Processing Algorithms (TensorFlow, Keras, Hugging Face Transformers)

[0190] Message Distribution API

[0191] Information terminal app:

[0192] Programming language: Swift (iOS), Kotlin (Android)

[0193] Database: Firebase Firestore

[0194] Notification system: Firebase Cloud Messaging (FCM)

[0195] Program processing

[0196] Server Processing

[0197] The server monitors social media accounts in real time to obtain timeline data, which it then stores and updates in a database. The server then uses a natural language processing algorithm to detect offensive messages from the obtained data. This algorithm identifies offensive content and negative sentiment.

[0198] If the number of offensive messages exceeds a certain threshold, the server uses a generative AI model to generate a defensive message. The generated message undergoes a quality check and is then sent to the collaborator's device. Notifications are sent using Firebase Cloud Messaging (FCM), allowing collaborators to respond quickly.

[0199] Information terminal processing

[0200] The collaborator's information device receives the advocacy message sent from the server. When the notification arrives, the message is saved in the device's local data store, and the application notifies the user to display the new message on the user interface, ready for the user to check and post to SNS with one click.

[0201] Specific examples

[0202] For example, if a user receives a large number of offensive messages on Twitter, the server detects these messages and generates a supportive message such as, "This user's actions are admirable. Let's support them together!" This message is sent to the collaborator's information terminal, and the collaborator can post this message to Twitter with one click, thereby mitigating the impact of the offensive messages.

[0203] Example prompts to input to a generative AI model:

[0204] A specific user is being attacked on social media. Please generate a positive, supportive message to defuse the situation. Please keep it general and supportive, without including specific usernames or examples of behavior.

[0205] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0206] Step 1:

[0207] The server starts monitoring the social media account in real time, obtains the user's timeline data, and saves and updates it in the database.

[0208] Input: User's social media account information, timeline data

[0209] Output: Timeline data stored in a database

[0210] Specific operation: The server retrieves the user's social media timeline through the API, updates it periodically, and saves this data in a database (e.g., Firebase Firestore).

[0211] Step 2:

[0212] The server analyzes the timeline data using natural language processing algorithms to detect offensive messages.

[0213] Input: Timeline data stored in a database, natural language processing algorithms (e.g., Hugging Face Transformers)

[0214] Output: List of detected offensive messages

[0215] Specific operation: The server uses a library for natural language processing to analyze the timeline data, extracting posts containing offensive messages or negative sentiment and creating a list of them.

[0216] Step 3:

[0217] If the server detects an offensive message that exceeds a certain threshold, it uses a generative AI model to generate a defensive message.

[0218] Input: List of offensive messages, generative AI model

[0219] Output: Generated advocacy message

[0220] Specific operation: The server inputs a prompt sentence to the generative AI model to create a supportive message to defuse an aggressive situation. Based on the prompt sentence, the generative AI model generates an appropriate supportive message.

[0221] Step 4:

[0222] The server distributes the generated advocacy message to the information terminals of the collaborators.

[0223] Input: Generated advocacy message, list of contributors

[0224] Output: Advocacy message received on the collaborator's information terminal

[0225] Specific operation: The server uses Firebase Cloud Messaging to send notifications of the advocacy message to all devices included in the contributor list.

[0226] Step 5:

[0227] The terminal receives the protection message sent from the server and stores it in a local data store.

[0228] Input: Advice message sent from the server

[0229] Output: Advocacy messages saved to the local data store

[0230] Specific behavior: The device receives the notification and saves it to a local data store (e.g., an SQLite database).

[0231] Step 6:

[0232] The device will display a notification prompting the user to acknowledge the advocacy message.

[0233] Input: Advocacy messages stored in the local data store

[0234] Output: User-visible notification message

[0235] What happens: The device application displays a new message notification on the user interface, and the user taps the notification to view the message details.

[0236] Step 7:

[0237] Users can view advocacy messages and post them to social media with one click.

[0238] Input: Advocacy message, SNS account information

[0239] Output: Advocacy messages posted on social media

[0240] Specific operation: When a user presses a button that allows them to post a support message to a social networking site with one click, the device application posts the message using the social networking site API (e.g., Twitter API).

[0241] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0242] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and in particular, an embodiment in which an emotion engine that recognizes the emotions of users is combined will be described.

[0243] Overall system overview

[0244] This system consists of a server, a terminal, and a user, and integrates an emotion engine to improve the effectiveness of advocacy messages. The server monitors the SNS accounts of service subscribers and detects offensive messages. The emotion engine recognizes the user's emotional state in real time and adjusts the generation of advocacy messages based on the results. The generated advocacy messages are distributed to the terminals of collaborators, who then post them on the SNS.

[0245] Server Roles

[0246] 1. Starting the monitoring module

[0247] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[0248] 2. Detecting offensive messages

[0249] The server analyzes the acquired data using a natural language processing algorithm to detect offensive messages. This algorithm performs sentiment analysis of the language and detects content that is deemed offensive. The detected offensive messages are stored in a dedicated log.

[0250] 3. Applying the Emotion Engine

[0251] The server uses an emotion engine to recognize the user's emotional state in real time, thereby determining whether the user is experiencing stress or anxiety.

[0252] 4. Threshold Checking and Generative Model Tuning

[0253] The server determines whether the number of detected offensive messages exceeds a set threshold and adjusts the generation of advocacy messages based on data from the emotion engine, using a generative model to generate appropriate message content according to the user's emotional state.

[0254] 5. Message Creation and Preparation for Distribution

[0255] The server uses the generative model to create an advocacy message that passes the quality check, prepares it for distribution to the collaborators' terminals, and adds it to the message queue for distribution.

[0256] 6. Send messages to collaborators

[0257] The server distributes the generated advocacy messages to the collaborators' terminals via APIs and notification systems based on the list of registered collaborators.

[0258] Device Role

[0259] 1. Receiving a message

[0260] The collaborator's terminal receives the advocacy message sent from the server and stores it in a local data store.

[0261] 2. User interface notifications

[0262] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[0263] User Roles

[0264] 1. Review and post advocacy messages

[0265] The collaborator user checks the notification on their device and confirms the content of the received advocacy message. Then, they post the received advocacy message to the SNS via the device's application. By pressing the post button, the device publishes the message using the SNS API.

[0266] Specific examples

[0267] Here's a concrete example: In a situation where a user is receiving a large number of aggressive messages, the emotion engine detects that the user's stress level is rising. Based on this information, the server generates a more emotionally sensitive supportive message, such as, "Your efforts are being heard by everyone. There is a lot of support for you." This message is sent to the collaborator's device, and the collaborator posts it on social media, effectively mitigating the impact of the aggressive messages.

[0268] In this way, the SNS Shield system applies an emotion engine to provide real-time support messages that correspond to the user's emotional state, providing an environment in which users can use SNS safely.

[0269] The processing flow will be explained below.

[0270] Step 1: Starting the monitoring module

[0271] The server runs a module that monitors subscribers' social media accounts, capturing their posts and comments in real time and storing them in a database. A dedicated database entry is created for each monitored account.

[0272] Step 2: Getting the Timeline Data

[0273] The server periodically collects subscribers' timeline data via SNS API, including posts, comments, replies, likes, etc. This data is stored in a database and prepared for subsequent processing.

[0274] Step 3: Detecting offensive messages

[0275] The server applies natural language processing algorithms to the collected timeline data, which are designed to identify messages containing offensive or negative content, and records the detected messages in a dedicated log.

[0276] Step 4: Applying the Emotion Engine

[0277] The server simultaneously monitors the collected offensive messages and activates an emotion engine that analyzes the user's current emotions based on their timeline data and past messages, and determines their stress and anxiety levels.

[0278] Step 5: Threshold check

[0279] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, the server considers the user's emotional state data obtained from the emotion engine and proceeds to the next step.

[0280] Step 6: Generating Advocacy Messages

[0281] The server uses a generative model to generate appropriate advocacy messages based on data from the emotion engine. For example, if a user's stress level is high, a message such as "This user is doing great things, so let's support them" is generated.

[0282] Step 7: Check message quality

[0283] The server quality-checks the content of generated advocacy messages, regenerating inappropriate messages, and fine-tuning the tone and content of messages based on feedback from the emotion engine.

[0284] Step 8: Preparing the Message for Distribution

[0285] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list. The messages are added to a message queue for distribution.

[0286] Step 9: Distribute messages to collaborators

[0287] The server distributes advocacy messages to registered collaborators' devices using APIs and notification systems. When a notification is received, the collaborator receives the message on their own device.

[0288] Step 10: Receiving a message

[0289] The terminal receives the protection message sent from the server and stores it in local storage, and the user interface displays the new message to notify the user.

[0290] Step 11: User Interface Notification

[0291] The terminal displays the received advocacy message on the user interface and notifies the collaborator that there is a new message. The user checks the contents of the message.

[0292] Step 12: Review your advocacy message

[0293] The user checks the notification on the device and confirms the content of the received advocacy message. If the message content is deemed appropriate, the user prepares to post it.

[0294] Step 13: Post an Advocacy Message

[0295] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[0296] Step 14: Feedback on submission results

[0297] The terminal notifies the server that the message has been successfully posted, and the server saves the posting log and uses it as data to evaluate the effectiveness of the system.

[0298] Step 15: Evaluate the effectiveness of the system

[0299] The server evaluates the effectiveness of the system based on the feedback data and improves the generative model, emotion engine, and natural language processing algorithm as needed.

[0300] Through the above steps, the system utilizes the emotion engine to generate a defense message according to the user's emotional state, and can effectively mitigate the impact of aggressive messages.

[0301] Example 2

[0302] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0303] Flaming on social networking services (SNS) caused by offensive messages places a significant psychological burden on users and has become a social problem. To solve this problem, a system that can detect offensive messages and take appropriate countermeasures is needed, but current technology is insufficient in taking countermeasures based on the user's real-time emotional state. In addition, there is a need to improve the accuracy of detecting offensive messages and the quality of the generated defense messages.

[0304] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0305] In this invention, the server includes a means for detecting offensive messages, a means for generating a defense message when a threshold is exceeded, a means for distributing the generated defense message to collaborators, a means for recognizing the emotional state of the user in real time, and a means for adjusting the content of the defense message based on the emotional state acquired in real time, thereby making it possible to quickly and effectively suppress flame wars caused by offensive messages and reduce the burden on users.

[0306] An "information terminal" is a device used by a user, and includes a smartphone, tablet, or personal computer.

[0307] An "offensive message" is a message that contains malicious or harassing words or content and causes a psychological burden to the user.

[0308] A "means for detecting" is a method or device that uses specific algorithms or techniques to identify and identify offensive messages.

[0309] A "protective message" is a message of support or encouragement that is generated to help a user who has received an offensive message and to reduce the mental burden.

[0310] A "generating means" is a method or device for automatically generating advocacy messages using a specific algorithm or model.

[0311] "Emotional state" refers to a particular psychological state or mood that a user is experiencing, including states such as anxiety, stress, and joy.

[0312] A "real-time recognition means" is a method or device for instantly analyzing and understanding a user's current emotional state.

[0313] An "adjusting means" is a method or device for modifying or optimizing generated messages or other output content based on acquired data or conditions.

[0314] The "distribution means" refers to a method or device for transmitting the generated advocacy message to the collaborator's information terminal and making it available.

[0315] "Allies" are users or groups selected to post advocacy messages on social media.

[0316] A "social networking service" is a platform that enables users to interact and share information with other users online.

[0317] MODE FOR CARRYING OUT THE INVENTION

[0318] The present invention is a system for suppressing flame wars caused by offensive messages on social networking services (SNS). The system is composed of a server, a terminal, and a user, and includes the following elements and means:

[0319] Server Roles

[0320] The server plays a key role and performs the following functions:

[0321] 1. Launching the social media account monitoring module and collecting data

[0322] The server periodically obtains timeline data of service subscribers using SNS APIs, for example, by using Twitter APIs to collect tweet data from users and store and update it in a database.

[0323] 2. Detecting offensive messages

[0324] The server analyzes the collected data using Python natural language processing libraries (such as NLTK and spaCy). It uses specific keywords and sentiment analysis techniques to detect offensive messages. Messages that are deemed offensive are recorded in a dedicated log file.

[0325] 3. Applying the Emotion Engine

[0326] The server uses IBM Watson's sentiment analysis API to extract emotions from users' posts, allowing it to understand in real time the level of stress or anxiety a user is experiencing. Emotion scores are stored in a database and updated in real time.

[0327] 4. Threshold Checking and Adjusting the Advocacy Message Generation Model

[0328] The server evaluates whether the number of detected offensive messages exceeds a pre-defined threshold, and adjusts the parameters of a defense message generation model (e.g., GPT-3) based on the user's emotional state data.

[0329] 5. Advocacy message generation and quality check

[0330] The server inputs a prompt into a generative AI model (such as GPT-3) to generate a supportive message tailored to the user's situation. For example, the prompt might read, "The user is currently receiving aggressive messages on social media and is experiencing high stress levels. Please generate a supportive message to alleviate this situation." The generated message passes a quality check and is then prepared for distribution.

[0331] 6. Preparing and implementing advocacy message distribution

[0332] The server adds the verified advocacy message to a message queue and sends it to the registered collaborators through Firebase Cloud Messaging or APIs, for example, by calling the "SendMessage" API to deliver the message to the collaborators' devices.

[0333] Device Role

[0334] The device provides an interface for collaborators to display the advocacy messages they receive and post them to social media.

[0335] 1. Receiving and notifying advocacy messages

[0336] The collaborator's device receives the message sent by the server and stores it in a local data store. The device uses a notification system to notify the collaborator that a new message has been received, for example by displaying a pop-up or banner notification.

[0337] 2. Review and post advocacy messages

[0338] The collaborator user checks the notification on their device. After checking the content of the advocacy message, they press the "Post" button to call the SNS API and post the message to SNS. For example, we will post the message using the "POST statuses / update" endpoint.

[0339] User Roles

[0340] When users receive an offensive message on SNS, they can receive support through the system, which reduces the psychological burden and allows them to continue using SNS with peace of mind.

[0341] Specific examples

[0342] For example, consider a situation where User A is receiving a large number of aggressive messages on social media. Based on this information, the server uses its emotion engine to detect that User A's stress level is rising. Based on this data, it generates a supportive message such as, "Your efforts are being noticed by everyone. There are many voices of support for you." This message is sent to the device of Collaborator B, who then posts it on social media to support User A.

[0343] Prompt Sentence Examples

[0344] "Currently, users are receiving aggressive messages on social media and their stress levels are high. Please generate advocacy messages to ease this situation."

[0345] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0346] Step 1:

[0347] The server starts the SNS account monitoring module and periodically calls the SNS API of the service subscriber. As input, it receives the SNS API authentication information and user ID. Specifically, it retrieves the latest 20 tweets using the Twitter API's "GET statuses / user_timeline" endpoint. The output is the retrieved tweet data, which is then stored and updated in the database.

[0348] Step 2:

[0349] The server analyzes the tweet data stored in the database using natural language processing algorithms. The stored tweet data is used as input. Specifically, text analysis is performed using Python's NLTK and spaCy. To detect offensive messages, specific keywords and sentiment scores are calculated. The output is a list of messages deemed offensive, which is recorded in a dedicated log file.

[0350] Step 3:

[0351] The server calls IBM Watson's sentiment analysis API to analyze the user's emotional state. It uses the analyzed tweet data as input. Specifically, it sends the text of each tweet to the sentiment analysis API and obtains a sentiment score. The output is a sentiment score for each tweet, which is stored in a database and updated in real time.

[0352] Step 4:

[0353] The server counts the number of offensive messages detected within a certain period of time and compares it with a set threshold. It uses the number of offensive messages recorded in the log file and the sentiment scores stored in the database as input. Specifically, if the number of offensive messages exceeds the threshold, it adjusts the parameters of a generative AI model (such as GPT-3) based on the sentiment data. The output is the adjusted results of the generative model.

[0354] Step 5:

[0355] The server inputs a prompt into the generative AI model to generate a supportive message. The input uses the adjusted generative model and a prompt: "Currently, the user is receiving abusive messages on social media, and their stress level is high. Please generate a supportive message to alleviate this situation." Specifically, the server invokes the generative AI model to generate a message, which is then reviewed by a human auditor. The output is a supportive message that has passed the quality check.

[0356] Step 6:

[0357] The server adds the generated advocacy message to a message queue and distributes the message based on the contributor list. It uses the advocacy message that passed the quality check and the contributor list as input. Specifically, it sends the message to the contributor's device via Firebase Cloud Messaging or an API. The output is the advocacy message delivered to the contributor's device.

[0358] Step 7:

[0359] The terminal receives the advocacy message sent from the server and stores it in a local data store. The advocacy message sent from the server is used as input. After receiving the advocacy message, the terminal displays a pop-up or banner notification to inform the collaborator of the existence of a new message. The output is the advocacy message sent to the collaborator.

[0360] Step 8:

[0361] The user checks the notification on the device and views the content of the defense message. The defense message stored on the device is used as input. The specific operation is to check the message content, press the "Post" button to call the SNS API, and publish the message. The output is the defense message posted to the SNS.

[0362] (Application example 2)

[0363] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0364] In addition to the problem of an increase in abusive messages on social networking services (SNS) and an increasing number of users being victimized, there is also the problem of stress and fatigue among factory workers, which can lead to reduced productivity and accidents. The current situation, where there is a lack of stress management and psychological support, reduces work efficiency and has a negative impact on the health of workers.

