system

Abstract

In recent years, slander on the Internet has become a social problem. Such slander causes serious psychological damage to the victims and has a negative impact on society as a whole. Therefore, there is a need for an effective system that can help emotional users regain a calm perspective and suppress and reduce slander. The system includes: means for receiving a user input; means for analyzing the received user input; means for generating a warning message in response to the input; and means for displaying the generated warning message to the user. A system including:

Description

[Technical Field]

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

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

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

[0004] In recent years, slander on the Internet has become a social problem. Such slander inflicts serious psychological damage on victims and has a negative impact on society as a whole. Existing systems lack sufficient means to detect and suppress offensive comments when users enter them, resulting in the problem of repeated slander. Therefore, there is a need for an effective system that can help emotional users regain a calm perspective and suppress and reduce slander. [Means for solving the problem]

[0005] The present invention provides a system that receives user input, analyzes the input content, and generates and displays a warning message to the user if the input content contains offensive content. The system first includes a means for receiving the user's input, and then a means for analyzing the received input content using natural language processing technology. If the analysis results in the input content being determined to be offensive, the system also includes a means for generating and displaying a warning message to the user. The system also includes a means for suggesting alternative phrases to the user when the warning message is displayed, allowing the user to regain a calm perspective and effectively curb slander.

[0006] "User" refers to a person who uses a computer system or application software.

[0007] "Input" refers to text data or instructions that a user sends to a system through an interface such as a keyboard or touchscreen.

[0008] "Means for receiving" refers to hardware or software components that obtain text or data entered by a user and temporarily store it.

[0009] "Means for analysis" refers to algorithms or programs that analyze input data received from users and classify and evaluate its contents.

[0010] "Means for generating a warning message" refers to a program or function that generates an appropriate warning message based on the analysis results and converts it into a displayable format.

[0011] The "displaying means" refers to a display, a window on a screen, or the like for visually presenting the generated warning message to the user.

[0012] "Natural language processing technology" refers to the field of technology that analyzes the meaning and emotions of words and phrases used by users and understands and generates human language.

[0013] "Means for suggesting alternative phrases" refers to a function that suggests more appropriate wording to the user when offensive language is detected.

[0014] A "system" is a collection of hardware and software components that work together to achieve a specific purpose. [Brief explanation of the drawings]

[0015] [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 illustrating 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

[0016] 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.

[0017] First, the terms used in the following description will be explained.

[0018] 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).

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 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.

[0026] 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).

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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."

[0036] The system of the present invention monitors user input in real time, analyzes the content of the input using natural language processing technology, and generates and displays a warning message to the user if the input contains offensive words or phrases. Specific embodiments will be described in detail below.

[0037] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and transmits the content to a server.

[0038] The server receives text data sent from the device, temporarily stores it in a buffer for analysis, and then passes it to a natural language processing (NLP) module.

[0039] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0040] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[0041] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user. The user can check the displayed warning message and correct the input content as necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[0042] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[0043] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content to the server. The server analyzes the received text using an NLP module and detects that "idiot" is an offensive word. The warning generation module generates a warning message and sends it to User A, saying, "Is this sentence offensive?" User A checks the warning and corrects the input, saying, "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[0044] In this way, the system can help users regain their composure even in emotional situations, thereby reducing slander.

[0045] The processing flow will be explained below.

[0046] Step 1:

[0047] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[0048] Step 2:

[0049] The device temporarily stores the user's input and creates an HTTP POST request to send the entered text data to the server.

[0050] Step 3:

[0051] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data is the input text "You're an idiot."

[0052] Step 4:

[0053] The server calls an internal method to pass the received text data to the NLP module, or executes an API call.

[0054] Step 5:

[0055] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[0056] Step 6:

[0057] The NLP module generates the analysis results and sends them back to the server's alert generation module. The resulting data includes a list of offensive keywords and a sentiment score.

[0058] Step 7:

[0059] The server's warning generation module receives the analysis results from the NLP module and generates an appropriate warning message, such as "Is the text offensive?"

[0060] Step 8:

[0061] The interface for sending the generated warning message from the warning generation module to the UI module is called, and the generated warning message is sent to the terminal as a response.

[0062] Step 9:

[0063] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[0064] Step 10:

[0065] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[0066] Step 11:

[0067] The device creates an HTTP POST request to send the corrected text data back to the server.

[0068] Step 12:

[0069] The server receives the corrected text data and checks it again using the same NLP module to ensure it does not contain any offensive content.

[0070] Step 13:

[0071] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[0072] Through these steps, you can respond carefully even when end users are emotional and prevent slander.

[0073] Example 1

[0074] 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."

[0075] In modern online communication, users may unconsciously use offensive language, potentially hurting the other person or a third party. This problem has led to online slander and cyberbullying, becoming a social issue. To prevent this, a system is needed that can monitor user input in real time, automatically detect offensive language, and issue appropriate warnings.

[0076] 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.

[0077] In this invention, the server includes means for acquiring user input, means for transmitting the acquired user input data to the server, means for the server to receive and temporarily store the input data transmitted from the user, means for analyzing the user input data using natural language processing technology, means for generating a warning message based on the input data, means for transmitting the generated warning message to the user, and means for re-analyzing the input data corrected by the user and determining the offensiveness of the input data. This enables the user to receive advice in real time to avoid offensive language without being swayed by emotion.

[0078] "User" refers to an individual or group that uses a terminal to input comments or messages.

[0079] "Input data" refers to text information entered into the system by a user via a terminal.

[0080] "Server" refers to a remote computer system for receiving, storing, and analyzing user-input data.

[0081] "Terminal" refers to a device (e.g., a smartphone, tablet, or personal computer) that a user uses to enter input data.

[0082] "Natural language processing (NLP)" refers to technology that enables computers to understand, interpret, and generate human language.

[0083] "Warning Message" refers to a notification that is generated when text entered by a user is determined to be offensive or inappropriate.

[0084] "Alternative phrases" refers to alternative suggestions for offensive or inappropriate text entered by a user.

[0085] "Aggression assessment" refers to the process of using natural language processing technology to determine whether input data has the potential to harm others.

[0086] "Real-time" refers to data being entered and processed nearly simultaneously, with minimal delay.

[0087] MODE FOR CARRYING OUT THE INVENTION

[0088] The system of the present invention monitors text data entered by users on their terminals in real time and analyzes the data using natural language processing technology. If the analysis detects offensive words or phrases, it generates a warning message and displays it to the user.

[0089] User Input

[0090] Users use devices such as smartphones, tablets, and personal computers to input comments on chat applications and social networking platforms. These inputs are temporarily stored on the device and then sent to a server.

[0091] Data transmission and reception

[0092] The terminal transmits the text data entered by the user. This data is sent to a server via the Internet, and the server temporarily stores the received data in a buffer for analysis. A database management system (e.g., MySQL (registered trademark), PostgreSQL) is used for storing the data.

