System

Abstract

A system is provided.SOLUTION: A system for automatically monitoring and managing safety of communication in a metaverse, comprising: means for collecting chat messages; means for analyzing the collected chat messages using a natural language processing model and detecting inappropriate words and expressions; means for automatically changing and displaying a corresponding message when an inappropriate word or expression is detected; means for collecting and analyzing an inappropriate image and deleting the detected inappropriate image; and means for converting collected voice data into text using a voice recognition model, the system includes a means for analyzing by a natural language processing model and detecting an inappropriate utterance, a means for transparently humanizing a speaker of a corresponding voice when the inappropriate voice is detected, and a means for notifying a user of the detected inappropriate content or action.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] As the metaverse becomes more widespread, "metahara," or slander and impersonation, has become a serious problem. While there is a demand for a safe and secure environment for users in the metaverse, the current lack of a legal system within the metaverse makes effective monitoring and countermeasures difficult. In particular, with the existence of multiple communication methods, such as chat, images, and voice, a monitoring and countermeasure system suited to each is required. Against this background, the present invention aims to provide a system that can detect inappropriate content for different communication methods within the metaverse and automatically deal with it. [Means for solving the problem]

[0005] The present invention is a system for automatically monitoring and managing the security of communications within the metaverse, which includes the following means:

[0006] 1. A means of collecting chat messages and a means of analyzing the collected chat messages using a natural language processing model to detect inappropriate words and expressions.

[0007] 2. A means of automatically modifying and displaying the appropriate message when inappropriate words or expressions are detected.

[0008] 3. A means of collecting and analyzing inappropriate images and removing any detected inappropriate images.

[0009] 4. A means of converting collected voice data into text using a speech recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks.

[0010] 5. A means to make the speaker of inappropriate audio invisible when it is detected.

[0011] 6. A means of notifying users of detected inappropriate content or behavior.

[0012] These measures allow appropriate measures to be taken immediately even if meta-harassment occurs, providing users with a safe and secure metaverse environment.

[0013] The "metaverse" is a three-dimensional virtual space that uses virtual reality and augmented reality, a digital environment where users can interact with other users using avatars.

[0014] "Metahara" refers to harassment such as slander and impersonation that occurs within the metaverse.

[0015] "Generative AI" is a system that uses artificial intelligence technology to analyze data and generate new content.

[0016] "Natural language processing" is a technology that uses computers to process, understand, analyze, and generate human language.

[0017] "Computer vision" is a technology that allows computer systems to acquire and analyze information from images and videos.

[0018] A "speech recognition model" is a machine learning model for analyzing voice data and converting it into text format.

[0019] "Invisibility" is a technique that makes a specific user invisible to other users, preventing them from recognizing their presence.

[0020] "Inappropriate content" refers to words, expressions, images, sounds, and other content that may be offensive or harmful to users in communications within the metaverse.

[0021] "Analysis queue" refers to a list of data awaiting their turn for analysis.

[0022] A "feedback report" is a report that shows evaluations of user behavior and content and areas for improvement. [Brief explanation of the drawings]

[0023] [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

[0024] 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.

[0025] First, the terms used in the following description will be explained.

[0026] 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).

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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."

[0031] [First embodiment]

[0032] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0033] 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.

[0034] 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).

[0035] 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.

[0036] 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.

[0037] 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.

[0038] 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.

[0039] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0040] 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.

[0041] 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.

[0042] 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.

[0043] 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."

[0044] The present invention relates to a system for automatically monitoring and managing the safety of communications within the metaverse. To implement this system, the following programs and processing means are included:

[0045] System configuration

[0046] 1. Data Collection Module

[0047] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[0048] 2. Data Analysis Module

[0049] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[0050] The server inputs the received image data into a computer vision model to detect inappropriate content.

[0051] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[0052] 3. Supported modules

[0053] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0054] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[0055] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0056] 4. Notification Module

[0057] The server notifies the affected users about any inappropriate content or behavior detected.

[0058] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[0059] Specific examples

[0060] Example 1: Monitoring chat messages

[0061] User A sends a chat message in the metaverse saying "I hope you die."

[0062] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0063] The server automatically converts this to "Shut up, you idiot" and displays it.

[0064] Example 2: Image monitoring

[0065] User B uploads an inappropriate image (sexually explicit image).

[0066] The terminal sends this image to the server.

[0067] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0068] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0069] Example 3: Monitoring voice activity

[0070] User C shouts insults such as "You idiot" in the voice chat.

[0071] The terminal transmits the voice data to the server.

[0072] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0073] The server makes user C invisible, making him invisible to other users.

[0074] This system will make communication within the metaverse safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[0075] The processing flow will be explained below.

[0076] Step 1:

[0077] A user types and sends a chat message within the metaverse.

[0078] The terminal collects the sent chat messages in real time and transmits them to the server.

[0079] Step 2:

[0080] A user uploads screenshots or image data within the metaverse.

[0081] The terminal transmits the uploaded image data to the server.

[0082] Step 3:

[0083] A user initiates a voice chat in the metaverse.

[0084] The device captures audio data in real time and sends it to the server.

[0085] Step 4:

[0086] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[0087] The server detects inappropriate words and expressions.

[0088] Step 5:

[0089] The image data received by the server is input into a computer vision model and the content is analyzed.

[0090] The server detects inappropriate content.

[0091] Step 6:

[0092] The server inputs the received voice data into a voice recognition model and converts it into text.

[0093] The server inputs the converted data into text into an NLP model and analyzes the content.

[0094] The server detects inappropriate comments.

[0095] Step 7:

[0096] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0097] Example: Converting the message "die" to "shut up."

[0098] Step 8:

[0099] The server immediately deletes any inappropriate image data that is detected.

[0100] The server sends a notification to the user.

[0101] Step 9:

[0102] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[0103] The fact that the server has become invisible is reflected to users other than the speaker.

[0104] Step 10:

[0105] The server will notify the user of any inappropriate content or violating behavior detected.

[0106] The server provides a detailed feedback report to the user.

[0107] Example 1

[0108] 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."

[0109] In the metaverse, inappropriate comments, images, and audio are shared between users, potentially compromising user safety and comfort. Conventional systems lack effective monitoring and management methods, and it takes time for inappropriate content to be detected, potentially adversely affecting users during that time. Therefore, there is a need for a system that can monitor communication in the metaverse in real time and respond immediately.

[0110] 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.

[0111] In this invention, the server includes means for collecting communication data from user devices, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for analyzing the collected image data using a computer vision model to detect and delete inappropriate content, means for converting the collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate remarks, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, and means for notifying the user of detected inappropriate content or behavior. This makes it possible to monitor communication within the metaverse in real time and respond immediately to inappropriate content or behavior.

[0112] A "user device" is an electronic device used by a user to access and communicate with the metaverse.

[0113] "Communication data" refers to information such as chat messages, image data, and audio data exchanged between users.

[0114] A "natural language processing model" is an artificial intelligence model for analyzing and understanding text data.

[0115] A "chat message" is a message sent in text format by a user within the metaverse.

[0116] "Inappropriate words and phrases" are words or phrases that may be offensive, insulting, or upsetting to a user in a communication.

[0117] A "computer vision model" is an artificial intelligence model for analyzing and understanding image data.

[0118] "Inappropriate Content" refers to images or videos that disrupt healthy communication within the Metaverse and may cause mental or emotional harm to other users.

[0119] A "voice recognition model" is an artificial intelligence model for converting voice data into text data.

[0120] "Invisibility" is a process in which the speaker or actor of detected inappropriate behavior is made invisible to other users.

[0121] A "notification" is a message or alert that informs a user that a particular action or statement is inappropriate.

[0122] This invention relates to a system that automatically monitors and manages the safety of communications within the metaverse. The system collects and analyzes chat messages, image data, and audio data exchanged between users in real time, and aims to immediately detect and respond to inappropriate content.

[0123] System configuration

[0124] To implement this system, the following hardware and software are used:

[0125] Hardware

[0126] 1. User's device

[0127] Communication-enabled devices such as personal computers, smartphones, and tablets

[0128] These devices are used to access the metaverse server and send and receive data such as chat, audio, and images.

[0129] 2. Server

[0130] High-performance server computers

[0131] It has the computing power necessary to analyze chat messages, perform voice recognition, and image analysis.

[0132] software

[0133] 1. Natural Language Processing Model (NLP Model)

[0134] Examples: BERT, GPT-3

[0135] It is used to analyze chat messages and detect inappropriate words and phrases.

[0136] 2. Computer Vision Models

[0137] Examples: ResNet, YOLO

[0138] Used to analyze image data and detect inappropriate content.

[0139] 3. Speech Recognition Model

[0140] Examples: Google Cloud Speech-to-Text, DeepSpeech

[0141] It is used to convert audio data into text format.

[0142] Example of operation

[0143] Example 1: Monitoring chat messages

[0144] User A sends a chat message in the metaverse saying "I hope you die."

[0145] The terminal immediately sends this message to the server.

[0146] The server inputs this message into a natural language processing model, which analyzes and detects that the word "die" is inappropriate.

[0147] The server automatically converts the message to "Shut up you" and displays it in an appropriate format.

[0148] Example 2: Image monitoring

[0149] User B uploads an inappropriate image (e.g., a sexually explicit image).

[0150] The terminal transmits this image to the server.

[0151] The server inputs the received images into a computer vision model to analyze and detect any inappropriate content.

[0152] The server immediately deletes the image and notifies User B that "inappropriate content has been posted."

[0153] Example 3: Monitoring voice activity

[0154] User C shouts insults such as "You idiot" in the voice chat.

[0155] The terminal transmits the voice data to the server.

[0156] The server converts the voice data into text using a voice recognition model and analyzes it using a natural language processing model.

[0157] The server recognizes the word "idiot" as inappropriate and turns User C invisible, making him invisible to other users.

[0158] Prompt Sentence Examples

[0159] prompt:

[0160] If I send a chat message in the Metaverse like "I hope you die," how does the system handle it?

[0161] response:

[0162] 1. The device sends a chat message saying "You should die" to the server in real time.

[0163] 2. The server analyzes the message using a natural language processing model and detects that the word "die" is inappropriate.

[0164] 3. The server automatically converts the message to "Shut up, you idiot" and displays it.

[0165] 4. The server notifies User A of the revised message and informs him that the original message was inappropriate.

[0166] This system collects and analyzes data in real time and responds immediately to maintain healthy communication within the metaverse, providing a safe and secure environment for users.

[0167] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0168] Step 1: Data collection

[0169] The terminal collects communication data from the user's device. The input data includes chat messages, image data, and voice data. The collected data is temporarily stored in memory and then the next step is performed.

[0170] Specific behavior:

[0171] The user types "I wish you would die" into the chat window and presses the send button.

[0172] A user says "You idiot" in voice chat.

[0173] A user posts an image using the image upload feature.

[0174] Step 2: Send data

[0175] The terminal sends the collected data to the server in real time. The input data is the collected communication data, which is sent directly to the server via the network. The server adds the received data to a processing queue.

[0176] Specific behavior:

[0177] The device sends the chat message "You should die" data to the server.

[0178] The device sends the recorded audio data to the server.

[0179] The terminal transmits the uploaded image file to the server.

[0180] Step 3: Analyzing chat messages

[0181] The server inputs the received chat message into a natural language processing model (NLP model) to analyze the meaning and keywords. The input data is the received chat message, and the output is the judgment result of whether it contains inappropriate words or expressions.

[0182] Specific behavior:

[0183] The server inputs the chat message "You should die" into the NLP model.

[0184] The server analyzes the word "die" as inappropriate and obtains the judgment result.

[0185] Step 4: Analyze the images

[0186] The server inputs the received image data into a computer vision model to detect inappropriate content. The input data is the received image data, and the output is a judgment result on whether the image is inappropriate or not.

[0187] Specific behavior:

[0188] The server inputs the received image data into a computer vision model.

[0189] The server uses image analysis models to detect explicit elements in the image.

[0190] Step 5: Analyze the audio data

[0191] The server inputs the received voice data into a voice recognition model, converts it into text format, and then analyzes it using a natural language processing model. The input data is the received voice data, and the output is a judgment result on whether or not it contains inappropriate remarks.

[0192] Specific behavior:

[0193] The server inputs the voice data into a speech recognition model and converts it into text.

[0194] The server inputs the converted text data into an NLP model and recognizes the word "idiot" as inappropriate.

[0195] Step 6: Implementing response measures

[0196] The server takes appropriate action based on the analysis results. The input data is the analysis results, and if any inappropriate content is detected, the server takes appropriate action. The output is an appropriately modified message, image deletion, user invisibility, etc.

[0197] Specific behavior:

[0198] The server converts "You should die" to "Shut up" and displays it.

[0199] The server deletes inappropriate image files.

[0200] The server makes user C invisible, making him invisible to other users.

[0201] Step 7: Notification

[0202] The server sends notifications to the relevant users about the detected inappropriate behavior. The input data is the result of the action taken, and the output is a notification message to the user.

[0203] Specific behavior:

[0204] The server notifies User A that "Your comment, 'You should die,' has been deemed inappropriate."

[0205] The server warns User B that the posted image has been deleted because it is inappropriate.

[0206] The server notifies User C that "Your comment 'You idiot' has been deemed inappropriate and made invisible."

[0207] (Application example 1)

[0208] 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."

[0209] Communication between users in the metaverse or virtual store may not always be secure. The free circulation of inappropriate comments, images, and audio not only impairs the user experience, but also risks reducing the reliability of the platform itself. Real-time communication, in particular, requires immediate response, necessitating efficient monitoring and management systems. Providing appropriate feedback to users in the event of inappropriate behavior is also important from the perspective of preventing recurrence and user education. Therefore, the present invention aims to provide a system that automatically monitors and manages the security of communication in the metaverse or virtual store, and provides appropriate responses and feedback.

[0210] 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.

[0211] In this invention, the server includes means for collecting chat messages, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for collecting and analyzing inappropriate images and deleting the detected inappropriate images, means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, means for notifying the user of detected inappropriate content or behavior, means for providing the user with a detailed feedback report on the detected inappropriate content, and means for monitoring communication within the virtual store in real time and taking specific measures immediately. This enables safe and smooth communication between users within the metaverse and virtual store, thereby improving the reliability of the platform.

[0212] The "metaverse" is a three-dimensional virtual space built on the Internet, a platform where users can interact with each other and engage in activities through avatars.

[0213] A "virtual store" is a virtual store that mimics a real store, where users can purchase digital items and services and interact with other users.

[0214] "Chat messages" are messages exchanged between users in real time in text format.

[0215] A "natural language processing model" is an artificial intelligence technology for understanding and analyzing human language, extracting meaning from text data.

[0216] "Inappropriate words and phrases" are words or phrases that contain offensive, discriminatory or objectionable content and are inappropriate for communication.

[0217] To "collect" means to obtain specific data or information and gather it in a designated location.

[0218] "Analysis" means examining data or information in detail to understand and interpret its content.

[0219] A "speech recognition model" is an artificial intelligence technology for converting voice data into text, recognizing speech as text information.

[0220] "Invisibility" refers to the process of making a specific user invisible to other users.

[0221] A "feedback report" is a document that provides detailed information about inappropriate behavior or content and guidelines for improvement.

[0222] "Real-time monitoring" refers to the process of instantly monitoring data and behavior and taking necessary action on the spot.

[0223] "Responding immediately" means immediately taking appropriate action or measures against any inappropriate behavior or content that is detected.

[0224] The present invention relates to a system for automatically monitoring and managing communications within a metaverse or virtual store. To implement this system, the following programs and processing means are included:

[0225] System configuration

[0226] 1. Data Collection Module

[0227] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[0228] 2. Data Analysis Module

[0229] The server inputs chat messages into a natural language processing (NLP) model using a generative AI model to perform semantic analysis and keyword extraction.

[0230] The server inputs the received image data into a computer vision model (e.g., OpenCV) to detect inappropriate content.

[0231] The server inputs the collected audio data into a speech recognition model (e.g., a Transformers speech recognition model) and converts it into text format, which is then analyzed for inappropriate speech using the same NLP model.

[0232] 3. Supported modules

[0233] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0234] The server detects inappropriate images, immediately deletes them, and notifies the user with a warning.

[0235] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0236] 4. Notification Module

[0237] The server sends notifications to the affected users about any inappropriate content or behavior detected.

[0238] The server provides the user with a detailed feedback report of any inappropriate content detected.

[0239] Specific examples

[0240] Example 1: Monitoring chat messages

[0241] User A sends a chat message in the metaverse saying "I hope you die."

[0242] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0243] The server automatically converts this to "Shut up, you idiot" and displays it.

[0244] Example 2: Image monitoring

[0245] User B uploads an inappropriate image (sexually explicit image).

[0246] The terminal sends this image to the server.

[0247] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0248] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0249] Example 3: Monitoring voice activity

[0250] User C shouts insults such as "You idiot" in the voice chat.

[0251] The terminal transmits the voice data to the server.

[0252] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0253] The server makes user C invisible, making him invisible to other users.

[0254] Prompt Sentence Examples

[0255] "Determine whether a particular chat message is inappropriate. Example: 'I hope you die.'"

[0256] "Please decide if this image is inappropriate. Here is the image data: [base64 encoded image data]"

[0257] This system will make communication within the metaverse and virtual stores safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[0258] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0259] Step 1:

[0260] Data collection

[0261] The device collects chat messages, uploaded image data, and voice chat data in real time, and transmits this data to a server.

[0262] Input: User chat messages, image data, and audio data

[0263] Output: Various data sent to the server

[0264] Step 2:

[0265] Chat message analysis

[0266] The server inputs the collected chat messages into a natural language processing (NLP) model using a generative AI model, where semantic analysis and keyword extraction are performed to detect inappropriate words and expressions.

[0267] Input: The chat message the user said

[0268] Data processing: Semantic analysis and keyword extraction using NLP models

[0269] Output: A list of inappropriate words and phrases

[0270] Step 3:

[0271] Correcting inappropriate messages

[0272] The server automatically converts chat messages that are deemed inappropriate into appropriate language and displays them accordingly, for example, changing the word "die" to "shut up."

[0273] Input: A list of inappropriate words and phrases

[0274] Data processing: Automatically convert inappropriate words and expressions into appropriate ones

[0275] Output: Modified chat message

[0276] Step 4:

[0277] Image data analysis

[0278] The server then feeds the collected image data into computer vision models to detect inappropriate content, for example, using libraries such as OpenCV to identify sexually explicit images.

[0279] Input: User uploaded image data

[0280] Data processing: Image analysis using computer vision models

[0281] Output: Result of whether or not inappropriate content is included

[0282] Step 5:

[0283] Inappropriate image removal and notification

[0284] The server immediately deletes image data that is detected as inappropriate and notifies the user of a warning.

[0285] Input: Inappropriate content detection result

[0286] Data processing: Deletion of image data and creation of warning notices

[0287] Output: Deleted image data and a warning notice to the user

[0288] Step 6:

[0289] Analysis of audio data

[0290] The server inputs the collected voice data into a speech recognition model, converts it into text format, and then analyzes it for inappropriate comments using an NLP model.