[0365] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for detecting offensive messages to the information terminal account, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the collaborator's information terminal, means for the collaborator to post the defense message to a social networking service via the information terminal, means for acquiring video of the work environment and recognizing the emotional state in real time, and means for executing appropriate support messages and actions based on the emotional state. This makes it possible to mitigate the impact of offensive messages while managing the stress of workers in the work environment, improving work efficiency and ensuring safety.

[0366] An "information terminal" is a device used to process information such as social networking services.

[0367] "Offensive messages" are messages with negative content that criticize or insult others.

[0368] "Threshold" means a specific standard, here referring to the volume or frequency of offensive messages.

[0369] A "protective message" is a message intended to provide emotional support and encouragement to a user who has been harmed by an offensive message.

[0370] A "collaborator" is an individual or group responsible for receiving and posting advocacy messages on a social networking service.

[0371] "Emotional state" refers to an individual's psychological state and includes feelings such as stress, anxiety, and happiness.

[0372] A "generative model" refers to an algorithm or template used to tailor the content of messages generated by AI.

[0373] A "natural language processing algorithm" is a computer algorithm that analyzes natural language and understands its meaning and emotions.

[0374] "Work environment" refers to the place where work is done, such as a factory or office.

[0375] "Support messages" are messages that provide appropriate instructions and words of encouragement to reduce fatigue and stress in workers.

[0376] "Real time" refers to a process or activity occurring immediately at the present time.

[0377] "Footage" refers to visual information captured by a camera or other imaging device.

[0378] "Action" refers to the behavior or operation that is performed according to the emotional state.

[0379] This invention is a system that recognizes users' emotional states in real time on social networking services (SNS) and in factory work environments, and generates and distributes appropriate messages. The system includes a server, terminals, cameras installed in the work environment, and an emotion recognition engine.

[0380] System Configuration

[0381] The server monitors social media accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators' devices. It also acquires video data from cameras in the factory and recognizes the emotional state of workers in real time.

[0382] Server Roles

[0383] The server has the ability to detect offensive messages sent to the information terminal account, using a natural language processing algorithm to identify offensive content through sentiment analysis of language.

[0384] If the number of offensive messages exceeds a certain threshold, the server automatically creates a defense message using the generative model, which is then distributed to the collaborators' devices, who then post the message on the SNS.

[0385] The server also analyzes video data acquired from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time, generating appropriate support messages and notifying them if workers become stressed or fatigued.

[0386] Device Role

[0387] The collaborator's device receives the advocacy message sent from the server and notifies the collaborator via a user interface. The collaborator then checks the message and posts it to the SNS.

[0388] The terminals will display support messages and instructions to workers in the factory, encouraging them to take breaks as needed. The terminals will include a message receiving system and a display interface.

[0389] The role of emotion recognition engines

[0390] The emotion recognition engine analyzes video data acquired from the camera and detects worker stress and fatigue in real time. Based on this, an appropriate support message is generated using a generative AI model. The support message is then delivered to the worker via a server.

[0391] Specific examples

[0392] For example, consider a situation where a worker is feeling stressed due to continuous, repetitive work. In this case, the emotion recognition engine detects a high stress level, and the server generates a supportive message such as, "Take a short break. Your efforts are appreciated." This message is then displayed on the worker's device.

[0393] Prompt Sentence Examples

[0394] Examples of prompts to input to a generative AI model include:

[0395] "If the emotion recognition engine detects high stress levels, generate an appropriate encouraging message for the worker, such as, 'Your efforts are appreciated. Take a short break.'"

[0396] In this way, the system not only mitigates the impact of offensive messages, but also manages worker stress in the work environment, improving work efficiency and ensuring safety.

[0397] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0398] Step 1:

[0399] The server activates a means for detecting offensive messages against the information terminal's account. This means acquires timeline data from the social media account and saves and updates it in a database. The input is the social media timeline data, and the output is the updated timeline data saved in the database. Specifically, the server acquires the social media data through an API and saves it in a database for analysis.

[0400] Step 2:

[0401] The server analyzes the acquired data using natural language processing algorithms to detect offensive messages. The input is the saved timeline data, and the output is a list of detected offensive messages. Specifically, it uses a natural language processing library to perform sentiment analysis of the text data and flag offensive messages.

[0402] Step 3:

[0403] The server applies an emotion recognition engine to recognize the user's emotional state when the number of offensive messages exceeds a certain threshold. The input is a list of offensive messages, and the output is the user's emotional state data. Specifically, the server starts the emotion recognition engine and evaluates the user's emotions in real time.

[0404] Step 4:

[0405] The server uses a generative AI model to generate an advocacy message based on the emotional state data. The input is the emotional state data, and the output is the generated advocacy message. Specifically, the generative AI model is used to create a message that is in tune with the user's emotions. An example of a prompt sentence is, "If the emotion recognition engine recognizes that the stress level is high, generate an appropriate encouraging message for the worker. For example, something like, 'Your efforts are appreciated. Please take a short break.'"

[0406] Step 5:

[0407] The server starts a means to distribute the generated advocacy message to the collaborator's terminal. The input is the generated advocacy message, and the output is the message sent to the collaborator's terminal. The specific operation is to distribute the message to the collaborator via an API or a notification system.

[0408] Step 6:

[0409] The collaborator's terminal saves the received advocacy message and displays it on the user interface to notify the collaborator. The input is the advocacy message received from the server, and the output is the message displayed on the user interface. The specific operation is to use the terminal's notification system to notify the collaborator that there is a new message.

[0410] Step 7:

[0411] The collaborator user checks the defense message notified on their device and posts it to the SNS. The input is the received defense message, and the output is the message posted to the SNS. Specifically, when the collaborator presses the post button on their device, the message is published through the SNS API.

[0412] Step 8:

[0413] The server acquires video data from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time. The input is video data from the cameras, and the output is data on the emotional state of the workers. Specifically, the server analyzes the video data, and the emotion recognition engine detects stress and fatigue.

[0414] Step 9:

[0415] The server generates appropriate support messages and actions based on the emotional state data and notifies the worker's device. The input is the emotional state data, and the output is the support message. Specifically, the server uses the generative AI model to create a message encouraging the worker to take a break and displays it on the device.

[0416] Step 10:

[0417] The worker's device displays the received support message and notifies the worker. The input is the generated support message, and the output is the message displayed on the device. Specifically, the message is displayed on the device's display or through the notification system, prompting the worker to take appropriate action.

[0418] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0419] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0420] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0421] [Second embodiment]

[0422] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0423] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0424] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0425] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0426] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0427] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0428] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0429] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0430] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0431] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0432] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0433] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0434] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and a method for implementing the system will be described using a specific example.

[0435] Overall system overview

[0436] This system consists of a server, a terminal, and a user. The server monitors the SNS accounts of service subscribers and detects offensive messages. For detected offensive messages, the server uses a generative model to generate defensive messages and distributes them to the terminals of collaborators. The collaborators then post defensive messages to the SNS via their terminals, thereby mitigating the impact of the offensive messages.

[0437] Server Roles

[0438] 1. Starting the monitoring module

[0439] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[0440] 2. Detecting offensive messages

[0441] The server analyzes the data it receives using a natural language processing algorithm to identify offensive messages. The algorithm performs sentiment analysis of the language and detects content that is deemed offensive. Detected offensive messages are stored in a dedicated log.

[0442] 3. Generating Advocacy Messages

[0443] If the number of offensive messages exceeds a certain threshold, the server uses a generative model to generate defensive messages, which undergo quality checks and are added to a message queue.

[0444] 4. Send messages to collaborators

[0445] The server accesses the list of registered contributors and distributes the generated advocacy messages to the contributors' devices, quickly using APIs and notification systems.

[0446] Device Role

[0447] 1. Receiving a message

[0448] The collaborator's device receives the advocacy message sent from the server, which causes the message to be stored in the device's local data store.

[0449] 2. Submission Preparation and Notification

[0450] The device application notifies the user that a new message has been sent to the user interface, and after confirmation, the user is ready to post the message to the SNS.

[0451] User Roles

[0452] 1. Posting a message of support

[0453] The contributor user checks the advocacy message sent from the server and posts it to the SNS via the application on their device. By pressing the post button, the message is made public using the SNS API.

[0454] Specific examples

[0455] A concrete example is given below. For example, suppose a user receives a large number of offensive messages. In this case, the server detects these messages and generates a supportive message such as, "This user's behavior is admirable. Let's support them together." This message is sent to the collaborator's device, and the collaborator posts it on SNS, mitigating the impact of the offensive messages.

[0456] In this way, the SNS Shield system counters offensive messages in real time, providing a safe environment for users to use SNS.

[0457] The processing flow will be explained below.

[0458] Step 1: Starting the monitoring module

[0459] The server starts a module for monitoring the social media accounts of service subscribers, allowing the server to obtain subscribers' posts and comments in real time.

[0460] Step 2: Getting the Timeline Data

[0461] The server periodically collects subscribers' latest posts and comments via the SNS's API and stores them in a database.

[0462] Step 3: Detecting offensive messages

[0463] The server runs natural language processing algorithms on the collected data to detect offensive messages, specifically identifying messages that contain negative sentiment or offensive content.

[0464] Step 4: Threshold check

[0465] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, proceeds to the next step.

[0466] Step 5: Generating an advocacy message

[0467] The server uses the generative model to generate advocacy messages, such as "This user is doing great work, let's support them."

[0468] Step 6: Check message quality

[0469] The server quality checks the generated advocacy message for correctness, and if incorrect, it is regenerated.

[0470] Step 7: Prepare the message for distribution

[0471] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list, and adds them to the message queue for distribution.

[0472] Step 8: Distribute messages to collaborators

[0473] The server distributes messages to the collaborators' devices. Messages are sent to the collaborators' devices through APIs or notification systems.

[0474] Step 9: Receiving a message

[0475] The terminal receives the advocacy message sent from the server and stores it in a local data store.

[0476] Step 10: User Interface Notification

[0477] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[0478] Step 11: Review your advocacy message

[0479] The user checks the notification on the terminal and confirms the content of the received advocacy message.

[0480] Step 12: Post an Advocacy Message

[0481] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[0482] Step 13: Feedback on submission results

[0483] The terminal notifies the server that the message has been successfully posted, which causes a posting log to be recorded on the server.

[0484] Step 14: Evaluate the effectiveness of the system

[0485] The server evaluates the effectiveness of the system based on the feedback and improves the generative model and algorithms as needed.

[0486] Example 1

[0487] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0488] On modern online platforms, online flame wars caused by offensive messages are becoming more frequent, placing a greater mental burden on users. This problem significantly undermines the environment in which users can use the platform safely. Conventional systems have not established a method for effectively suppressing these offensive messages, and it is particularly difficult to respond in real time. Therefore, there is a need to provide a safe and secure environment for users by quickly detecting offensive messages on online platforms and responding appropriately.

[0489] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0490] In this invention, the server includes a means for monitoring abusive messages to the account of the information processing device, a means for generating a defense message when the abusive messages exceed a certain threshold, and a means for distributing the generated defense message to the information processing device of the collaborator. This makes it possible to detect abusive messages in real time and respond quickly to prevent online flaming and provide a safe and secure environment for users.

[0491] "Information processing device" generally refers to electronic devices capable of data processing, such as computers and smartphones.

[0492] An "account" is data including identification information and authentication information that allows a user to be individually managed on an online platform.

[0493] "Offensive messages" are messages that are intended to hurt, insult, or cause anxiety to others.

[0494] "Generative AI models" refer to algorithms or frameworks for generating text or data using artificial intelligence techniques.

[0495] "Natural language processing algorithms" refer to computational techniques for understanding, generating, and manipulating human language.

[0496] A "threshold" is a value that meets a particular condition or criterion, and when exceeded, triggers a particular action.

[0497] "Advocacy messages" refer to positive messages that are generated to mitigate the impact of aggressive messages and protect the target audience.

[0498] "Online Platform" refers to a website or application that enables users to share information, exchange, or interact over the Internet.

[0499] "Distribution" refers to the sharing of specific information or data with multiple recipients.

[0500] The present invention relates to a system for suppressing flame wars caused by offensive messages on online platforms. The system is composed of an information processing device, a server, and a user. Specific embodiments for carrying out the invention are described below.

[0501] Configuration of information processing device

[0502] The information processing device is an electronic device capable of data processing, such as a computer or smartphone, and functions as a terminal used by collaborators. This information processing device includes a network connection function for receiving messages and a local data store for saving and displaying received messages. The specific implementation uses an SQLite database.

[0503] Server Configuration

[0504] The server consists of the following main modules:

[0505] Monitoring module: Implemented using Python language and Django framework, it monitors the accounts of service subscribers.

[0506] Natural language processing module: Analyzes and detects offensive messages using TensorFlow and PyTorch.

[0507] Generative module: Creates advocacy messages using a generative AI model (e.g., OpenAI GPT-3).

[0508] Notification module: Distributes generated advocacy messages to collaborators' information processing devices using RESTful APIs and the Firebase notification system.

[0509] User Roles

[0510] The collaborating user checks the advocacy message received from the server and posts it to the online platform via an information processing device. The user uses the application on their device to press the post button, which publishes the message through the SNS API (e.g., Twitter API).

[0511] Overview of the invention

[0512] The server monitors the subscriber's account on the online platform and detects offensive messages using a natural language processing module. A generation module generates a defensive message for the detected offensive message. The defensive message is distributed to the collaborator's information processing device, and after the user has reviewed it, the user can post it on the online platform to mitigate the impact of the offensive message.

[0513] Specific examples

[0514] For example, if a user receives an offensive message such as "I completely disagree with your opinion. I think it's wrong," the server detects this message and uses a generative AI model (e.g., GPT-3) to generate a defensive message such as "I think it's okay to have different opinions. Let's respect this user's opinion." This message is then sent to the collaborator's information processing device, and the collaborator posts it on an online platform to mitigate the impact of the offensive message.

[0515] Examples of prompt statements

[0516] Offensive messages:

[0517] "I don't agree with you at all. I think you're wrong."

[0518] Advocacy message:

[0519] "It's good to have different opinions. Let's respect this user's opinion."

[0520] In this way, the system responds to offensive messages in real time, providing a safe and secure environment for users to use online platforms.

[0521] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0522] Step 1:

[0523] The server activates a monitoring module to monitor the service subscriber's account.

[0524] Input: Service subscriber account information

[0525] Specific operation: Uses the Django framework and executes the management command python manage.py runserver.

[0526] Data processing: Send a request to the SNS API to retrieve timeline data.

[0527] Output: Timeline data is saved to a database.

[0528] Step 2:

[0529] The server analyzes the acquired timeline data using a natural language processing algorithm to detect offensive messages.

[0530] Input: Saved timeline data

[0531] Specific operation: Loads a TensorFlow sentiment analysis model and performs analysis on the data.

[0532] Data processing: Processing data with sentiment analysis algorithms to identify offensive messages.

[0533] Output: Offensive messages are saved to a dedicated log file.

[0534] Step 3:

[0535] The server uses a generative AI model to generate a defensive message when an offensive message exceeds a certain threshold.

[0536] Input: offensive message

[0537] Specific behavior: Send an offensive message as a prompt to the GPT-3 API.

[0538] Data processing: The generated advocacy messages are reviewed in a quality check module.

[0539] Output: Advocacy messages that meet the criteria are added to the message queue.

[0540] Step 4:

[0541] The server distributes the advocacy message generated based on the collaborator list to the information processing device.

[0542] Input: Deferred message in message queue

[0543] Specific behavior: Sends notifications using RESTful APIs and the Firebase notification system.

[0544] Data processing: Extract the message from the message queue.

[0545] Output: The message is distributed to the collaborators' information processing devices.

[0546] Step 5:

[0547] The collaborator's terminal receives the advocacy message sent from the server.

[0548] Input: Notification from the server

[0549] Specific operation: Set up a listener to receive notifications from Firebase.

[0550] Data processing: Received messages are saved in a SQLite database.

[0551] Output: Saved advocacy messages are logged to a local data store.

[0552] Step 6:

[0553] The terminal notifies the user of the arrival of a new message and displays it on the user interface, allowing the user to check the message and prepare it for posting.

[0554] Input: Saved advocacy message

[0555] Specific operation: Notifies using the notification manager (Notification Manager for Android, UNUserNotificationCenter for iOS).

[0556] Data processing: Display the message content within the app.

[0557] Output: The user has reviewed the message and is ready to hit the post button.

[0558] Step 7:

[0559] A user posts an advocacy message to an online platform via an application on the terminal.

[0560] Input: Saved advocacy messages and user actions

[0561] Specific operation: By pressing the post button, a POST request is sent to the SNS API (e.g. Twitter API).

[0562] Data processing: Generates request data to be sent to the SNS API.

[0563] Output: The advocacy message is published on an online platform.

[0564] (Application example 1)

[0565] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0566] In recent years, there has been an increase in offensive and abusive messages on social networking services (SNS). These messages can have a negative impact on users' mental health and can lead to a phenomenon known as "flaming." While methods to prevent and mitigate this are needed, effective countermeasures are currently lacking. Furthermore, because manual responses are time-consuming and labor-intensive, real-time automated countermeasures are needed.