[0093] Data analysis using natural language processing (NLP) modules

[0094] The server passes the stored text data to an NLP module. Specifically, it uses natural language processing services such as Google® Cloud Natural Language API and IBM Watson® Natural Language Understanding. The NLP module analyzes the text data and extracts sentiment scores and offensive keywords. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0095] Generate a warning message

[0096] Once the analysis results are returned to the server, a warning generation module receives them and generates an appropriate warning message. For example, the input "You're an idiot" generates a warning message such as "Is this sentence aggressive?". Additionally, an alternative phrase such as "Let's speak from a more rational perspective" can be included.

[0097] Displaying a warning message

[0098] The generated warning message is sent from the server to the terminal, and the terminal displays the warning message to the user. By displaying the warning message using a pop-up window, the user can check the message in real time.

[0099] Correcting input information

[0100] The user can check the displayed warning message and correct their input if necessary. For example, they could make a correction such as, "You seem to be trying too hard, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is deemed to be non-offensive, the server allows it to be posted.

[0101] Specific scenario example

[0102] 1. User A types "You're an idiot" into a chat app and presses the send button.

[0103] 2. The device sends the input information to the server.

[0104] 3. The server analyzes the text data using an NLP module to detect offensive keywords.

[0105] 4. The warning generation module generates a message saying "Is the text offensive?" and sends it to User A.

[0106] 5. User A sees the warning and corrects his input, saying, "You seem to be pushing yourself too hard. What's wrong?"

[0107] 6. The corrections are sent to the server, inspected again, and if there are no problems, the post is allowed to be posted.

[0108] Prompt Sentence Examples

[0109] Design a system that generates and displays a warning message to users who enter offensive comments, as follows:

[0110] It monitors the text entered by the user in real time, performs sentiment analysis and keyword extraction, and if it detects offensive words such as "idiot" or "die," it displays a message asking, "Is your writing offensive?" and suggests an alternative phrase, such as, "Let's speak from a more calm perspective."

[0111] This system will help curb online slander in real time, allowing users to maintain a calm perspective.

[0112] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0113] Step 1:

[0114] A user inputs a comment on the device. For example, user A inputs "You're an idiot." This input is temporarily stored in the device's memory.

[0115] Input: Text entered by the user (e.g., "You're an idiot")

[0116] Output: Temporarily saved text data

[0117] Step 2:

[0118] The device sends the saved text data to the server using HTTP or WebSocket as the communication protocol.

[0119] Input: Temporarily saved text data

[0120] Output: Text data sent to the server

[0121] Step 3:

[0122] The server stores the received text data in a buffer for analysis, using a database management system (e.g., MySQL or PostgreSQL).

[0123] Input: Text data sent to the server

[0124] Output: Text data saved in the analysis buffer

[0125] Step 4:

[0126] The server passes the stored text data to a natural language processing (NLP) module, which sends the data through an API. The NLP module uses Google Cloud Natural Language API or IBM Watson Natural Language Understanding.

[0127] Input: Text data stored in the parsing buffer

[0128] Output: Text data passed to the NLP module

[0129] Step 5:

[0130] The NLP module analyzes the text data and extracts sentiment scores and offensive keywords, for example, if the word "idiot" is included, it will be determined that this word is offensive.

[0131] Input: Text data passed to the NLP module

[0132] Output: Sentiment score and offensive keywords

[0133] Step 6:

[0134] The server's warning generation module receives the analysis results returned by the NLP module and generates an appropriate warning message. For example, for the input "You're an idiot," it generates the warning message "Is this sentence aggressive?". Additionally, it can include an alternative phrase, "Let's speak from a more rational perspective."

[0135] Input: Sentiment score and offensive keywords

[0136] Output: Warning message and alternative phrase

[0137] Step 7:

[0138] The server then sends the generated alert messages to the terminal, again using HTTP or WebSocket to transmit data in real time.

[0139] Input: warning message and alternative phrase

[0140] Output: Warning message sent to terminal

[0141] Step 8:

[0142] The terminal displays a warning message to the user, for example, by using a pop-up window.

[0143] Input: The warning message sent to the terminal

[0144] Output: A warning message that is displayed to the user.

[0145] Step 9:

[0146] The user checks the displayed warning message and corrects the input content if necessary. For example, the user might say, "You seem to be pushing yourself too hard. What's wrong?" The corrected text data is then sent to the server again.

[0147] Input: The warning message displayed to the user

[0148] Output: Corrected text data

[0149] Step 10:

[0150] The server then passes the corrected text data back to the NLP module for re-analysis. If the corrected input is deemed non-offensive, the server allows the edited content to be posted as is.

[0151] Input: Modified text data

[0152] Output: Text data that is allowed to be posted

[0153] This series of steps allows users to regain a calm perspective even in emotional situations and suppress or reduce slander.

[0154] (Application example 1)

[0155] 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."

[0156] In recent years, customer service in brick-and-mortar stores has become increasingly important, but there is a challenge in properly managing responses based on human emotions and reactions. In particular, if staff become emotional or use offensive language, it can worsen relationships with customers and damage the credibility of the business. There is a need for a multi-functional system that can prevent such situations and ensure calm and courteous customer service at all times.

[0157] 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.

[0158] In this invention, the server includes means for receiving user input, means for analyzing the received user input, means for generating a warning message for the input content, means for displaying the generated warning message to the user, means for converting the voice input into text data, and means for monitoring emotional words and offensive phrases in real time using the text data. This enables store staff to automatically detect offensive words and content containing negative emotions during conversations with customers, receive appropriate warnings, and encourage calm responses.

[0159] "User" refers to the entity that operates the system or makes input.

[0160] "Input" refers to information or data that a user provides to a system.

[0161] "Voice input" refers to information provided to the system by a user speaking.

[0162] "Text data" refers to written information generated by voice input or other means.

[0163] "Analysis" refers to the process of interpreting input information and understanding its content and intent.

[0164] "Emotional words" refer to words that strongly express the user's emotions.

[0165] "Offensive phrases" refer to words or expressions that can have a negative impact on others.

[0166] A "warning message" refers to a notification that notifies the user that there is a problem with the input content.

[0167] "Voice recognition technology" refers to the technology that converts voice input into text data.

[0168] "Natural language processing technology" refers to technology for understanding and analyzing human language.

[0169] "Smart glasses" refers to glasses-type devices that are worn by the user to display information and receive input.

[0170] "Server" refers to a central processing unit that processes information and stores data.

[0171] "Monitoring" refers to the act of continuously observing specific information or behavior to detect changes or problems.

[0172] The present invention relates to a system for supporting customer service in brick-and-mortar stores, and in particular to a system that detects conversations containing offensive language or negative emotions in real time and generates and displays appropriate warning messages, thereby enabling staff to respond calmly. An embodiment of the present invention will be described in detail below.

[0173] System Configuration

[0174] The system consists of the following main components:

[0175] 1. Smart Glasses

[0176] Staff wear smart glasses and use voice input, which is collected through the smart glasses' microphone.

[0177] 2. Voice recognition function

[0178] Voice input is converted to text in real time using cloud-based or built-in speech recognition technology, such as the Google Speech-to-Text API.

[0179] 3. Server

[0180] The server has the following roles:

[0181] Data reception: Receives text data sent from the smart glasses.