[0291] Input: User voice chat data

[0292] Data processing: Converting voice data into text using a speech recognition model and analyzing it using an NLP model

[0293] Output: Translated audio data and a list of inappropriate words

[0294] Step 7:

[0295] Making inappropriate audio speakers invisible

[0296] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0297] Input: A list of inappropriate words

[0298] Data processing: Making users invisible

[0299] Output: Invisible user

[0300] Step 8:

[0301] Providing detailed feedback reports

[0302] The server provides the user with a detailed feedback report of the inappropriate content and behavior detected, including specific details of the behavior that was inappropriate.

[0303] Input: Data on inappropriate content and behavior detected

[0304] Data processing: Creating detailed feedback reports

[0305] Output: A feedback report provided to the user

[0306] 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.

[0307] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and combines an emotion engine that recognizes user emotions to provide more advanced responses. To implement this system, the following programs and processing means are included.

[0308] System configuration

[0309] 1. Data Collection Module

[0310] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[0311] 2. Data Analysis Module

[0312] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[0313] The server inputs the received image data into a computer vision model to detect inappropriate content.

[0314] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[0315] 3. Sentiment Analysis Module

[0316] The server inputs the collected data into an emotion engine to analyze the user's emotions.

[0317] The server enhances the detection of inappropriate content based on the results of sentiment analysis.

[0318] 4. Supported modules

[0319] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0320] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[0321] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0322] If the emotion engine recognizes a negative emotion, it provides mental support to the user.

[0323] 5. Notification Module

[0324] The server will then notify the user of any inappropriate content or behavior that has been detected.

[0325] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[0326] Specific examples

[0327] Example 1: Monitoring chat messages and sentiment

[0328] User A sends a chat message in the metaverse saying "I hope you die."

[0329] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0330] The server automatically converts this to "Shut up, you idiot" and displays it.

[0331] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[0332] Example 2: Image and Emotion Monitoring

[0333] User B uploads an inappropriate image (sexually explicit image).

[0334] The terminal sends this image to the server.

[0335] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0336] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0337] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[0338] Example 3: Monitoring voice utterances and emotions

[0339] User C shouts insults such as "You idiot" in the voice chat.

[0340] The terminal transmits the voice data to the server.

[0341] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0342] The server makes user C invisible, making him invisible to other users.

[0343] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[0344] This system will not only make communication within the metaverse safer and more comfortable, but will also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[0345] The processing flow will be explained below.

[0346] Step 1:

[0347] A user types and sends a chat message within the metaverse.

[0348] The terminal collects the sent chat messages in real time and transmits them to the server.

[0349] Step 2:

[0350] A user uploads screenshots or image data within the metaverse.

[0351] The terminal transmits the uploaded image data to the server.

[0352] Step 3:

[0353] A user initiates a voice chat in the metaverse.

[0354] The device captures audio data in real time and sends it to the server.

[0355] Step 4:

[0356] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[0357] The server detects inappropriate words and expressions.

[0358] Step 5:

[0359] The image data received by the server is input into a computer vision model and the content is analyzed.

[0360] The server detects inappropriate content.

[0361] Step 6:

[0362] The server inputs the received voice data into a voice recognition model and converts it into text.

[0363] The server inputs the converted data into text into an NLP model and analyzes the content.

[0364] The server detects inappropriate comments.

[0365] Step 7:

[0366] The server inputs the collected data into an emotion engine to analyze the user's emotional state.

[0367] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[0368] Step 8:

[0369] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0370] Example: Converting the message "die" to "shut up."

[0371] At the same time, the server uses an emotion engine to evaluate the emotional state of user A, and if negative emotions are detected, provides mental support.

[0372] Step 9:

[0373] The server immediately deletes any inappropriate image data that is detected.

[0374] The server sends a notification to the user.

[0375] The server uses an emotion engine to evaluate User B's emotional state and considers an appropriate response.

[0376] Step 10:

[0377] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[0378] The fact that the server has become invisible is reflected to users other than the speaker.

[0379] The server uses an emotion engine to assess the emotional state of user C and provides support as needed.

[0380] Step 11:

[0381] The server notifies the user of any inappropriate content or behavior detected.

[0382] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[0383] Example 2

[0384] 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."

[0385] While communication within the metaverse is increasing, there is a risk that inappropriate words and actions may exist, preventing users from interacting safely in a healthy environment. There is a need for a system that can respond quickly and effectively to inappropriate words and actions and provide feedback that takes into account the user's emotional state.

[0386] 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.

[0387] In this invention, the server

[0388] A means for collecting communication data (chat messages, images, audio) sent by users within the metaverse;

[0389] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[0390] A means for automatically changing and displaying detected inappropriate words and expressions;

[0391] A means for analyzing the collected image data using a computer vision model to detect and remove inappropriate images;

[0392] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[0393] A means for making the speaker of the detected inappropriate voice invisible;

[0394] A means for inputting the collected communication data into an emotion analysis engine and analyzing the user's emotions;

[0395] A means for detecting negative emotions based on the emotion analysis results and providing mental support;

[0396] a means for notifying users of detected inappropriate content or behavior;

[0397] and a means for generating detailed feedback reports and providing them to users. This allows appropriate management of inappropriate speech and behavior within the metaverse, enabling users to interact with peace of mind in a healthy and safe communication environment.

[0398] "Communication data" is a general term that includes chat messages, image data, and voice data sent by users within the metaverse.

[0399] A "natural language processing model" is a computer algorithm or machine learning model that analyzes text data and understands and extracts grammar, context, and keywords.

[0400] A "computer vision model" is an algorithm or deep learning network that analyzes image data to detect inappropriate content or specific features.

[0401] A "voice recognition model" is a technology for converting voice data into text data and analyzing the content of speech.

[0402] An "emotion analysis engine" is an algorithm or AI technology that analyzes a user's text, image, and voice data to identify and evaluate their emotional state.

[0403] "Mental support" refers to the means and services that provide psychological and emotional care and assistance in response to negative emotions detected through analysis.

[0404] "Invisibility" is a function that restricts the actions of a user who has behaved inappropriately by making that user invisible to other users.

[0405] "Feedback Report" means a written or digital message that provides detailed explanation and advice regarding detected inappropriate behavior or content.

[0406] "Inappropriate content" refers to chat messages, images, audio, etc. within the metaverse that are deemed socially or ethically problematic.

[0407] "User notification" is a function that notifies the relevant user about inappropriate behavior or content detected by the system.

[0408] MODE FOR CARRYING OUT THE INVENTION

[0409] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and provides advanced responses that take into account the emotional state of the user. To implement this system, the following programs and processing means are included.

[0410] System Configuration

[0411] 1. Data Collection Module

[0412] The terminal collects chat messages, image data, and voice data sent by users within the metaverse in real time and transmits them to a server.

[0413] 2. Data Analysis Module

[0414] The server inputs the collected chat messages into a natural language processing (NLP) model to perform semantic analysis and keyword extraction, using Python's NLTK and spaCy.

[0415] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content.

[0416] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API), converts it into text, and then analyzes the text data with an NLP model to detect inappropriate speech.

[0417] 3. Sentiment Analysis Module

[0418] The server inputs the collected data into an emotion engine (e.g., Microsoft Azure emotion analysis API, snownlp, etc.) to analyze the user's emotions.

[0419] Based on the results of sentiment analysis, we will strengthen the detection of inappropriate content and consider taking action as necessary.

[0420] 4. Supported modules

[0421] The server automatically converts detected inappropriate chat messages into appropriate expressions, immediately deletes inappropriate images, and notifies users with a warning. If inappropriate voice comments are detected, the person making the comment becomes invisible to other users.

[0422] If the emotion engine recognizes a negative emotion, the server provides mental support to the user.

[0423] 5. Notification Module

[0424] The server notifies users of any inappropriate content or behavior detected and provides detailed feedback reports to indicate which actions were inappropriate.

[0425] Specific examples

[0426] Example 1: Monitoring chat messages and sentiment

[0427] User A sends a chat message in the metaverse saying "I hope you die."

[0428] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0429] The server automatically converts this to "Shut up, you idiot" and displays it.

[0430] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[0431] Example 2: Image and Emotion Monitoring

[0432] User B uploads an inappropriate image (sexually explicit image).

[0433] The terminal sends this image to the server.

[0434] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0435] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0436] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[0437] Example 3: Monitoring voice utterances and emotions

[0438] User C shouts insults such as "You idiot" in the voice chat.

[0439] The terminal transmits the voice data to the server.

[0440] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0441] The server makes user C invisible, making him invisible to other users.

[0442] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[0443] This system will not only make communication within the metaverse safer and more comfortable, but will also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[0444] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0445] Step 1: Data collection

[0446] The terminal collects chat messages, image data, and voice data sent by users in the metaverse in real time. This data is input and sent to a server via a network.

[0447] Specific behavior:

[0448] The user types "Hello" in the chat window and presses the send button.

[0449] The device captures the text data "Hello" and sends it to the server.

[0450] Users take screenshots of their avatars and upload them to a shared gallery in the metaverse.

[0451] The device captures the screenshot image data and sends it to the server.

[0452] Step 2: Chat message analysis

[0453] The server inputs the received chat messages into a natural language processing (NLP) model to extract keywords and perform semantic analysis. The input is the chat message, and the output is the analysis result. Specifically, Python's NLTK and spaCy are used.

[0454] Specific behavior:

[0455] The server receives the message "Hello" and asks the natural language processing model to analyze it.

[0456] The NLP model tags "hello" as a greeting and returns the result to the server.

[0457] Step 3: Image data analysis

[0458] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content. The input is the image data, and the output is the detection result of inappropriate content.

[0459] Specific behavior:

[0460] The server receives the screenshots uploaded by the user.

[0461] Computer vision models scan images to determine if they contain inappropriate content.

[0462] The server receives the inappropriate content detection result and removes the image.

[0463] Step 4: Audio data analysis

[0464] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API) and converts it into text format. The text data is then analyzed using an NLP model to detect inappropriate speech. The input is the audio data, and the output is the analyzed text data.

[0465] Specific behavior:

[0466] A user says "You're such an idiot" in voice chat.

[0467] The device captures the audio data and sends it to the server.

[0468] The server converts the voice data into text and generates the text data "You're an idiot."

[0469] The NLP model determines that "baka" is an inappropriate word and returns the result to the server.

[0470] Step 5: Sentiment Analysis

[0471] The server inputs the collected text, image, and voice data into an emotion engine (e.g., Microsoft Azure's emotion analysis API, snownlp, etc.) to analyze the user's emotions. The input is data (text, image, voice), and the output is an evaluation of the user's emotional state.

[0472] Specific behavior:

[0473] The server inputs the chat message "You're such an idiot" into an emotion engine, which analyzes the emotional state of the message.

[0474] The emotion engine detects "anger" and returns the result to the server.

[0475] Step 6: Address inappropriate content

[0476] The server automatically takes appropriate action as needed based on the results of the sentiment analysis. The inputs are the detection results of inappropriate content and the sentiment analysis results, and the outputs are the modified message, the deletion of inappropriate content, and the results of making the user invisible.

[0477] Specific behavior:

[0478] The server automatically converts "You're an idiot" to "Shut up you" and displays it to the sending user.

[0479] If an inappropriate image is detected, it will be removed and a warning message will be sent to the uploader.

[0480] The server turns users who make inappropriate vocal comments into invisible people, making them invisible to other users.

[0481] Step 7: Providing mental support

[0482] The server provides mental support to the user when a negative emotion is recognized by the emotion engine. The input is the emotion analysis result, and the output is the support provided.

[0483] Specific behavior:

[0484] If the server recognizes "anger," it provides the user with a message explaining how to relax and a link to a counseling service.

[0485] Step 8: User Notification

[0486] The server notifies the user of any detected inappropriate content or behavior and also generates and provides a detailed feedback report. The input is the detection result of inappropriate content, and the output is the user notification and feedback report.

[0487] Specific behavior:

[0488] The server sends a notification to the user about the detected inappropriate content.

[0489] The report states, "Your comment, 'You're an idiot,' was deemed inappropriate. Please be careful next time."

[0490] (Application example 2)

[0491] 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."

[0492] In current brick-and-mortar stores, communication between customers and store staff can be inappropriate, and customers' feelings can be poorly understood, leading to delayed responses. This can lead to lower customer satisfaction, which can have a negative impact on store ratings and sales. Furthermore, because it is difficult to respond appropriately in real time, advanced management of the customer experience in brick-and-mortar stores is required.

[0493] The identification processing 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 collecting chat messages; means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions; means for automatically changing and displaying the corresponding message when inappropriate words or expressions are detected; means for collecting and analyzing inappropriate images and deleting the detected inappropriate images; means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments; means for making the speaker of the corresponding voice invisible when inappropriate voice is detected; means for notifying the user of detected inappropriate content or behavior; means for analyzing the user's emotions, including an emotion analysis engine; means for providing an appropriate response when the user's emotions are recognized as negative; and means for monitoring conversations and behaviors between customers and store clerks in a physical store and notifying the store clerk in real time based on the emotion analysis results. This enables real-time monitoring of customer experiences in physical stores and enables prompt and appropriate responses.

[0494] The "metaverse" is a virtual space that exists on the Internet, a virtual reality environment in which users can interact using avatars.

[0495] "Communication safety" refers to the standards and measures to ensure that interactions between users are appropriate and healthy.

[0496] "Monitoring" refers to the act of watching users' behavior and messages in real time to detect inappropriate behavior or expressions.

[0497] "Management" refers to the act of taking appropriate action against detected inappropriate behavior or content and operating the entire system.

[0498] A "chat message" is a communication message sent by a user in text format.

[0499] "Means of collection" refers to the equipment or software mechanisms used to collect data sent by users on a server.

[0500] A "natural language processing model" refers to the algorithms and technologies that allow computers to understand and analyze human language.

[0501] "Inappropriate words and phrases" means language or phrases that may be offensive, insulting, or otherwise objectionable.

[0502] "Means for automatically changing and displaying" refers to technology that instantly corrects detected inappropriate content and replaces it with safe expressions before displaying it.

[0503] "Inappropriate Images" are images that contain sexually explicit material, violence, or other inappropriate visual content.

[0504] "Means for analyzing and deleting detected inappropriate images" refers to technology that analyzes collected images and removes them if they are deemed inappropriate.

[0505] "Voice data" refers to data that is a digital recording of a user's voice.

[0506] A "speech recognition model" refers to the algorithms and techniques used to convert speech into text.

[0507] "Invisibility methods" are techniques that make inappropriate commenters invisible to other users.

[0508] "Means of notification" refers to the mechanism by which the system communicates warnings and information to the user.

[0509] "Sentiment analysis engine" refers to software or hardware for analyzing user emotions from collected data.

[0510] "Means for providing appropriate responses when emotions are perceived as negative" refers to technology for providing appropriate support or notifications when a user's emotions are determined to be negative.

[0511] "Means for monitoring conversations and behavior between customers and store clerks in physical stores and notifying store clerks in real time based on the results of emotion analysis" is a system that monitors interactions between customers and store clerks in physical stores and immediately notifies store clerks based on the analyzed emotion results.

[0512] The present invention is a system for monitoring customer communications in real time in a brick-and-mortar store and providing safe and effective responses. The system includes the following main components:

[0513] 1. Data Collection Module

[0514] The server uses smart glasses or head-mounted displays to monitor the conversations and behavior of customers and store staff in real time. These devices collect customer voice and facial expression data and send it to the server. Specific hardware examples include Google Glass and Microsoft HoloLens.

[0515] 2. Data Analysis Module

[0516] The server converts the collected voice data into text using a speech recognition model. This speech recognition is performed using the Google Cloud Speech-to-Text API. The converted text is then analyzed using a natural language processing (NLP) model to recognize the content and tone of the conversation. This analysis is performed using Google Cloud Natural Language and OpenAI GPT-3.

[0517] The server then inputs the collected image data into a sentiment analysis engine to analyze the customer's emotional state, using the Affectiva SDK and Amazon Rekognition.

[0518] 3. Supported modules

[0519] If a customer shows signs of dissatisfaction or discomfort, the server will immediately send a notification to the store clerk's smart glasses, displaying an alert to prompt appropriate action, for example, a notification such as: "The customer is dissatisfied. Please take action."

[0520] It also includes a function that automatically corrects inappropriate comments or behavior when it is detected, converting them into safe language and displaying it to customers.

[0521] 4. Notification Module

[0522] At the end of the day, the server automatically generates a customer feedback report and sends a notification to the salesperson and manager, allowing the salesperson to identify specific areas for improvement and strive to improve customer service. Example: "Today's feedback report: List of complaints and areas for improvement."

[0523] Example prompt sentences

[0524] Customer facial expression prompt: "What emotions do you see in the customer's facial expressions?"

[0525] Example input to a generative AI model: "This customer's mouth corners are downturned and their eyes are narrowed. What emotion is this customer experiencing?"

[0526] Conversation analysis prompt: "Does this conversation suggest that the customer is unhappy?"

[0527] Example input to a generative AI model: "Customer: 'Hey, isn't this too expensive? It's a lot more than I expected.' Salesperson: 'Sorry, but we're not currently offering any special discounts, so this is the price.'"

[0528] This allows for real-time monitoring of customer experiences in physical stores, enabling quick and appropriate responses, ultimately improving customer satisfaction.

[0529] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0530] Step 1:

[0531] The server collects real-time conversation and behavior data between customers and store staff in a physical store through smart glasses or a head-mounted display. The device captures voice and image data and sends them to the server. The input is voice and image data, and the output is the raw data sent to the server.

[0532] Step 2:

[0533] The server inputs the received voice data into a voice recognition model and converts it into text format. The Google Cloud Speech-to-Text API is used for voice recognition. The input is voice data and the output is text data. In this step, the voice data is converted into conversational content.

[0534] Step 3:

[0535] The server inputs the converted text data into a natural language processing (NLP) model to analyze the semantics of the conversation. Google Cloud Natural Language and OpenAI GPT-3 are used for the NLP model. The input is text data, and the output is the tone and meaning of the analyzed conversation. In this step, the content and tone of the conversation are analyzed.

[0536] Step 4:

[0537] The server inputs the collected image data into an emotion analysis engine to analyze the customer's emotional state. Emotion analysis uses the Affectiva SDK and Amazon Rekognition. The input is image data, and the output is the analysis result of the customer's emotional state. In this step, emotions are read from the customer's facial expressions.

[0538] Step 5:

[0539] The server integrates the analyzed conversation content with the customer's emotional state, and if the customer shows signs of dissatisfaction or discomfort, it sends a notification to the store clerk's smart glasses. The input is the analysis result, and the output is a notification to the store clerk. Specifically, the clerk will see a message such as, "The customer is dissatisfied. Please take action."

[0540] Step 6:

[0541] When the server detects inappropriate comments or behavior, it automatically modifies the content, converts it into safer language, and displays it. The input is the data of the inappropriate comments or behavior, and the output is the modified content. Specifically, offensive language is changed to softer language.

[0542] Step 7:

[0543] At the end of each day, the server automatically generates a customer feedback report and sends notifications to store staff and managers. The input is the customer interaction data for the day, and the output is a feedback report. Specifically, the report contains content such as "Today's feedback report: list of complaints and areas for improvement."

[0544] 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.

[0545] 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.