[0567] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0568] In this invention, the server includes means for detecting offensive messages to the account of the information terminal, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the information terminal of the collaborator, means for the collaborator to post the defense message to the social networking service via the information terminal, means for monitoring the SNS account in real time, means for notifying the generated defense message, and means for the notified user to post the message with one click after confirmation. This makes it possible to respond quickly and automatically to offensive messages and minimize the impact of flame wars on the SNS.

[0569] An "information terminal" is an electronic device that a user uses to access the Internet.

[0570] An "offensive message" is a message that contains negative or offensive content and may cause psychological harm to a user.

[0571] The "threshold" is a criterion for determining whether an offensive message exceeds a certain standard value.

[0572] "Advocacy messages" are messages that positively evaluate and support users and are generated to counter aggressive messages.

[0573] A "collaborator" is a user whose role is to receive advocacy messages and post them on social media from their own account.

[0574] "Real time" is a time concept in which information is acquired and processed immediately, meaning a nearly simultaneous response.

[0575] "Notifications" are alerts or messages sent to collaborators' information terminals when a specific event occurs.

[0576] "One-click posting" is a feature that allows users to post messages on social media with minimal operations, i.e., a single click or tap.

[0577] A "natural language processing algorithm" is a technology that analyzes human language and understands meaning and emotions.

[0578] A "generative AI model" is an artificial intelligence technology that automatically generates new text based on existing data.

[0579] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and is implemented as follows.

[0580] Overall system configuration

[0581] This system consists of a server, the collaborator's information terminal, and the user who operates it. The server monitors SNS accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators. The information terminal receives defensive messages, and posts them to the SNS after user confirmation.

[0582] Hardware and software configuration

[0583] Hardware:

[0584] Server: a high-performance computer

[0585] Information device: Smartphone (iPhone or Android device)

[0586] software:

[0587] Server-side program:

[0588] Natural Language Processing Algorithms (TensorFlow, Keras, Hugging Face Transformers)

[0589] Message Distribution API

[0590] Information terminal app:

[0591] Programming language: Swift (iOS), Kotlin (Android)

[0592] Database: Firebase Firestore

[0593] Notification system: Firebase Cloud Messaging (FCM)

[0594] Program processing

[0595] Server Processing

[0596] The server monitors social media accounts in real time to obtain timeline data, which it then stores and updates in a database. The server then uses a natural language processing algorithm to detect offensive messages from the obtained data. This algorithm identifies offensive content and negative sentiment.

[0597] If the number of offensive messages exceeds a certain threshold, the server uses a generative AI model to generate a defensive message. The generated message undergoes a quality check and is then sent to the collaborator's device. Notifications are sent using Firebase Cloud Messaging (FCM), allowing collaborators to respond quickly.

[0598] Information terminal processing

[0599] The collaborator's information device receives the advocacy message sent from the server. When the notification arrives, the message is saved in the device's local data store, and the application notifies the user to display the new message on the user interface, ready for the user to check and post to SNS with one click.

[0600] Specific examples

[0601] For example, if a user receives a large number of offensive messages on Twitter, the server detects these messages and generates a supportive message such as, "This user's actions are admirable. Let's support them together!" This message is sent to the collaborator's information terminal, and the collaborator can post this message to Twitter with one click, thereby mitigating the impact of the offensive messages.

[0602] Example prompts to input to a generative AI model:

[0603] A specific user is being attacked on social media. Please generate a positive, supportive message to defuse the situation. Please keep it general and supportive, without including specific usernames or examples of behavior.

[0604] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0605] Step 1:

[0606] The server starts monitoring the social media account in real time, obtains the user's timeline data, and saves and updates it in the database.

[0607] Input: User's social media account information, timeline data

[0608] Output: Timeline data stored in a database

[0609] Specific operation: The server retrieves the user's social media timeline through the API, updates it periodically, and saves this data in a database (e.g., Firebase Firestore).

[0610] Step 2:

[0611] The server analyzes the timeline data using natural language processing algorithms to detect offensive messages.

[0612] Input: Timeline data stored in a database, natural language processing algorithms (e.g., Hugging Face Transformers)

[0613] Output: List of detected offensive messages

[0614] Specific operation: The server uses a library for natural language processing to analyze the timeline data, extracting posts containing offensive messages or negative sentiment and creating a list of them.

[0615] Step 3:

[0616] If the server detects an offensive message that exceeds a certain threshold, it uses a generative AI model to generate a defensive message.

[0617] Input: List of offensive messages, generative AI model

[0618] Output: Generated advocacy message

[0619] Specific operation: The server inputs a prompt sentence to the generative AI model to create a supportive message to defuse an aggressive situation. Based on the prompt sentence, the generative AI model generates an appropriate supportive message.

[0620] Step 4:

[0621] The server distributes the generated advocacy message to the information terminals of the collaborators.

[0622] Input: Generated advocacy message, list of contributors

[0623] Output: Advocacy message received on the collaborator's information terminal

[0624] Specific operation: The server uses Firebase Cloud Messaging to send notifications of the advocacy message to all devices included in the contributor list.

[0625] Step 5:

[0626] The terminal receives the protection message sent from the server and stores it in a local data store.

[0627] Input: Advice message sent from the server

[0628] Output: Advocacy messages saved to the local data store

[0629] Specific behavior: The device receives the notification and saves it to a local data store (e.g., an SQLite database).

[0630] Step 6:

[0631] The device will display a notification prompting the user to acknowledge the advocacy message.

[0632] Input: Advocacy messages stored in the local data store

[0633] Output: User-visible notification message

[0634] What happens: The device application displays a new message notification on the user interface, and the user taps the notification to view the message details.

[0635] Step 7:

[0636] Users can view advocacy messages and post them to social media with one click.

[0637] Input: Advocacy message, SNS account information

[0638] Output: Advocacy messages posted on social media

[0639] Specific operation: When a user presses a button that allows them to post a support message to a social networking site with one click, the device application posts the message using the social networking site API (e.g., Twitter API).

[0640] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0641] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and in particular, an embodiment in which an emotion engine that recognizes the emotions of users is combined will be described.

[0642] Overall system overview

[0643] This system consists of a server, a terminal, and a user, and integrates an emotion engine to improve the effectiveness of advocacy messages. The server monitors the SNS accounts of service subscribers and detects offensive messages. The emotion engine recognizes the user's emotional state in real time and adjusts the generation of advocacy messages based on the results. The generated advocacy messages are distributed to the terminals of collaborators, who then post them on the SNS.

[0644] Server Roles

[0645] 1. Starting the monitoring module

[0646] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[0647] 2. Detecting offensive messages

[0648] The server analyzes the acquired data using a natural language processing algorithm to detect offensive messages. This algorithm performs sentiment analysis of the language and detects content that is deemed offensive. The detected offensive messages are stored in a dedicated log.

[0649] 3. Applying the Emotion Engine

[0650] The server uses an emotion engine to recognize the user's emotional state in real time, thereby determining whether the user is experiencing stress or anxiety.

[0651] 4. Threshold Checking and Generative Model Tuning

[0652] The server determines whether the number of detected offensive messages exceeds a set threshold and adjusts the generation of advocacy messages based on data from the emotion engine, using a generative model to generate appropriate message content according to the user's emotional state.

[0653] 5. Message Creation and Preparation for Distribution

[0654] The server uses the generative model to create an advocacy message that passes the quality check, prepares it for distribution to the collaborators' terminals, and adds it to the message queue for distribution.

[0655] 6. Send messages to collaborators

[0656] The server distributes the generated advocacy messages to the collaborators' terminals via APIs and notification systems based on the list of registered collaborators.

[0657] Device Role

[0658] 1. Receiving a message

[0659] The collaborator's terminal receives the advocacy message sent from the server and stores it in a local data store.

[0660] 2. User interface notifications

[0661] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[0662] User Roles

[0663] 1. Review and post advocacy messages

[0664] The collaborator user checks the notification on their device and confirms the content of the received advocacy message. Then, they post the received advocacy message to the SNS via the device's application. By pressing the post button, the device publishes the message using the SNS API.

[0665] Specific examples

[0666] Here's a concrete example: In a situation where a user is receiving a large number of aggressive messages, the emotion engine detects that the user's stress level is rising. Based on this information, the server generates a more emotionally sensitive supportive message, such as, "Your efforts are being heard by everyone. There is a lot of support for you." This message is sent to the collaborator's device, and the collaborator posts it on social media, effectively mitigating the impact of the aggressive messages.

[0667] In this way, the SNS Shield system applies an emotion engine to provide real-time support messages that correspond to the user's emotional state, providing an environment in which users can use SNS safely.

[0668] The processing flow will be explained below.

[0669] Step 1: Starting the monitoring module

[0670] The server runs a module that monitors subscribers' social media accounts, capturing their posts and comments in real time and storing them in a database. A dedicated database entry is created for each monitored account.

[0671] Step 2: Getting the Timeline Data

[0672] The server periodically collects subscribers' timeline data via SNS API, including posts, comments, replies, likes, etc. This data is stored in a database and prepared for subsequent processing.

[0673] Step 3: Detecting offensive messages

[0674] The server applies natural language processing algorithms to the collected timeline data, which are designed to identify messages containing offensive or negative content, and records the detected messages in a dedicated log.

[0675] Step 4: Applying the Emotion Engine

[0676] The server simultaneously monitors the collected offensive messages and activates an emotion engine that analyzes the user's current emotions based on their timeline data and past messages, and determines their stress and anxiety levels.

[0677] Step 5: Threshold check

[0678] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, the server considers the user's emotional state data obtained from the emotion engine and proceeds to the next step.

[0679] Step 6: Generating Advocacy Messages

[0680] The server uses a generative model to generate appropriate advocacy messages based on data from the emotion engine. For example, if a user's stress level is high, a message such as "This user is doing great things, so let's support them" is generated.

[0681] Step 7: Check message quality

[0682] The server quality-checks the content of generated advocacy messages, regenerating inappropriate messages, and fine-tuning the tone and content of messages based on feedback from the emotion engine.

[0683] Step 8: Preparing the Message for Distribution

[0684] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list. The messages are added to a message queue for distribution.

[0685] Step 9: Distribute messages to collaborators

[0686] The server distributes advocacy messages to registered collaborators' devices using APIs and notification systems. When a notification is received, the collaborator receives the message on their own device.

[0687] Step 10: Receiving a message

[0688] The terminal receives the protection message sent from the server and stores it in local storage, and the user interface displays the new message to notify the user.

[0689] Step 11: User Interface Notification

[0690] The terminal displays the received advocacy message on the user interface and notifies the collaborator that there is a new message. The user checks the contents of the message.

[0691] Step 12: Review your advocacy message

[0692] The user checks the notification on the device and confirms the content of the received advocacy message. If the message content is deemed appropriate, the user prepares to post it.

[0693] Step 13: Post an Advocacy Message

[0694] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[0695] Step 14: Feedback on submission results

[0696] The terminal notifies the server that the message has been successfully posted, and the server saves the posting log and uses it as data to evaluate the effectiveness of the system.

[0697] Step 15: Evaluate the effectiveness of the system

[0698] The server evaluates the effectiveness of the system based on the feedback data and improves the generative model, emotion engine, and natural language processing algorithm as needed.

[0699] Through the above steps, the system utilizes the emotion engine to generate a defense message according to the user's emotional state, and can effectively mitigate the impact of aggressive messages.

[0700] Example 2

[0701] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0702] Flaming on social networking services (SNS) caused by offensive messages places a significant psychological burden on users and has become a social problem. To solve this problem, a system that can detect offensive messages and take appropriate countermeasures is needed, but current technology is insufficient in taking countermeasures based on the user's real-time emotional state. In addition, there is a need to improve the accuracy of detecting offensive messages and the quality of the generated defense messages.

[0703] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0704] In this invention, the server includes a means for detecting offensive messages, a means for generating a defense message when a threshold is exceeded, a means for distributing the generated defense message to collaborators, a means for recognizing the emotional state of the user in real time, and a means for adjusting the content of the defense message based on the emotional state acquired in real time, thereby making it possible to quickly and effectively suppress flame wars caused by offensive messages and reduce the burden on users.

[0705] An "information terminal" is a device used by a user, and includes a smartphone, tablet, or personal computer.

[0706] An "offensive message" is a message that contains malicious or harassing words or content and causes a psychological burden to the user.

[0707] A "means for detecting" is a method or device that uses specific algorithms or techniques to identify and identify offensive messages.

[0708] A "protective message" is a message of support or encouragement that is generated to help a user who has received an offensive message and to reduce the mental burden.

[0709] A "generating means" is a method or device for automatically generating advocacy messages using a specific algorithm or model.

[0710] "Emotional state" refers to a particular psychological state or mood that a user is experiencing, including states such as anxiety, stress, and joy.

[0711] A "real-time recognition means" is a method or device for instantly analyzing and understanding a user's current emotional state.

[0712] An "adjusting means" is a method or device for modifying or optimizing generated messages or other output content based on acquired data or conditions.

[0713] The "distribution means" refers to a method or device for transmitting the generated advocacy message to the collaborator's information terminal and making it available.

[0714] "Allies" are users or groups selected to post advocacy messages on social media.

[0715] A "social networking service" is a platform that enables users to interact and share information with other users online.

[0716] MODE FOR CARRYING OUT THE INVENTION

[0717] The present invention is a system for suppressing flame wars caused by offensive messages on social networking services (SNS). The system is composed of a server, a terminal, and a user, and includes the following elements and means:

[0718] Server Roles

[0719] The server plays a key role and performs the following functions:

[0720] 1. Launching the social media account monitoring module and collecting data

[0721] The server periodically obtains timeline data of service subscribers using SNS APIs, for example, by using Twitter APIs to collect tweet data from users and store and update it in a database.

[0722] 2. Detecting offensive messages

[0723] The server analyzes the collected data using Python natural language processing libraries (such as NLTK and spaCy). It uses specific keywords and sentiment analysis techniques to detect offensive messages. Messages that are deemed offensive are recorded in a dedicated log file.

[0724] 3. Applying the Emotion Engine

[0725] The server uses IBM Watson's sentiment analysis API to extract emotions from users' posts, allowing it to understand in real time the level of stress or anxiety a user is experiencing. Emotion scores are stored in a database and updated in real time.

[0726] 4. Threshold Checking and Adjusting the Advocacy Message Generation Model

[0727] The server evaluates whether the number of detected offensive messages exceeds a pre-defined threshold, and adjusts the parameters of a defense message generation model (e.g., GPT-3) based on the user's emotional state data.

[0728] 5. Advocacy message generation and quality check

[0729] The server inputs a prompt into a generative AI model (such as GPT-3) to generate a supportive message tailored to the user's situation. For example, the prompt might read, "The user is currently receiving aggressive messages on social media and is experiencing high stress levels. Please generate a supportive message to alleviate this situation." The generated message passes a quality check and is then prepared for distribution.

[0730] 6. Preparing and implementing advocacy message distribution

[0731] The server adds the verified advocacy message to a message queue and sends it to the registered collaborators through Firebase Cloud Messaging or APIs, for example, by calling the "SendMessage" API to deliver the message to the collaborators' devices.

[0732] Device Role

[0733] The device provides an interface for collaborators to display the advocacy messages they receive and post them to social media.

[0734] 1. Receiving and notifying advocacy messages

[0735] The collaborator's device receives the message sent by the server and stores it in a local data store. The device uses a notification system to notify the collaborator that a new message has been received, for example by displaying a pop-up or banner notification.

[0736] 2. Review and post advocacy messages

[0737] The collaborator user checks the notification on their device. After checking the content of the advocacy message, they press the "Post" button to call the SNS API and post the message to SNS. For example, we will post the message using the "POST statuses / update" endpoint.

[0738] User Roles

[0739] When users receive an offensive message on SNS, they can receive support through the system, which reduces the psychological burden and allows them to continue using SNS with peace of mind.

[0740] Specific examples

[0741] For example, consider a situation where User A is receiving a large number of aggressive messages on social media. Based on this information, the server uses its emotion engine to detect that User A's stress level is rising. Based on this data, it generates a supportive message such as, "Your efforts are being noticed by everyone. There are many voices of support for you." This message is sent to the device of Collaborator B, who then posts it on social media to support User A.

[0742] Prompt Sentence Examples

[0743] "Currently, users are receiving aggressive messages on social media and their stress levels are high. Please generate advocacy messages to ease this situation."

[0744] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0745] Step 1:

[0746] The server starts the SNS account monitoring module and periodically calls the SNS API of the service subscriber. As input, it receives the SNS API authentication information and user ID. Specifically, it retrieves the latest 20 tweets using the Twitter API's "GET statuses / user_timeline" endpoint. The output is the retrieved tweet data, which is then stored and updated in the database.

[0747] Step 2:

[0748] The server analyzes the tweet data stored in the database using natural language processing algorithms. The stored tweet data is used as input. Specifically, text analysis is performed using Python's NLTK and spaCy. To detect offensive messages, specific keywords and sentiment scores are calculated. The output is a list of messages deemed offensive, which is recorded in a dedicated log file.

[0749] Step 3:

[0750] The server calls IBM Watson's sentiment analysis API to analyze the user's emotional state. It uses the analyzed tweet data as input. Specifically, it sends the text of each tweet to the sentiment analysis API and obtains a sentiment score. The output is a sentiment score for each tweet, which is stored in a database and updated in real time.