[0182] Natural Language Processing (NLP): Analyzes incoming text data to detect emotional words and offensive phrases. NLP technologies available include spacy and TextBlob.

[0183] Warning generation: When emotive words or offensive phrases are detected, appropriate warning messages and alternative phrases are generated.

[0184] Message sending: The generated warning message is sent to the smart glasses and displayed to staff in real time.

[0185] Data processing and calculation

[0186] 1. Convert voice input to text data:

[0187] The voice data collected by the smart glasses is converted into text data in real time using a voice recognition API. For example, it can be converted into the following format:

[0188] Voice input: "You're an idiot."

[0189] Speech recognition API output: Text data "You're an idiot"

[0190] 2. Text data analysis:

[0191] The server receives the text data and analyzes it using natural language processing technology. Specifically, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is detected as an offensive word.

[0192] 3. Generate a warning message:

[0193] If offensive language or negative sentiment is detected, the server uses a warning generation module to generate an appropriate warning message and alternative phrases. For example, the following warning message may be generated:

[0194] Warning message: "Is the conversation getting aggressive? Try to stay calm."

[0195] 4. Displaying messages:

[0196] The generated warning messages are sent to smart glasses and displayed to staff in real time.

[0197] Specific examples

[0198] For example, this system works if a staff member in a physical store says "You're an idiot" while interacting with a customer. The voice input is recognized by the smart glasses and converted into text data such as "You're an idiot" via the Google Speech-to-Text API. This text data is sent to the server and analyzed by the NLP module. Here, "idiot" is detected as an offensive word, and the server generates a warning message saying, "Is the conversation becoming aggressive? Please try to respond calmly." This message is displayed on the smart glasses, and the staff member can immediately correct their response by asking, "Can you tell me what's wrong?"

[0199] Prompt Sentence Examples

[0200] If a customer types "You're an idiot," create a warning message that this system will display and a suggested calm response.

[0201] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0202] Step 1:

[0203] The user wears the smart glasses and inputs voice data. The microphone in the smart glasses collects the voice data.

[0204] Input: Audio data (e.g., "You're an idiot").

[0205] Output: The audio data is kept intact in the smart glasses and is ready to be sent to the server.

[0206] Step 2:

[0207] The smart glasses send the collected voice data to a server, which then passes the received voice data to a voice recognition API.

[0208] Input: Audio data transmitted from smart glasses.

[0209] Output: A request is sent to the speech recognition API.

[0210] Step 3:

[0211] The server uses a speech recognition API (e.g., Google Speech-to-Text API) to convert the voice data into text data in real time.

[0212] Input: The audio data received by the speech recognition API.

[0213] Output: Text data (e.g., "You're an idiot") is generated.

[0214] Step 4:

[0215] The server passes the generated text data to a natural language processing (NLP) module for analysis, specifically detecting offensive words and negative sentiment scores within the text.

[0216] Input: Text data (e.g., "You're an idiot").

[0217] Output: Analysis results (e.g. "idiot" is detected as an offensive word).

[0218] Step 5:

[0219] Based on the analysis results from the NLP module, the server uses the warning generation module to generate appropriate warning messages and alternative phrases.

[0220] Input: Analysis results (offensive language detection results).

[0221] Output: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.") is generated.

[0222] Step 6:

[0223] The generated warning message is again sent from the server to the smart glasses, which then display the warning message to the user in real time.

[0224] Input: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.").

[0225] Output: A warning message will be displayed on the smart glasses display.

[0226] 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.

[0227] The system according to the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and an emotion engine, and generates and displays a warning message to the user if the input content contains offensive words or phrases. Specific embodiments are described in detail below.

[0228] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and metadata such as input speed and pattern, and then transmits the contents to the server.

[0229] The server receives text data and metadata from the device. This data is temporarily stored in a buffer for analysis. The server then passes the text data to the NLP module and emotion engine.

[0230] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0231] In parallel, the emotion engine analyzes the user's input speed and patterns to estimate the user's emotional state. For example, a sudden increase in input speed may suggest that the user is emotional. This estimated emotional state is then integrated with the analysis results of the NLP module to make a final assessment.

[0232] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[0233] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user as a pop-up window. The user can check the displayed warning message and correct the input content if necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[0234] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[0235] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and an emotion engine, and detects that "idiot" is an offensive word and that the user's typing speed is increasing. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[0236] In this way, a system that combines an emotion engine can respond carefully even in situations where the user is emotional, thereby reducing slander.

[0237] The processing flow will be explained below.

[0238] Step 1:

[0239] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[0240] Step 2:

[0241] The device temporarily stores the user's input and makes an HTTP POST request to send the entered text data and metadata such as typing speed and pattern to the server.

[0242] Step 3:

[0243] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data consists of the input text "You're an idiot" and metadata such as input speed and pattern.

[0244] Step 4:

[0245] The server calls an internal method to pass the received text data to the NLP module, and the metadata is passed to the emotion engine.

[0246] Step 5:

[0247] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[0248] Step 6:

[0249] The emotion engine analyzes the metadata and analyzes the user's typing speed and patterns. Based on this data, it infers the user's emotional state. For example, if the user's typing speed suddenly increases, it infers that the user is angry.

[0250] Step 7:

[0251] The NLP module and the sentiment engine combine their respective analysis results to generate a comprehensive assessment result, which includes a list of offensive keywords, a sentiment score, and an estimated emotional state.

[0252] Step 8:

[0253] The server's warning generation module receives the integrated evaluation results and generates appropriate warning messages, such as "Is the sentence offensive?" and "How about this wording?" suggestions.

[0254] Step 9:

[0255] The warning generation module calls an interface to send the generated warning message and alternative phrase to the UI module, which then sends the generated warning message and alternative phrase to the terminal as a response.

[0256] Step 10:

[0257] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[0258] Step 11:

[0259] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[0260] Step 12:

[0261] The device creates an HTTP POST request to send the corrected text data back to the server.

[0262] Step 13:

[0263] The server receives the corrected text data and runs it through the same NLP module and emotion engine again to ensure it does not contain any offensive content.

[0264] Step 14:

[0265] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[0266] Through the above steps, the system can carefully respond to even emotional end users and prevent slander, making communication on the Internet healthier and more constructive.

[0267] Example 2

[0268] 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."

[0269] Current online communication platforms have a problem where users become emotional and send offensive messages, which can worsen dialogue within the community. Furthermore, the mental stress and defamation caused by such offensive messages are also serious problems. The problem with existing methods is that they do not adequately implement mechanisms to detect and warn users about offensive messages in advance.

[0270] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: means for receiving a user input; means for temporarily storing the received user input and transmitting it to the server together with metadata such as input speed and pattern; means for saving the text data and metadata received by the server in an analysis buffer and passing them to a natural language processing module and a sentiment analysis engine; means for the natural language processing module to perform sentiment analysis and keyword extraction on the text data and detect offensive words and negative sentiment scores; means for the sentiment analysis engine to estimate the user's emotional state from the user's input speed and pattern and integrate the results; means for generating a warning message based on the integrated analysis results; and means for displaying the generated warning message to the user. This makes it possible to issue a warning before a user becomes emotional and sends an offensive message, thereby improving the quality of communication.