[0546] 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.

[0547] [Second embodiment]

[0548] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0549] 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.

[0550] 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).

[0551] 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.

[0552] 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.

[0553] 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).

[0554] 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.

[0555] 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.

[0556] 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.

[0557] 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.

[0558] In the smart glasses 214, 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.

[0559] 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."

[0560] The present invention relates to a system for automatically monitoring and managing the safety of communications within the metaverse. To implement this system, the following programs and processing means are included:

[0561] System configuration

[0562] 1. Data Collection Module

[0563] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[0564] 2. Data Analysis Module

[0565] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[0566] The server inputs the received image data into a computer vision model to detect inappropriate content.

[0567] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[0568] 3. Supported modules

[0569] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0570] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[0571] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0572] 4. Notification Module

[0573] The server notifies the affected users about any inappropriate content or behavior detected.

[0574] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[0575] Specific examples

[0576] Example 1: Monitoring chat messages

[0577] User A sends a chat message in the metaverse saying "I hope you die."

[0578] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0579] The server automatically converts this to "Shut up, you idiot" and displays it.

[0580] Example 2: Image monitoring

[0581] User B uploads an inappropriate image (sexually explicit image).

[0582] The terminal sends this image to the server.

[0583] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0584] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0585] Example 3: Monitoring voice activity

[0586] User C shouts insults such as "You idiot" in the voice chat.

[0587] The terminal transmits the voice data to the server.

[0588] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0589] The server makes user C invisible, making him invisible to other users.

[0590] This system will make communication within the metaverse safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[0591] The processing flow will be explained below.

[0592] Step 1:

[0593] A user types and sends a chat message within the metaverse.

[0594] The terminal collects the sent chat messages in real time and transmits them to the server.

[0595] Step 2:

[0596] A user uploads screenshots or image data within the metaverse.

[0597] The terminal transmits the uploaded image data to the server.

[0598] Step 3:

[0599] A user initiates a voice chat in the metaverse.

[0600] The device captures audio data in real time and sends it to the server.

[0601] Step 4:

[0602] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[0603] The server detects inappropriate words and expressions.

[0604] Step 5:

[0605] The image data received by the server is input into a computer vision model and the content is analyzed.

[0606] The server detects inappropriate content.

[0607] Step 6:

[0608] The server inputs the received voice data into a voice recognition model and converts it into text.

[0609] The server inputs the converted data into text into an NLP model and analyzes the content.

[0610] The server detects inappropriate comments.

[0611] Step 7:

[0612] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0613] Example: Converting the message "die" to "shut up."

[0614] Step 8:

[0615] The server immediately deletes any inappropriate image data that is detected.

[0616] The server sends a notification to the user.

[0617] Step 9:

[0618] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[0619] The fact that the server has become invisible is reflected to users other than the speaker.

[0620] Step 10:

[0621] The server will notify the user of any inappropriate content or violating behavior detected.

[0622] The server provides a detailed feedback report to the user.

[0623] Example 1

[0624] 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."

[0625] In the metaverse, inappropriate comments, images, and audio are shared between users, potentially compromising user safety and comfort. Conventional systems lack effective monitoring and management methods, and it takes time for inappropriate content to be detected, potentially adversely affecting users during that time. Therefore, there is a need for a system that can monitor communication in the metaverse in real time and respond immediately.

[0626] 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.

[0627] In this invention, the server includes means for collecting communication data from user devices, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for analyzing the collected image data using a computer vision model to detect and delete inappropriate content, means for converting the collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate remarks, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, and means for notifying the user of detected inappropriate content or behavior. This makes it possible to monitor communication within the metaverse in real time and respond immediately to inappropriate content or behavior.

[0628] A "user device" is an electronic device used by a user to access and communicate with the metaverse.

[0629] "Communication data" refers to information such as chat messages, image data, and audio data exchanged between users.

[0630] A "natural language processing model" is an artificial intelligence model for analyzing and understanding text data.

[0631] A "chat message" is a message sent in text format by a user within the metaverse.

[0632] "Inappropriate words and phrases" are words or phrases that may be offensive, insulting, or upsetting to a user in a communication.

[0633] A "computer vision model" is an artificial intelligence model for analyzing and understanding image data.

[0634] "Inappropriate Content" refers to images or videos that disrupt healthy communication within the Metaverse and may cause mental or emotional harm to other users.

[0635] A "voice recognition model" is an artificial intelligence model for converting voice data into text data.

[0636] "Invisibility" is a process in which the speaker or actor of detected inappropriate behavior is made invisible to other users.

[0637] A "notification" is a message or alert that informs a user that a particular action or statement is inappropriate.

[0638] This invention relates to a system that automatically monitors and manages the safety of communications within the metaverse. The system collects and analyzes chat messages, image data, and audio data exchanged between users in real time, and aims to immediately detect and respond to inappropriate content.

[0639] System configuration

[0640] To implement this system, the following hardware and software are used:

[0641] Hardware

[0642] 1. User's device

[0643] Communication-enabled devices such as personal computers, smartphones, and tablets

[0644] These devices are used to access the metaverse server and send and receive data such as chat, audio, and images.

[0645] 2. Server

[0646] High-performance server computers

[0647] It has the computing power necessary to analyze chat messages, perform voice recognition, and image analysis.

[0648] software

[0649] 1. Natural Language Processing Model (NLP Model)

[0650] Examples: BERT, GPT-3

[0651] It is used to analyze chat messages and detect inappropriate words and phrases.

[0652] 2. Computer Vision Models

[0653] Examples: ResNet, YOLO

[0654] Used to analyze image data and detect inappropriate content.

[0655] 3. Speech Recognition Model

[0656] Examples: Google Cloud Speech-to-Text, DeepSpeech

[0657] It is used to convert audio data into text format.

[0658] Example of operation

[0659] Example 1: Monitoring chat messages

[0660] User A sends a chat message in the metaverse saying "I hope you die."

[0661] The terminal immediately sends this message to the server.

[0662] The server inputs this message into a natural language processing model, which analyzes and detects that the word "die" is inappropriate.

[0663] The server automatically converts the message to "Shut up you" and displays it in an appropriate format.

[0664] Example 2: Image monitoring

[0665] User B uploads an inappropriate image (e.g., a sexually explicit image).

[0666] The terminal transmits this image to the server.

[0667] The server inputs the received images into a computer vision model to analyze and detect any inappropriate content.

[0668] The server immediately deletes the image and notifies User B that "inappropriate content has been posted."

[0669] Example 3: Monitoring voice activity

[0670] User C shouts insults such as "You idiot" in the voice chat.

[0671] The terminal transmits the voice data to the server.

[0672] The server converts the voice data into text using a voice recognition model and analyzes it using a natural language processing model.

[0673] The server recognizes the word "idiot" as inappropriate and turns User C invisible, making him invisible to other users.

[0674] Prompt Sentence Examples

[0675] prompt:

[0676] If I send a chat message in the Metaverse like "I hope you die," how does the system handle it?

[0677] response:

[0678] 1. The device sends a chat message saying "You should die" to the server in real time.

[0679] 2. The server analyzes the message using a natural language processing model and detects that the word "die" is inappropriate.

[0680] 3. The server automatically converts the message to "Shut up, you idiot" and displays it.

[0681] 4. The server notifies User A of the revised message and informs him that the original message was inappropriate.

[0682] This system collects and analyzes data in real time and responds immediately to maintain healthy communication within the metaverse, providing a safe and secure environment for users.

[0683] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0684] Step 1: Data collection

[0685] The terminal collects communication data from the user's device. The input data includes chat messages, image data, and voice data. The collected data is temporarily stored in memory and then the next step is performed.

[0686] Specific behavior:

[0687] The user types "I wish you would die" into the chat window and presses the send button.

[0688] A user says "You idiot" in voice chat.

[0689] A user posts an image using the image upload feature.

[0690] Step 2: Send data

[0691] The terminal sends the collected data to the server in real time. The input data is the collected communication data, which is sent directly to the server via the network. The server adds the received data to a processing queue.

[0692] Specific behavior:

[0693] The device sends the chat message "You should die" data to the server.

[0694] The device sends the recorded audio data to the server.

[0695] The terminal transmits the uploaded image file to the server.

[0696] Step 3: Analyzing chat messages

[0697] The server inputs the received chat message into a natural language processing model (NLP model) to analyze the meaning and keywords. The input data is the received chat message, and the output is the judgment result of whether it contains inappropriate words or expressions.

[0698] Specific behavior:

[0699] The server inputs the chat message "You should die" into the NLP model.

[0700] The server analyzes the word "die" as inappropriate and obtains the judgment result.

[0701] Step 4: Analyze the images

[0702] The server inputs the received image data into a computer vision model to detect inappropriate content. The input data is the received image data, and the output is a judgment result on whether the image is inappropriate or not.

[0703] Specific behavior:

[0704] The server inputs the received image data into a computer vision model.

[0705] The server uses image analysis models to detect explicit elements in the image.

[0706] Step 5: Analyze the audio data

[0707] The server inputs the received voice data into a voice recognition model, converts it into text format, and then analyzes it using a natural language processing model. The input data is the received voice data, and the output is a judgment result on whether or not it contains inappropriate remarks.

[0708] Specific behavior:

[0709] The server inputs the voice data into a speech recognition model and converts it into text.

[0710] The server inputs the converted text data into an NLP model and recognizes the word "idiot" as inappropriate.

[0711] Step 6: Implementing response measures

[0712] The server takes appropriate action based on the analysis results. The input data is the analysis results, and if any inappropriate content is detected, the server takes appropriate action. The output is an appropriately modified message, image deletion, user invisibility, etc.

[0713] Specific behavior:

[0714] The server converts "You should die" to "Shut up" and displays it.

[0715] The server deletes inappropriate image files.

[0716] The server makes user C invisible, making him invisible to other users.

[0717] Step 7: Notification

[0718] The server sends notifications to the relevant users about the detected inappropriate behavior. The input data is the result of the action taken, and the output is a notification message to the user.

[0719] Specific behavior:

[0720] The server notifies User A that "Your comment, 'You should die,' has been deemed inappropriate."

[0721] The server warns User B that the posted image has been deleted because it is inappropriate.

[0722] The server notifies User C that "Your comment 'You idiot' has been deemed inappropriate and made invisible."

[0723] (Application example 1)

[0724] 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."

[0725] Communication between users in the metaverse or virtual store may not always be secure. The free circulation of inappropriate comments, images, and audio not only impairs the user experience, but also risks reducing the reliability of the platform itself. Real-time communication, in particular, requires immediate response, necessitating efficient monitoring and management systems. Providing appropriate feedback to users in the event of inappropriate behavior is also important from the perspective of preventing recurrence and user education. Therefore, the present invention aims to provide a system that automatically monitors and manages the security of communication in the metaverse or virtual store, and provides appropriate responses and feedback.

[0726] 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.

[0727] In this invention, the server includes means for collecting chat messages, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for collecting and analyzing inappropriate images and deleting the detected inappropriate images, means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, means for notifying the user of detected inappropriate content or behavior, means for providing the user with a detailed feedback report on the detected inappropriate content, and means for monitoring communication within the virtual store in real time and taking specific measures immediately. This enables safe and smooth communication between users within the metaverse and virtual store, thereby improving the reliability of the platform.

[0728] The "metaverse" is a three-dimensional virtual space built on the Internet, a platform where users can interact with each other and engage in activities through avatars.

[0729] A "virtual store" is a virtual store that mimics a real store, where users can purchase digital items and services and interact with other users.

[0730] "Chat messages" are messages exchanged between users in real time in text format.

[0731] A "natural language processing model" is an artificial intelligence technology for understanding and analyzing human language, extracting meaning from text data.

[0732] "Inappropriate words and phrases" are words or phrases that contain offensive, discriminatory or objectionable content and are inappropriate for communication.

[0733] To "collect" means to obtain specific data or information and gather it in a designated location.

[0734] "Analysis" means examining data or information in detail to understand and interpret its content.

[0735] A "speech recognition model" is an artificial intelligence technology for converting voice data into text, recognizing speech as text information.

[0736] "Invisibility" refers to the process of making a specific user invisible to other users.

[0737] A "feedback report" is a document that provides detailed information about inappropriate behavior or content and guidelines for improvement.

[0738] "Real-time monitoring" refers to the process of instantly monitoring data and behavior and taking necessary action on the spot.

[0739] "Responding immediately" means immediately taking appropriate action or measures against any inappropriate behavior or content that is detected.

[0740] The present invention relates to a system for automatically monitoring and managing communications within a metaverse or virtual store. To implement this system, the following programs and processing means are included:

[0741] System configuration

[0742] 1. Data Collection Module

[0743] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[0744] 2. Data Analysis Module

[0745] The server inputs chat messages into a natural language processing (NLP) model using a generative AI model to perform semantic analysis and keyword extraction.

[0746] The server inputs the received image data into a computer vision model (e.g., OpenCV) to detect inappropriate content.

[0747] The server inputs the collected audio data into a speech recognition model (e.g., a Transformers speech recognition model) and converts it into text format, which is then analyzed for inappropriate speech using the same NLP model.

[0748] 3. Supported modules

[0749] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0750] The server detects inappropriate images, immediately deletes them, and notifies the user with a warning.

[0751] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0752] 4. Notification Module

[0753] The server sends notifications to the affected users about any inappropriate content or behavior detected.

[0754] The server provides the user with a detailed feedback report of any inappropriate content detected.

[0755] Specific examples

[0756] Example 1: Monitoring chat messages

[0757] User A sends a chat message in the metaverse saying "I hope you die."

[0758] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0759] The server automatically converts this to "Shut up, you idiot" and displays it.

[0760] Example 2: Image monitoring

[0761] User B uploads an inappropriate image (sexually explicit image).

[0762] The terminal sends this image to the server.

[0763] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0764] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0765] Example 3: Monitoring voice activity

[0766] User C shouts insults such as "You idiot" in the voice chat.

[0767] The terminal transmits the voice data to the server.

[0768] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0769] The server makes user C invisible, making him invisible to other users.

[0770] Prompt Sentence Examples

[0771] "Determine whether a particular chat message is inappropriate. Example: 'I hope you die.'"

[0772] "Please decide if this image is inappropriate. Here is the image data: [base64 encoded image data]"

[0773] This system will make communication within the metaverse and virtual stores safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[0774] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0775] Step 1:

[0776] Data collection

[0777] The device collects chat messages, uploaded image data, and voice chat data in real time, and transmits this data to a server.

[0778] Input: User chat messages, image data, and audio data

[0779] Output: Various data sent to the server

[0780] Step 2:

[0781] Chat message analysis

[0782] The server inputs the collected chat messages into a natural language processing (NLP) model using a generative AI model, where semantic analysis and keyword extraction are performed to detect inappropriate words and expressions.

[0783] Input: The chat message the user said

[0784] Data processing: Semantic analysis and keyword extraction using NLP models

[0785] Output: A list of inappropriate words and phrases

[0786] Step 3:

[0787] Correcting inappropriate messages

[0788] The server automatically converts chat messages that are deemed inappropriate into appropriate language and displays them accordingly, for example, changing the word "die" to "shut up."

[0789] Input: A list of inappropriate words and phrases

[0790] Data processing: Automatically convert inappropriate words and expressions into appropriate ones

[0791] Output: Modified chat message

[0792] Step 4:

[0793] Image data analysis

[0794] The server then feeds the collected image data into computer vision models to detect inappropriate content, for example, using libraries such as OpenCV to identify sexually explicit images.

[0795] Input: User uploaded image data

[0796] Data processing: Image analysis using computer vision models

[0797] Output: Result of whether or not inappropriate content is included

[0798] Step 5:

[0799] Inappropriate image removal and notification

[0800] The server immediately deletes image data that is detected as inappropriate and notifies the user of a warning.

[0801] Input: Inappropriate content detection result

[0802] Data processing: Deletion of image data and creation of warning notices

[0803] Output: Deleted image data and a warning notice to the user

[0804] Step 6:

[0805] Analysis of audio data

[0806] The server inputs the collected voice data into a speech recognition model, converts it into text format, and then analyzes it for inappropriate comments using an NLP model.

[0807] Input: User voice chat data

[0808] Data processing: Converting voice data into text using a speech recognition model and analyzing it using an NLP model

[0809] Output: Translated audio data and a list of inappropriate words

[0810] Step 7:

[0811] Making inappropriate audio speakers invisible

[0812] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0813] Input: A list of inappropriate words

[0814] Data processing: Making users invisible

[0815] Output: Invisible user

[0816] Step 8:

[0817] Providing detailed feedback reports

[0818] The server provides the user with a detailed feedback report of the inappropriate content and behavior detected, including specific details of the behavior that was inappropriate.

[0819] Input: Data on inappropriate content and behavior detected

[0820] Data processing: Creating detailed feedback reports

[0821] Output: A feedback report provided to the user

[0822] 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.

[0823] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and combines an emotion engine that recognizes user emotions to provide more advanced responses. To implement this system, the following programs and processing means are included.

[0824] System configuration

[0825] 1. Data Collection Module

[0826] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[0827] 2. Data Analysis Module

[0828] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[0829] The server inputs the received image data into a computer vision model to detect inappropriate content.

[0830] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[0831] 3. Sentiment Analysis Module

[0832] The server inputs the collected data into an emotion engine to analyze the user's emotions.

[0833] The server enhances the detection of inappropriate content based on the results of sentiment analysis.

[0834] 4. Supported modules

[0835] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0836] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[0837] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[0838] If the emotion engine recognizes a negative emotion, it provides mental support to the user.

[0839] 5. Notification Module

[0840] The server will then notify the user of any inappropriate content or behavior that has been detected.

[0841] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[0842] Specific examples

[0843] Example 1: Monitoring chat messages and sentiment

[0844] User A sends a chat message in the metaverse saying "I hope you die."

[0845] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0846] The server automatically converts this to "Shut up, you idiot" and displays it.

[0847] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[0848] Example 2: Image and Emotion Monitoring

[0849] User B uploads an inappropriate image (sexually explicit image).

[0850] The terminal sends this image to the server.

[0851] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0852] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0853] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[0854] Example 3: Monitoring voice utterances and emotions

[0855] User C shouts insults such as "You idiot" in the voice chat.

[0856] The terminal transmits the voice data to the server.

[0857] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0858] The server makes user C invisible, making him invisible to other users.

[0859] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[0860] This system will not only make communication within the metaverse safer and more comfortable, but will also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[0861] The processing flow will be explained below.

[0862] Step 1:

[0863] A user types and sends a chat message within the metaverse.

[0864] The terminal collects the sent chat messages in real time and transmits them to the server.

[0865] Step 2:

[0866] A user uploads screenshots or image data within the metaverse.

[0867] The terminal transmits the uploaded image data to the server.

[0868] Step 3:

[0869] A user initiates a voice chat in the metaverse.

[0870] The device captures audio data in real time and sends it to the server.

[0871] Step 4:

[0872] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[0873] The server detects inappropriate words and expressions.

[0874] Step 5:

[0875] The image data received by the server is input into a computer vision model and the content is analyzed.

[0876] The server detects inappropriate content.

[0877] Step 6:

[0878] The server inputs the received voice data into a voice recognition model and converts it into text.