[0751] Step 4:

[0752] The server counts the number of offensive messages detected within a certain period of time and compares it with a set threshold. It uses the number of offensive messages recorded in the log file and the sentiment scores stored in the database as input. Specifically, if the number of offensive messages exceeds the threshold, it adjusts the parameters of a generative AI model (such as GPT-3) based on the sentiment data. The output is the adjusted results of the generative model.

[0753] Step 5:

[0754] The server inputs a prompt into the generative AI model to generate a supportive message. The input uses the adjusted generative model and a prompt: "Currently, the user is receiving abusive messages on social media, and their stress level is high. Please generate a supportive message to alleviate this situation." Specifically, the server invokes the generative AI model to generate a message, which is then reviewed by a human auditor. The output is a supportive message that has passed the quality check.

[0755] Step 6:

[0756] The server adds the generated advocacy message to a message queue and distributes the message based on the contributor list. It uses the advocacy message that passed the quality check and the contributor list as input. Specifically, it sends the message to the contributor's device via Firebase Cloud Messaging or an API. The output is the advocacy message delivered to the contributor's device.

[0757] Step 7:

[0758] The terminal receives the advocacy message sent from the server and stores it in a local data store. The advocacy message sent from the server is used as input. After receiving the advocacy message, the terminal displays a pop-up or banner notification to inform the collaborator of the existence of a new message. The output is the advocacy message sent to the collaborator.

[0759] Step 8:

[0760] The user checks the notification on the device and views the content of the defense message. The defense message stored on the device is used as input. The specific operation is to check the message content, press the "Post" button to call the SNS API, and publish the message. The output is the defense message posted to the SNS.

[0761] (Application example 2)

[0762] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0763] In addition to the problem of an increase in abusive messages on social networking services (SNS) and an increasing number of users being victimized, there is also the problem of stress and fatigue among factory workers, which can lead to reduced productivity and accidents. The current situation, where there is a lack of stress management and psychological support, reduces work efficiency and has a negative impact on the health of workers.

[0764] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for detecting offensive messages to the information terminal account, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the collaborator's information terminal, means for the collaborator to post the defense message to a social networking service via the information terminal, means for acquiring video of the work environment and recognizing the emotional state in real time, and means for executing appropriate support messages and actions based on the emotional state. This makes it possible to mitigate the impact of offensive messages while managing the stress of workers in the work environment, improving work efficiency and ensuring safety.

[0765] An "information terminal" is a device used to process information such as social networking services.

[0766] "Offensive messages" are messages with negative content that criticize or insult others.

[0767] "Threshold" means a specific standard, here referring to the volume or frequency of offensive messages.

[0768] A "protective message" is a message intended to provide emotional support and encouragement to a user who has been harmed by an offensive message.

[0769] A "collaborator" is an individual or group responsible for receiving and posting advocacy messages on a social networking service.

[0770] "Emotional state" refers to an individual's psychological state and includes feelings such as stress, anxiety, and happiness.

[0771] A "generative model" refers to an algorithm or template used to tailor the content of messages generated by AI.

[0772] A "natural language processing algorithm" is a computer algorithm that analyzes natural language and understands its meaning and emotions.

[0773] "Work environment" refers to the place where work is done, such as a factory or office.

[0774] "Support messages" are messages that provide appropriate instructions and words of encouragement to reduce fatigue and stress in workers.

[0775] "Real time" refers to a process or activity occurring immediately at the present time.

[0776] "Footage" refers to visual information captured by a camera or other imaging device.

[0777] "Action" refers to the behavior or operation that is performed according to the emotional state.

[0778] This invention is a system that recognizes users' emotional states in real time on social networking services (SNS) and in factory work environments, and generates and distributes appropriate messages. The system includes a server, terminals, cameras installed in the work environment, and an emotion recognition engine.

[0779] System Configuration

[0780] The server monitors social media accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators' devices. It also acquires video data from cameras in the factory and recognizes the emotional state of workers in real time.

[0781] Server Roles

[0782] The server has the ability to detect offensive messages sent to the information terminal account, using a natural language processing algorithm to identify offensive content through sentiment analysis of language.

[0783] If the number of offensive messages exceeds a certain threshold, the server automatically creates a defense message using the generative model, which is then distributed to the collaborators' devices, who then post the message on the SNS.

[0784] The server also analyzes video data acquired from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time, generating appropriate support messages and notifying them if workers become stressed or fatigued.

[0785] Device Role

[0786] The collaborator's device receives the advocacy message sent from the server and notifies the collaborator via a user interface. The collaborator then checks the message and posts it to the SNS.

[0787] The terminals will display support messages and instructions to workers in the factory, encouraging them to take breaks as needed. The terminals will include a message receiving system and a display interface.

[0788] The role of emotion recognition engines

[0789] The emotion recognition engine analyzes video data acquired from the camera and detects worker stress and fatigue in real time. Based on this, an appropriate support message is generated using a generative AI model. The support message is then delivered to the worker via a server.

[0790] Specific examples

[0791] For example, consider a situation where a worker is feeling stressed due to continuous, repetitive work. In this case, the emotion recognition engine detects a high stress level, and the server generates a supportive message such as, "Take a short break. Your efforts are appreciated." This message is then displayed on the worker's device.

[0792] Prompt Sentence Examples

[0793] Examples of prompts to input to a generative AI model include:

[0794] "If the emotion recognition engine detects high stress levels, generate an appropriate encouraging message for the worker, such as, 'Your efforts are appreciated. Take a short break.'"

[0795] In this way, the system not only mitigates the impact of offensive messages, but also manages worker stress in the work environment, improving work efficiency and ensuring safety.

[0796] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0797] Step 1:

[0798] The server activates a means for detecting offensive messages against the information terminal's account. This means acquires timeline data from the social media account and saves and updates it in a database. The input is the social media timeline data, and the output is the updated timeline data saved in the database. Specifically, the server acquires the social media data through an API and saves it in a database for analysis.

[0799] Step 2:

[0800] The server analyzes the acquired data using natural language processing algorithms to detect offensive messages. The input is the saved timeline data, and the output is a list of detected offensive messages. Specifically, it uses a natural language processing library to perform sentiment analysis of the text data and flag offensive messages.

[0801] Step 3:

[0802] The server applies an emotion recognition engine to recognize the user's emotional state when the number of offensive messages exceeds a certain threshold. The input is a list of offensive messages, and the output is the user's emotional state data. Specifically, the server starts the emotion recognition engine and evaluates the user's emotions in real time.

[0803] Step 4:

[0804] The server uses a generative AI model to generate an advocacy message based on the emotional state data. The input is the emotional state data, and the output is the generated advocacy message. Specifically, the generative AI model is used to create a message that is in tune with the user's emotions. An example of a prompt sentence is, "If the emotion recognition engine recognizes that the stress level is high, generate an appropriate encouraging message for the worker. For example, something like, 'Your efforts are appreciated. Please take a short break.'"

[0805] Step 5:

[0806] The server starts a means to distribute the generated advocacy message to the collaborator's terminal. The input is the generated advocacy message, and the output is the message sent to the collaborator's terminal. The specific operation is to distribute the message to the collaborator via an API or a notification system.

[0807] Step 6:

[0808] The collaborator's terminal saves the received advocacy message and displays it on the user interface to notify the collaborator. The input is the advocacy message received from the server, and the output is the message displayed on the user interface. The specific operation is to use the terminal's notification system to notify the collaborator that there is a new message.

[0809] Step 7:

[0810] The collaborator user checks the defense message notified on their device and posts it to the SNS. The input is the received defense message, and the output is the message posted to the SNS. Specifically, when the collaborator presses the post button on their device, the message is published through the SNS API.

[0811] Step 8:

[0812] The server acquires video data from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time. The input is video data from the cameras, and the output is data on the emotional state of the workers. Specifically, the server analyzes the video data, and the emotion recognition engine detects stress and fatigue.

[0813] Step 9:

[0814] The server generates appropriate support messages and actions based on the emotional state data and notifies the worker's device. The input is the emotional state data, and the output is the support message. Specifically, the server uses the generative AI model to create a message encouraging the worker to take a break and displays it on the device.

[0815] Step 10:

[0816] The worker's device displays the received support message and notifies the worker. The input is the generated support message, and the output is the message displayed on the device. Specifically, the message is displayed on the device's display or through the notification system, prompting the worker to take appropriate action.

[0817] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0818] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0819] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0820] [Third embodiment]

[0821] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0822] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0823] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0824] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0825] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0826] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0827] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0828] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0829] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0830] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0831] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0832] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0833] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and a method for implementing the system will be described using a specific example.

[0834] Overall system overview

[0835] This system consists of a server, a terminal, and a user. The server monitors the SNS accounts of service subscribers and detects offensive messages. For detected offensive messages, the server uses a generative model to generate defensive messages and distributes them to the terminals of collaborators. The collaborators then post defensive messages to the SNS via their terminals, thereby mitigating the impact of the offensive messages.

[0836] Server Roles

[0837] 1. Starting the monitoring module

[0838] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[0839] 2. Detecting offensive messages

[0840] The server analyzes the data it receives using a natural language processing algorithm to identify offensive messages. The algorithm performs sentiment analysis of the language and detects content that is deemed offensive. Detected offensive messages are stored in a dedicated log.

[0841] 3. Generating Advocacy Messages

[0842] If the number of offensive messages exceeds a certain threshold, the server uses a generative model to generate defensive messages, which undergo quality checks and are added to a message queue.

[0843] 4. Send messages to collaborators

[0844] The server accesses the list of registered contributors and distributes the generated advocacy messages to the contributors' devices, quickly using APIs and notification systems.

[0845] Device Role

[0846] 1. Receiving a message

[0847] The collaborator's device receives the advocacy message sent from the server, which causes the message to be stored in the device's local data store.

[0848] 2. Submission Preparation and Notification

[0849] The device application notifies the user that a new message has been sent to the user interface, and after confirmation, the user is ready to post the message to the SNS.

[0850] User Roles

[0851] 1. Posting a message of support

[0852] The contributor user checks the advocacy message sent from the server and posts it to the SNS via the application on their device. By pressing the post button, the message is made public using the SNS API.

[0853] Specific examples

[0854] A concrete example is given below. For example, suppose a user receives a large number of offensive messages. In this case, the server detects these messages and generates a supportive message such as, "This user's behavior is admirable. Let's support them together." This message is sent to the collaborator's device, and the collaborator posts it on SNS, mitigating the impact of the offensive messages.

[0855] In this way, the SNS Shield system counters offensive messages in real time, providing a safe environment for users to use SNS.

[0856] The processing flow will be explained below.

[0857] Step 1: Starting the monitoring module

[0858] The server starts a module for monitoring the social media accounts of service subscribers, allowing the server to obtain subscribers' posts and comments in real time.

[0859] Step 2: Getting the Timeline Data

[0860] The server periodically collects subscribers' latest posts and comments via the SNS's API and stores them in a database.

[0861] Step 3: Detecting offensive messages

[0862] The server runs natural language processing algorithms on the collected data to detect offensive messages, specifically identifying messages that contain negative sentiment or offensive content.

[0863] Step 4: Threshold check

[0864] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, proceeds to the next step.

[0865] Step 5: Generating an advocacy message

[0866] The server uses the generative model to generate advocacy messages, such as "This user is doing great work, let's support them."

[0867] Step 6: Check message quality

[0868] The server quality checks the generated advocacy message for correctness, and if incorrect, it is regenerated.

[0869] Step 7: Prepare the message for distribution

[0870] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list, and adds them to the message queue for distribution.

[0871] Step 8: Distribute messages to collaborators

[0872] The server distributes messages to the collaborators' devices. Messages are sent to the collaborators' devices through APIs or notification systems.

[0873] Step 9: Receiving a message

[0874] The terminal receives the advocacy message sent from the server and stores it in a local data store.

[0875] Step 10: User Interface Notification

[0876] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[0877] Step 11: Review your advocacy message

[0878] The user checks the notification on the terminal and confirms the content of the received advocacy message.

[0879] Step 12: Post an Advocacy Message

[0880] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[0881] Step 13: Feedback on submission results

[0882] The terminal notifies the server that the message has been successfully posted, which causes a posting log to be recorded on the server.

[0883] Step 14: Evaluate the effectiveness of the system

[0884] The server evaluates the effectiveness of the system based on the feedback and improves the generative model and algorithms as needed.

[0885] Example 1

[0886] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0887] On modern online platforms, online flame wars caused by offensive messages are becoming more frequent, placing a greater mental burden on users. This problem significantly undermines the environment in which users can use the platform safely. Conventional systems have not established a method for effectively suppressing these offensive messages, and it is particularly difficult to respond in real time. Therefore, there is a need to provide a safe and secure environment for users by quickly detecting offensive messages on online platforms and responding appropriately.

[0888] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0889] In this invention, the server includes a means for monitoring abusive messages to the account of the information processing device, a means for generating a defense message when the abusive messages exceed a certain threshold, and a means for distributing the generated defense message to the information processing device of the collaborator. This makes it possible to detect abusive messages in real time and respond quickly to prevent online flaming and provide a safe and secure environment for users.

[0890] "Information processing device" generally refers to electronic devices capable of data processing, such as computers and smartphones.

[0891] An "account" is data including identification information and authentication information that allows a user to be individually managed on an online platform.

[0892] "Offensive messages" are messages that are intended to hurt, insult, or cause anxiety to others.

[0893] "Generative AI models" refer to algorithms or frameworks for generating text or data using artificial intelligence techniques.

[0894] "Natural language processing algorithms" refer to computational techniques for understanding, generating, and manipulating human language.

[0895] A "threshold" is a value that meets a particular condition or criterion, and when exceeded, triggers a particular action.

[0896] "Advocacy messages" refer to positive messages that are generated to mitigate the impact of aggressive messages and protect the target audience.

[0897] "Online Platform" refers to a website or application that enables users to share information, exchange, or interact over the Internet.

[0898] "Distribution" refers to the sharing of specific information or data with multiple recipients.

[0899] The present invention relates to a system for suppressing flame wars caused by offensive messages on online platforms. The system is composed of an information processing device, a server, and a user. Specific embodiments for carrying out the invention are described below.

[0900] Configuration of information processing device

[0901] The information processing device is an electronic device capable of data processing, such as a computer or smartphone, and functions as a terminal used by collaborators. This information processing device includes a network connection function for receiving messages and a local data store for saving and displaying received messages. The specific implementation uses an SQLite database.

[0902] Server Configuration

[0903] The server consists of the following main modules:

[0904] Monitoring module: Implemented using Python language and Django framework, it monitors the accounts of service subscribers.

[0905] Natural language processing module: Analyzes and detects offensive messages using TensorFlow and PyTorch.

[0906] Generative module: Creates advocacy messages using a generative AI model (e.g., OpenAI GPT-3).

[0907] Notification module: Distributes generated advocacy messages to collaborators' information processing devices using RESTful APIs and the Firebase notification system.

[0908] User Roles

[0909] The collaborating user checks the advocacy message received from the server and posts it to the online platform via an information processing device. The user uses the application on their device to press the post button, which publishes the message through the SNS API (e.g., Twitter API).

[0910] Overview of the invention

[0911] The server monitors the subscriber's account on the online platform and detects offensive messages using a natural language processing module. A generation module generates a defensive message for the detected offensive message. The defensive message is distributed to the collaborator's information processing device, and after the user has reviewed it, the user can post it on the online platform to mitigate the impact of the offensive message.

[0912] Specific examples

[0913] For example, if a user receives an offensive message such as "I completely disagree with your opinion. I think it's wrong," the server detects this message and uses a generative AI model (e.g., GPT-3) to generate a defensive message such as "I think it's okay to have different opinions. Let's respect this user's opinion." This message is then sent to the collaborator's information processing device, and the collaborator posts it on an online platform to mitigate the impact of the offensive message.

[0914] Examples of prompt statements

[0915] Offensive messages:

[0916] "I don't agree with you at all. I think you're wrong."

[0917] Advocacy message:

[0918] "It's good to have different opinions. Let's respect this user's opinion."

[0919] In this way, the system responds to offensive messages in real time, providing a safe and secure environment for users to use online platforms.

[0920] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0921] Step 1:

[0922] The server activates a monitoring module to monitor the service subscriber's account.

[0923] Input: Service subscriber account information

[0924] Specific operation: Uses the Django framework and executes the management command python manage.py runserver.

[0925] Data processing: Send a request to the SNS API to retrieve timeline data.

[0926] Output: Timeline data is saved to a database.

[0927] Step 2:

[0928] The server analyzes the acquired timeline data using a natural language processing algorithm to detect offensive messages.

[0929] Input: Saved timeline data

[0930] Specific operation: Loads a TensorFlow sentiment analysis model and performs analysis on the data.

[0931] Data processing: Processing data with sentiment analysis algorithms to identify offensive messages.

[0932] Output: Offensive messages are saved to a dedicated log file.

[0933] Step 3:

[0934] The server uses a generative AI model to generate a defensive message when an offensive message exceeds a certain threshold.

[0935] Input: offensive message

[0936] Specific behavior: Send an offensive message as a prompt to the GPT-3 API.

[0937] Data processing: The generated advocacy messages are reviewed in a quality check module.

[0938] Output: Advocacy messages that meet the criteria are added to the message queue.

[0939] Step 4:

[0940] The server distributes the advocacy message generated based on the collaborator list to the information processing device.