[0271] "User" means a person who enters comments or messages on the Online Platform.

[0272] "Input speed" refers to the speed at which a user types characters, measured in characters per second.

[0273] "Metadata" refers to various data related to user input, including input speed, pattern, timestamp, and the like.

[0274] "Server" refers to a computer system that receives and analyzes user input data.

[0275] The "analysis buffer" refers to a memory area that temporarily stores received data.

[0276] "Natural language processing module" refers to a software component that analyzes received text data and performs sentiment scoring and keyword extraction.

[0277] "Sentiment analysis engine" refers to a software component that infers a user's emotional state from their typing speed and patterns.

[0278] "Warning Generation Module" means the software component that generates a warning message based on the analysis results of the Natural Language Processing Module and the Sentiment Analysis Engine.

[0279] A "warning message" refers to a message that includes a warning when a user enters emotional and offensive words.

[0280] "Alternative phrases" refer to suggested milder versions of offensive language.

[0281] MODE FOR CARRYING OUT THE INVENTION

[0282] The system of the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and a sentiment analysis engine, and generates and displays a warning message to the user if the input contains offensive words or phrases. This system aims to improve the quality of communication by issuing a warning before the user becomes emotional and sends an offensive message.

[0283] Hardware and software used

[0284] Server: A computer system that receives, analyzes, and generates alert messages.

[0285] Terminal: The device on which the user inputs information (e.g., smartphone, PC)

[0286] Natural language processing module: A software component that performs sentiment analysis and keyword extraction on text data (e.g., Hugging Face's Transformers library, Google Cloud Natural Language API).

[0287] Sentiment analysis engine: A software component that infers emotional states from input speed and patterns (e.g., Apache Kafka)

[0288] Alert Generation Module: A software component that generates an alert message (e.g., AlertManager).

[0289] Specific operation of the system

[0290] First, a user enters a comment using a device such as a chat application or a social networking platform. For example, the user enters "You're an idiot" in a text field. The device temporarily stores the text data entered by the user and metadata such as typing speed and pattern, and then sends the contents to the server. The metadata includes keystroke speed (e.g., 5 characters per second).

[0291] The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis. The received data is in the format of "Text: You're an idiot" and "Speed: 5 characters per second." The server then passes this text data and metadata to the natural language processing module and sentiment analysis engine.

[0292] The natural language processing module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative sentiment scores. At this point, the offensive keyword "idiot" is extracted from the phrase "You're an idiot," and the sentiment score is calculated as "aggressive." The sentiment analysis engine estimates the user's emotional state from their typing speed and patterns. If their typing speed increases suddenly (for example, from 5 characters per second to 8 characters per second), it is assumed that the user is emotional. The server combines the results of the NLP module and the sentiment analysis engine to make a final assessment. For example, the results may be "Text: Aggressive," "Sentiment Score: High," and "Speed: Rapid Increase."

[0293] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, it may generate a warning message such as "Is the text offensive?" with an alternative phrase such as "How about this wording?" The generated warning message is then sent back to the terminal from the server.

[0294] The terminal displays this warning message to the user as a pop-up window. The user can review the warning message and correct the input if necessary. For example, they could correct it to "You seem to be overdoing it, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is not offensive, the server will not generate any further warnings and will allow the posting.

[0295] Examples and prompts

[0296] Specifically, User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and a sentiment analysis engine, detecting that the word "idiot" is offensive and that the user's typing speed has suddenly increased. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard, what's wrong?" The correction is sent to the server and inspected again. If there are no problems, the post is allowed to proceed.

[0297] In this way, by detecting offensive messages in real time and displaying a warning to the user, the quality of communication can be improved and abusive behavior can be reduced.

[0298] Examples of prompts include:

[0299] "Judge whether this message is offensive and issue a warning if necessary: ​​'You're an idiot.'"

[0300] "The user's typing speed has suddenly increased. Please rate whether this typing is emotional."

[0301] Using these prompts, the system can properly detect offensive messages and generate warnings.

[0302] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0303] Program processing steps

[0304] Step 1: Receive user input

[0305] 1. A user types a comment into a chat application or social media platform.

[0306] Example: A user types "You're an idiot" into a text field.

[0307] 2. Input: User's text data (e.g., "You're an idiot")

[0308] 3. Output: Input text data and metadata temporarily saved on the device

[0309] Step 2: The device sends the input

[0310] 1. The device temporarily stores the text data entered by the user and metadata (such as input speed and pattern).

[0311] Example: Temporarily save the text data "You're an idiot" with an input speed of 5 characters per second.

[0312] 2. The device sends the input and metadata to the server.

[0313] An interface API may be used for transmission.

[0314] 3. Input: User text data and metadata

[0315] 4. Output: The input text data and metadata sent to the server.

[0316] Step 3: The server receives the data and stores it in a buffer for analysis.

[0317] 1. The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis.

[0318] Example: Received data "Text: You're an idiot" and "Speed: 5 characters / second" is saved in a buffer.

[0319] 2. Input: Text data and metadata sent from the device

[0320] 3. Output: Data saved in the analysis buffer

[0321] Step 4: Text analysis using the natural language processing module

[0322] 1. The server passes the text data and metadata to a natural language processing (NLP) module.

[0323] Example: Send the text data "You're an idiot" to the NLP module.

[0324] 2. The NLP module analyzes the received text data, calculates sentiment scores, and extracts keywords.

[0325] Example: Extract the keyword "idiot" from the text "You're an idiot" and calculate the emotion score of "aggressive."

[0326] 3. Input: Text data

[0327] 4. Output: Keywords and sentiment scores

[0328] Step 5: Metadata analysis with sentiment analysis engine

[0329] 1. The server passes the metadata to the sentiment analysis engine.

[0330] Example: Sending metadata to a sentiment analysis engine with an input rate of 5 characters per second.

[0331] 2. The sentiment analysis engine infers the user's emotional state from their typing speed and patterns.

[0332] Example: Detecting a sudden increase in typing speed and inferring that the user is becoming "emotional."

[0333] 3. Input: Metadata

[0334] 4. Output: Estimation of emotional state

[0335] Step 6: Synthesis of the analysis results

[0336] 1. The server integrates the results of the NLP module and the sentiment analysis engine and performs the final evaluation.

[0337] Combined results: "Text: Aggressive" "Sentiment score: High" "Speed: Rapid"

[0338] 2. Input: Keywords and sentiment scores from the NLP module, emotional states from the sentiment analysis engine

[0339] 3. Output: Integrated analysis results

[0340] Step 7: Generate a warning message

[0341] 1. The server's warning generation module generates appropriate warning messages based on the integrated analysis results.

[0342] For example, generate a warning message such as "Is this sentence offensive?" with an alternative phrase such as "How about this phrase?"

[0343] 2. Input: Integrated analysis results

[0344] 3. Output: Generated warning messages

[0345] Step 8: Sending a warning message

[0346] 1. The server sends the generated warning message back to the terminal.