[0879] The server inputs the converted data into text into an NLP model and analyzes the content.

[0880] The server detects inappropriate comments.

[0881] Step 7:

[0882] The server inputs the collected data into an emotion engine to analyze the user's emotional state.

[0883] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[0884] Step 8:

[0885] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[0886] Example: Converting the message "die" to "shut up."

[0887] At the same time, the server uses an emotion engine to evaluate the emotional state of user A, and if negative emotions are detected, provides mental support.

[0888] Step 9:

[0889] The server immediately deletes any inappropriate image data that is detected.

[0890] The server sends a notification to the user.

[0891] The server uses an emotion engine to evaluate User B's emotional state and considers an appropriate response.

[0892] Step 10:

[0893] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[0894] The fact that the server has become invisible is reflected to users other than the speaker.

[0895] The server uses an emotion engine to assess the emotional state of user C and provides support as needed.

[0896] Step 11:

[0897] The server notifies the user of any inappropriate content or behavior detected.

[0898] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[0899] Example 2

[0900] 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."

[0901] While communication within the metaverse is increasing, there is a risk that inappropriate words and actions may exist, preventing users from interacting safely in a healthy environment. There is a need for a system that can respond quickly and effectively to inappropriate words and actions and provide feedback that takes into account the user's emotional state.

[0902] 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.

[0903] In this invention, the server

[0904] A means for collecting communication data (chat messages, images, audio) sent by users within the metaverse;

[0905] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[0906] A means for automatically changing and displaying detected inappropriate words and expressions;

[0907] A means for analyzing the collected image data using a computer vision model to detect and remove inappropriate images;

[0908] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[0909] A means for making the speaker of the detected inappropriate voice invisible;

[0910] A means for inputting the collected communication data into an emotion analysis engine and analyzing the user's emotions;

[0911] A means for detecting negative emotions based on the emotion analysis results and providing mental support;

[0912] a means for notifying users of detected inappropriate content or behavior;

[0913] and a means for generating detailed feedback reports and providing them to users. This allows appropriate management of inappropriate speech and behavior within the metaverse, enabling users to interact with peace of mind in a healthy and safe communication environment.

[0914] "Communication data" is a general term that includes chat messages, image data, and voice data sent by users within the metaverse.

[0915] A "natural language processing model" is a computer algorithm or machine learning model that analyzes text data and understands and extracts grammar, context, and keywords.

[0916] A "computer vision model" is an algorithm or deep learning network that analyzes image data to detect inappropriate content or specific features.

[0917] A "voice recognition model" is a technology for converting voice data into text data and analyzing the content of speech.

[0918] An "emotion analysis engine" is an algorithm or AI technology that analyzes a user's text, image, and voice data to identify and evaluate their emotional state.

[0919] "Mental support" refers to the means and services that provide psychological and emotional care and assistance in response to negative emotions detected through analysis.

[0920] "Invisibility" is a function that restricts the actions of a user who has behaved inappropriately by making that user invisible to other users.

[0921] "Feedback Report" means a written or digital message that provides detailed explanation and advice regarding detected inappropriate behavior or content.

[0922] "Inappropriate content" refers to chat messages, images, audio, etc. within the metaverse that are deemed socially or ethically problematic.

[0923] "User notification" is a function that notifies the relevant user about inappropriate behavior or content detected by the system.

[0924] MODE FOR CARRYING OUT THE INVENTION

[0925] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and provides advanced responses that take into account the emotional state of the user. To implement this system, the following programs and processing means are included.

[0926] System Configuration

[0927] 1. Data Collection Module

[0928] The terminal collects chat messages, image data, and voice data sent by users within the metaverse in real time and transmits them to a server.

[0929] 2. Data Analysis Module

[0930] The server inputs the collected chat messages into a natural language processing (NLP) model to perform semantic analysis and keyword extraction, using Python's NLTK and spaCy.

[0931] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content.

[0932] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API), converts it into text, and then analyzes the text data with an NLP model to detect inappropriate speech.

[0933] 3. Sentiment Analysis Module

[0934] The server inputs the collected data into an emotion engine (e.g., Microsoft Azure emotion analysis API, snownlp, etc.) to analyze the user's emotions.

[0935] Based on the results of sentiment analysis, we will strengthen the detection of inappropriate content and consider taking action as necessary.

[0936] 4. Supported modules

[0937] The server automatically converts detected inappropriate chat messages into appropriate expressions, immediately deletes inappropriate images, and notifies users with a warning. If inappropriate voice comments are detected, the person making the comment becomes invisible to other users.

[0938] If the emotion engine recognizes a negative emotion, the server provides mental support to the user.

[0939] 5. Notification Module

[0940] The server notifies users of any inappropriate content or behavior detected and provides detailed feedback reports to indicate which actions were inappropriate.

[0941] Specific examples

[0942] Example 1: Monitoring chat messages and sentiment

[0943] User A sends a chat message in the metaverse saying "I hope you die."

[0944] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[0945] The server automatically converts this to "Shut up, you idiot" and displays it.

[0946] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[0947] Example 2: Image and Emotion Monitoring

[0948] User B uploads an inappropriate image (sexually explicit image).

[0949] The terminal sends this image to the server.

[0950] The server analyzes the image using computer vision models and detects that it is inappropriate.

[0951] The server deletes the image and notifies User B that "inappropriate content has been posted."

[0952] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[0953] Example 3: Monitoring voice utterances and emotions

[0954] User C shouts insults such as "You idiot" in the voice chat.

[0955] The terminal transmits the voice data to the server.

[0956] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[0957] The server makes user C invisible, making him invisible to other users.

[0958] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[0959] This system will not only make communication within the metaverse safer and more comfortable, but will also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[0960] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0961] Step 1: Data collection

[0962] The terminal collects chat messages, image data, and voice data sent by users in the metaverse in real time. This data is input and sent to a server via a network.

[0963] Specific behavior:

[0964] The user types "Hello" in the chat window and presses the send button.

[0965] The device captures the text data "Hello" and sends it to the server.

[0966] Users take screenshots of their avatars and upload them to a shared gallery in the metaverse.

[0967] The device captures the screenshot image data and sends it to the server.

[0968] Step 2: Chat message analysis

[0969] The server inputs the received chat messages into a natural language processing (NLP) model to extract keywords and perform semantic analysis. The input is the chat message, and the output is the analysis result. Specifically, Python's NLTK and spaCy are used.

[0970] Specific behavior:

[0971] The server receives the message "Hello" and asks the natural language processing model to analyze it.

[0972] The NLP model tags "hello" as a greeting and returns the result to the server.

[0973] Step 3: Image data analysis

[0974] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content. The input is the image data, and the output is the detection result of inappropriate content.

[0975] Specific behavior:

[0976] The server receives the screenshots uploaded by the user.

[0977] Computer vision models scan images to determine if they contain inappropriate content.

[0978] The server receives the inappropriate content detection result and removes the image.

[0979] Step 4: Audio data analysis

[0980] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API) and converts it into text format. The text data is then analyzed using an NLP model to detect inappropriate speech. The input is the audio data, and the output is the analyzed text data.

[0981] Specific behavior:

[0982] A user says "You're such an idiot" in voice chat.

[0983] The device captures the audio data and sends it to the server.

[0984] The server converts the voice data into text and generates the text data "You're an idiot."

[0985] The NLP model determines that "baka" is an inappropriate word and returns the result to the server.

[0986] Step 5: Sentiment Analysis

[0987] The server inputs the collected text, image, and voice data into an emotion engine (e.g., Microsoft Azure's emotion analysis API, snownlp, etc.) to analyze the user's emotions. The input is data (text, image, voice), and the output is an evaluation of the user's emotional state.

[0988] Specific behavior:

[0989] The server inputs the chat message "You're such an idiot" into an emotion engine, which analyzes the emotional state of the message.

[0990] The emotion engine detects "anger" and returns the result to the server.

[0991] Step 6: Address inappropriate content

[0992] The server automatically takes appropriate action as needed based on the results of the sentiment analysis. The inputs are the detection results of inappropriate content and the sentiment analysis results, and the outputs are the modified message, the deletion of inappropriate content, and the results of making the user invisible.

[0993] Specific behavior:

[0994] The server automatically converts "You're an idiot" to "Shut up you" and displays it to the sending user.

[0995] If an inappropriate image is detected, it will be removed and a warning message will be sent to the uploader.

[0996] The server turns users who make inappropriate vocal comments into invisible people, making them invisible to other users.

[0997] Step 7: Providing mental support

[0998] The server provides mental support to the user when a negative emotion is recognized by the emotion engine. The input is the emotion analysis result, and the output is the support provided.

[0999] Specific behavior:

[1000] If the server recognizes "anger," it provides the user with a message explaining how to relax and a link to a counseling service.

[1001] Step 8: User Notification

[1002] The server notifies the user of any detected inappropriate content or behavior and also generates and provides a detailed feedback report. The input is the detection result of inappropriate content, and the output is the user notification and feedback report.

[1003] Specific behavior:

[1004] The server sends a notification to the user about the detected inappropriate content.

[1005] The report states, "Your comment, 'You're an idiot,' was deemed inappropriate. Please be careful next time."

[1006] (Application example 2)

[1007] 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."

[1008] In current brick-and-mortar stores, communication between customers and store staff can be inappropriate, and customers' feelings can be poorly understood, leading to delayed responses. This can lead to lower customer satisfaction, which can have a negative impact on store ratings and sales. Furthermore, because it is difficult to respond appropriately in real time, advanced management of the customer experience in brick-and-mortar stores is required.

[1009] The identification processing 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 collecting chat messages; means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions; means for automatically changing and displaying the corresponding message when inappropriate words or expressions are detected; means for collecting and analyzing inappropriate images and deleting the detected inappropriate images; means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments; means for making the speaker of the corresponding voice invisible when inappropriate voice is detected; means for notifying the user of detected inappropriate content or behavior; means for analyzing the user's emotions, including an emotion analysis engine; means for providing an appropriate response when the user's emotions are recognized as negative; and means for monitoring conversations and behaviors between customers and store clerks in a physical store and notifying the store clerk in real time based on the emotion analysis results. This enables real-time monitoring of customer experiences in physical stores and enables prompt and appropriate responses.

[1010] The "metaverse" is a virtual space that exists on the Internet, a virtual reality environment in which users can interact using avatars.

[1011] "Communication safety" refers to the standards and measures to ensure that interactions between users are appropriate and healthy.

[1012] "Monitoring" refers to the act of watching users' behavior and messages in real time to detect inappropriate behavior or expressions.

[1013] "Management" refers to the act of taking appropriate action against detected inappropriate behavior or content and operating the entire system.

[1014] A "chat message" is a communication message sent by a user in text format.

[1015] "Means of collection" refers to the equipment or software mechanisms used to collect data sent by users on a server.

[1016] A "natural language processing model" refers to the algorithms and technologies that allow computers to understand and analyze human language.

[1017] "Inappropriate words and phrases" means language or phrases that may be offensive, insulting, or otherwise objectionable.

[1018] "Means for automatically changing and displaying" refers to technology that instantly corrects detected inappropriate content and replaces it with safe expressions before displaying it.

[1019] "Inappropriate Images" are images that contain sexually explicit material, violence, or other inappropriate visual content.

[1020] "Means for analyzing and deleting detected inappropriate images" refers to technology that analyzes collected images and removes them if they are deemed inappropriate.

[1021] "Voice data" refers to data that is a digital recording of a user's voice.

[1022] A "speech recognition model" refers to the algorithms and techniques used to convert speech into text.

[1023] "Invisibility methods" are techniques that make inappropriate commenters invisible to other users.

[1024] "Means of notification" refers to the mechanism by which the system communicates warnings and information to the user.

[1025] "Sentiment analysis engine" refers to software or hardware for analyzing user emotions from collected data.

[1026] "Means for providing appropriate responses when emotions are perceived as negative" refers to technology for providing appropriate support or notifications when a user's emotions are determined to be negative.

[1027] "Means for monitoring conversations and behavior between customers and store clerks in physical stores and notifying store clerks in real time based on the results of emotion analysis" is a system that monitors interactions between customers and store clerks in physical stores and immediately notifies store clerks based on the analyzed emotion results.

[1028] The present invention is a system for monitoring customer communications in real time in a brick-and-mortar store and providing safe and effective responses. The system includes the following main components:

[1029] 1. Data Collection Module

[1030] The server uses smart glasses or head-mounted displays to monitor the conversations and behavior of customers and store staff in real time. These devices collect customer voice and facial expression data and send it to the server. Specific hardware examples include Google Glass and Microsoft HoloLens.

[1031] 2. Data Analysis Module

[1032] The server converts the collected voice data into text using a speech recognition model. This speech recognition is performed using the Google Cloud Speech-to-Text API. The converted text is then analyzed using a natural language processing (NLP) model to recognize the content and tone of the conversation. This analysis is performed using Google Cloud Natural Language and OpenAI GPT-3.

[1033] The server then inputs the collected image data into a sentiment analysis engine to analyze the customer's emotional state, using the Affectiva SDK and Amazon Rekognition.

[1034] 3. Supported modules

[1035] If a customer shows signs of dissatisfaction or discomfort, the server will immediately send a notification to the store clerk's smart glasses, displaying an alert to prompt appropriate action, for example, a notification such as: "The customer is dissatisfied. Please take action."

[1036] It also includes a function that automatically corrects inappropriate comments or behavior when it is detected, converting them into safe language and displaying it to customers.

[1037] 4. Notification Module

[1038] At the end of the day, the server automatically generates a customer feedback report and sends a notification to the salesperson and manager, allowing the salesperson to identify specific areas for improvement and strive to improve customer service. Example: "Today's feedback report: List of complaints and areas for improvement."

[1039] Example prompt sentences

[1040] Customer facial expression prompt: "What emotions do you see in the customer's facial expressions?"

[1041] Example input to a generative AI model: "This customer's mouth corners are downturned and their eyes are narrowed. What emotion is this customer experiencing?"

[1042] Conversation analysis prompt: "Does this conversation suggest that the customer is unhappy?"

[1043] Example input to a generative AI model: "Customer: 'Hey, isn't this too expensive? It's a lot more than I expected.' Salesperson: 'Sorry, but we're not currently offering any special discounts, so this is the price.'"

[1044] This allows for real-time monitoring of customer experiences in physical stores, enabling quick and appropriate responses, ultimately improving customer satisfaction.

[1045] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1046] Step 1:

[1047] The server collects real-time conversation and behavior data between customers and store staff in a physical store through smart glasses or a head-mounted display. The device captures voice and image data and sends them to the server. The input is voice and image data, and the output is the raw data sent to the server.

[1048] Step 2:

[1049] The server inputs the received voice data into a voice recognition model and converts it into text format. The Google Cloud Speech-to-Text API is used for voice recognition. The input is voice data and the output is text data. In this step, the voice data is converted into conversational content.

[1050] Step 3:

[1051] The server inputs the converted text data into a natural language processing (NLP) model to analyze the semantics of the conversation. Google Cloud Natural Language and OpenAI GPT-3 are used for the NLP model. The input is text data, and the output is the tone and meaning of the analyzed conversation. In this step, the content and tone of the conversation are analyzed.

[1052] Step 4:

[1053] The server inputs the collected image data into an emotion analysis engine to analyze the customer's emotional state. Emotion analysis uses the Affectiva SDK and Amazon Rekognition. The input is image data, and the output is the analysis result of the customer's emotional state. In this step, emotions are read from the customer's facial expressions.

[1054] Step 5:

[1055] The server integrates the analyzed conversation content with the customer's emotional state, and if the customer shows signs of dissatisfaction or discomfort, it sends a notification to the store clerk's smart glasses. The input is the analysis result, and the output is a notification to the store clerk. Specifically, the clerk will see a message such as, "The customer is dissatisfied. Please take action."

[1056] Step 6:

[1057] When the server detects inappropriate comments or behavior, it automatically modifies the content, converts it into safer language, and displays it. The input is the data of the inappropriate comments or behavior, and the output is the modified content. Specifically, offensive language is changed to softer language.

[1058] Step 7:

[1059] At the end of each day, the server automatically generates a customer feedback report and sends notifications to store staff and managers. The input is the customer interaction data for the day, and the output is a feedback report. Specifically, the report contains content such as "Today's feedback report: list of complaints and areas for improvement."

[1060] 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.

[1061] 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.

[1062] 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.

[1063] [Third embodiment]

[1064] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1065] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1066] 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).

[1067] 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.

[1068] 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.

[1069] 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).

[1070] 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.

[1071] 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.

[1072] 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.

[1073] 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.

[1074] 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.

[1075] 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."

[1076] The present invention relates to a system for automatically monitoring and managing the safety of communications within the metaverse. To implement this system, the following programs and processing means are included:

[1077] System configuration

[1078] 1. Data Collection Module

[1079] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[1080] 2. Data Analysis Module

[1081] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[1082] The server inputs the received image data into a computer vision model to detect inappropriate content.

[1083] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[1084] 3. Supported modules

[1085] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1086] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[1087] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1088] 4. Notification Module

[1089] The server notifies the affected users about any inappropriate content or behavior detected.

[1090] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[1091] Specific examples

[1092] Example 1: Monitoring chat messages

[1093] User A sends a chat message in the metaverse saying "I hope you die."

[1094] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1095] The server automatically converts this to "Shut up, you idiot" and displays it.

[1096] Example 2: Image monitoring

[1097] User B uploads an inappropriate image (sexually explicit image).

[1098] The terminal sends this image to the server.

[1099] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1100] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1101] Example 3: Monitoring voice activity

[1102] User C shouts insults such as "You idiot" in the voice chat.

[1103] The terminal transmits the voice data to the server.

[1104] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1105] The server makes user C invisible, making him invisible to other users.

[1106] This system will make communication within the metaverse safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[1107] The processing flow will be explained below.

[1108] Step 1:

[1109] A user types and sends a chat message within the metaverse.

[1110] The terminal collects the sent chat messages in real time and transmits them to the server.

[1111] Step 2:

[1112] A user uploads screenshots or image data within the metaverse.

[1113] The terminal transmits the uploaded image data to the server.

[1114] Step 3:

[1115] A user initiates a voice chat in the metaverse.

[1116] The device captures audio data in real time and sends it to the server.

[1117] Step 4:

[1118] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[1119] The server detects inappropriate words and expressions.

[1120] Step 5:

[1121] The image data received by the server is input into a computer vision model and the content is analyzed.

[1122] The server detects inappropriate content.

[1123] Step 6:

[1124] The server inputs the received voice data into a voice recognition model and converts it into text.

[1125] The server inputs the converted data into text into an NLP model and analyzes the content.

[1126] The server detects inappropriate comments.

[1127] Step 7:

[1128] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1129] Example: Converting the message "die" to "shut up."

[1130] Step 8:

[1131] The server immediately deletes any inappropriate image data that is detected.

[1132] The server sends a notification to the user.

[1133] Step 9:

[1134] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[1135] The fact that the server has become invisible is reflected to users other than the speaker.

[1136] Step 10:

[1137] The server will notify the user of any inappropriate content or violating behavior detected.

[1138] The server provides a detailed feedback report to the user.

[1139] Example 1

[1140] 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."