[0941] Input: Deferred message in message queue

[0942] Specific behavior: Sends notifications using RESTful APIs and the Firebase notification system.

[0943] Data processing: Extract the message from the message queue.

[0944] Output: The message is distributed to the collaborators' information processing devices.

[0945] Step 5:

[0946] The collaborator's terminal receives the advocacy message sent from the server.

[0947] Input: Notification from the server

[0948] Specific operation: Set up a listener to receive notifications from Firebase.

[0949] Data processing: Received messages are saved in a SQLite database.

[0950] Output: Saved advocacy messages are logged to a local data store.

[0951] Step 6:

[0952] The terminal notifies the user of the arrival of a new message and displays it on the user interface, allowing the user to check the message and prepare it for posting.

[0953] Input: Saved advocacy message

[0954] Specific operation: Notifies using the notification manager (Notification Manager for Android, UNUserNotificationCenter for iOS).

[0955] Data processing: Display the message content within the app.

[0956] Output: The user has reviewed the message and is ready to hit the post button.

[0957] Step 7:

[0958] A user posts an advocacy message to an online platform via an application on the terminal.

[0959] Input: Saved advocacy messages and user actions

[0960] Specific operation: By pressing the post button, a POST request is sent to the SNS API (e.g. Twitter API).

[0961] Data processing: Generates request data to be sent to the SNS API.

[0962] Output: The advocacy message is published on an online platform.

[0963] (Application example 1)

[0964] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0965] In recent years, there has been an increase in offensive and abusive messages on social networking services (SNS). These messages can have a negative impact on users' mental health and can lead to a phenomenon known as "flaming." While methods to prevent and mitigate this are needed, effective countermeasures are currently lacking. Furthermore, because manual responses are time-consuming and labor-intensive, real-time automated countermeasures are needed.

[0966] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0967] In this invention, the server includes means for detecting offensive messages to the account of the information terminal, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the information terminal of the collaborator, means for the collaborator to post the defense message to the social networking service via the information terminal, means for monitoring the SNS account in real time, means for notifying the generated defense message, and means for the notified user to post the message with one click after confirmation. This makes it possible to respond quickly and automatically to offensive messages and minimize the impact of flame wars on the SNS.

[0968] An "information terminal" is an electronic device that a user uses to access the Internet.

[0969] An "offensive message" is a message that contains negative or offensive content and may cause psychological harm to a user.

[0970] The "threshold" is a criterion for determining whether an offensive message exceeds a certain standard value.

[0971] "Advocacy messages" are messages that positively evaluate and support users and are generated to counter aggressive messages.

[0972] A "collaborator" is a user whose role is to receive advocacy messages and post them on social media from their own account.

[0973] "Real time" is a time concept in which information is acquired and processed immediately, meaning a nearly simultaneous response.

[0974] "Notifications" are alerts or messages sent to collaborators' information terminals when a specific event occurs.

[0975] "One-click posting" is a feature that allows users to post messages on social media with minimal operations, i.e., a single click or tap.

[0976] A "natural language processing algorithm" is a technology that analyzes human language and understands meaning and emotions.

[0977] A "generative AI model" is an artificial intelligence technology that automatically generates new text based on existing data.

[0978] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and is implemented as follows.

[0979] Overall system configuration

[0980] This system consists of a server, the collaborator's information terminal, and the user who operates it. The server monitors SNS accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators. The information terminal receives defensive messages, and posts them to the SNS after user confirmation.

[0981] Hardware and software configuration

[0982] Hardware:

[0983] Server: a high-performance computer

[0984] Information device: Smartphone (iPhone or Android device)

[0985] software:

[0986] Server-side program:

[0987] Natural Language Processing Algorithms (TensorFlow, Keras, Hugging Face Transformers)

[0988] Message Distribution API

[0989] Information terminal app:

[0990] Programming language: Swift (iOS), Kotlin (Android)

[0991] Database: Firebase Firestore

[0992] Notification system: Firebase Cloud Messaging (FCM)

[0993] Program processing

[0994] Server Processing

[0995] The server monitors social media accounts in real time to obtain timeline data, which it then stores and updates in a database. The server then uses a natural language processing algorithm to detect offensive messages from the obtained data. This algorithm identifies offensive content and negative sentiment.

[0996] If the number of offensive messages exceeds a certain threshold, the server uses a generative AI model to generate a defensive message. The generated message undergoes a quality check and is then sent to the collaborator's device. Notifications are sent using Firebase Cloud Messaging (FCM), allowing collaborators to respond quickly.

[0997] Information terminal processing

[0998] The collaborator's information device receives the advocacy message sent from the server. When the notification arrives, the message is saved in the device's local data store, and the application notifies the user to display the new message on the user interface, ready for the user to check and post to SNS with one click.

[0999] Specific examples

[1000] For example, if a user receives a large number of offensive messages on Twitter, the server detects these messages and generates a supportive message such as, "This user's actions are admirable. Let's support them together!" This message is sent to the collaborator's information terminal, and the collaborator can post this message to Twitter with one click, thereby mitigating the impact of the offensive messages.

[1001] Example prompts to input to a generative AI model:

[1002] A specific user is being attacked on social media. Please generate a positive, supportive message to defuse the situation. Please keep it general and supportive, without including specific usernames or examples of behavior.

[1003] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1004] Step 1:

[1005] The server starts monitoring the social media account in real time, obtains the user's timeline data, and saves and updates it in the database.

[1006] Input: User's social media account information, timeline data

[1007] Output: Timeline data stored in a database

[1008] Specific operation: The server retrieves the user's social media timeline through the API, updates it periodically, and saves this data in a database (e.g., Firebase Firestore).

[1009] Step 2:

[1010] The server analyzes the timeline data using natural language processing algorithms to detect offensive messages.

[1011] Input: Timeline data stored in a database, natural language processing algorithms (e.g., Hugging Face Transformers)

[1012] Output: List of detected offensive messages

[1013] Specific operation: The server uses a library for natural language processing to analyze the timeline data, extracting posts containing offensive messages or negative sentiment and creating a list of them.

[1014] Step 3:

[1015] If the server detects an offensive message that exceeds a certain threshold, it uses a generative AI model to generate a defensive message.

[1016] Input: List of offensive messages, generative AI model

[1017] Output: Generated advocacy message

[1018] Specific operation: The server inputs a prompt sentence to the generative AI model to create a supportive message to defuse an aggressive situation. Based on the prompt sentence, the generative AI model generates an appropriate supportive message.

[1019] Step 4:

[1020] The server distributes the generated advocacy message to the information terminals of the collaborators.

[1021] Input: Generated advocacy message, list of contributors

[1022] Output: Advocacy message received on the collaborator's information terminal

[1023] Specific operation: The server uses Firebase Cloud Messaging to send notifications of the advocacy message to all devices included in the contributor list.

[1024] Step 5:

[1025] The terminal receives the protection message sent from the server and stores it in a local data store.

[1026] Input: Advice message sent from the server

[1027] Output: Advocacy messages saved to the local data store

[1028] Specific behavior: The device receives the notification and saves it to a local data store (e.g., an SQLite database).

[1029] Step 6:

[1030] The device will display a notification prompting the user to acknowledge the advocacy message.

[1031] Input: Advocacy messages stored in the local data store

[1032] Output: User-visible notification message

[1033] What happens: The device application displays a new message notification on the user interface, and the user taps the notification to view the message details.

[1034] Step 7:

[1035] Users can view advocacy messages and post them to social media with one click.

[1036] Input: Advocacy message, SNS account information

[1037] Output: Advocacy messages posted on social media

[1038] Specific operation: When a user presses a button that allows them to post a support message to a social networking site with one click, the device application posts the message using the social networking site API (e.g., Twitter API).

[1039] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1040] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and in particular, an embodiment in which an emotion engine that recognizes the emotions of users is combined will be described.

[1041] Overall system overview

[1042] This system consists of a server, a terminal, and a user, and integrates an emotion engine to improve the effectiveness of advocacy messages. The server monitors the SNS accounts of service subscribers and detects offensive messages. The emotion engine recognizes the user's emotional state in real time and adjusts the generation of advocacy messages based on the results. The generated advocacy messages are distributed to the terminals of collaborators, who then post them on the SNS.

[1043] Server Roles

[1044] 1. Starting the monitoring module

[1045] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[1046] 2. Detecting offensive messages

[1047] The server analyzes the acquired data using a natural language processing algorithm to detect offensive messages. This algorithm performs sentiment analysis of the language and detects content that is deemed offensive. The detected offensive messages are stored in a dedicated log.

[1048] 3. Applying the Emotion Engine

[1049] The server uses an emotion engine to recognize the user's emotional state in real time, thereby determining whether the user is experiencing stress or anxiety.

[1050] 4. Threshold Checking and Generative Model Tuning

[1051] The server determines whether the number of detected offensive messages exceeds a set threshold and adjusts the generation of advocacy messages based on data from the emotion engine, using a generative model to generate appropriate message content according to the user's emotional state.

[1052] 5. Message Creation and Preparation for Distribution

[1053] The server uses the generative model to create an advocacy message that passes the quality check, prepares it for distribution to the collaborators' terminals, and adds it to the message queue for distribution.

[1054] 6. Send messages to collaborators

[1055] The server distributes the generated advocacy messages to the collaborators' terminals via APIs and notification systems based on the list of registered collaborators.

[1056] Device Role

[1057] 1. Receiving a message

[1058] The collaborator's terminal receives the advocacy message sent from the server and stores it in a local data store.

[1059] 2. User interface notifications

[1060] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[1061] User Roles

[1062] 1. Review and post advocacy messages

[1063] The collaborator user checks the notification on their device and confirms the content of the received advocacy message. Then, they post the received advocacy message to the SNS via the device's application. By pressing the post button, the device publishes the message using the SNS API.

[1064] Specific examples

[1065] Here's a concrete example: In a situation where a user is receiving a large number of aggressive messages, the emotion engine detects that the user's stress level is rising. Based on this information, the server generates a more emotionally sensitive supportive message, such as, "Your efforts are being heard by everyone. There is a lot of support for you." This message is sent to the collaborator's device, and the collaborator posts it on social media, effectively mitigating the impact of the aggressive messages.

[1066] In this way, the SNS Shield system applies an emotion engine to provide real-time support messages that correspond to the user's emotional state, providing an environment in which users can use SNS safely.

[1067] The processing flow will be explained below.

[1068] Step 1: Starting the monitoring module

[1069] The server runs a module that monitors subscribers' social media accounts, capturing their posts and comments in real time and storing them in a database. A dedicated database entry is created for each monitored account.

[1070] Step 2: Getting the Timeline Data

[1071] The server periodically collects subscribers' timeline data via SNS API, including posts, comments, replies, likes, etc. This data is stored in a database and prepared for subsequent processing.

[1072] Step 3: Detecting offensive messages

[1073] The server applies natural language processing algorithms to the collected timeline data, which are designed to identify messages containing offensive or negative content, and records the detected messages in a dedicated log.

[1074] Step 4: Applying the Emotion Engine

[1075] The server simultaneously monitors the collected offensive messages and activates an emotion engine that analyzes the user's current emotions based on their timeline data and past messages, and determines their stress and anxiety levels.

[1076] Step 5: Threshold check

[1077] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, the server considers the user's emotional state data obtained from the emotion engine and proceeds to the next step.

[1078] Step 6: Generating Advocacy Messages

[1079] The server uses a generative model to generate appropriate advocacy messages based on data from the emotion engine. For example, if a user's stress level is high, a message such as "This user is doing great things, so let's support them" is generated.

[1080] Step 7: Check message quality

[1081] The server quality-checks the content of generated advocacy messages, regenerating inappropriate messages, and fine-tuning the tone and content of messages based on feedback from the emotion engine.

[1082] Step 8: Preparing the Message for Distribution

[1083] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list. The messages are added to a message queue for distribution.

[1084] Step 9: Distribute messages to collaborators

[1085] The server distributes advocacy messages to registered collaborators' devices using APIs and notification systems. When a notification is received, the collaborator receives the message on their own device.

[1086] Step 10: Receiving a message

[1087] The terminal receives the protection message sent from the server and stores it in local storage, and the user interface displays the new message to notify the user.

[1088] Step 11: User Interface Notification

[1089] The terminal displays the received advocacy message on the user interface and notifies the collaborator that there is a new message. The user checks the contents of the message.

[1090] Step 12: Review your advocacy message

[1091] The user checks the notification on the device and confirms the content of the received advocacy message. If the message content is deemed appropriate, the user prepares to post it.

[1092] Step 13: Post an Advocacy Message

[1093] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[1094] Step 14: Feedback on submission results

[1095] The terminal notifies the server that the message has been successfully posted, and the server saves the posting log and uses it as data to evaluate the effectiveness of the system.

[1096] Step 15: Evaluate the effectiveness of the system

[1097] The server evaluates the effectiveness of the system based on the feedback data and improves the generative model, emotion engine, and natural language processing algorithm as needed.

[1098] Through the above steps, the system utilizes the emotion engine to generate a defense message according to the user's emotional state, and can effectively mitigate the impact of aggressive messages.

[1099] Example 2

[1100] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1101] Flaming on social networking services (SNS) caused by offensive messages places a significant psychological burden on users and has become a social problem. To solve this problem, a system that can detect offensive messages and take appropriate countermeasures is needed, but current technology is insufficient in taking countermeasures based on the user's real-time emotional state. In addition, there is a need to improve the accuracy of detecting offensive messages and the quality of the generated defense messages.

[1102] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1103] In this invention, the server includes a means for detecting offensive messages, a means for generating a defense message when a threshold is exceeded, a means for distributing the generated defense message to collaborators, a means for recognizing the emotional state of the user in real time, and a means for adjusting the content of the defense message based on the emotional state acquired in real time, thereby making it possible to quickly and effectively suppress flame wars caused by offensive messages and reduce the burden on users.

[1104] An "information terminal" is a device used by a user, and includes a smartphone, tablet, or personal computer.

[1105] An "offensive message" is a message that contains malicious or harassing words or content and causes a psychological burden to the user.

[1106] A "means for detecting" is a method or device that uses specific algorithms or techniques to identify and identify offensive messages.

[1107] A "protective message" is a message of support or encouragement that is generated to help a user who has received an offensive message and to reduce the mental burden.

[1108] A "generating means" is a method or device for automatically generating advocacy messages using a specific algorithm or model.

[1109] "Emotional state" refers to a particular psychological state or mood that a user is experiencing, including states such as anxiety, stress, and joy.

[1110] A "real-time recognition means" is a method or device for instantly analyzing and understanding a user's current emotional state.

[1111] An "adjusting means" is a method or device for modifying or optimizing generated messages or other output content based on acquired data or conditions.

[1112] The "distribution means" refers to a method or device for transmitting the generated advocacy message to the collaborator's information terminal and making it available.

[1113] "Allies" are users or groups selected to post advocacy messages on social media.

[1114] A "social networking service" is a platform that enables users to interact and share information with other users online.

[1115] MODE FOR CARRYING OUT THE INVENTION

[1116] The present invention is a system for suppressing flame wars caused by offensive messages on social networking services (SNS). The system is composed of a server, a terminal, and a user, and includes the following elements and means:

[1117] Server Roles

[1118] The server plays a key role and performs the following functions:

[1119] 1. Launching the social media account monitoring module and collecting data

[1120] The server periodically obtains timeline data of service subscribers using SNS APIs, for example, by using Twitter APIs to collect tweet data from users and store and update it in a database.

[1121] 2. Detecting offensive messages

[1122] The server analyzes the collected data using Python natural language processing libraries (such as NLTK and spaCy). It uses specific keywords and sentiment analysis techniques to detect offensive messages. Messages that are deemed offensive are recorded in a dedicated log file.

[1123] 3. Applying the Emotion Engine

[1124] The server uses IBM Watson's sentiment analysis API to extract emotions from users' posts, allowing it to understand in real time the level of stress or anxiety a user is experiencing. Emotion scores are stored in a database and updated in real time.

[1125] 4. Threshold Checking and Adjusting the Advocacy Message Generation Model

[1126] The server evaluates whether the number of detected offensive messages exceeds a pre-defined threshold, and adjusts the parameters of a defense message generation model (e.g., GPT-3) based on the user's emotional state data.

[1127] 5. Advocacy message generation and quality check

[1128] The server inputs a prompt into a generative AI model (such as GPT-3) to generate a supportive message tailored to the user's situation. For example, the prompt might read, "The user is currently receiving aggressive messages on social media and is experiencing high stress levels. Please generate a supportive message to alleviate this situation." The generated message passes a quality check and is then prepared for distribution.

[1129] 6. Preparing and implementing advocacy message distribution

[1130] The server adds the verified advocacy message to a message queue and sends it to the registered collaborators through Firebase Cloud Messaging or APIs, for example, by calling the "SendMessage" API to deliver the message to the collaborators' devices.

[1131] Device Role

[1132] The device provides an interface for collaborators to display the advocacy messages they receive and post them to social media.

[1133] 1. Receiving and notifying advocacy messages

[1134] The collaborator's device receives the message sent by the server and stores it in a local data store. The device uses a notification system to notify the collaborator that a new message has been received, for example by displaying a pop-up or banner notification.

[1135] 2. Review and post advocacy messages

[1136] The collaborator user checks the notification on their device. After checking the content of the advocacy message, they press the "Post" button to call the SNS API and post the message to SNS. For example, we will post the message using the "POST statuses / update" endpoint.