[0347] For example: Sending a message saying, "Is the text becoming aggressive?"

[0348] 2. Input: Generated warning message

[0349] 3. Output: Warning message sent to terminal

[0350] Step 9: Your device will display a warning message

[0351] 1. The terminal displays the received warning message to the user as a pop-up window.

[0352] For example, a warning message may appear on the user's screen asking, "Is the text offensive?"

[0353] 2. Input: Received warning message

[0354] 3. Output: The displayed warning message

[0355] Step 10: User corrects input

[0356] 1. The user checks the displayed warning message and corrects the input as necessary.

[0357] For example, correct them with, "You seem like you're overdoing it, what's wrong?"

[0358] 2. Input: Correct the input of the user who saw the warning message.

[0359] 3. Output: Corrected text data

[0360] Step 11: Resubmit your corrections

[0361] 1. The terminal sends the corrected text back to the server.

[0362] Example: Send the modified text "You seem to be overdoing it, what's wrong?" to the server.

[0363] 2. Input: Corrected text data

[0364] 3. Output: The corrected text sent to the server

[0365] Step 12: Reanalysis and final check

[0366] 1. The server analyzes the corrected text again using the NLP module and sentiment analysis engine, and if there are no problems, allows it to be posted.

[0367] Example: The corrected text "You seem to be overdoing it, what's wrong?" is determined to be non-aggressive.

[0368] 2. Input: Corrected text data

[0369] 3. Output: Based on the analysis results, if there are no problems, the submission is permitted.

[0370] In this way, the system can monitor user input in real time and immediately issue a warning if it contains offensive content, allowing users to regain their composure and reduce abusive behavior.

[0371] (Application example 2)

[0372] 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."

[0373] In electronic payment services, there is a need to reduce offensive language and emotional exchanges that can occur during communication between users and support staff. However, conventional systems lack the means to monitor such offensive language in real time and generate and present appropriate warnings or alternative phrases. This poses a challenge in maintaining the dignity of communication and reducing the mental burden on users.

[0374] The specific processing by the specific 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 receiving user input, means for analyzing the received user input, means for generating a warning message in response to the input content, means for displaying the generated warning message to the user, and means for analyzing the user's input pattern and estimating the user's emotional state. This makes it possible to monitor and correct offensive words in real time in communications for electronic payment services and suggest alternative phrases, thereby preventing slander and reducing the mental burden on users.

[0375] "User input" refers to text data sent by a user to support staff or the system in an electronic payment service.

[0376] "Means for receiving" refers to an interface that allows the system to recognize text data entered by the user and then send it to the server.

[0377] "Means for analyzing" refers to the process of analyzing received text data using natural language processing techniques and emotion engines to assess offensive language and the user's emotional state.

[0378] The "means for generating a warning message" refers to a module for creating a message to warn or caution the user based on the results of the analysis.

[0379] "Means for displaying to the user" refers to an interface for displaying the generated warning message as a pop-up or notification on the user's device.

[0380] "User input patterns" refers to metadata such as the speed and timing at which a user types text, and are used to infer emotional states.

[0381] "Means for inferring emotional state" refers to an engine or algorithm that analyzes a user's input pattern and speed to determine whether the user is emotional.

[0382] "Means for suggesting alternative phrases" refers to a module that suggests replacing offensive language with more appropriate and calmer expressions when a user uses it.

[0383] "Natural language processing technology" refers to technology for analyzing text data and understanding and extracting its meaning and emotions.

[0384] "Input speed and pattern metadata" refers to additional information, such as the speed and frequency at which a user actually types text, that can be used to infer a user's emotional state.

[0385] The present invention is a system that monitors communication between users and support staff in electronic payment services in real time, detects offensive words and phrases, and displays a warning message. The system analyzes the user's input data and its metadata (input speed and patterns) to prevent excessive remarks. Specific embodiments for implementing the present invention are described below.

[0386] Hardware / Software used

[0387] Hardware:

[0388] Smartphone

[0389] tablet

[0390] personal computer

[0391] server

[0392] software:

[0393] Natural Language Processing (NLP) modules (e.g., spaCy, NLTK)

[0394] Emotion engine (proprietary or public library)

[0395] Interfaces (e.g. React, Vue.js)

[0396] Means of implementation

[0397] server

[0398] The server receives the text data sent by the user and its metadata (input speed, pattern) and temporarily stores it in a buffer for analysis.

[0399] The server then passes the received data to the NLP module and emotion engine. The NLP module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative emotion scores. The emotion engine also analyzes the user's typing speed and patterns to estimate the user's emotional state.

[0400] The warning generation module combines these analysis results and generates an appropriate warning message. For example, if the user's input is an offensive one such as "You're an idiot," the system generates a warning message saying "Is the sentence offensive?" along with an alternative phrase such as "Let's speak from a more rational perspective."

[0401] Terminal

[0402] The device, such as a smartphone or tablet, collects text data and its metadata entered by the user in real time and sends it to a server, which then displays a warning message and alternative phrases received from the server as a pop-up window.

[0403] User

[0404] The user checks the warning message and corrects the input content as necessary. For example, the user sends the corrected text "You seem to be pushing yourself too hard. What's wrong?" to the server again.

[0405] Specific examples

[0406] If a user types "This system is completely unusable, it's stupid!" into the support section of an electronic payment service, the system analyzes the input data in real time. The server detects the offensive word "stupid" and infers from the metadata that the user is emotional. The warning generation module displays messages on the terminal such as "Is this sentence offensive?" and "Let's speak from a more rational perspective." The user confirms this and corrects their input by saying, "There seems to be a slight problem with this system. How can I fix it?"

[0407] Prompt Sentence Examples

[0408] "Evaluate the offensiveness of messages entered by users in real time and estimate the user's emotional state using an emotion engine. If offensive language is included, generate and display an appropriate warning message."

[0409] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0410] Step 1:

[0411] The device receives the user's text input in real time and collects the input data and metadata such as input speed and timing. The device temporarily stores this information and sends it to the server. Input: User's text data and metadata. Output: Transfer of received text data and metadata to the server.

[0412] Step 2:

[0413] The server receives the text data and metadata sent from the terminal and temporarily stores them in a buffer for analysis. Input: Text data and metadata from the terminal. Output: Data stored in the buffer for analysis.

[0414] Step 3:

[0415] The server analyzes the text data stored in the buffer using a natural language processing (NLP) module to extract sentiment and keywords from the text. The NLP module performs calculations to detect offensive words and negative sentiment scores. Input: Text data in the buffer. Output: Sentiment analysis results and keyword extraction results.

[0416] Step 4:

[0417] In parallel, the server passes the metadata to the emotion engine, which analyzes the user's input pattern and speed to estimate their emotional state. If the input speed suddenly increases, the emotion engine estimates that the user is emotional. Input: Metadata in the buffer. Output: Estimated emotional state.