[1141] In the metaverse, inappropriate comments, images, and audio are shared between users, potentially compromising user safety and comfort. Conventional systems lack effective monitoring and management methods, and it takes time for inappropriate content to be detected, potentially adversely affecting users during that time. Therefore, there is a need for a system that can monitor communication in the metaverse in real time and respond immediately.

[1142] 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.

[1143] In this invention, the server includes means for collecting communication data from user devices, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for analyzing the collected image data using a computer vision model to detect and delete inappropriate content, means for converting the collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate remarks, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, and means for notifying the user of detected inappropriate content or behavior. This makes it possible to monitor communication within the metaverse in real time and respond immediately to inappropriate content or behavior.

[1144] A "user device" is an electronic device used by a user to access and communicate with the metaverse.

[1145] "Communication data" refers to information such as chat messages, image data, and audio data exchanged between users.

[1146] A "natural language processing model" is an artificial intelligence model for analyzing and understanding text data.

[1147] A "chat message" is a message sent in text format by a user within the metaverse.

[1148] "Inappropriate words and phrases" are words or phrases that may be offensive, insulting, or upsetting to a user in a communication.

[1149] A "computer vision model" is an artificial intelligence model for analyzing and understanding image data.

[1150] "Inappropriate Content" refers to images or videos that disrupt healthy communication within the Metaverse and may cause mental or emotional harm to other users.

[1151] A "voice recognition model" is an artificial intelligence model for converting voice data into text data.

[1152] "Invisibility" is a process in which the speaker or actor of detected inappropriate behavior is made invisible to other users.

[1153] A "notification" is a message or alert that informs a user that a particular action or statement is inappropriate.

[1154] This invention relates to a system that automatically monitors and manages the safety of communications within the metaverse. The system collects and analyzes chat messages, image data, and audio data exchanged between users in real time, and aims to immediately detect and respond to inappropriate content.

[1155] System configuration

[1156] To implement this system, the following hardware and software are used:

[1157] Hardware

[1158] 1. User's device

[1159] Communication-enabled devices such as personal computers, smartphones, and tablets

[1160] These devices are used to access the metaverse server and send and receive data such as chat, audio, and images.

[1161] 2. Server

[1162] High-performance server computers

[1163] It has the computing power necessary to analyze chat messages, perform voice recognition, and image analysis.

[1164] software

[1165] 1. Natural Language Processing Model (NLP Model)

[1166] Examples: BERT, GPT-3

[1167] It is used to analyze chat messages and detect inappropriate words and phrases.

[1168] 2. Computer Vision Models

[1169] Examples: ResNet, YOLO

[1170] Used to analyze image data and detect inappropriate content.

[1171] 3. Speech Recognition Model

[1172] Examples: Google Cloud Speech-to-Text, DeepSpeech

[1173] It is used to convert audio data into text format.

[1174] Example of operation

[1175] Example 1: Monitoring chat messages

[1176] User A sends a chat message in the metaverse saying "I hope you die."

[1177] The terminal immediately sends this message to the server.

[1178] The server inputs this message into a natural language processing model, which analyzes and detects that the word "die" is inappropriate.

[1179] The server automatically converts the message to "Shut up you" and displays it in an appropriate format.

[1180] Example 2: Image monitoring

[1181] User B uploads an inappropriate image (e.g., a sexually explicit image).

[1182] The terminal transmits this image to the server.

[1183] The server inputs the received images into a computer vision model to analyze and detect any inappropriate content.

[1184] The server immediately deletes the image and notifies User B that "inappropriate content has been posted."

[1185] Example 3: Monitoring voice activity

[1186] User C shouts insults such as "You idiot" in the voice chat.

[1187] The terminal transmits the voice data to the server.

[1188] The server converts the voice data into text using a voice recognition model and analyzes it using a natural language processing model.

[1189] The server recognizes the word "idiot" as inappropriate and turns User C invisible, making him invisible to other users.

[1190] Prompt Sentence Examples

[1191] prompt:

[1192] If I send a chat message in the Metaverse like "I hope you die," how does the system handle it?

[1193] response:

[1194] 1. The device sends a chat message saying "You should die" to the server in real time.

[1195] 2. The server analyzes the message using a natural language processing model and detects that the word "die" is inappropriate.

[1196] 3. The server automatically converts the message to "Shut up, you idiot" and displays it.

[1197] 4. The server notifies User A of the revised message and informs him that the original message was inappropriate.

[1198] This system collects and analyzes data in real time and responds immediately to maintain healthy communication within the metaverse, providing a safe and secure environment for users.

[1199] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1200] Step 1: Data collection

[1201] The terminal collects communication data from the user's device. The input data includes chat messages, image data, and voice data. The collected data is temporarily stored in memory and then the next step is performed.

[1202] Specific behavior:

[1203] The user types "I wish you would die" into the chat window and presses the send button.

[1204] A user says "You idiot" in voice chat.

[1205] A user posts an image using the image upload feature.

[1206] Step 2: Send data

[1207] The terminal transmits the collected data to the server in real time. The input data is the collected communication data, which is transmitted as is to the server via the network. The server adds the received data to a processing queue.

[1208] Specific behavior:

[1209] The device sends the chat message "You should die" data to the server.

[1210] The device sends the recorded audio data to the server.

[1211] The terminal transmits the uploaded image file to the server.

[1212] Step 3: Analyzing chat messages

[1213] The server inputs the received chat message into a natural language processing model (NLP model) to analyze the meaning and keywords. The input data is the received chat message, and the output is the judgment result of whether it contains inappropriate words or expressions.

[1214] Specific behavior:

[1215] The server inputs the chat message "You should die" into the NLP model.

[1216] The server analyzes the word "die" as inappropriate and obtains the judgment result.

[1217] Step 4: Analyze the images

[1218] The server inputs the received image data into a computer vision model to detect inappropriate content. The input data is the received image data, and the output is a judgment result on whether the image is inappropriate or not.

[1219] Specific behavior:

[1220] The server inputs the received image data into a computer vision model.

[1221] The server uses image analysis models to detect explicit elements in the image.

[1222] Step 5: Analyze the audio data

[1223] The server inputs the received voice data into a voice recognition model, converts it into text format, and then analyzes it using a natural language processing model. The input data is the received voice data, and the output is a judgment result on whether or not it contains inappropriate remarks.

[1224] Specific behavior:

[1225] The server inputs the voice data into a speech recognition model and converts it into text.

[1226] The server inputs the converted text data into an NLP model and recognizes the word "idiot" as inappropriate.

[1227] Step 6: Implementing response measures

[1228] The server takes appropriate action based on the analysis results. The input data is the analysis results, and if any inappropriate content is detected, the server takes appropriate action. The output is an appropriately modified message, image deletion, user invisibility, etc.

[1229] Specific behavior:

[1230] The server converts "You should die" to "Shut up" and displays it.

[1231] The server deletes inappropriate image files.

[1232] The server makes user C invisible, making him invisible to other users.

[1233] Step 7: Notification

[1234] The server sends notifications to the relevant users about the detected inappropriate behavior. The input data is the result of the action taken, and the output is a notification message to the user.

[1235] Specific behavior:

[1236] The server notifies User A that "Your comment, 'You should die,' has been deemed inappropriate."

[1237] The server warns User B that the posted image has been deleted because it is inappropriate.

[1238] The server notifies User C that "Your comment 'You idiot' has been deemed inappropriate and made invisible."

[1239] (Application example 1)

[1240] 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."

[1241] Communication between users in the metaverse or virtual store may not always be secure. The free circulation of inappropriate comments, images, and audio not only impairs the user experience, but also risks reducing the reliability of the platform itself. Real-time communication, in particular, requires immediate response, necessitating efficient monitoring and management systems. Providing appropriate feedback to users in the event of inappropriate behavior is also important from the perspective of preventing recurrence and user education. Therefore, the present invention aims to provide a system that automatically monitors and manages the security of communication in the metaverse or virtual store, and provides appropriate responses and feedback.

[1242] 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.

[1243] In this invention, the server includes means for collecting chat messages, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for collecting and analyzing inappropriate images and deleting the detected inappropriate images, means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, means for notifying the user of detected inappropriate content or behavior, means for providing the user with a detailed feedback report on the detected inappropriate content, and means for monitoring communication within the virtual store in real time and taking specific measures immediately. This enables safe and smooth communication between users within the metaverse and virtual store, thereby improving the reliability of the platform.

[1244] The "metaverse" is a three-dimensional virtual space built on the Internet, a platform where users can interact with each other and engage in activities through avatars.

[1245] A "virtual store" is a virtual store that mimics a real store, where users can purchase digital items and services and interact with other users.

[1246] "Chat messages" are messages exchanged between users in real time in text format.

[1247] A "natural language processing model" is an artificial intelligence technology for understanding and analyzing human language, extracting meaning from text data.

[1248] "Inappropriate words and phrases" are words or phrases that contain offensive, discriminatory or objectionable content and are inappropriate for communication.

[1249] To "collect" means to obtain specific data or information and gather it in a designated location.

[1250] "Analysis" means examining data or information in detail to understand and interpret its content.

[1251] A "voice recognition model" is an artificial intelligence technology for converting voice data into text, recognizing speech as text information.

[1252] "Invisibility" refers to the process of making a specific user invisible to other users.

[1253] A "feedback report" is a document that provides detailed information about inappropriate behavior or content and guidelines for improvement.

[1254] "Real-time monitoring" refers to the process of instantly monitoring data and behavior and taking necessary action on the spot.

[1255] "Responding immediately" means immediately taking appropriate action or measures against any inappropriate behavior or content that is detected.

[1256] The present invention relates to a system for automatically monitoring and managing communications within a metaverse or virtual store. To implement this system, the following programs and processing means are included:

[1257] System configuration

[1258] 1. Data Collection Module

[1259] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[1260] 2. Data Analysis Module

[1261] The server inputs chat messages into a natural language processing (NLP) model using a generative AI model to perform semantic analysis and keyword extraction.

[1262] The server inputs the received image data into a computer vision model (e.g., OpenCV) to detect inappropriate content.

[1263] The server inputs the collected audio data into a speech recognition model (e.g., a Transformers speech recognition model) and converts it into text format, which is then analyzed for inappropriate speech using the same NLP model.

[1264] 3. Supported modules

[1265] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1266] The server detects inappropriate images, immediately deletes them, and notifies the user with a warning.

[1267] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1268] 4. Notification Module

[1269] The server sends notifications to the affected users about any inappropriate content or behavior detected.

[1270] The server provides the user with a detailed feedback report of any inappropriate content detected.

[1271] Specific examples

[1272] Example 1: Monitoring chat messages

[1273] User A sends a chat message in the metaverse saying "I hope you die."

[1274] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1275] The server automatically converts this to "Shut up, you idiot" and displays it.

[1276] Example 2: Image monitoring

[1277] User B uploads an inappropriate image (sexually explicit image).

[1278] The terminal sends this image to the server.

[1279] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1280] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1281] Example 3: Monitoring voice activity

[1282] User C shouts insults such as "You idiot" in the voice chat.

[1283] The terminal transmits the voice data to the server.

[1284] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1285] The server makes user C invisible, making him invisible to other users.

[1286] Prompt Sentence Examples

[1287] "Determine whether a particular chat message is inappropriate. Example: 'I hope you die.'"

[1288] "Please decide if this image is inappropriate. Here is the image data: [base64 encoded image data]"

[1289] This system will make communication within the metaverse and virtual stores safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[1290] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1291] Step 1:

[1292] Data collection

[1293] The device collects chat messages, uploaded image data, and voice chat data in real time, and transmits this data to a server.

[1294] Input: User chat messages, image data, and audio data

[1295] Output: Various data sent to the server

[1296] Step 2:

[1297] Chat message analysis

[1298] The server then inputs the collected chat messages into a natural language processing (NLP) model using a generative AI model, where semantic analysis and keyword extraction are performed to detect inappropriate words and expressions.

[1299] Input: The chat message the user said

[1300] Data processing: Semantic analysis and keyword extraction using NLP models

[1301] Output: A list of inappropriate words and phrases

[1302] Step 3:

[1303] Correcting inappropriate messages

[1304] The server automatically converts chat messages that are deemed inappropriate into appropriate language and displays them accordingly, for example, changing the word "die" to "shut up."

[1305] Input: A list of inappropriate words and phrases

[1306] Data processing: Automatically convert inappropriate words and expressions into appropriate ones

[1307] Output: Modified chat message

[1308] Step 4:

[1309] Image data analysis

[1310] The server then feeds the collected image data into computer vision models to detect inappropriate content, for example, using libraries such as OpenCV to identify sexually explicit images.

[1311] Input: User uploaded image data

[1312] Data processing: Image analysis using computer vision models

[1313] Output: Result of whether or not inappropriate content is included

[1314] Step 5:

[1315] Inappropriate image removal and notification

[1316] The server immediately deletes image data that is detected as inappropriate and notifies the user of a warning.

[1317] Input: Inappropriate content detection result

[1318] Data processing: Deletion of image data and creation of warning notices

[1319] Output: Deleted image data and a warning notice to the user

[1320] Step 6:

[1321] Analysis of audio data

[1322] The server inputs the collected voice data into a speech recognition model, converts it into text format, and then analyzes it for inappropriate comments using an NLP model.

[1323] Input: User voice chat data

[1324] Data processing: Converting voice data into text using a speech recognition model and analyzing it using an NLP model

[1325] Output: Translated audio data and a list of inappropriate words

[1326] Step 7:

[1327] Making inappropriate audio speakers invisible

[1328] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1329] Input: A list of inappropriate words

[1330] Data processing: Making users invisible

[1331] Output: Invisible user

[1332] Step 8:

[1333] Providing detailed feedback reports

[1334] The server provides the user with a detailed feedback report of the inappropriate content and behavior detected, including specific details of the behavior that was inappropriate.

[1335] Input: Data on inappropriate content and behavior detected

[1336] Data processing: Creating detailed feedback reports

[1337] Output: A feedback report provided to the user

[1338] 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.

[1339] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and combines an emotion engine that recognizes user emotions to provide more advanced responses. To implement this system, the following programs and processing means are included.

[1340] System configuration

[1341] 1. Data Collection Module

[1342] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[1343] 2. Data Analysis Module

[1344] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[1345] The server inputs the received image data into a computer vision model to detect inappropriate content.

[1346] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[1347] 3. Sentiment Analysis Module

[1348] The server inputs the collected data into an emotion engine to analyze the user's emotions.

[1349] The server enhances the detection of inappropriate content based on the results of sentiment analysis.

[1350] 4. Supported modules

[1351] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1352] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[1353] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1354] If the emotion engine recognizes a negative emotion, it provides mental support to the user.

[1355] 5. Notification Module

[1356] The server will then notify the user of any inappropriate content or behavior that has been detected.

[1357] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[1358] Specific examples

[1359] Example 1: Monitoring chat messages and sentiment

[1360] User A sends a chat message in the metaverse saying "I hope you die."

[1361] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1362] The server automatically converts this to "Shut up, you idiot" and displays it.

[1363] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[1364] Example 2: Image and Emotion Monitoring

[1365] User B uploads an inappropriate image (sexually explicit image).

[1366] The terminal sends this image to the server.

[1367] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1368] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1369] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[1370] Example 3: Monitoring voice utterances and emotions

[1371] User C shouts insults such as "You idiot" in the voice chat.

[1372] The terminal transmits the voice data to the server.

[1373] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1374] The server makes user C invisible, making him invisible to other users.

[1375] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[1376] This system will not only make communication within the metaverse safer and more comfortable, but also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[1377] The processing flow will be explained below.

[1378] Step 1:

[1379] A user types and sends a chat message within the metaverse.

[1380] The terminal collects the sent chat messages in real time and transmits them to the server.

[1381] Step 2:

[1382] A user uploads screenshots or image data within the metaverse.

[1383] The terminal transmits the uploaded image data to the server.

[1384] Step 3:

[1385] A user initiates a voice chat in the metaverse.

[1386] The device captures audio data in real time and sends it to the server.

[1387] Step 4:

[1388] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[1389] The server detects inappropriate words and expressions.

[1390] Step 5:

[1391] The image data received by the server is input into a computer vision model and the content is analyzed.

[1392] The server detects inappropriate content.

[1393] Step 6:

[1394] The server inputs the received voice data into a voice recognition model and converts it into text.

[1395] The server inputs the converted data into text into an NLP model and analyzes the content.

[1396] The server detects inappropriate comments.

[1397] Step 7:

[1398] The server inputs the collected data into an emotion engine to analyze the user's emotional state.

[1399] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[1400] Step 8:

[1401] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1402] Example: Converting the message "die" to "shut up."

[1403] At the same time, the server uses an emotion engine to evaluate the emotional state of user A, and if negative emotions are detected, provides mental support.

[1404] Step 9:

[1405] The server immediately deletes any inappropriate image data that is detected.

[1406] The server sends a notification to the user.

[1407] The server uses an emotion engine to evaluate User B's emotional state and considers an appropriate response.

[1408] Step 10:

[1409] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[1410] The fact that the server has become invisible is reflected to users other than the speaker.

[1411] The server uses an emotion engine to assess the emotional state of user C and provides support as needed.

[1412] Step 11:

[1413] The server notifies the user of any inappropriate content or behavior detected.

[1414] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[1415] Example 2

[1416] 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."

[1417] While communication within the metaverse is increasing, there is a risk that inappropriate words and actions may exist, preventing users from interacting safely in a healthy environment. There is a need for a system that can respond quickly and effectively to inappropriate words and actions and provide feedback that takes into account the user's emotional state.

[1418] 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.

[1419] In this invention, the server

[1420] A means for collecting communication data (chat messages, images, audio) sent by users within the metaverse;

[1421] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[1422] A means for automatically changing and displaying detected inappropriate words and expressions;

[1423] A means for analyzing the collected image data using a computer vision model to detect and remove inappropriate images;

[1424] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[1425] A means for making the speaker of the detected inappropriate voice invisible;

[1426] A means for inputting the collected communication data into an emotion analysis engine and analyzing the user's emotions;

[1427] A means for detecting negative emotions based on the emotion analysis results and providing mental support;

[1428] a means for notifying users of detected inappropriate content or behavior;

[1429] and a means for generating detailed feedback reports and providing them to users. This allows appropriate management of inappropriate speech and behavior within the metaverse, enabling users to interact with peace of mind in a healthy and safe communication environment.

[1430] "Communication data" is a general term that includes chat messages, image data, and voice data sent by users within the metaverse.

[1431] A "natural language processing model" is a computer algorithm or machine learning model that analyzes text data and understands and extracts grammar, context, and keywords.

[1432] A "computer vision model" is an algorithm or deep learning network that analyzes image data to detect inappropriate content or specific features.

[1433] A "voice recognition model" is a technology for converting voice data into text data and analyzing the content of speech.

[1434] An "emotion analysis engine" is an algorithm or AI technology that analyzes a user's text, image, and voice data to identify and evaluate their emotional state.

[1435] "Mental support" refers to the means and services that provide psychological and emotional care and assistance in response to negative emotions detected through analysis.

[1436] "Invisibility" is a function that restricts the actions of a user who has behaved inappropriately by making that user invisible to other users.