[1137] User Roles

[1138] When users receive an offensive message on SNS, they can receive support through the system, which reduces the psychological burden and allows them to continue using SNS with peace of mind.

[1139] Specific examples

[1140] For example, consider a situation where User A is receiving a large number of aggressive messages on social media. Based on this information, the server uses its emotion engine to detect that User A's stress level is rising. Based on this data, it generates a supportive message such as, "Your efforts are being noticed by everyone. There are many voices of support for you." This message is sent to the device of Collaborator B, who then posts it on social media to support User A.

[1141] Prompt Sentence Examples

[1142] "Currently, users are receiving aggressive messages on social media and their stress levels are high. Please generate advocacy messages to ease this situation."

[1143] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1144] Step 1:

[1145] The server starts the SNS account monitoring module and periodically calls the SNS API of the service subscriber. As input, it receives the SNS API authentication information and user ID. Specifically, it retrieves the latest 20 tweets using the Twitter API's "GET statuses / user_timeline" endpoint. The output is the retrieved tweet data, which is then stored and updated in the database.

[1146] Step 2:

[1147] The server analyzes the tweet data stored in the database using natural language processing algorithms. The stored tweet data is used as input. Specifically, text analysis is performed using Python's NLTK and spaCy. To detect offensive messages, specific keywords and sentiment scores are calculated. The output is a list of messages deemed offensive, which is recorded in a dedicated log file.

[1148] Step 3:

[1149] The server calls IBM Watson's sentiment analysis API to analyze the user's emotional state. It uses the analyzed tweet data as input. Specifically, it sends the text of each tweet to the sentiment analysis API and obtains a sentiment score. The output is a sentiment score for each tweet, which is stored in a database and updated in real time.

[1150] Step 4:

[1151] The server counts the number of offensive messages detected within a certain period of time and compares it with a set threshold. It uses the number of offensive messages recorded in the log file and the sentiment scores stored in the database as input. Specifically, if the number of offensive messages exceeds the threshold, it adjusts the parameters of a generative AI model (such as GPT-3) based on the sentiment data. The output is the adjusted results of the generative model.

[1152] Step 5:

[1153] The server inputs a prompt into the generative AI model to generate a supportive message. The input uses the adjusted generative model and a prompt: "Currently, the user is receiving abusive messages on social media, and their stress level is high. Please generate a supportive message to alleviate this situation." Specifically, the server invokes the generative AI model to generate a message, which is then reviewed by a human auditor. The output is a supportive message that has passed the quality check.

[1154] Step 6:

[1155] The server adds the generated advocacy message to a message queue and distributes the message based on the contributor list. It uses the advocacy message that passed the quality check and the contributor list as input. Specifically, it sends the message to the contributor's device via Firebase Cloud Messaging or an API. The output is the advocacy message delivered to the contributor's device.

[1156] Step 7:

[1157] The terminal receives the advocacy message sent from the server and stores it in a local data store. The advocacy message sent from the server is used as input. After receiving the advocacy message, the terminal displays a pop-up or banner notification to inform the collaborator of the existence of a new message. The output is the advocacy message sent to the collaborator.

[1158] Step 8:

[1159] The user checks the notification on the device and views the content of the defense message. The defense message stored on the device is used as input. The specific operation is to check the message content, press the "Post" button to call the SNS API, and publish the message. The output is the defense message posted to the SNS.

[1160] (Application example 2)

[1161] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1162] In addition to the problem of an increase in abusive messages on social networking services (SNS) and an increasing number of users being victimized, there is also the problem of stress and fatigue among factory workers, which can lead to reduced productivity and accidents. The current situation, where there is a lack of stress management and psychological support, reduces work efficiency and has a negative impact on the health of workers.

[1163] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for detecting offensive messages to the information terminal account, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the collaborator's information terminal, means for the collaborator to post the defense message to a social networking service via the information terminal, means for acquiring video of the work environment and recognizing the emotional state in real time, and means for executing appropriate support messages and actions based on the emotional state. This makes it possible to mitigate the impact of offensive messages while managing the stress of workers in the work environment, improving work efficiency and ensuring safety.

[1164] An "information terminal" is a device used to process information such as social networking services.

[1165] "Offensive messages" are messages with negative content that criticize or insult others.

[1166] "Threshold" means a specific standard, here referring to the volume or frequency of offensive messages.

[1167] A "protective message" is a message intended to provide emotional support and encouragement to a user who has been harmed by an offensive message.

[1168] A "collaborator" is an individual or group responsible for receiving and posting advocacy messages on a social networking service.

[1169] "Emotional state" refers to an individual's psychological state and includes feelings such as stress, anxiety, and happiness.

[1170] A "generative model" refers to an algorithm or template used to tailor the content of messages generated by AI.

[1171] A "natural language processing algorithm" is a computer algorithm that analyzes natural language and understands its meaning and emotions.

[1172] "Work environment" refers to the place where work is done, such as a factory or office.

[1173] "Support messages" are messages that provide appropriate instructions and words of encouragement to reduce fatigue and stress in workers.

[1174] "Real time" refers to a process or activity occurring immediately at the present time.

[1175] "Footage" refers to visual information captured by a camera or other imaging device.

[1176] "Action" refers to the behavior or operation that is performed according to the emotional state.

[1177] This invention is a system that recognizes users' emotional states in real time on social networking services (SNS) and in factory work environments, and generates and distributes appropriate messages. The system includes a server, terminals, cameras installed in the work environment, and an emotion recognition engine.

[1178] System Configuration

[1179] The server monitors social media accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators' devices. It also acquires video data from cameras in the factory and recognizes the emotional state of workers in real time.

[1180] Server Roles

[1181] The server has the ability to detect offensive messages sent to the information terminal account, using a natural language processing algorithm to identify offensive content through sentiment analysis of language.

[1182] If the number of offensive messages exceeds a certain threshold, the server automatically creates a defense message using the generative model, which is then distributed to the collaborators' devices, who then post the message on the SNS.

[1183] The server also analyzes video data acquired from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time, generating appropriate support messages and notifying them if workers become stressed or fatigued.

[1184] Device Role

[1185] The collaborator's device receives the advocacy message sent from the server and notifies the collaborator via a user interface. The collaborator then checks the message and posts it to the SNS.

[1186] The terminals will display support messages and instructions to workers in the factory, encouraging them to take breaks as needed. The terminals will include a message receiving system and a display interface.

[1187] The role of emotion recognition engines

[1188] The emotion recognition engine analyzes video data acquired from the camera and detects worker stress and fatigue in real time. Based on this, an appropriate support message is generated using a generative AI model. The support message is then delivered to the worker via a server.

[1189] Specific examples

[1190] For example, consider a situation where a worker is feeling stressed due to continuous, repetitive work. In this case, the emotion recognition engine detects a high stress level, and the server generates a supportive message such as, "Take a short break. Your efforts are appreciated." This message is then displayed on the worker's device.

[1191] Prompt Sentence Examples

[1192] Examples of prompts to input to a generative AI model include:

[1193] "If the emotion recognition engine detects high stress levels, generate an appropriate encouraging message for the worker, such as, 'Your efforts are appreciated. Take a short break.'"

[1194] In this way, the system not only mitigates the impact of offensive messages, but also manages worker stress in the work environment, improving work efficiency and ensuring safety.

[1195] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1196] Step 1:

[1197] The server activates a means for detecting offensive messages against the information terminal's account. This means acquires timeline data from the social media account and saves and updates it in a database. The input is the social media timeline data, and the output is the updated timeline data saved in the database. Specifically, the server acquires the social media data through an API and saves it in a database for analysis.

[1198] Step 2:

[1199] The server analyzes the acquired data using natural language processing algorithms to detect offensive messages. The input is the saved timeline data, and the output is a list of detected offensive messages. Specifically, it uses a natural language processing library to perform sentiment analysis of the text data and flag offensive messages.

[1200] Step 3:

[1201] The server applies an emotion recognition engine to recognize the user's emotional state when the number of offensive messages exceeds a certain threshold. The input is a list of offensive messages, and the output is the user's emotional state data. Specifically, the server starts the emotion recognition engine and evaluates the user's emotions in real time.

[1202] Step 4:

[1203] The server uses a generative AI model to generate an advocacy message based on the emotional state data. The input is the emotional state data, and the output is the generated advocacy message. Specifically, the generative AI model is used to create a message that is in tune with the user's emotions. An example of a prompt sentence is, "If the emotion recognition engine recognizes that the stress level is high, generate an appropriate encouraging message for the worker. For example, something like, 'Your efforts are appreciated. Please take a short break.'"

[1204] Step 5:

[1205] The server starts a means to distribute the generated advocacy message to the collaborator's terminal. The input is the generated advocacy message, and the output is the message sent to the collaborator's terminal. The specific operation is to distribute the message to the collaborator via an API or a notification system.

[1206] Step 6:

[1207] The collaborator's terminal saves the received advocacy message and displays it on the user interface to notify the collaborator. The input is the advocacy message received from the server, and the output is the message displayed on the user interface. The specific operation is to use the terminal's notification system to notify the collaborator that there is a new message.

[1208] Step 7:

[1209] The collaborator user checks the defense message notified on their device and posts it to the SNS. The input is the received defense message, and the output is the message posted to the SNS. Specifically, when the collaborator presses the post button on their device, the message is published through the SNS API.

[1210] Step 8:

[1211] The server acquires video data from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time. The input is video data from the cameras, and the output is data on the emotional state of the workers. Specifically, the server analyzes the video data, and the emotion recognition engine detects stress and fatigue.

[1212] Step 9:

[1213] The server generates appropriate support messages and actions based on the emotional state data and notifies the worker's device. The input is the emotional state data, and the output is the support message. Specifically, the server uses the generative AI model to create a message encouraging the worker to take a break and displays it on the device.

[1214] Step 10:

[1215] The worker's device displays the received support message and notifies the worker. The input is the generated support message, and the output is the message displayed on the device. Specifically, the message is displayed on the device's display or through the notification system, prompting the worker to take appropriate action.

[1216] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1217] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1218] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1219] [Fourth embodiment]

[1220] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1221] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1222] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1223] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1224] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1225] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1226] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1227] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1228] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1229] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1230] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1231] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1232] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1233] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and a method for implementing the system will be described using a specific example.

[1234] Overall system overview

[1235] This system consists of a server, a terminal, and a user. The server monitors the SNS accounts of service subscribers and detects offensive messages. For detected offensive messages, the server uses a generative model to generate defensive messages and distributes them to the terminals of collaborators. The collaborators then post defensive messages to the SNS via their terminals, thereby mitigating the impact of the offensive messages.

[1236] Server Roles

[1237] 1. Starting the monitoring module

[1238] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[1239] 2. Detecting offensive messages

[1240] The server analyzes the data it receives using a natural language processing algorithm to identify offensive messages. The algorithm performs sentiment analysis of the language and detects content that is deemed offensive. Detected offensive messages are stored in a dedicated log.

[1241] 3. Generating Advocacy Messages

[1242] If the number of offensive messages exceeds a certain threshold, the server uses a generative model to generate defensive messages, which undergo quality checks and are added to a message queue.

[1243] 4. Send messages to collaborators

[1244] The server accesses the list of registered contributors and distributes the generated advocacy messages to the contributors' devices, quickly using APIs and notification systems.

[1245] Device Role

[1246] 1. Receiving a message

[1247] The collaborator's device receives the advocacy message sent from the server, which causes the message to be stored in the device's local data store.

[1248] 2. Submission Preparation and Notification

[1249] The device application notifies the user that a new message has been sent to the user interface, and after confirmation, the user is ready to post the message to the SNS.

[1250] User Roles

[1251] 1. Posting a message of support

[1252] The contributor user checks the advocacy message sent from the server and posts it to the SNS via the application on their device. By pressing the post button, the message is made public using the SNS API.

[1253] Specific examples

[1254] A concrete example is given below. For example, suppose a user receives a large number of offensive messages. In this case, the server detects these messages and generates a supportive message such as, "This user's behavior is admirable. Let's support them together." This message is sent to the collaborator's device, and the collaborator posts it on SNS, mitigating the impact of the offensive messages.

[1255] In this way, the SNS Shield system counters offensive messages in real time, providing a safe environment for users to use SNS.

[1256] The processing flow will be explained below.

[1257] Step 1: Starting the monitoring module

[1258] The server starts a module for monitoring the social media accounts of service subscribers, allowing the server to obtain subscribers' posts and comments in real time.

[1259] Step 2: Getting the Timeline Data

[1260] The server periodically collects subscribers' latest posts and comments via the SNS's API and stores them in a database.

[1261] Step 3: Detecting offensive messages

[1262] The server runs natural language processing algorithms on the collected data to detect offensive messages, specifically identifying messages that contain negative sentiment or offensive content.

[1263] Step 4: Threshold check

[1264] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, proceeds to the next step.

[1265] Step 5: Generating an advocacy message

[1266] The server uses the generative model to generate advocacy messages, such as "This user is doing great work, let's support them."

[1267] Step 6: Check message quality

[1268] The server quality checks the generated advocacy message for correctness, and if incorrect, it is regenerated.

[1269] Step 7: Prepare the message for distribution

[1270] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list, and adds them to the message queue for distribution.

[1271] Step 8: Distribute messages to collaborators

[1272] The server distributes messages to the collaborators' devices. Messages are sent to the collaborators' devices through APIs or notification systems.

[1273] Step 9: Receiving a message

[1274] The terminal receives the advocacy message sent from the server and stores it in a local data store.

[1275] Step 10: User Interface Notification

[1276] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[1277] Step 11: Review your advocacy message

[1278] The user checks the notification on the terminal and confirms the content of the received advocacy message.

[1279] Step 12: Post an Advocacy Message

[1280] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[1281] Step 13: Feedback on submission results

[1282] The terminal notifies the server that the message has been successfully posted, which causes a posting log to be recorded on the server.

[1283] Step 14: Evaluate the effectiveness of the system

[1284] The server evaluates the effectiveness of the system based on the feedback and improves the generative model and algorithms as needed.

[1285] Example 1

[1286] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1287] On modern online platforms, online flame wars caused by offensive messages are becoming more frequent, placing a greater mental burden on users. This problem significantly undermines the environment in which users can use the platform safely. Conventional systems have not established a method for effectively suppressing these offensive messages, and it is particularly difficult to respond in real time. Therefore, there is a need to provide a safe and secure environment for users by quickly detecting offensive messages on online platforms and responding appropriately.

[1288] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1289] In this invention, the server includes a means for monitoring abusive messages to the account of the information processing device, a means for generating a defense message when the abusive messages exceed a certain threshold, and a means for distributing the generated defense message to the information processing device of the collaborator. This makes it possible to detect abusive messages in real time and respond quickly to prevent online flaming and provide a safe and secure environment for users.

[1290] "Information processing device" generally refers to electronic devices capable of data processing, such as computers and smartphones.

[1291] An "account" is data including identification information and authentication information that allows a user to be individually managed on an online platform.

[1292] "Offensive messages" are messages that are intended to hurt, insult, or cause anxiety to others.

[1293] "Generative AI models" refer to algorithms or frameworks for generating text or data using artificial intelligence techniques.

[1294] "Natural language processing algorithms" refer to computational techniques for understanding, generating, and manipulating human language.

[1295] A "threshold" is a value that meets a particular condition or criterion, and when exceeded, triggers a particular action.

[1296] "Advocacy messages" refer to positive messages that are generated to mitigate the impact of aggressive messages and protect the target audience.

[1297] "Online Platform" refers to a website or application that enables users to share information, exchange, or interact over the Internet.

[1298] "Distribution" refers to the sharing of specific information or data with multiple recipients.

[1299] The present invention relates to a system for suppressing flame wars caused by offensive messages on online platforms. The system is composed of an information processing device, a server, and a user. Specific embodiments for carrying out the invention are described below.

[1300] Configuration of information processing device

[1301] The information processing device is an electronic device capable of data processing, such as a computer or smartphone, and functions as a terminal used by collaborators. This information processing device includes a network connection function for receiving messages and a local data store for saving and displaying received messages. The specific implementation uses an SQLite database.

[1302] Server Configuration

[1303] The server consists of the following main modules:

[1304] Monitoring module: Implemented using Python language and Django framework, it monitors the accounts of service subscribers.

[1305] Natural language processing module: Analyzes and detects offensive messages using TensorFlow and PyTorch.

[1306] Generative module: Creates advocacy messages using a generative AI model (e.g., OpenAI GPT-3).

[1307] Notification module: Distributes generated advocacy messages to collaborators' information processing devices using RESTful APIs and the Firebase notification system.

[1308] User Roles

[1309] The collaborating user checks the advocacy message received from the server and posts it to the online platform via an information processing device. The user uses the application on their device to press the post button, which publishes the message through the SNS API (e.g., Twitter API).

[1310] Overview of the invention

[1311] The server monitors the subscriber's account on the online platform and detects offensive messages using a natural language processing module. A generation module generates a defensive message for the detected offensive message. The defensive message is distributed to the collaborator's information processing device, and after the user has reviewed it, the user can post it on the online platform to mitigate the impact of the offensive message.