[0418] Step 5:

[0419] The server integrates the analysis results of the NLP module with the emotional state of the emotion engine and passes them to the warning generation module. The warning generation module generates an appropriate warning message and alternative phrases if the message contains offensive language or has a high negative emotion score. Input: Sentiment analysis results, keyword extraction results, estimated emotional state. Output: Generated warning message and alternative phrases.

[0420] Step 6:

[0421] The server sends the generated warning message and alternative phrase to the terminal. The terminal displays these messages to the user as a popup window. Input: Generated warning message and alternative phrase. Output: Sent warning message to the user terminal.

[0422] Step 7:

[0423] The user checks the displayed warning message and corrects the input as necessary. For example, the user might correct the input by saying, "You seem to be pushing yourself too hard. What's wrong?" Input: The user's reaction after receiving the warning message. Output: The corrected text data.

[0424] Step 8:

[0425] The terminal sends the text data corrected by the user to the server again, and the server repeats the same process to re-analyze it. If there are no offensive words, the server allows the post. Input: Corrected text data. Output: Final decision on whether the post is allowed or not.

[0426] 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.

[0427] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (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.

[0428] 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.

[0429] [Second embodiment]

[0430] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0431] 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.

[0432] 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).

[0433] 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.

[0434] 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.

[0435] 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).

[0436] 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.

[0437] 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.

[0438] 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.

[0439] 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.

[0440] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0441] 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."

[0442] The system of the present invention monitors user input in real time, analyzes the content of the input using natural language processing technology, and generates and displays a warning message to the user if the input contains offensive words or phrases. Specific embodiments will be described in detail below.

[0443] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and transmits the content to a server.

[0444] The server receives text data sent from the device, temporarily stores it in a buffer for analysis, and then passes it to a natural language processing (NLP) module.

[0445] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0446] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[0447] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user. The user can check the displayed warning message and correct the input content as necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[0448] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[0449] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content to the server. The server analyzes the received text using an NLP module and detects that "idiot" is an offensive word. The warning generation module generates a warning message and sends it to User A, saying, "Is this sentence offensive?" User A checks the warning and corrects the input, saying, "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[0450] In this way, the system can help users regain their composure even in emotional situations, thereby reducing slander.

[0451] The processing flow will be explained below.

[0452] Step 1:

[0453] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[0454] Step 2:

[0455] The device temporarily stores the user's input and creates an HTTP POST request to send the entered text data to the server.

[0456] Step 3:

[0457] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data is the input text "You're an idiot."

[0458] Step 4:

[0459] The server calls an internal method to pass the received text data to the NLP module, or executes an API call.

[0460] Step 5:

[0461] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[0462] Step 6:

[0463] The NLP module generates the analysis results and sends them back to the server's alert generation module. The resulting data includes a list of offensive keywords and a sentiment score.

[0464] Step 7:

[0465] The server's warning generation module receives the analysis results from the NLP module and generates an appropriate warning message, such as "Is the text offensive?"

[0466] Step 8:

[0467] The interface for sending the generated warning message from the warning generation module to the UI module is called, and the generated warning message is sent to the terminal as a response.

[0468] Step 9:

[0469] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[0470] Step 10:

[0471] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[0472] Step 11:

[0473] The device creates an HTTP POST request to send the corrected text data back to the server.

[0474] Step 12:

[0475] The server receives the corrected text data and checks it again using the same NLP module to ensure it does not contain any offensive content.

[0476] Step 13:

[0477] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[0478] Through these steps, you can respond carefully even when end users are emotional and prevent slander.

[0479] Example 1

[0480] 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."

[0481] In modern online communication, users may unconsciously use offensive language, potentially hurting the other person or a third party. This problem has led to online slander and cyberbullying, becoming a social issue. To prevent this, a system is needed that can monitor user input in real time, automatically detect offensive language, and issue appropriate warnings.

[0482] 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.

[0483] In this invention, the server includes means for acquiring user input, means for transmitting the acquired user input data to the server, means for the server to receive and temporarily store the input data transmitted from the user, means for analyzing the user input data using natural language processing technology, means for generating a warning message based on the input data, means for transmitting the generated warning message to the user, and means for re-analyzing the input data corrected by the user and determining the offensiveness of the input data. This enables the user to receive advice in real time to avoid offensive language without being swayed by emotion.

[0484] "User" refers to an individual or group that uses a terminal to input comments or messages.

[0485] "Input data" refers to text information entered into the system by a user via a terminal.

[0486] "Server" refers to a remote computer system for receiving, storing, and analyzing user-input data.

[0487] "Terminal" refers to a device (e.g., a smartphone, tablet, or personal computer) that a user uses to enter input data.

[0488] "Natural language processing (NLP)" refers to technology that enables computers to understand, interpret, and generate human language.

[0489] "Warning Message" refers to a notification that is generated when text entered by a user is determined to be offensive or inappropriate.

[0490] "Alternative phrases" refers to alternative suggestions for offensive or inappropriate text entered by a user.

[0491] "Aggression assessment" refers to the process of using natural language processing technology to determine whether input data has the potential to harm others.

[0492] "Real-time" refers to data being entered and processed nearly simultaneously, with minimal delay.

[0493] MODE FOR CARRYING OUT THE INVENTION

[0494] The system of the present invention monitors text data entered by users on their terminals in real time and analyzes the data using natural language processing technology. If the analysis detects offensive words or phrases, it generates a warning message and displays it to the user.

[0495] User Input

[0496] Users use devices such as smartphones, tablets, and personal computers to input comments on chat applications and social networking platforms. These inputs are temporarily stored on the device and then sent to a server.

[0497] Data transmission and reception

[0498] The terminal transmits the text data entered by the user. This data is sent to a server via the Internet, and the server temporarily stores the received data in a buffer for analysis. A database management system (e.g., MySQL or PostgreSQL) is used for storage.

[0499] Data analysis using natural language processing (NLP) modules

[0500] The server passes the stored text data to an NLP module. Specifically, it uses natural language processing services such as Google Cloud Natural Language API and IBM Watson Natural Language Understanding. The NLP module analyzes the text data and extracts sentiment scores and offensive keywords. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0501] Generate a warning message

[0502] Once the analysis results are returned to the server, a warning generation module receives them and generates an appropriate warning message. For example, the input "You're an idiot" generates a warning message such as "Is this sentence aggressive?". Additionally, an alternative phrase such as "Let's speak from a more rational perspective" can be included.

[0503] Displaying a warning message

[0504] The generated warning message is sent from the server to the terminal, and the terminal displays the warning message to the user. By displaying the warning message using a pop-up window, the user can check the message in real time.

[0505] Correcting input information

[0506] The user can check the displayed warning message and correct their input if necessary. For example, they could make a correction such as, "You seem to be trying too hard, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is deemed to be non-offensive, the server allows it to be posted.

[0507] Specific scenario example

[0508] 1. User A types "You're an idiot" into a chat app and presses the send button.

[0509] 2. The device sends the input information to the server.

[0510] 3. The server analyzes the text data using an NLP module to detect offensive keywords.

[0511] 4. The warning generation module generates a message saying "Is the text offensive?" and sends it to User A.

[0512] 5. User A sees the warning and corrects his input, saying, "You seem to be pushing yourself too hard. What's wrong?"