[1437] "Feedback Report" means a written or digital message that provides detailed explanation and advice regarding detected inappropriate behavior or content.

[1438] "Inappropriate content" refers to chat messages, images, audio, etc. within the metaverse that are deemed socially or ethically problematic.

[1439] "User notification" is a function that notifies the relevant user about inappropriate behavior or content detected by the system.

[1440] MODE FOR CARRYING OUT THE INVENTION

[1441] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and provides advanced responses that take into account the emotional state of the user. To implement this system, the following programs and processing means are included.

[1442] System Configuration

[1443] 1. Data Collection Module

[1444] The terminal collects chat messages, image data, and voice data sent by users within the metaverse in real time and transmits them to a server.

[1445] 2. Data Analysis Module

[1446] The server inputs the collected chat messages into a natural language processing (NLP) model to perform semantic analysis and keyword extraction, using Python's NLTK and spaCy.

[1447] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content.

[1448] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API), converts it into text, and then analyzes the text data with an NLP model to detect inappropriate speech.

[1449] 3. Sentiment Analysis Module

[1450] The server inputs the collected data into an emotion engine (e.g., Microsoft Azure emotion analysis API, snownlp, etc.) to analyze the user's emotions.

[1451] Based on the results of sentiment analysis, we will strengthen the detection of inappropriate content and consider taking action as necessary.

[1452] 4. Supported modules

[1453] The server automatically converts detected inappropriate chat messages into appropriate expressions, immediately deletes inappropriate images, and notifies users with a warning. If inappropriate voice comments are detected, the person making the comment becomes invisible to other users.

[1454] If the emotion engine recognizes a negative emotion, the server provides mental support to the user.

[1455] 5. Notification Module

[1456] The server notifies users of any inappropriate content or behavior detected and provides detailed feedback reports to indicate which actions were inappropriate.

[1457] Specific examples

[1458] Example 1: Monitoring chat messages and sentiment

[1459] User A sends a chat message in the metaverse saying "I hope you die."

[1460] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1461] The server automatically converts this to "Shut up, you idiot" and displays it.

[1462] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[1463] Example 2: Image and Emotion Monitoring

[1464] User B uploads an inappropriate image (sexually explicit image).

[1465] The terminal sends this image to the server.

[1466] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1467] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1468] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[1469] Example 3: Monitoring voice utterances and emotions

[1470] User C shouts insults such as "You idiot" in the voice chat.

[1471] The terminal transmits the voice data to the server.

[1472] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1473] The server makes user C invisible, making him invisible to other users.

[1474] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[1475] This system will not only make communication within the metaverse safer and more comfortable, but will also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[1476] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1477] Step 1: Data collection

[1478] The terminal collects chat messages, image data, and voice data sent by users in the metaverse in real time. This data is input and sent to a server via a network.

[1479] Specific behavior:

[1480] The user types "Hello" in the chat window and presses the send button.

[1481] The device captures the text data "Hello" and sends it to the server.

[1482] Users take screenshots of their avatars and upload them to a shared gallery in the metaverse.

[1483] The device captures the screenshot image data and sends it to the server.

[1484] Step 2: Chat message analysis

[1485] The server inputs the received chat messages into a natural language processing (NLP) model to extract keywords and perform semantic analysis. The input is the chat message, and the output is the analysis result. Specifically, Python's NLTK and spaCy are used.

[1486] Specific behavior:

[1487] The server receives the message "Hello" and asks the natural language processing model to analyze it.

[1488] The NLP model tags "hello" as a greeting and returns the result to the server.

[1489] Step 3: Image data analysis

[1490] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content. The input is the image data, and the output is the detection result of inappropriate content.

[1491] Specific behavior:

[1492] The server receives the screenshots uploaded by the user.

[1493] Computer vision models scan images to determine if they contain inappropriate content.

[1494] The server receives the inappropriate content detection result and removes the image.

[1495] Step 4: Audio data analysis

[1496] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API) and converts it into text format. The text data is then analyzed using an NLP model to detect inappropriate speech. The input is the audio data, and the output is the analyzed text data.

[1497] Specific behavior:

[1498] A user says "You're such an idiot" in voice chat.

[1499] The device captures the audio data and sends it to the server.

[1500] The server converts the voice data into text and generates the text data "You're an idiot."

[1501] The NLP model determines that "baka" is an inappropriate word and returns the result to the server.

[1502] Step 5: Sentiment Analysis

[1503] The server inputs the collected text, image, and voice data into an emotion engine (e.g., Microsoft Azure's emotion analysis API, snownlp, etc.) to analyze the user's emotions. The input is data (text, image, voice), and the output is an evaluation of the user's emotional state.

[1504] Specific behavior:

[1505] The server inputs the chat message "You're such an idiot" into an emotion engine, which analyzes the emotional state of the message.

[1506] The emotion engine detects "anger" and returns the result to the server.

[1507] Step 6: Address inappropriate content

[1508] The server automatically takes appropriate action as needed based on the results of the sentiment analysis. The inputs are the detection results of inappropriate content and the sentiment analysis results, and the outputs are the modified message, the deletion of inappropriate content, and the results of making the user invisible.

[1509] Specific behavior:

[1510] The server automatically converts "You're an idiot" to "Shut up you" and displays it to the sending user.

[1511] If an inappropriate image is detected, it will be removed and a warning message will be sent to the uploader.

[1512] The server turns users who make inappropriate vocal comments into invisible people, making them invisible to other users.

[1513] Step 7: Providing mental support

[1514] The server provides mental support to the user when a negative emotion is recognized by the emotion engine. The input is the emotion analysis result, and the output is the support provided.

[1515] Specific behavior:

[1516] If the server recognizes "anger," it provides the user with a message explaining how to relax and a link to a counseling service.

[1517] Step 8: User Notification

[1518] The server notifies the user of any detected inappropriate content or behavior and also generates and provides a detailed feedback report. The input is the detection result of inappropriate content, and the output is the user notification and feedback report.

[1519] Specific behavior:

[1520] The server sends a notification to the user about the detected inappropriate content.

[1521] The report states, "Your comment, 'You're an idiot,' was deemed inappropriate. Please be careful next time."

[1522] (Application example 2)

[1523] 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."

[1524] In current brick-and-mortar stores, communication between customers and store staff can be inappropriate, and customers' feelings can be poorly understood, leading to delayed responses. This can lead to lower customer satisfaction, which can have a negative impact on store ratings and sales. Furthermore, because it is difficult to respond appropriately in real time, advanced management of the customer experience in brick-and-mortar stores is required.

[1525] The identification processing 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 collecting chat messages; means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions; means for automatically changing and displaying the corresponding message when inappropriate words or expressions are detected; means for collecting and analyzing inappropriate images and deleting the detected inappropriate images; means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments; means for making the speaker of the corresponding voice invisible when inappropriate voice is detected; means for notifying the user of detected inappropriate content or behavior; means for analyzing the user's emotions, including an emotion analysis engine; means for providing an appropriate response when the user's emotions are recognized as negative; and means for monitoring conversations and behaviors between customers and store clerks in a physical store and notifying the store clerk in real time based on the emotion analysis results. This enables real-time monitoring of customer experiences in physical stores and enables prompt and appropriate responses.

[1526] The "metaverse" is a virtual space that exists on the Internet, a virtual reality environment in which users can interact using avatars.

[1527] "Communication safety" refers to the standards and measures to ensure that interactions between users are appropriate and healthy.

[1528] "Monitoring" refers to the act of watching users' behavior and messages in real time to detect inappropriate behavior or expressions.

[1529] "Management" refers to the act of taking appropriate action against detected inappropriate behavior or content and operating the entire system.

[1530] A "chat message" is a communication message sent by a user in text format.

[1531] "Means of collection" refers to the equipment or software mechanisms used to collect data sent by users on a server.

[1532] A "natural language processing model" refers to the algorithms and technologies that allow computers to understand and analyze human language.

[1533] "Inappropriate words and phrases" means language or phrases that may be offensive, insulting, or otherwise objectionable.

[1534] "Means for automatically changing and displaying" refers to technology that instantly corrects detected inappropriate content and replaces it with safe expressions before displaying it.

[1535] "Inappropriate Images" are images that contain sexually explicit material, violence, or other inappropriate visual content.

[1536] "Means for analyzing and deleting detected inappropriate images" refers to technology that analyzes collected images and removes them if they are deemed inappropriate.

[1537] "Voice data" refers to data that is a digital recording of a user's voice.

[1538] A "speech recognition model" refers to the algorithms and techniques used to convert speech into text.

[1539] "Invisibility methods" are techniques that make inappropriate commenters invisible to other users.

[1540] "Means of notification" refers to the mechanism by which the system communicates warnings and information to the user.

[1541] "Sentiment analysis engine" refers to software or hardware for analyzing user emotions from collected data.

[1542] "Means for providing appropriate responses when emotions are perceived as negative" refers to technology for providing appropriate support or notifications when a user's emotions are determined to be negative.

[1543] "Means for monitoring conversations and behavior between customers and store clerks in physical stores and notifying store clerks in real time based on the results of emotion analysis" is a system that monitors interactions between customers and store clerks in physical stores and immediately notifies store clerks based on the analyzed emotion results.

[1544] The present invention is a system for monitoring customer communications in real time in a brick-and-mortar store and providing safe and effective responses. The system includes the following main components:

[1545] 1. Data Collection Module

[1546] The server uses smart glasses or head-mounted displays to monitor the conversations and behavior of customers and store staff in real time. These devices collect customer voice and facial expression data and send it to the server. Specific hardware examples include Google Glass and Microsoft HoloLens.

[1547] 2. Data Analysis Module

[1548] The server converts the collected voice data into text using a speech recognition model. This speech recognition is performed using the Google Cloud Speech-to-Text API. The converted text is then analyzed using a natural language processing (NLP) model to recognize the content and tone of the conversation. This analysis is performed using Google Cloud Natural Language and OpenAI GPT-3.

[1549] The server then inputs the collected image data into a sentiment analysis engine to analyze the customer's emotional state, using the Affectiva SDK and Amazon Rekognition.

[1550] 3. Supported modules

[1551] If a customer shows signs of dissatisfaction or discomfort, the server will immediately send a notification to the store clerk's smart glasses, displaying an alert to prompt appropriate action, for example, a notification such as: "The customer is dissatisfied. Please take action."

[1552] It also includes a function that automatically corrects inappropriate comments or behavior when it is detected, converting them into safe language and displaying it to customers.

[1553] 4. Notification Module

[1554] At the end of the day, the server automatically generates a customer feedback report and sends a notification to the salesperson and manager, allowing the salesperson to identify specific areas for improvement and strive to improve customer service. Example: "Today's feedback report: List of complaints and areas for improvement."

[1555] Example prompt sentences

[1556] Customer facial expression prompt: "What emotions do you see in the customer's facial expressions?"

[1557] Example input to a generative AI model: "This customer's mouth corners are downturned and their eyes are narrowed. What emotion is this customer experiencing?"

[1558] Conversation analysis prompt: "Does this conversation suggest that the customer is unhappy?"

[1559] Example input to a generative AI model: "Customer: 'Hey, isn't this too expensive? It's a lot more than I expected.' Salesperson: 'Sorry, but we're not currently offering any special discounts, so this is the price.'"

[1560] This allows for real-time monitoring of customer experiences in physical stores, enabling quick and appropriate responses, ultimately improving customer satisfaction.

[1561] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1562] Step 1:

[1563] The server collects real-time conversation and behavior data between customers and store staff in a physical store through smart glasses or a head-mounted display. The device captures voice and image data and sends them to the server. The input is voice and image data, and the output is the raw data sent to the server.

[1564] Step 2:

[1565] The server inputs the received voice data into a voice recognition model and converts it into text format. The Google Cloud Speech-to-Text API is used for voice recognition. The input is voice data and the output is text data. In this step, the voice data is converted into conversational content.

[1566] Step 3:

[1567] The server inputs the converted text data into a natural language processing (NLP) model to analyze the semantics of the conversation. Google Cloud Natural Language and OpenAI GPT-3 are used for the NLP model. The input is text data, and the output is the tone and meaning of the analyzed conversation. In this step, the content and tone of the conversation are analyzed.

[1568] Step 4:

[1569] The server inputs the collected image data into an emotion analysis engine to analyze the customer's emotional state. Emotion analysis uses the Affectiva SDK and Amazon Rekognition. The input is image data, and the output is the analysis result of the customer's emotional state. In this step, emotions are read from the customer's facial expressions.

[1570] Step 5:

[1571] The server integrates the analyzed conversation content with the customer's emotional state, and if the customer shows signs of dissatisfaction or discomfort, it sends a notification to the store clerk's smart glasses. The input is the analysis result, and the output is a notification to the store clerk. Specifically, the clerk will see a message such as, "The customer is dissatisfied. Please take action."

[1572] Step 6:

[1573] When the server detects inappropriate comments or behavior, it automatically modifies the content, converts it into safer language, and displays it. The input is the data of the inappropriate comments or behavior, and the output is the modified content. Specifically, offensive language is changed to softer language.

[1574] Step 7:

[1575] At the end of each day, the server automatically generates a customer feedback report and sends notifications to store staff and managers. The input is the customer interaction data for the day, and the output is a feedback report. Specifically, the report contains content such as "Today's feedback report: list of complaints and areas for improvement."

[1576] 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.

[1577] 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.

[1578] 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.

[1579] [Fourth embodiment]

[1580] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1581] 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.

[1582] 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).

[1583] 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.

[1584] 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.

[1585] 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).

[1586] 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.

[1587] 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.

[1588] 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.

[1589] 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.

[1590] 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.

[1591] 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.

[1592] 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."

[1593] The present invention relates to a system for automatically monitoring and managing the safety of communications within the metaverse. To implement this system, the following programs and processing means are included:

[1594] System configuration

[1595] 1. Data Collection Module

[1596] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[1597] 2. Data Analysis Module

[1598] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[1599] The server inputs the received image data into a computer vision model to detect inappropriate content.

[1600] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[1601] 3. Supported modules

[1602] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1603] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[1604] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1605] 4. Notification Module

[1606] The server notifies the affected users about any inappropriate content or behavior detected.

[1607] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[1608] Specific examples

[1609] Example 1: Monitoring chat messages

[1610] User A sends a chat message in the metaverse saying "I hope you die."

[1611] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1612] The server automatically converts this to "Shut up, you idiot" and displays it.

[1613] Example 2: Image monitoring

[1614] User B uploads an inappropriate image (sexually explicit image).

[1615] The terminal sends this image to the server.

[1616] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1617] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1618] Example 3: Monitoring voice activity

[1619] User C shouts insults such as "You idiot" in the voice chat.

[1620] The terminal transmits the voice data to the server.

[1621] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1622] The server makes user C invisible, making him invisible to other users.

[1623] This system will make communication within the metaverse safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[1624] The processing flow will be explained below.

[1625] Step 1:

[1626] A user types and sends a chat message within the metaverse.

[1627] The terminal collects the sent chat messages in real time and transmits them to the server.

[1628] Step 2:

[1629] A user uploads screenshots or image data within the metaverse.

[1630] The terminal transmits the uploaded image data to the server.

[1631] Step 3:

[1632] A user initiates a voice chat in the metaverse.

[1633] The device captures audio data in real time and sends it to the server.

[1634] Step 4:

[1635] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[1636] The server detects inappropriate words and expressions.

[1637] Step 5:

[1638] The image data received by the server is input into a computer vision model and the content is analyzed.

[1639] The server detects inappropriate content.

[1640] Step 6:

[1641] The server inputs the received voice data into a voice recognition model and converts it into text.

[1642] The server inputs the converted data into text into an NLP model and analyzes the content.

[1643] The server detects inappropriate comments.

[1644] Step 7:

[1645] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1646] Example: Converting the message "die" to "shut up."

[1647] Step 8:

[1648] The server immediately deletes any inappropriate image data that is detected.

[1649] The server sends a notification to the user.

[1650] Step 9:

[1651] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[1652] The fact that the server has become invisible is reflected to users other than the speaker.

[1653] Step 10:

[1654] The server will notify the user of any inappropriate content or violating behavior detected.

[1655] The server provides a detailed feedback report to the user.

[1656] Example 1

[1657] 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."

[1658] In the metaverse, inappropriate comments, images, and audio are shared between users, potentially compromising user safety and comfort. Conventional systems lack effective monitoring and management methods, and it takes time for inappropriate content to be detected, potentially adversely affecting users during that time. Therefore, there is a need for a system that can monitor communication in the metaverse in real time and respond immediately.

[1659] 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.

[1660] In this invention, the server includes means for collecting communication data from user devices, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for analyzing the collected image data using a computer vision model to detect and delete inappropriate content, means for converting the collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate remarks, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, and means for notifying the user of detected inappropriate content or behavior. This makes it possible to monitor communication within the metaverse in real time and respond immediately to inappropriate content or behavior.

[1661] A "user device" is an electronic device used by a user to access and communicate with the metaverse.

[1662] "Communication data" refers to information such as chat messages, image data, and audio data exchanged between users.

[1663] A "natural language processing model" is an artificial intelligence model for analyzing and understanding text data.

[1664] A "chat message" is a message sent in text format by a user within the metaverse.

[1665] "Inappropriate words and phrases" are words or phrases that may be offensive, insulting, or upsetting to a user in a communication.

[1666] A "computer vision model" is an artificial intelligence model for analyzing and understanding image data.

[1667] "Inappropriate Content" refers to images or videos that disrupt healthy communication within the Metaverse and may cause mental or emotional harm to other users.

[1668] A "voice recognition model" is an artificial intelligence model for converting voice data into text data.

[1669] "Invisibility" is a process in which the speaker or actor of detected inappropriate behavior is made invisible to other users.

[1670] A "notification" is a message or alert that informs a user that a particular action or statement is inappropriate.

[1671] This invention relates to a system that automatically monitors and manages the safety of communications within the metaverse. The system collects and analyzes chat messages, image data, and audio data exchanged between users in real time, and aims to immediately detect and respond to inappropriate content.

[1672] System configuration

[1673] To implement this system, the following hardware and software are used:

[1674] Hardware

[1675] 1. User's device

[1676] Communication-enabled devices such as personal computers, smartphones, and tablets

[1677] These devices are used to access the metaverse server and send and receive data such as chat, audio, and images.

[1678] 2. Server

[1679] High-performance server computers

[1680] It has the computing power necessary to analyze chat messages, perform voice recognition, and image analysis.

[1681] software

[1682] 1. Natural Language Processing Model (NLP Model)

[1683] Examples: BERT, GPT-3

[1684] It is used to analyze chat messages and detect inappropriate words and phrases.

[1685] 2. Computer Vision Models

[1686] Examples: ResNet, YOLO

[1687] Used to analyze image data and detect inappropriate content.

[1688] 3. Speech Recognition Model

[1689] Examples: Google Cloud Speech-to-Text, DeepSpeech

[1690] It is used to convert audio data into text format.

[1691] Example of operation

[1692] Example 1: Monitoring chat messages

[1693] User A sends a chat message in the metaverse saying "I hope you die."

[1694] The terminal immediately sends this message to the server.

[1695] The server inputs this message into a natural language processing model, which analyzes and detects that the word "die" is inappropriate.