[1312] Specific examples

[1313] For example, if a user receives an offensive message such as "I completely disagree with your opinion. I think it's wrong," the server detects this message and uses a generative AI model (e.g., GPT-3) to generate a defensive message such as "I think it's okay to have different opinions. Let's respect this user's opinion." This message is then sent to the collaborator's information processing device, and the collaborator posts it on an online platform to mitigate the impact of the offensive message.

[1314] Examples of prompt statements

[1315] Offensive messages:

[1316] "I don't agree with you at all. I think you're wrong."

[1317] Advocacy message:

[1318] "It's good to have different opinions. Let's respect this user's opinion."

[1319] In this way, the system responds to offensive messages in real time, providing a safe and secure environment for users to use online platforms.

[1320] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1321] Step 1:

[1322] The server activates a monitoring module to monitor the service subscriber's account.

[1323] Input: Service subscriber account information

[1324] Specific operation: Uses the Django framework and executes the management command python manage.py runserver.

[1325] Data processing: Send a request to the SNS API to retrieve timeline data.

[1326] Output: Timeline data is saved to a database.

[1327] Step 2:

[1328] The server analyzes the acquired timeline data using a natural language processing algorithm to detect offensive messages.

[1329] Input: Saved timeline data

[1330] Specific operation: Loads a TensorFlow sentiment analysis model and performs analysis on the data.

[1331] Data processing: Processing data with sentiment analysis algorithms to identify offensive messages.

[1332] Output: Offensive messages are saved to a dedicated log file.

[1333] Step 3:

[1334] The server uses a generative AI model to generate a defensive message when an offensive message exceeds a certain threshold.

[1335] Input: offensive message

[1336] Specific behavior: Send an offensive message as a prompt to the GPT-3 API.

[1337] Data processing: The generated advocacy messages are reviewed in a quality check module.

[1338] Output: Advocacy messages that meet the criteria are added to the message queue.

[1339] Step 4:

[1340] The server distributes the advocacy message generated based on the collaborator list to the information processing device.

[1341] Input: Deferred message in message queue

[1342] Specific behavior: Sends notifications using RESTful APIs and the Firebase notification system.

[1343] Data processing: Extract the message from the message queue.

[1344] Output: The message is distributed to the collaborators' information processing devices.

[1345] Step 5:

[1346] The collaborator's terminal receives the advocacy message sent from the server.

[1347] Input: Notification from the server

[1348] Specific operation: Set up a listener to receive notifications from Firebase.

[1349] Data processing: Received messages are saved in a SQLite database.

[1350] Output: Saved advocacy messages are logged to a local data store.

[1351] Step 6:

[1352] The terminal notifies the user of the arrival of a new message and displays it on the user interface, allowing the user to check the message and prepare it for posting.

[1353] Input: Saved advocacy message

[1354] Specific operation: Notifies using the notification manager (Notification Manager for Android, UNUserNotificationCenter for iOS).

[1355] Data processing: Display the message content within the app.

[1356] Output: The user has reviewed the message and is ready to hit the post button.

[1357] Step 7:

[1358] A user posts an advocacy message to an online platform via an application on the terminal.

[1359] Input: Saved advocacy messages and user actions

[1360] Specific operation: By pressing the post button, a POST request is sent to the SNS API (e.g. Twitter API).

[1361] Data processing: Generates request data to be sent to the SNS API.

[1362] Output: The advocacy message is published on an online platform.

[1363] (Application example 1)

[1364] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1365] In recent years, there has been an increase in offensive and abusive messages on social networking services (SNS). These messages can have a negative impact on users' mental health and can lead to a phenomenon known as "flaming." While methods to prevent and mitigate this are needed, effective countermeasures are currently lacking. Furthermore, because manual responses are time-consuming and labor-intensive, real-time automated countermeasures are needed.

[1366] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1367] In this invention, the server includes means for detecting offensive messages to the account of the information terminal, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the information terminal of the collaborator, means for the collaborator to post the defense message to the social networking service via the information terminal, means for monitoring the SNS account in real time, means for notifying the generated defense message, and means for the notified user to post the message with one click after confirmation. This makes it possible to respond quickly and automatically to offensive messages and minimize the impact of flame wars on the SNS.

[1368] An "information terminal" is an electronic device that a user uses to access the Internet.

[1369] An "offensive message" is a message that contains negative or offensive content and may cause psychological harm to a user.

[1370] The "threshold" is a criterion for determining whether an offensive message exceeds a certain standard value.

[1371] "Advocacy messages" are messages that positively evaluate and support users and are generated to counter aggressive messages.

[1372] A "collaborator" is a user whose role is to receive advocacy messages and post them on social media from their own account.

[1373] "Real time" is a time concept in which information is acquired and processed immediately, meaning a nearly simultaneous response.

[1374] "Notifications" are alerts or messages sent to collaborators' information terminals when a specific event occurs.

[1375] "One-click posting" is a feature that allows users to post messages on social media with minimal operations, i.e., a single click or tap.

[1376] A "natural language processing algorithm" is a technology that analyzes human language and understands meaning and emotions.

[1377] A "generative AI model" is an artificial intelligence technology that automatically generates new text based on existing data.

[1378] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and is implemented as follows.

[1379] Overall system configuration

[1380] This system consists of a server, the collaborator's information terminal, and the user who operates it. The server monitors SNS accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators. The information terminal receives defensive messages, and posts them to the SNS after user confirmation.

[1381] Hardware and software configuration

[1382] Hardware:

[1383] Server: a high-performance computer

[1384] Information device: Smartphone (iPhone or Android device)

[1385] software:

[1386] Server-side program:

[1387] Natural Language Processing Algorithms (TensorFlow, Keras, Hugging Face Transformers)

[1388] Message Distribution API

[1389] Information terminal app:

[1390] Programming language: Swift (iOS), Kotlin (Android)

[1391] Database: Firebase Firestore

[1392] Notification system: Firebase Cloud Messaging (FCM)

[1393] Program processing

[1394] Server Processing

[1395] The server monitors social media accounts in real time to obtain timeline data, which it then stores and updates in a database. The server then uses a natural language processing algorithm to detect offensive messages from the obtained data. This algorithm identifies offensive content and negative sentiment.

[1396] If the number of offensive messages exceeds a certain threshold, the server uses a generative AI model to generate a defensive message. The generated message undergoes a quality check and is then sent to the collaborator's device. Notifications are sent using Firebase Cloud Messaging (FCM), allowing collaborators to respond quickly.

[1397] Information terminal processing

[1398] The collaborator's information device receives the advocacy message sent from the server. When the notification arrives, the message is saved in the device's local data store, and the application notifies the user to display the new message on the user interface, ready for the user to check and post to SNS with one click.

[1399] Specific examples

[1400] For example, if a user receives a large number of offensive messages on Twitter, the server detects these messages and generates a supportive message such as, "This user's actions are admirable. Let's support them together!" This message is sent to the collaborator's information terminal, and the collaborator can post this message to Twitter with one click, thereby mitigating the impact of the offensive messages.

[1401] Example prompts to input to a generative AI model:

[1402] A specific user is being attacked on social media. Please generate a positive, supportive message to defuse the situation. Please keep it general and supportive, without including specific usernames or examples of behavior.

[1403] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1404] Step 1:

[1405] The server starts monitoring the social media account in real time, obtains the user's timeline data, and saves and updates it in the database.

[1406] Input: User's social media account information, timeline data

[1407] Output: Timeline data stored in a database

[1408] Specific operation: The server retrieves the user's social media timeline through the API, updates it periodically, and saves this data in a database (e.g., Firebase Firestore).

[1409] Step 2:

[1410] The server analyzes the timeline data using natural language processing algorithms to detect offensive messages.

[1411] Input: Timeline data stored in a database, natural language processing algorithms (e.g., Hugging Face Transformers)

[1412] Output: List of detected offensive messages

[1413] Specific operation: The server uses a library for natural language processing to analyze the timeline data, extracting posts containing offensive messages or negative sentiment and creating a list of them.

[1414] Step 3:

[1415] If the server detects an offensive message that exceeds a certain threshold, it uses a generative AI model to generate a defensive message.

[1416] Input: List of offensive messages, generative AI model

[1417] Output: Generated advocacy message

[1418] Specific operation: The server inputs a prompt sentence to the generative AI model to create a supportive message to defuse an aggressive situation. Based on the prompt sentence, the generative AI model generates an appropriate supportive message.

[1419] Step 4:

[1420] The server distributes the generated advocacy message to the information terminals of the collaborators.

[1421] Input: Generated advocacy message, list of contributors

[1422] Output: Advocacy message received on the collaborator's information terminal

[1423] Specific operation: The server uses Firebase Cloud Messaging to send notifications of the advocacy message to all devices included in the contributor list.

[1424] Step 5:

[1425] The terminal receives the protection message sent from the server and stores it in a local data store.

[1426] Input: Advice message sent from the server

[1427] Output: Advocacy messages saved to the local data store

[1428] Specific behavior: The device receives the notification and saves it to a local data store (e.g., an SQLite database).

[1429] Step 6:

[1430] The device will display a notification prompting the user to acknowledge the advocacy message.

[1431] Input: Advocacy messages stored in the local data store

[1432] Output: User-visible notification message

[1433] What happens: The device application displays a new message notification on the user interface, and the user taps the notification to view the message details.

[1434] Step 7:

[1435] Users can view advocacy messages and post them to social media with one click.

[1436] Input: Advocacy message, SNS account information

[1437] Output: Advocacy messages posted on social media

[1438] Specific operation: When a user presses a button that allows them to post a support message to a social networking site with one click, the device application posts the message using the social networking site API (e.g., Twitter API).

[1439] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1440] The present invention relates to a system for suppressing flame wars caused by offensive messages on social networking services (SNS), and in particular, an embodiment in which an emotion engine that recognizes the emotions of users is combined will be described.

[1441] Overall system overview

[1442] This system consists of a server, a terminal, and a user, and integrates an emotion engine to improve the effectiveness of advocacy messages. The server monitors the SNS accounts of service subscribers and detects offensive messages. The emotion engine recognizes the user's emotional state in real time and adjusts the generation of advocacy messages based on the results. The generated advocacy messages are distributed to the terminals of collaborators, who then post them on the SNS.

[1443] Server Roles

[1444] 1. Starting the monitoring module

[1445] The server starts a module to monitor the SNS accounts of service subscribers, thereby obtaining subscriber timeline data in real time and saving / updating it in the database.

[1446] 2. Detecting offensive messages

[1447] The server analyzes the acquired data using a natural language processing algorithm to detect offensive messages. This algorithm performs sentiment analysis of the language and detects content that is deemed offensive. The detected offensive messages are stored in a dedicated log.

[1448] 3. Applying the Emotion Engine

[1449] The server uses an emotion engine to recognize the user's emotional state in real time, thereby determining whether the user is experiencing stress or anxiety.

[1450] 4. Threshold Checking and Generative Model Tuning

[1451] The server determines whether the number of detected offensive messages exceeds a set threshold and adjusts the generation of advocacy messages based on data from the emotion engine, using a generative model to generate appropriate message content according to the user's emotional state.

[1452] 5. Message Creation and Preparation for Distribution

[1453] The server uses the generative model to create an advocacy message that passes the quality check, prepares it for distribution to the collaborators' terminals, and adds it to the message queue for distribution.

[1454] 6. Send messages to collaborators

[1455] The server distributes the generated advocacy messages to the collaborators' terminals via APIs and notification systems based on the list of registered collaborators.

[1456] Device Role

[1457] 1. Receiving a message

[1458] The collaborator's terminal receives the advocacy message sent from the server and stores it in a local data store.

[1459] 2. User interface notifications

[1460] The terminal displays the received message on the user interface and notifies the collaborator that there is a new message.

[1461] User Roles

[1462] 1. Review and post advocacy messages

[1463] The collaborator user checks the notification on their device and confirms the content of the received advocacy message. Then, they post the received advocacy message to the SNS via the device's application. By pressing the post button, the device publishes the message using the SNS API.

[1464] Specific examples

[1465] Here's a concrete example: In a situation where a user is receiving a large number of aggressive messages, the emotion engine detects that the user's stress level is rising. Based on this information, the server generates a more emotionally sensitive supportive message, such as, "Your efforts are being heard by everyone. There is a lot of support for you." This message is sent to the collaborator's device, and the collaborator posts it on social media, effectively mitigating the impact of the aggressive messages.

[1466] In this way, the SNS Shield system applies an emotion engine to provide real-time support messages that correspond to the user's emotional state, providing an environment in which users can use SNS safely.

[1467] The processing flow will be explained below.

[1468] Step 1: Starting the monitoring module

[1469] The server runs a module that monitors subscribers' social media accounts, capturing their posts and comments in real time and storing them in a database. A dedicated database entry is created for each monitored account.

[1470] Step 2: Getting the Timeline Data

[1471] The server periodically collects subscribers' timeline data via SNS API, including posts, comments, replies, likes, etc. This data is stored in a database and prepared for subsequent processing.

[1472] Step 3: Detecting offensive messages

[1473] The server applies natural language processing algorithms to the collected timeline data, which are designed to identify messages containing offensive or negative content, and records the detected messages in a dedicated log.

[1474] Step 4: Applying the Emotion Engine

[1475] The server simultaneously monitors the collected offensive messages and activates an emotion engine that analyzes the user's current emotions based on their timeline data and past messages, and determines their stress and anxiety levels.

[1476] Step 5: Threshold check

[1477] The server determines whether the number of detected offensive messages exceeds a pre-set threshold, and if so, the server considers the user's emotional state data obtained from the emotion engine and proceeds to the next step.

[1478] Step 6: Generating Advocacy Messages

[1479] The server uses a generative model to generate appropriate advocacy messages based on data from the emotion engine. For example, if a user's stress level is high, a message such as "This user is doing great things, so let's support them" is generated.

[1480] Step 7: Check message quality

[1481] The server quality-checks the content of generated advocacy messages, regenerating inappropriate messages, and fine-tuning the tone and content of messages based on feedback from the emotion engine.

[1482] Step 8: Preparing the Message for Distribution

[1483] The server prepares the advocacy messages that pass the quality check for distribution based on the contributor list. The messages are added to a message queue for distribution.

[1484] Step 9: Distribute messages to collaborators

[1485] The server distributes advocacy messages to registered collaborators' devices using APIs and notification systems. When a notification is received, the collaborator receives the message on their own device.

[1486] Step 10: Receiving a message

[1487] The terminal receives the protection message sent from the server and stores it in local storage, and the user interface displays the new message to notify the user.

[1488] Step 11: User Interface Notification

[1489] The terminal displays the received advocacy message on the user interface and notifies the collaborator that there is a new message. The user checks the contents of the message.

[1490] Step 12: Review your advocacy message

[1491] The user checks the notification on the device and confirms the content of the received advocacy message. If the message content is deemed appropriate, the user prepares to post it.

[1492] Step 13: Post an Advocacy Message

[1493] The user posts the received advocacy message to the SNS via the device application. When the user presses the post button, the device publishes the message using the SNS API.

[1494] Step 14: Feedback on submission results

[1495] The terminal notifies the server that the message has been successfully posted, and the server saves the posting log and uses it as data to evaluate the effectiveness of the system.

[1496] Step 15: Evaluate the effectiveness of the system

[1497] The server evaluates the effectiveness of the system based on the feedback data and improves the generative model, emotion engine, and natural language processing algorithm as needed.

[1498] Through the above steps, the system utilizes the emotion engine to generate a defense message according to the user's emotional state, and can effectively mitigate the impact of aggressive messages.

[1499] Example 2

[1500] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1501] Flaming on social networking services (SNS) caused by offensive messages places a significant psychological burden on users and has become a social problem. To solve this problem, a system that can detect offensive messages and take appropriate countermeasures is needed, but current technology is insufficient in taking countermeasures based on the user's real-time emotional state. In addition, there is a need to improve the accuracy of detecting offensive messages and the quality of the generated defense messages.

[1502] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1503] In this invention, the server includes a means for detecting offensive messages, a means for generating a defense message when a threshold is exceeded, a means for distributing the generated defense message to collaborators, a means for recognizing the emotional state of the user in real time, and a means for adjusting the content of the defense message based on the emotional state acquired in real time, thereby making it possible to quickly and effectively suppress flame wars caused by offensive messages and reduce the burden on users.

[1504] An "information terminal" is a device used by a user, and includes a smartphone, tablet, or personal computer.

[1505] An "offensive message" is a message that contains malicious or harassing words or content and causes a psychological burden to the user.

[1506] A "means for detecting" is a method or device that uses specific algorithms or techniques to identify and identify offensive messages.

[1507] A "protective message" is a message of support or encouragement that is generated to help a user who has received an offensive message and to reduce the mental burden.

[1508] A "generating means" is a method or device for automatically generating advocacy messages using a specific algorithm or model.

[1509] "Emotional state" refers to a particular psychological state or mood that a user is experiencing, including states such as anxiety, stress, and joy.

[1510] A "real-time recognition means" is a method or device for instantly analyzing and understanding a user's current emotional state.

[1511] An "adjusting means" is a method or device for modifying or optimizing generated messages or other output content based on acquired data or conditions.

[1512] The "distribution means" refers to a method or device for transmitting the generated advocacy message to the collaborator's information terminal and making it available.

[1513] "Allies" are users or groups selected to post advocacy messages on social media.

[1514] A "social networking service" is a platform that enables users to interact and share information with other users online.