[0513] 6. The corrections are sent to the server, inspected again, and if there are no problems, the post is allowed to be posted.

[0514] Prompt Sentence Examples

[0515] Design a system that generates and displays a warning message to users who enter offensive comments, as follows:

[0516] It monitors the text entered by the user in real time, performs sentiment analysis and keyword extraction, and if it detects offensive words such as "idiot" or "die," it displays a message asking, "Is your writing offensive?" and suggests an alternative phrase, such as, "Let's speak from a more calm perspective."

[0517] This system will help curb online slander in real time, allowing users to maintain a calm perspective.

[0518] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0519] Step 1:

[0520] A user inputs a comment on the device. For example, user A inputs "You're an idiot." This input is temporarily stored in the device's memory.

[0521] Input: Text entered by the user (e.g., "You're an idiot")

[0522] Output: Temporarily saved text data

[0523] Step 2:

[0524] The device sends the saved text data to the server using HTTP or WebSocket as the communication protocol.

[0525] Input: Temporarily saved text data

[0526] Output: Text data sent to the server

[0527] Step 3:

[0528] The server stores the received text data in a buffer for analysis, using a database management system (e.g., MySQL or PostgreSQL).

[0529] Input: Text data sent to the server

[0530] Output: Text data saved in the analysis buffer

[0531] Step 4:

[0532] The server passes the stored text data to a natural language processing (NLP) module, which sends the data through an API. The NLP module uses Google Cloud Natural Language API or IBM Watson Natural Language Understanding.

[0533] Input: Text data stored in the parsing buffer

[0534] Output: Text data passed to the NLP module

[0535] Step 5:

[0536] The NLP module analyzes the text data and extracts sentiment scores and offensive keywords, for example, if the word "idiot" is included, it will be determined that this word is offensive.

[0537] Input: Text data passed to the NLP module

[0538] Output: Sentiment score and offensive keywords

[0539] Step 6:

[0540] The server's warning generation module receives the analysis results returned by the NLP module and generates an appropriate warning message. For example, for the input "You're an idiot," it generates the warning message "Is this sentence aggressive?". Additionally, it can include an alternative phrase, "Let's speak from a more rational perspective."

[0541] Input: Sentiment score and offensive keywords

[0542] Output: Warning message and alternative phrase

[0543] Step 7:

[0544] The server then sends the generated alert messages to the terminal, again using HTTP or WebSocket to transmit data in real time.

[0545] Input: warning message and alternative phrase

[0546] Output: Warning message sent to terminal

[0547] Step 8:

[0548] The terminal displays a warning message to the user, for example, by using a pop-up window.

[0549] Input: The warning message sent to the terminal

[0550] Output: A warning message that is displayed to the user.

[0551] Step 9:

[0552] The user checks the displayed warning message and corrects the input content if necessary. For example, the user might say, "You seem to be pushing yourself too hard. What's wrong?" The corrected text data is then sent to the server again.

[0553] Input: The warning message displayed to the user

[0554] Output: Corrected text data

[0555] Step 10:

[0556] The server then passes the corrected text data back to the NLP module for re-analysis. If the corrected input is deemed non-offensive, the server allows the edited content to be posted as is.

[0557] Input: Modified text data

[0558] Output: Text data that is allowed to be posted

[0559] This series of steps allows users to regain a calm perspective even in emotional situations and suppress or reduce slander.

[0560] (Application example 1)

[0561] 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."

[0562] In recent years, customer service in brick-and-mortar stores has become increasingly important, but there is a challenge in properly managing responses based on human emotions and reactions. In particular, if staff become emotional or use offensive language, it can worsen relationships with customers and damage the credibility of the business. There is a need for a multi-functional system that can prevent such situations and ensure calm and courteous customer service at all times.

[0563] 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.

[0564] In this invention, the server includes means for receiving user input, means for analyzing the received user input, means for generating a warning message for the input content, means for displaying the generated warning message to the user, means for converting the voice input into text data, and means for monitoring emotional words and offensive phrases in real time using the text data. This enables store staff to automatically detect offensive words and content containing negative emotions during conversations with customers, receive appropriate warnings, and encourage calm responses.

[0565] "User" refers to the entity that operates the system or makes input.

[0566] "Input" refers to information or data that a user provides to a system.

[0567] "Voice input" refers to information provided to the system by a user speaking.

[0568] "Text data" refers to written information generated by voice input or other means.

[0569] "Analysis" refers to the process of interpreting input information and understanding its content and intent.

[0570] "Emotional words" refer to words that strongly express the user's emotions.

[0571] "Offensive phrases" refer to words or expressions that can have a negative impact on others.

[0572] A "warning message" refers to a notification that notifies the user that there is a problem with the input content.

[0573] "Voice recognition technology" refers to the technology that converts voice input into text data.

[0574] "Natural language processing technology" refers to technology for understanding and analyzing human language.

[0575] "Smart glasses" refers to glasses-type devices that are worn by the user to display information and receive input.

[0576] "Server" refers to a central processing unit that processes information and stores data.

[0577] "Monitoring" refers to the act of continuously observing specific information or behavior to detect changes or problems.

[0578] The present invention relates to a system for supporting customer service in brick-and-mortar stores, and in particular to a system that detects conversations containing offensive language or negative emotions in real time and generates and displays appropriate warning messages, thereby enabling staff to respond calmly. An embodiment of the present invention will be described in detail below.

[0579] System Configuration

[0580] The system consists of the following main components:

[0581] 1. Smart Glasses

[0582] Staff wear smart glasses and use voice input, which is collected through the smart glasses' microphone.

[0583] 2. Voice recognition function

[0584] Voice input is converted to text in real time using cloud-based or built-in speech recognition technology, such as the Google Speech-to-Text API.

[0585] 3. Server

[0586] The server has the following roles:

[0587] Data reception: Receives text data sent from the smart glasses.

[0588] Natural Language Processing (NLP): Analyzes incoming text data to detect emotional words and offensive phrases. NLP technologies available include spacy and TextBlob.

[0589] Warning generation: When emotive words or offensive phrases are detected, appropriate warning messages and alternative phrases are generated.

[0590] Message sending: The generated warning message is sent to the smart glasses and displayed to staff in real time.

[0591] Data processing and calculation

[0592] 1. Convert voice input to text data:

[0593] The voice data collected by the smart glasses is converted into text data in real time using a voice recognition API. For example, it can be converted into the following format:

[0594] Voice input: "You're an idiot."

[0595] Speech recognition API output: Text data "You're an idiot"

[0596] 2. Text data analysis:

[0597] The server receives the text data and analyzes it using natural language processing technology. Specifically, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is detected as an offensive word.

[0598] 3. Generate a warning message:

[0599] If offensive language or negative sentiment is detected, the server uses a warning generation module to generate an appropriate warning message and alternative phrases. For example, the following warning message may be generated:

[0600] Warning message: "Is the conversation getting aggressive? Try to stay calm."

[0601] 4. Displaying messages:

[0602] The generated warning messages are sent to smart glasses and displayed to staff in real time.