[1696] The server automatically converts the message to "Shut up you" and displays it in an appropriate format.

[1697] Example 2: Image monitoring

[1698] User B uploads an inappropriate image (e.g., a sexually explicit image).

[1699] The terminal transmits this image to the server.

[1700] The server inputs the received images into a computer vision model to analyze and detect any inappropriate content.

[1701] The server immediately deletes the image and notifies User B that "inappropriate content has been posted."

[1702] Example 3: Monitoring voice activity

[1703] User C shouts insults such as "You idiot" in the voice chat.

[1704] The terminal transmits the voice data to the server.

[1705] The server converts the voice data into text using a voice recognition model and analyzes it using a natural language processing model.

[1706] The server recognizes the word "idiot" as inappropriate and turns User C invisible, making him invisible to other users.

[1707] Prompt Sentence Examples

[1708] prompt:

[1709] If I send a chat message in the Metaverse like "I hope you die," how does the system handle it?

[1710] response:

[1711] 1. The device sends a chat message saying "You should die" to the server in real time.

[1712] 2. The server analyzes the message using a natural language processing model and detects that the word "die" is inappropriate.

[1713] 3. The server automatically converts the message to "Shut up, you idiot" and displays it.

[1714] 4. The server notifies User A of the revised message and informs him that the original message was inappropriate.

[1715] This system collects and analyzes data in real time and responds immediately to maintain healthy communication within the metaverse, providing a safe and secure environment for users.

[1716] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1717] Step 1: Data collection

[1718] The terminal collects communication data from the user's device. The input data includes chat messages, image data, and voice data. The collected data is temporarily stored in memory and then the next step is performed.

[1719] Specific behavior:

[1720] The user types "I wish you would die" into the chat window and presses the send button.

[1721] A user says "You idiot" in voice chat.

[1722] A user posts an image using the image upload feature.

[1723] Step 2: Send data

[1724] The terminal transmits the collected data to the server in real time. The input data is the collected communication data, which is transmitted as is to the server via the network. The server adds the received data to a processing queue.

[1725] Specific behavior:

[1726] The device sends the chat message "You should die" data to the server.

[1727] The device sends the recorded audio data to the server.

[1728] The terminal transmits the uploaded image file to the server.

[1729] Step 3: Analyzing chat messages

[1730] The server inputs the received chat message into a natural language processing model (NLP model) to analyze the meaning and keywords. The input data is the received chat message, and the output is the judgment result of whether it contains inappropriate words or expressions.

[1731] Specific behavior:

[1732] The server inputs the chat message "You should die" into the NLP model.

[1733] The server analyzes the word "die" as inappropriate and obtains the judgment result.

[1734] Step 4: Analyze the images

[1735] The server inputs the received image data into a computer vision model to detect inappropriate content. The input data is the received image data, and the output is a judgment result on whether the image is inappropriate or not.

[1736] Specific behavior:

[1737] The server inputs the received image data into a computer vision model.

[1738] The server uses image analysis models to detect explicit elements in the image.

[1739] Step 5: Analyze the audio data

[1740] The server inputs the received voice data into a voice recognition model, converts it into text format, and then analyzes it using a natural language processing model. The input data is the received voice data, and the output is a judgment result on whether or not it contains inappropriate remarks.

[1741] Specific behavior:

[1742] The server inputs the voice data into a speech recognition model and converts it into text.

[1743] The server inputs the converted text data into an NLP model and recognizes the word "idiot" as inappropriate.

[1744] Step 6: Implementing response measures

[1745] The server takes appropriate action based on the analysis results. The input data is the analysis results, and if any inappropriate content is detected, the server takes appropriate action. The output is an appropriately modified message, image deletion, user invisibility, etc.

[1746] Specific behavior:

[1747] The server converts "You should die" to "Shut up" and displays it.

[1748] The server deletes inappropriate image files.

[1749] The server makes user C invisible, making him invisible to other users.

[1750] Step 7: Notification

[1751] The server sends notifications to the relevant users about the detected inappropriate behavior. The input data is the result of the action taken, and the output is a notification message to the user.

[1752] Specific behavior:

[1753] The server notifies User A that "Your comment, 'You should die,' has been deemed inappropriate."

[1754] The server warns User B that the posted image has been deleted because it is inappropriate.

[1755] The server notifies User C that "Your comment 'You idiot' has been deemed inappropriate and made invisible."

[1756] (Application example 1)

[1757] 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."

[1758] Communication between users in the metaverse or virtual store may not always be secure. The free circulation of inappropriate comments, images, and audio not only impairs the user experience, but also risks reducing the reliability of the platform itself. Real-time communication, in particular, requires immediate response, necessitating efficient monitoring and management systems. Providing appropriate feedback to users in the event of inappropriate behavior is also important from the perspective of preventing recurrence and user education. Therefore, the present invention aims to provide a system that automatically monitors and manages the security of communication in the metaverse or virtual store, and provides appropriate responses and feedback.

[1759] 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.

[1760] In this invention, the server includes means for collecting chat messages, means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions, means for automatically changing and displaying the corresponding messages when inappropriate words or expressions are detected, means for collecting and analyzing inappropriate images and deleting the detected inappropriate images, means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments, means for making the speaker of the corresponding voice invisible when inappropriate voice is detected, means for notifying the user of detected inappropriate content or behavior, means for providing the user with a detailed feedback report on the detected inappropriate content, and means for monitoring communication within the virtual store in real time and taking specific measures immediately. This enables safe and smooth communication between users within the metaverse and virtual store, thereby improving the reliability of the platform.

[1761] The "metaverse" is a three-dimensional virtual space built on the Internet, a platform where users can interact with each other and engage in activities through avatars.

[1762] A "virtual store" is a virtual store that mimics a real store, where users can purchase digital items and services and interact with other users.

[1763] "Chat messages" are messages exchanged between users in real time in text format.

[1764] A "natural language processing model" is an artificial intelligence technology for understanding and analyzing human language, extracting meaning from text data.

[1765] "Inappropriate words and phrases" are words or phrases that contain offensive, discriminatory or objectionable content and are inappropriate for communication.

[1766] To "collect" means to obtain specific data or information and gather it in a designated location.

[1767] "Analysis" means examining data or information in detail to understand and interpret its content.

[1768] A "voice recognition model" is an artificial intelligence technology for converting voice data into text, recognizing speech as text information.

[1769] "Invisibility" refers to the process of making a specific user invisible to other users.

[1770] A "feedback report" is a document that provides detailed information about inappropriate behavior or content and guidelines for improvement.

[1771] "Real-time monitoring" refers to the process of instantly monitoring data and behavior and taking necessary action on the spot.

[1772] "Responding immediately" means immediately taking appropriate action or measures against any inappropriate behavior or content that is detected.

[1773] The present invention relates to a system for automatically monitoring and managing communications within a metaverse or virtual store. To implement this system, the following programs and processing means are included:

[1774] System configuration

[1775] 1. Data Collection Module

[1776] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[1777] 2. Data Analysis Module

[1778] The server inputs chat messages into a natural language processing (NLP) model using a generative AI model to perform semantic analysis and keyword extraction.

[1779] The server inputs the received image data into a computer vision model (e.g., OpenCV) to detect inappropriate content.

[1780] The server inputs the collected audio data into a speech recognition model (e.g., a Transformers speech recognition model) and converts it into text format, which is then analyzed for inappropriate speech using the same NLP model.

[1781] 3. Supported modules

[1782] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1783] The server detects inappropriate images, immediately deletes them, and notifies the user with a warning.

[1784] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1785] 4. Notification Module

[1786] The server sends notifications to the affected users about any inappropriate content or behavior detected.

[1787] The server provides the user with a detailed feedback report of any inappropriate content detected.

[1788] Specific examples

[1789] Example 1: Monitoring chat messages

[1790] User A sends a chat message in the metaverse saying "I hope you die."

[1791] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1792] The server automatically converts this to "Shut up, you idiot" and displays it.

[1793] Example 2: Image monitoring

[1794] User B uploads an inappropriate image (sexually explicit image).

[1795] The terminal sends this image to the server.

[1796] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1797] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1798] Example 3: Monitoring voice activity

[1799] User C shouts insults such as "You idiot" in the voice chat.

[1800] The terminal transmits the voice data to the server.

[1801] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1802] The server makes user C invisible, making him invisible to other users.

[1803] Prompt Sentence Examples

[1804] "Determine whether a particular chat message is inappropriate. Example: 'I hope you die.'"

[1805] "Please decide if this image is inappropriate. Here is the image data: [base64 encoded image data]"

[1806] This system will make communication within the metaverse and virtual stores safer and more comfortable, allowing users to interact with other users in a healthy environment, which will promote the use of the metaverse.

[1807] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1808] Step 1:

[1809] Data collection

[1810] The device collects chat messages, uploaded image data, and voice chat data in real time, and transmits this data to a server.

[1811] Input: User chat messages, image data, and audio data

[1812] Output: Various data sent to the server

[1813] Step 2:

[1814] Chat message analysis

[1815] The server then inputs the collected chat messages into a natural language processing (NLP) model using a generative AI model, where semantic analysis and keyword extraction are performed to detect inappropriate words and expressions.

[1816] Input: The chat message the user said

[1817] Data processing: Semantic analysis and keyword extraction using NLP models

[1818] Output: A list of inappropriate words and phrases

[1819] Step 3:

[1820] Correcting inappropriate messages

[1821] The server automatically converts chat messages that are deemed inappropriate into appropriate language and displays them accordingly, for example, changing the word "die" to "shut up."

[1822] Input: A list of inappropriate words and phrases

[1823] Data processing: Automatically convert inappropriate words and expressions into appropriate ones

[1824] Output: Modified chat message

[1825] Step 4:

[1826] Image data analysis

[1827] The server then feeds the collected image data into computer vision models to detect inappropriate content, for example, using libraries such as OpenCV to identify sexually explicit images.

[1828] Input: User uploaded image data

[1829] Data processing: Image analysis using computer vision models

[1830] Output: Result of whether or not inappropriate content is included

[1831] Step 5:

[1832] Inappropriate image removal and notification

[1833] The server immediately deletes image data that is detected as inappropriate and notifies the user of a warning.

[1834] Input: Inappropriate content detection result

[1835] Data processing: Deletion of image data and creation of warning notices

[1836] Output: Deleted image data and a warning notice to the user

[1837] Step 6:

[1838] Analysis of audio data

[1839] The server inputs the collected voice data into a speech recognition model, converts it into text format, and then analyzes it for inappropriate comments using an NLP model.

[1840] Input: User voice chat data

[1841] Data processing: Converting voice data into text using a speech recognition model and analyzing it using an NLP model

[1842] Output: Translated audio data and a list of inappropriate words

[1843] Step 7:

[1844] Making inappropriate audio speakers invisible

[1845] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1846] Input: A list of inappropriate words

[1847] Data processing: Making users invisible

[1848] Output: Invisible user

[1849] Step 8:

[1850] Providing detailed feedback reports

[1851] The server provides the user with a detailed feedback report of the inappropriate content and behavior detected, including specific details of the behavior that was inappropriate.

[1852] Input: Data on inappropriate content and behavior detected

[1853] Data processing: Creating detailed feedback reports

[1854] Output: A feedback report provided to the user

[1855] 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.

[1856] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and combines an emotion engine that recognizes user emotions to provide more advanced responses. To implement this system, the following programs and processing means are included.

[1857] System configuration

[1858] 1. Data Collection Module

[1859] The terminal collects chat messages, image data, and voice data sent by the user within the metaverse in real time and transmits them to the server.

[1860] 2. Data Analysis Module

[1861] The server inputs chat messages into a natural language processing (NLP) model for semantic analysis and keyword extraction.

[1862] The server inputs the received image data into a computer vision model to detect inappropriate content.

[1863] The server feeds the audio data into a speech recognition model, converts it into text, and then uses an NLP model to analyze it for inappropriate comments.

[1864] 3. Sentiment Analysis Module

[1865] The server inputs the collected data into an emotion engine to analyze the user's emotions.

[1866] The server enhances the detection of inappropriate content based on the results of sentiment analysis.

[1867] 4. Supported modules

[1868] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1869] The server immediately deletes any inappropriate images detected and notifies the user with a warning.

[1870] When an inappropriate voice comment is detected, the server turns the person making the comment into an invisible person, making them invisible to other users.

[1871] If the emotion engine recognizes a negative emotion, it provides mental support to the user.

[1872] 5. Notification Module

[1873] The server will then notify the user of any inappropriate content or behavior that has been detected.

[1874] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[1875] Specific examples

[1876] Example 1: Monitoring chat messages and sentiment

[1877] User A sends a chat message in the metaverse saying "I hope you die."

[1878] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1879] The server automatically converts this to "Shut up, you idiot" and displays it.

[1880] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[1881] Example 2: Image and Emotion Monitoring

[1882] User B uploads an inappropriate image (sexually explicit image).

[1883] The terminal sends this image to the server.

[1884] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1885] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1886] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[1887] Example 3: Monitoring voice utterances and emotions

[1888] User C shouts insults such as "You idiot" in the voice chat.

[1889] The terminal transmits the voice data to the server.

[1890] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1891] The server makes user C invisible, making him invisible to other users.

[1892] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[1893] This system will not only make communication within the metaverse safer and more comfortable, but also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[1894] The processing flow will be explained below.

[1895] Step 1:

[1896] A user types and sends a chat message within the metaverse.

[1897] The terminal collects the sent chat messages in real time and transmits them to the server.

[1898] Step 2:

[1899] A user uploads screenshots or image data within the metaverse.

[1900] The terminal transmits the uploaded image data to the server.

[1901] Step 3:

[1902] A user initiates a voice chat in the metaverse.

[1903] The device captures audio data in real time and sends it to the server.

[1904] Step 4:

[1905] The server inputs the received chat messages into a natural language processing (NLP) model to analyze the content.

[1906] The server detects inappropriate words and expressions.

[1907] Step 5:

[1908] The image data received by the server is input into a computer vision model and the content is analyzed.

[1909] The server detects inappropriate content.

[1910] Step 6:

[1911] The server inputs the received voice data into a voice recognition model and converts it into text.

[1912] The server inputs the converted data into text into an NLP model and analyzes the content.

[1913] The server detects inappropriate comments.

[1914] Step 7:

[1915] The server inputs the collected data into an emotion engine to analyze the user's emotional state.

[1916] The server evaluates the user's emotional state based on the analysis results of the emotion engine.

[1917] Step 8:

[1918] The server automatically converts detected inappropriate chat messages into appropriate expressions.

[1919] Example: Converting the message "die" to "shut up."

[1920] At the same time, the server uses an emotion engine to evaluate the emotional state of user A, and if negative emotions are detected, provides mental support.

[1921] Step 9:

[1922] The server immediately deletes any inappropriate image data that is detected.

[1923] The server sends a notification to the user.

[1924] The server uses an emotion engine to evaluate User B's emotional state and considers an appropriate response.

[1925] Step 10:

[1926] The server turns the speaker of the detected inappropriate voice utterance into an invisible person.

[1927] The fact that the server has become invisible is reflected to users other than the speaker.

[1928] The server uses an emotion engine to assess the emotional state of user C and provides support as needed.

[1929] Step 11:

[1930] The server notifies the user of any inappropriate content or behavior detected.

[1931] The server provides a detailed feedback report to the user, indicating which actions were inappropriate.

[1932] Example 2

[1933] 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."

[1934] While communication within the metaverse is increasing, there is a risk that inappropriate words and actions may exist, preventing users from interacting safely in a healthy environment. There is a need for a system that can respond quickly and effectively to inappropriate words and actions and provide feedback that takes into account the user's emotional state.

[1935] 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.

[1936] In this invention, the server

[1937] A means for collecting communication data (chat messages, images, audio) sent by users within the metaverse;

[1938] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[1939] A means for automatically changing and displaying detected inappropriate words and expressions;

[1940] A means for analyzing the collected image data using a computer vision model to detect and remove inappropriate images;

[1941] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[1942] A means for making the speaker of the detected inappropriate voice invisible;

[1943] A means for inputting the collected communication data into an emotion analysis engine and analyzing the user's emotions;

[1944] A means for detecting negative emotions based on the emotion analysis results and providing mental support;

[1945] a means for notifying users of detected inappropriate content or behavior;

[1946] and a means for generating detailed feedback reports and providing them to users. This allows appropriate management of inappropriate speech and behavior within the metaverse, enabling users to interact with peace of mind in a healthy and safe communication environment.

[1947] "Communication data" is a general term that includes chat messages, image data, and voice data sent by users within the metaverse.

[1948] A "natural language processing model" is a computer algorithm or machine learning model that analyzes text data and understands and extracts grammar, context, and keywords.

[1949] A "computer vision model" is an algorithm or deep learning network that analyzes image data to detect inappropriate content or specific features.

[1950] A "voice recognition model" is a technology for converting voice data into text data and analyzing the content of speech.

[1951] An "emotion analysis engine" is an algorithm or AI technology that analyzes a user's text, image, and voice data to identify and evaluate their emotional state.

[1952] "Mental support" refers to the means and services that provide psychological and emotional care and assistance in response to negative emotions detected through analysis.

[1953] "Invisibility" is a function that restricts the actions of a user who has behaved inappropriately by making that user invisible to other users.

[1954] "Feedback Report" means a written or digital message that provides detailed explanation and advice regarding detected inappropriate behavior or content.

[1955] "Inappropriate content" refers to chat messages, images, audio, etc. within the metaverse that are deemed to be socially or ethically problematic.

[1956] "User notification" is a function that notifies the relevant user about inappropriate behavior or content detected by the system.

[1957] MODE FOR CARRYING OUT THE INVENTION

[1958] The present invention is a system that automatically monitors and manages the safety of communication within the metaverse, and provides advanced responses that take into account the emotional state of the user. To implement this system, the following programs and processing means are included.

[1959] System Configuration

[1960] 1. Data Collection Module

[1961] The terminal collects chat messages, image data, and voice data sent by users within the metaverse in real time and transmits them to a server.

[1962] 2. Data Analysis Module

[1963] The server inputs the collected chat messages into a natural language processing (NLP) model to perform semantic analysis and keyword extraction, using Python's NLTK and spaCy.

[1964] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content.

[1965] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API), converts it into text, and then analyzes the text data with an NLP model to detect inappropriate speech.

[1966] 3. Sentiment Analysis Module

[1967] The server inputs the collected data into an emotion engine (e.g., Microsoft Azure emotion analysis API, snownlp, etc.) to analyze the user's emotions.

[1968] Based on the results of sentiment analysis, we will strengthen the detection of inappropriate content and consider taking action as necessary.

[1969] 4. Supported modules

[1970] The server automatically converts detected inappropriate chat messages into appropriate expressions, immediately deletes inappropriate images, and notifies users with a warning. If inappropriate voice comments are detected, the person making the comment becomes invisible to other users.

[1971] If the emotion engine recognizes a negative emotion, the server provides mental support to the user.

[1972] 5. Notification Module

[1973] The server notifies users of any inappropriate content or behavior detected and provides detailed feedback reports to indicate which actions were inappropriate.

[1974] Specific examples

[1975] Example 1: Monitoring chat messages and sentiment

[1976] User A sends a chat message in the metaverse saying "I hope you die."