[1515] MODE FOR CARRYING OUT THE INVENTION

[1516] The present invention is a system for suppressing flame wars caused by offensive messages on social networking services (SNS). The system is composed of a server, a terminal, and a user, and includes the following elements and means:

[1517] Server Roles

[1518] The server plays a key role and performs the following functions:

[1519] 1. Launching the social media account monitoring module and collecting data

[1520] The server periodically obtains timeline data of service subscribers using SNS APIs, for example, by using Twitter APIs to collect tweet data from users and store and update it in a database.

[1521] 2. Detecting offensive messages

[1522] The server analyzes the collected data using Python natural language processing libraries (such as NLTK and spaCy). It uses specific keywords and sentiment analysis techniques to detect offensive messages. Messages that are deemed offensive are recorded in a dedicated log file.

[1523] 3. Applying the Emotion Engine

[1524] The server uses IBM Watson's sentiment analysis API to extract emotions from users' posts, allowing it to understand in real time the level of stress or anxiety a user is experiencing. Emotion scores are stored in a database and updated in real time.

[1525] 4. Threshold Checking and Adjusting the Advocacy Message Generation Model

[1526] The server evaluates whether the number of detected offensive messages exceeds a pre-defined threshold, and adjusts the parameters of a defense message generation model (e.g., GPT-3) based on the user's emotional state data.

[1527] 5. Advocacy message generation and quality check

[1528] The server inputs a prompt into a generative AI model (such as GPT-3) to generate a supportive message tailored to the user's situation. For example, the prompt might read, "The user is currently receiving aggressive messages on social media and is experiencing high stress levels. Please generate a supportive message to alleviate this situation." The generated message passes a quality check and is then prepared for distribution.

[1529] 6. Preparing and implementing advocacy message distribution

[1530] The server adds the verified advocacy message to a message queue and sends it to the registered collaborators through Firebase Cloud Messaging or APIs, for example, by calling the "SendMessage" API to deliver the message to the collaborators' devices.

[1531] Device Role

[1532] The device provides an interface for collaborators to display the advocacy messages they receive and post them to social media.

[1533] 1. Receiving and notifying advocacy messages

[1534] The collaborator's device receives the message sent by the server and stores it in a local data store. The device uses a notification system to notify the collaborator that a new message has been received, for example by displaying a pop-up or banner notification.

[1535] 2. Review and post advocacy messages

[1536] The collaborator user checks the notification on their device. After checking the content of the advocacy message, they press the "Post" button to call the SNS API and post the message to SNS. For example, we will post the message using the "POST statuses / update" endpoint.

[1537] User Roles

[1538] When users receive an offensive message on SNS, they can receive support through the system, which reduces the psychological burden and allows them to continue using SNS with peace of mind.

[1539] Specific examples

[1540] For example, consider a situation where User A is receiving a large number of aggressive messages on social media. Based on this information, the server uses its emotion engine to detect that User A's stress level is rising. Based on this data, it generates a supportive message such as, "Your efforts are being noticed by everyone. There are many voices of support for you." This message is sent to the device of Collaborator B, who then posts it on social media to support User A.

[1541] Prompt Sentence Examples

[1542] "Currently, users are receiving aggressive messages on social media and their stress levels are high. Please generate advocacy messages to ease this situation."

[1543] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1544] Step 1:

[1545] The server starts the SNS account monitoring module and periodically calls the SNS API of the service subscriber. As input, it receives the SNS API authentication information and user ID. Specifically, it retrieves the latest 20 tweets using the Twitter API's "GET statuses / user_timeline" endpoint. The output is the retrieved tweet data, which is then stored and updated in the database.

[1546] Step 2:

[1547] The server analyzes the tweet data stored in the database using natural language processing algorithms. The stored tweet data is used as input. Specifically, text analysis is performed using Python's NLTK and spaCy. To detect offensive messages, specific keywords and sentiment scores are calculated. The output is a list of messages deemed offensive, which is recorded in a dedicated log file.

[1548] Step 3:

[1549] The server calls IBM Watson's sentiment analysis API to analyze the user's emotional state. It uses the analyzed tweet data as input. Specifically, it sends the text of each tweet to the sentiment analysis API and obtains a sentiment score. The output is a sentiment score for each tweet, which is stored in a database and updated in real time.

[1550] Step 4:

[1551] The server counts the number of offensive messages detected within a certain period of time and compares it with a set threshold. It uses the number of offensive messages recorded in the log file and the sentiment scores stored in the database as input. Specifically, if the number of offensive messages exceeds the threshold, it adjusts the parameters of a generative AI model (such as GPT-3) based on the sentiment data. The output is the adjusted results of the generative model.

[1552] Step 5:

[1553] The server inputs a prompt into the generative AI model to generate a supportive message. The input uses the adjusted generative model and a prompt: "Currently, the user is receiving abusive messages on social media, and their stress level is high. Please generate a supportive message to alleviate this situation." Specifically, the server invokes the generative AI model to generate a message, which is then reviewed by a human auditor. The output is a supportive message that has passed the quality check.

[1554] Step 6:

[1555] The server adds the generated advocacy message to a message queue and distributes the message based on the contributor list. It uses the advocacy message that passed the quality check and the contributor list as input. Specifically, it sends the message to the contributor's device via Firebase Cloud Messaging or an API. The output is the advocacy message delivered to the contributor's device.

[1556] Step 7:

[1557] The terminal receives the advocacy message sent from the server and stores it in a local data store. The advocacy message sent from the server is used as input. After receiving the advocacy message, the terminal displays a pop-up or banner notification to inform the collaborator of the existence of a new message. The output is the advocacy message sent to the collaborator.

[1558] Step 8:

[1559] The user checks the notification on the device and views the content of the defense message. The defense message stored on the device is used as input. The specific operation is to check the message content, press the "Post" button to call the SNS API, and publish the message. The output is the defense message posted to the SNS.

[1560] (Application example 2)

[1561] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1562] In addition to the problem of an increase in abusive messages on social networking services (SNS) and an increasing number of users being victimized, there is also the problem of stress and fatigue among factory workers, which can lead to reduced productivity and accidents. The current situation, where there is a lack of stress management and psychological support, reduces work efficiency and has a negative impact on the health of workers.

[1563] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for detecting offensive messages to the information terminal account, means for generating a defense message when the number of offensive messages exceeds a certain threshold, means for distributing the generated defense message to the collaborator's information terminal, means for the collaborator to post the defense message to a social networking service via the information terminal, means for acquiring video of the work environment and recognizing the emotional state in real time, and means for executing appropriate support messages and actions based on the emotional state. This makes it possible to mitigate the impact of offensive messages while managing the stress of workers in the work environment, improving work efficiency and ensuring safety.

[1564] An "information terminal" is a device used to process information such as social networking services.

[1565] "Offensive messages" are messages with negative content that criticize or insult others.

[1566] "Threshold" means a specific standard, here referring to the volume or frequency of offensive messages.

[1567] A "protective message" is a message intended to provide emotional support and encouragement to a user who has been harmed by an offensive message.

[1568] A "collaborator" is an individual or group responsible for receiving and posting advocacy messages on a social networking service.

[1569] "Emotional state" refers to an individual's psychological state and includes feelings such as stress, anxiety, and happiness.

[1570] A "generative model" refers to an algorithm or template used to tailor the content of messages generated by AI.

[1571] A "natural language processing algorithm" is a computer algorithm that analyzes natural language and understands its meaning and emotions.

[1572] "Work environment" refers to the place where work is done, such as a factory or office.

[1573] "Support messages" are messages that provide appropriate instructions and words of encouragement to reduce fatigue and stress in workers.

[1574] "Real time" refers to a process or activity occurring immediately at the present time.

[1575] "Footage" refers to visual information captured by a camera or other imaging device.

[1576] "Action" refers to the behavior or operation that is performed according to the emotional state.

[1577] This invention is a system that recognizes users' emotional states in real time on social networking services (SNS) and in factory work environments, and generates and distributes appropriate messages. The system includes a server, terminals, cameras installed in the work environment, and an emotion recognition engine.

[1578] System Configuration

[1579] The server monitors social media accounts, detects offensive messages, generates defensive messages, and distributes them to collaborators' devices. It also acquires video data from cameras in the factory and recognizes the emotional state of workers in real time.

[1580] Server Roles

[1581] The server has the ability to detect offensive messages sent to the information terminal account, using a natural language processing algorithm to identify offensive content through sentiment analysis of language.

[1582] If the number of offensive messages exceeds a certain threshold, the server automatically creates a defense message using the generative model, which is then distributed to the collaborators' devices, who then post the message on the SNS.

[1583] The server also analyzes video data acquired from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time, generating appropriate support messages and notifying them if workers become stressed or fatigued.

[1584] Device Role

[1585] The collaborator's device receives the advocacy message sent from the server and notifies the collaborator via a user interface. The collaborator then checks the message and posts it to the SNS.

[1586] The terminals will display support messages and instructions to workers in the factory, encouraging them to take breaks as needed. The terminals will include a message receiving system and a display interface.

[1587] The role of emotion recognition engines

[1588] The emotion recognition engine analyzes video data acquired from the camera and detects worker stress and fatigue in real time. Based on this, an appropriate support message is generated using a generative AI model. The support message is then delivered to the worker via a server.

[1589] Specific examples

[1590] For example, consider a situation where a worker is feeling stressed due to continuous, repetitive work. In this case, the emotion recognition engine detects a high stress level, and the server generates a supportive message such as, "Take a short break. Your efforts are appreciated." This message is then displayed on the worker's device.

[1591] Prompt Sentence Examples

[1592] Examples of prompts to input to a generative AI model include:

[1593] "If the emotion recognition engine detects high stress levels, generate an appropriate encouraging message for the worker, such as, 'Your efforts are appreciated. Take a short break.'"

[1594] In this way, the system not only mitigates the impact of offensive messages, but also manages worker stress in the work environment, improving work efficiency and ensuring safety.

[1595] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1596] Step 1:

[1597] The server activates a means for detecting offensive messages against the information terminal's account. This means acquires timeline data from the social media account and saves and updates it in a database. The input is the social media timeline data, and the output is the updated timeline data saved in the database. Specifically, the server acquires the social media data through an API and saves it in a database for analysis.

[1598] Step 2:

[1599] The server analyzes the acquired data using natural language processing algorithms to detect offensive messages. The input is the saved timeline data, and the output is a list of detected offensive messages. Specifically, it uses a natural language processing library to perform sentiment analysis of the text data and flag offensive messages.

[1600] Step 3:

[1601] The server applies an emotion recognition engine to recognize the user's emotional state when the number of offensive messages exceeds a certain threshold. The input is a list of offensive messages, and the output is the user's emotional state data. Specifically, the server starts the emotion recognition engine and evaluates the user's emotions in real time.

[1602] Step 4:

[1603] The server uses a generative AI model to generate an advocacy message based on the emotional state data. The input is the emotional state data, and the output is the generated advocacy message. Specifically, the generative AI model is used to create a message that is in tune with the user's emotions. An example of a prompt sentence is, "If the emotion recognition engine recognizes that the stress level is high, generate an appropriate encouraging message for the worker. For example, something like, 'Your efforts are appreciated. Please take a short break.'"

[1604] Step 5:

[1605] The server starts a means to distribute the generated advocacy message to the collaborator's terminal. The input is the generated advocacy message, and the output is the message sent to the collaborator's terminal. The specific operation is to distribute the message to the collaborator via an API or a notification system.

[1606] Step 6:

[1607] The collaborator's terminal saves the received advocacy message and displays it on the user interface to notify the collaborator. The input is the advocacy message received from the server, and the output is the message displayed on the user interface. The specific operation is to use the terminal's notification system to notify the collaborator that there is a new message.

[1608] Step 7:

[1609] The collaborator user checks the defense message notified on their device and posts it to the SNS. The input is the received defense message, and the output is the message posted to the SNS. Specifically, when the collaborator presses the post button on their device, the message is published through the SNS API.

[1610] Step 8:

[1611] The server acquires video data from cameras in the factory and uses an emotion recognition engine to recognize the emotional state of workers in real time. The input is video data from the cameras, and the output is data on the emotional state of the workers. Specifically, the server analyzes the video data, and the emotion recognition engine detects stress and fatigue.

[1612] Step 9:

[1613] The server generates appropriate support messages and actions based on the emotional state data and notifies the worker's device. The input is the emotional state data, and the output is the support message. Specifically, the server uses the generative AI model to create a message encouraging the worker to take a break and displays it on the device.

[1614] Step 10:

[1615] The worker's device displays the received support message and notifies the worker. The input is the generated support message, and the output is the message displayed on the device. Specifically, the message is displayed on the device's display or through the notification system, prompting the worker to take appropriate action.

[1616] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1617] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1618] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1619] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1620] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1621] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1622] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1623] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1624] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1625] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1626] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1627] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1628] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1629] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1630] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1631] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1632] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1633] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1634] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1635] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1636] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1637] The following is further disclosed regarding the above embodiment.

[1638] (Claim 1)

[1639] a means for detecting offensive messages directed to an account on the information terminal;

[1640] means for generating a defensive message when the offensive message exceeds a certain threshold;

[1641] means for distributing the generated advocacy message to the information terminals of the collaborators;

[1642] A means for a collaborator to post a support message on a social networking service via an information terminal;

[1643] A system including:

[1644] (Claim 2)

[1645] 10. The system of claim 1, wherein the means for generating an advocacy message automatically generates the advocacy message using a generative model.

[1646] (Claim 3)

[1647] 10. The system of claim 1, wherein the means for detecting offensive messages uses a natural language processing algorithm to identify offensive content and negative sentiment.

[1648] "Example 1"

[1649] (Claim 1)

[1650] means for monitoring an account of the information processing device for offensive messages;

[1651] means for generating a defensive message when the offensive message exceeds a certain threshold;

[1652] means for distributing the generated advocacy message to the collaborator's information processing device;

[1653] a means for a collaborator to post an advocacy message to the online platform via an information processing device;

[1654] A system including:

[1655] (Claim 2)

[1656] 2. The system of claim 1, wherein the means for generating an advocacy message automatically creates the advocacy message using a generative AI model.

[1657] (Claim 3)

[1658] 10. The system of claim 1, wherein the means for monitoring for offensive messages identifies offensive content and negative sentiment based on natural language processing algorithms.

[1659] "Application Example 1"

[1660] (Claim 1)

[1661] a means for detecting offensive messages directed to an account on the information terminal;

[1662] means for generating a defensive message when the offensive message exceeds a certain threshold;

[1663] means for distributing the generated advocacy message to the information terminals of the collaborators;

[1664] A means for a collaborator to post a support message on a social networking service via an information terminal;

[1665] A way to monitor social media accounts in real time,

[1666] means for notifying the generated advocacy message;

[1667] A way for users who receive notifications to post a message with one click after checking.

[1668] A system including:

[1669] (Claim 2)

[1670] 2. The system of claim 1, wherein the means for generating an advocacy message automatically creates the advocacy message using a generative AI model.

[1671] (Claim 3)

[1672] 10. The system of claim 1, wherein the means for detecting offensive messages uses a natural language processing algorithm to identify offensive content and negative sentiment.

[1673] "Example 2: Combining Emotion Engines"

[1674] (Claim 1)

[1675] a means for detecting offensive messages directed to an account on the information terminal;

[1676] means for generating a defensive message when the offensive message exceeds a certain threshold;

[1677] means for distributing the generated advocacy message to the information terminals of the collaborators;

[1678] A means for a collaborator to post a support message on a social networking service via an information terminal;

[1679] means for recognizing a user's emotional state in real time;

[1680] means for adjusting the content of the advocacy message based on the emotional state of the user obtained in real time;

[1681] A system including:

[1682] (Claim 2)

[1683] 10. The system of claim 1, wherein the means for generating an advocacy message automatically generates the advocacy message using a generative model.

[1684] (Claim 3)

[1685] 10. The system of claim 1, wherein the means for detecting offensive messages uses a natural language processing algorithm to identify offensive content and negative sentiment.

[1686] "Application example 2 when combining emotion engines"

[1687] (Claim 1)

[1688] a means for detecting offensive messages directed to an account on the information terminal;

[1689] means for generating a defensive message when the offensive message exceeds a certain threshold;

[1690] means for distributing the generated advocacy message to the information terminals of the collaborators;

[1691] A means for a collaborator to post a support message on a social networking service via an information terminal;

[1692] A means for capturing video of the work environment and recognizing the emotional state in real time;

[1693] A means to deliver appropriate support messages and actions based on emotional state;

[1694] A system including:

[1695] (Claim 2)

[1696] 10. The system of claim 1, wherein the means for generating an advocacy message automatically generates the advocacy message using a generative model.

[1697] (Claim 3)

[1698] 10. The system of claim 1, wherein the means for detecting offensive messages uses a natural language processing algorithm to identify offensive content and negative sentiment. [Explanation of symbols]

[1699] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a means for detecting offensive messages directed to an account on the information terminal; means for generating a defensive message when the offensive message exceeds a certain threshold; means for distributing the generated advocacy message to the information terminals of the collaborators; A means for a collaborator to post a support message on a social networking service via an information terminal; A system including:

2. 10. The system of claim 1, wherein the means for generating an advocacy message automatically generates the advocacy message using a generative model.

3. 10. The system of claim 1, wherein the means for detecting offensive messages uses a natural language processing algorithm to identify offensive content and negative sentiment.

Images