[1977] The server analyzes this message using an NLP model and detects that the word "die" is inappropriate.

[1978] The server automatically converts this to "Shut up, you idiot" and displays it.

[1979] At the same time, the server uses an emotion engine to analyze the emotional state of user A, and if it detects negative emotions, it provides mental support.

[1980] Example 2: Image and Emotion Monitoring

[1981] User B uploads an inappropriate image (sexually explicit image).

[1982] The terminal sends this image to the server.

[1983] The server analyzes the image using computer vision models and detects that it is inappropriate.

[1984] The server deletes the image and notifies User B that "inappropriate content has been posted."

[1985] The server uses an emotion engine to analyze User B's emotional state and considers an appropriate response.

[1986] Example 3: Monitoring voice utterances and emotions

[1987] User C shouts insults such as "You idiot" in the voice chat.

[1988] The terminal transmits the voice data to the server.

[1989] The server converts the speech to text and uses an NLP model to recognize the word "idiot" as inappropriate.

[1990] The server makes user C invisible, making him invisible to other users.

[1991] The server uses an emotion engine to analyze the emotional state of user C and provides support as needed.

[1992] This system will not only make communication within the metaverse safer and more comfortable, but also enable advanced responses that take into account the user's emotional state, allowing users to interact with other users in a healthy environment with peace of mind.

[1993] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1994] Step 1: Data collection

[1995] The device collects chat messages, image data, and voice data sent by users in the metaverse in real time. The collected data is then sent to a server via a network.

[1996] Specific behavior:

[1997] The user types "Hello" in the chat window and presses the send button.

[1998] The device captures the text data "Hello" and sends it to the server.

[1999] Users take screenshots of their avatars and upload them to a shared gallery in the metaverse.

[2000] The device captures the screenshot image data and sends it to the server.

[2001] Step 2: Chat message analysis

[2002] The server inputs the received chat messages into a natural language processing (NLP) model to extract keywords and perform semantic analysis. The input is the chat message, and the output is the analysis result. Specifically, Python's NLTK and spaCy are used.

[2003] Specific behavior:

[2004] The server receives the message "Hello" and asks the natural language processing model to analyze it.

[2005] The NLP model tags "hello" as a greeting and returns the result to the server.

[2006] Step 3: Image data analysis

[2007] The server inputs the received image data into a computer vision model (e.g., TensorFlow, OpenCV) to detect inappropriate content. The input is the image data, and the output is the detection result of inappropriate content.

[2008] Specific behavior:

[2009] The server receives the screenshots uploaded by the user.

[2010] Computer vision models scan images to determine if they contain inappropriate content.

[2011] The server receives the inappropriate content detection result and removes the image.

[2012] Step 4: Audio data analysis

[2013] The server inputs the audio data into a speech recognition model (e.g., PyTorch, Google's speech recognition API) and converts it into text format. The text data is then analyzed using an NLP model to detect inappropriate speech. The input is the audio data, and the output is the analyzed text data.

[2014] Specific behavior:

[2015] A user says "You're such an idiot" in voice chat.

[2016] The device captures the audio data and sends it to the server.

[2017] The server converts the voice data into text and generates the text data "You're an idiot."

[2018] The NLP model determines that "baka" is an inappropriate word and returns the result to the server.

[2019] Step 5: Sentiment Analysis

[2020] The server inputs the collected text, image, and voice data into an emotion engine (e.g., Microsoft Azure's emotion analysis API, snownlp, etc.) to analyze the user's emotions. The input is data (text, image, voice), and the output is an evaluation of the user's emotional state.

[2021] Specific behavior:

[2022] The server inputs the chat message "You're such an idiot" into an emotion engine, which analyzes the emotional state of the message.

[2023] The emotion engine detects "anger" and returns the result to the server.

[2024] Step 6: Address inappropriate content

[2025] The server automatically takes appropriate action as needed based on the results of the sentiment analysis. The inputs are the detection results of inappropriate content and the sentiment analysis results, and the outputs are the modified message, the deletion of inappropriate content, and the results of making the user invisible.

[2026] Specific behavior:

[2027] The server automatically converts "You're an idiot" to "Shut up you" and displays it to the sending user.

[2028] If an inappropriate image is detected, it will be removed and a warning message will be sent to the uploader.

[2029] The server turns users who make inappropriate vocal comments into invisible people, making them invisible to other users.

[2030] Step 7: Providing mental support

[2031] The server provides mental support to the user when a negative emotion is recognized by the emotion engine. The input is the emotion analysis result, and the output is the support provided.

[2032] Specific behavior:

[2033] If the server recognizes "anger," it provides the user with a message explaining how to relax and a link to a counseling service.

[2034] Step 8: User Notification

[2035] The server notifies the user of any detected inappropriate content or behavior and also generates and provides a detailed feedback report. The input is the detection result of inappropriate content, and the output is the user notification and feedback report.

[2036] Specific behavior:

[2037] The server sends a notification to the user about the detected inappropriate content.

[2038] The report states, "Your comment, 'You're an idiot,' was deemed inappropriate. Please be careful next time."

[2039] (Application example 2)

[2040] 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."

[2041] In current brick-and-mortar stores, communication between customers and store staff can be inappropriate, and customers' feelings can be poorly understood, leading to delayed responses. This can lead to lower customer satisfaction, which can have a negative impact on store ratings and sales. Furthermore, because it is difficult to respond appropriately in real time, advanced management of the customer experience in brick-and-mortar stores is required.

[2042] The identification processing 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 collecting chat messages; means for analyzing the collected chat messages using a natural language processing model to detect inappropriate words or expressions; means for automatically changing and displaying the corresponding message when inappropriate words or expressions are detected; means for collecting and analyzing inappropriate images and deleting the detected inappropriate images; means for converting collected voice data into text using a voice recognition model and analyzing it using a natural language processing model to detect inappropriate comments; means for making the speaker of the corresponding voice invisible when inappropriate voice is detected; means for notifying the user of detected inappropriate content or behavior; means for analyzing the user's emotions, including an emotion analysis engine; means for providing an appropriate response when the user's emotions are recognized as negative; and means for monitoring conversations and behaviors between customers and store clerks in a physical store and notifying the store clerk in real time based on the emotion analysis results. This enables real-time monitoring of customer experiences in physical stores and enables prompt and appropriate responses.

[2043] The "metaverse" is a virtual space that exists on the Internet, a virtual reality environment in which users can interact using avatars.

[2044] "Communication safety" refers to the standards and measures to ensure that interactions between users are appropriate and healthy.

[2045] "Monitoring" refers to the act of watching users' behavior and messages in real time to detect inappropriate behavior or expressions.

[2046] "Management" refers to the act of taking appropriate action against detected inappropriate behavior or content and operating the entire system.

[2047] A "chat message" is a communication message sent by a user in text format.

[2048] "Means of collection" refers to the equipment or software mechanisms used to collect data sent by users on a server.

[2049] A "natural language processing model" refers to the algorithms and technologies that allow computers to understand and analyze human language.

[2050] "Inappropriate words and phrases" means language or phrases that may be offensive, insulting, or otherwise objectionable.

[2051] "Means for automatically changing and displaying" refers to technology that instantly corrects detected inappropriate content and replaces it with safe expressions before displaying it.

[2052] "Inappropriate Images" are images that contain sexually explicit material, violence, or other inappropriate visual content.

[2053] "Means for analyzing and deleting detected inappropriate images" refers to technology that analyzes collected images and removes them if they are deemed inappropriate.

[2054] "Voice data" refers to data that is a digital recording of a user's voice.

[2055] A "speech recognition model" refers to the algorithms and techniques used to convert speech into text.

[2056] "Invisibility methods" are techniques that make inappropriate commenters invisible to other users.

[2057] "Means of notification" refers to the mechanism by which the system communicates warnings and information to the user.

[2058] "Sentiment analysis engine" refers to software or hardware for analyzing user emotions from collected data.

[2059] "Means for providing appropriate responses when emotions are perceived as negative" refers to technology for providing appropriate support or notifications when a user's emotions are determined to be negative.

[2060] "Means for monitoring conversations and behavior between customers and store clerks in physical stores and notifying store clerks in real time based on the results of emotion analysis" is a system that monitors interactions between customers and store clerks in physical stores and immediately notifies store clerks based on the analyzed emotion results.

[2061] The present invention is a system for monitoring customer communications in real time in a brick-and-mortar store and providing safe and effective responses. The system includes the following main components:

[2062] 1. Data Collection Module

[2063] The server uses smart glasses or head-mounted displays to monitor the conversations and behavior of customers and store staff in real time. These devices collect customer voice and facial expression data and send it to the server. Specific hardware examples include Google Glass and Microsoft HoloLens.

[2064] 2. Data Analysis Module

[2065] The server converts the collected voice data into text using a speech recognition model. This speech recognition is performed using the Google Cloud Speech-to-Text API. The converted text is then analyzed using a natural language processing (NLP) model to recognize the content and tone of the conversation. This analysis is performed using Google Cloud Natural Language and OpenAI GPT-3.

[2066] The server then inputs the collected image data into a sentiment analysis engine to analyze the customer's emotional state, using the Affectiva SDK and Amazon Rekognition.

[2067] 3. Supported modules

[2068] If a customer shows signs of dissatisfaction or discomfort, the server will immediately send a notification to the store clerk's smart glasses, displaying an alert to prompt appropriate action, for example, a notification such as: "The customer is dissatisfied. Please take action."

[2069] It also includes a function that automatically corrects inappropriate comments or behavior when it is detected, converting them into safe language and displaying it to customers.

[2070] 4. Notification Module

[2071] At the end of the day, the server automatically generates a customer feedback report and sends a notification to the salesperson and manager, allowing the salesperson to identify specific areas for improvement and strive to improve customer service. Example: "Today's feedback report: List of complaints and areas for improvement."

[2072] Example prompt sentences

[2073] Customer facial expression prompt: "What emotions do you see in the customer's facial expressions?"

[2074] Example input to a generative AI model: "This customer's mouth corners are downturned and their eyes are narrowed. What emotion is this customer experiencing?"

[2075] Conversation analysis prompt: "Does this conversation suggest that the customer is unhappy?"

[2076] Example input to a generative AI model: "Customer: 'Hey, isn't this too expensive? It's a lot more than I expected.' Salesperson: 'Sorry, but we're not currently offering any special discounts, so this is the price.'"

[2077] This allows for real-time monitoring of customer experiences in physical stores, enabling quick and appropriate responses, ultimately improving customer satisfaction.

[2078] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2079] Step 1:

[2080] The server collects real-time conversation and behavior data between customers and store staff in a physical store through smart glasses or a head-mounted display. The device captures voice and image data and sends them to the server. The input is voice and image data, and the output is raw data sent to the server.

[2081] Step 2:

[2082] The server inputs the received voice data into a voice recognition model and converts it into text format. The Google Cloud Speech-to-Text API is used for voice recognition. The input is voice data and the output is text data. In this step, the voice data is converted into conversational content.

[2083] Step 3:

[2084] The server inputs the converted text data into a natural language processing (NLP) model to analyze the semantics of the conversation. Google Cloud Natural Language and OpenAI GPT-3 are used for the NLP model. The input is text data, and the output is the tone and meaning of the analyzed conversation. In this step, the content and tone of the conversation are analyzed.

[2085] Step 4:

[2086] The server inputs the collected image data into an emotion analysis engine to analyze the customer's emotional state. Emotion analysis uses the Affectiva SDK and Amazon Rekognition. The input is image data, and the output is the analysis result of the customer's emotional state. In this step, emotions are read from the customer's facial expressions.

[2087] Step 5:

[2088] The server integrates the analyzed conversation content with the customer's emotional state, and if the customer shows signs of dissatisfaction or discomfort, it sends a notification to the store clerk's smart glasses. The input is the analysis result, and the output is a notification to the store clerk. Specifically, the clerk will see a message such as, "The customer is dissatisfied. Please take action."

[2089] Step 6:

[2090] When the server detects inappropriate comments or behavior, it automatically modifies the content, converts it into safer language, and displays it. The input is the data of the inappropriate comments or behavior, and the output is the modified content. Specifically, offensive language is changed to softer language.

[2091] Step 7:

[2092] At the end of each day, the server automatically generates a customer feedback report and sends notifications to store staff and managers. The input is the customer interaction data for the day, and the output is a feedback report. Specifically, the report contains content such as "Today's feedback report: list of complaints and areas for improvement."

[2093] 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.

[2094] 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.

[2095] 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.

[2096] 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.

[2097] FIG. 9 illustrates 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 behaviors 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.

[2098] 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.

[2099] 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).

[2100] 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.

[2101] 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."

[2102] 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.

[2103] 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).

[2104] 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.

[2105] 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.

[2106] 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.

[2107] 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.

[2108] 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.

[2109] 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.

[2110] 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.

[2111] 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.

[2112] 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.

[2113] 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.

[2114] The following is further disclosed regarding the above embodiment.

[2115] (Claim 1)

[2116] A system that automatically monitors and manages the safety of communications within the metaverse,

[2117] a means for collecting chat messages;

[2118] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[2119] A means for automatically changing and displaying the message when inappropriate words or expressions are detected;

[2120] A means for collecting and analyzing inappropriate images and removing detected inappropriate images;

[2121] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[2122] A means for making the speaker of the inappropriate voice invisible when the inappropriate voice is detected;

[2123] a means for notifying users of detected inappropriate content or behavior;

[2124] A system including:

[2125] (Claim 2)

[2126] 10. The system of claim 1,

[2127] The system further includes means for collecting image data from the user's device and transmitting it to the server.

[2128] (Claim 3)

[2129] 10. The system of claim 1,

[2130] The system further includes means for adding a chat message to an analysis queue when the message is sent.

[2131] (Claim 4)

[2132] 10. The system of claim 1,

[2133] The system further includes means for notifying a warning to a user when an inappropriate remark is detected.

[2134] (Claim 5)

[2135] 10. The system of claim 1,

[2136] The system further includes means for providing a detailed feedback report to the affected user when inappropriate content is detected.

[2137] (Claim 6)

[2138] 10. The system of claim 1,

[2139] The system further includes means for automatically converting detected inappropriate content or behavior into corresponding appropriate content.

[2140] "Example 1"

[2141] (Claim 1)

[2142] means for collecting communication data from a user's device;

[2143] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[2144] A means for automatically changing and displaying the message when inappropriate words or expressions are detected;

[2145] A means for analyzing the collected image data using a computer vision model to detect and remove inappropriate content; and

[2146] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[2147] A means for making the speaker of the inappropriate voice invisible when the inappropriate voice is detected;

[2148] a means for notifying users of detected inappropriate content or behavior;

[2149] A system including:

[2150] (Claim 2)

[2151] 10. The system of claim 1, further comprising means for transmitting the collected image data from the user's device to a server.

[2152] (Claim 3)

[2153] 10. The system of claim 1, further comprising means for adding a chat message to an analysis queue when the user sends the message.

[2154] "Application Example 1"

[2155] (Claim 1)

[2156] A system that automatically monitors and manages the safety of communications within the metaverse,

[2157] a means for collecting chat messages;

[2158] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[2159] A means for automatically changing and displaying the message when inappropriate words or expressions are detected;

[2160] A means for collecting and analyzing inappropriate images and removing detected inappropriate images;

[2161] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[2162] A means for making the speaker of the inappropriate voice invisible when the inappropriate voice is detected;

[2163] a means for notifying users of detected inappropriate content or behavior;

[2164] a means for providing users with detailed feedback reports of detected inappropriate content;

[2165] A means to monitor communication within the virtual store in real time and take specific action immediately,

[2166] A system including:

[2167] (Claim 2)

[2168] 10. The system of claim 1, further comprising means for collecting image data from a user's device and transmitting it to a server.

[2169] (Claim 3)

[2170] 10. The system of claim 1, further comprising means for adding a chat message to an analysis queue when the message is sent.

[2171] "Example 2: Combining Emotion Engines"

[2172] (Claim 1)

[2173] A means for collecting communication data (chat messages, images, audio) sent by users within the metaverse;

[2174] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[2175] A means for automatically changing and displaying the message when inappropriate words or expressions are detected;

[2176] means for analyzing the collected image data using a computer vision model to detect inappropriate images;

[2177] a means for removing the detected inappropriate images;

[2178] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[2179] A means for making the speaker invisible when inappropriate audio is detected;

[2180] A means for inputting the collected communication data into an emotion analysis engine and analyzing the user's emotions;

[2181] A means for detecting negative emotions based on the emotion analysis results and providing mental support;

[2182] a means for notifying users of detected inappropriate content or behavior;

[2183] means for generating and providing detailed feedback reports to users;

[2184] A system including:

[2185] (Claim 2)

[2186] 10. The system of claim 1, further comprising means for collecting image data from a user's device and transmitting it to a server.

[2187] (Claim 3)

[2188] 10. The system of claim 1, further comprising means for adding a chat message to an analysis queue when the message is sent.

[2189] "Application example 2 when combining emotion engines"

[2190] (Claim 1)

[2191] A system that automatically monitors and manages the safety of communications within the metaverse,

[2192] a means for collecting chat messages;

[2193] A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions;

[2194] A means for automatically changing and displaying the message when inappropriate words or expressions are detected;

[2195] A means for collecting and analyzing inappropriate images and removing detected inappropriate images;

[2196] A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks;

[2197] A means for making the speaker of the inappropriate voice invisible when the inappropriate voice is detected;

[2198] a means for notifying users of detected inappropriate content or behavior;

[2199] means for analyzing user emotions, the means including an emotion analysis engine;

[2200] means for providing an appropriate response when a user's emotion is recognized as negative;

[2201] A means of monitoring conversations and behavior between customers and store clerks in physical stores and notifying store clerks in real time based on the results of emotion analysis;

[2202] A system including:

[2203] (Claim 2)

[2204] 10. The system of claim 1, further comprising means for collecting image data from a user's device and transmitting it to a server.

[2205] (Claim 3)

[2206] 10. The system of claim 1, further comprising means for adding a chat message to an analysis queue when the message is sent. [Explanation of symbols]

[2207] 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 system that automatically monitors and manages the safety of communications within the metaverse, a means for collecting chat messages; A means for analyzing collected chat messages using a natural language processing model to detect inappropriate words and expressions; A means for automatically changing and displaying the message when inappropriate words or expressions are detected; A means for collecting and analyzing inappropriate images and removing detected inappropriate images; A means for converting the collected voice data into text using a voice recognition model, analyzing it using a natural language processing model, and detecting inappropriate remarks; A means for making the speaker of the inappropriate voice invisible when the inappropriate voice is detected; a means for notifying users of detected inappropriate content or behavior; A system including:

2. 10. The system of claim 1, The system further includes means for collecting image data from the user's device and transmitting it to the server.

3. 10. The system of claim 1, The system further includes means for adding a chat message to an analysis queue when the message is sent.

4. 10. The system of claim 1, The system further includes means for notifying a warning to a user when an inappropriate remark is detected.

5. 10. The system of claim 1, The system further includes means for providing a detailed feedback report to the affected user when inappropriate content is detected.

6. 10. The system of claim 1, The system further includes means for automatically converting detected inappropriate content or behavior into corresponding appropriate content.

Images