[0603] Specific examples

[0604] For example, this system works if a staff member in a physical store says "You're an idiot" while interacting with a customer. The voice input is recognized by the smart glasses and converted into text data such as "You're an idiot" via the Google Speech-to-Text API. This text data is sent to the server and analyzed by the NLP module. Here, "idiot" is detected as an offensive word, and the server generates a warning message saying, "Is the conversation becoming aggressive? Please try to respond calmly." This message is displayed on the smart glasses, and the staff member can immediately correct their response by asking, "Can you tell me what's wrong?"

[0605] Prompt Sentence Examples

[0606] If a customer types "You're an idiot," create a warning message that this system will display and a suggested calm response.

[0607] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0608] Step 1:

[0609] The user wears the smart glasses and inputs voice data. The microphone in the smart glasses collects the voice data.

[0610] Input: Audio data (e.g., "You're an idiot").

[0611] Output: The audio data is kept intact in the smart glasses and is ready to be sent to the server.

[0612] Step 2:

[0613] The smart glasses send the collected voice data to a server, which then passes the received voice data to a voice recognition API.

[0614] Input: Audio data transmitted from smart glasses.

[0615] Output: A request is sent to the speech recognition API.

[0616] Step 3:

[0617] The server uses a speech recognition API (e.g., Google Speech-to-Text API) to convert the voice data into text data in real time.

[0618] Input: The audio data received by the speech recognition API.

[0619] Output: Text data (e.g., "You're an idiot") is generated.

[0620] Step 4:

[0621] The server passes the generated text data to a natural language processing (NLP) module for analysis, specifically detecting offensive words and negative sentiment scores within the text.

[0622] Input: Text data (e.g., "You're an idiot").

[0623] Output: Analysis results (e.g. "idiot" is detected as an offensive word).

[0624] Step 5:

[0625] Based on the analysis results from the NLP module, the server uses the warning generation module to generate appropriate warning messages and alternative phrases.

[0626] Input: Analysis results (offensive language detection results).

[0627] Output: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.") is generated.

[0628] Step 6:

[0629] The generated warning message is again sent from the server to the smart glasses, which then display the warning message to the user in real time.

[0630] Input: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.").

[0631] Output: A warning message will be displayed on the smart glasses display.

[0632] 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.

[0633] The system according to the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and an emotion engine, and generates and displays a warning message to the user if the input content contains offensive words or phrases. Specific embodiments are described in detail below.

[0634] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and metadata such as input speed and pattern, and then transmits the contents to the server.

[0635] The server receives text data and metadata from the device. This data is temporarily stored in a buffer for analysis. The server then passes the text data to the NLP module and emotion engine.

[0636] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0637] In parallel, the emotion engine analyzes the user's input speed and patterns to estimate the user's emotional state. For example, a sudden increase in input speed may suggest that the user is emotional. This estimated emotional state is then integrated with the analysis results of the NLP module to make a final assessment.

[0638] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[0639] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user as a pop-up window. The user can check the displayed warning message and correct the input content if necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[0640] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[0641] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and an emotion engine, and detects that "idiot" is an offensive word and that the user's typing speed is increasing. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[0642] In this way, a system that combines an emotion engine can respond carefully even in situations where the user is emotional, thereby reducing slander.

[0643] The processing flow will be explained below.

[0644] Step 1:

[0645] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[0646] Step 2:

[0647] The device temporarily stores the user's input and makes an HTTP POST request to send the entered text data and metadata such as typing speed and pattern to the server.

[0648] Step 3:

[0649] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data consists of the input text "You're an idiot" and metadata such as input speed and pattern.

[0650] Step 4:

[0651] The server calls an internal method to pass the received text data to the NLP module, and the metadata is passed to the emotion engine.

[0652] Step 5:

[0653] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[0654] Step 6:

[0655] The emotion engine analyzes the metadata and analyzes the user's typing speed and patterns. Based on this data, it infers the user's emotional state. For example, if the user's typing speed suddenly increases, it infers that the user is angry.

[0656] Step 7:

[0657] The NLP module and the sentiment engine combine their respective analysis results to generate a comprehensive assessment result, which includes a list of offensive keywords, a sentiment score, and an estimated emotional state.

[0658] Step 8:

[0659] The server's warning generation module receives the integrated evaluation results and generates appropriate warning messages, such as "Is the sentence offensive?" and "How about this wording?" suggestions.

[0660] Step 9:

[0661] The warning generation module calls an interface to send the generated warning message and alternative phrase to the UI module, which then sends the generated warning message and alternative phrase to the terminal as a response.

[0662] Step 10:

[0663] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[0664] Step 11:

[0665] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[0666] Step 12:

[0667] The device creates an HTTP POST request to send the corrected text data back to the server.

[0668] Step 13:

[0669] The server receives the corrected text data and runs it through the same NLP module and emotion engine again to ensure it does not contain any offensive content.

[0670] Step 14:

[0671] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[0672] Through the above steps, the system can carefully respond to even emotional end users and prevent slander, making communication on the Internet healthier and more constructive.

[0673] Example 2

[0674] 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."

[0675] Current online communication platforms have a problem where users become emotional and send offensive messages, which can worsen dialogue within the community. Furthermore, the mental stress and defamation caused by such offensive messages are also serious problems. The problem with existing methods is that they do not adequately implement mechanisms to detect and warn users about offensive messages in advance.

[0676] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: means for receiving a user input; means for temporarily storing the received user input and transmitting it to the server together with metadata such as input speed and pattern; means for saving the text data and metadata received by the server in an analysis buffer and passing them to a natural language processing module and a sentiment analysis engine; means for the natural language processing module to perform sentiment analysis and keyword extraction on the text data and detect offensive words and negative sentiment scores; means for the sentiment analysis engine to estimate the user's emotional state from the user's input speed and pattern and integrate the results; means for generating a warning message based on the integrated analysis results; and means for displaying the generated warning message to the user. This makes it possible to issue a warning before a user becomes emotional and sends an offensive message, thereby improving the quality of communication.

[0677] "User" means a person who enters comments or messages on the Online Platform.

[0678] "Input speed" refers to the speed at which a user types characters, measured in characters per second.

[0679] "Metadata" refers to various data related to user input, including input speed, pattern, timestamp, and the like.

[0680] "Server" refers to a computer system that receives and analyzes user input data.

[0681] The "analysis buffer" refers to a memory area that temporarily stores received data.

[0682] "Natural language processing module" refers to a software component that analyzes received text data and performs sentiment scoring and keyword extraction.

[0683] "Sentiment analysis engine" refers to a software component that infers a user's emotional state from their typing speed and patterns.

[0684] "Warning Generation Module" means the software component that generates a warning message based on the analysis results of the Natural Language Processing Module and the Sentiment Analysis Engine.

[0685] A "warning message" refers to a message that includes a warning when a user enters emotional and offensive words.

[0686] "Alternative phrases" refer to suggested milder versions of offensive language.

[0687] MODE FOR CARRYING OUT THE INVENTION

[0688] The system of the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and a sentiment analysis engine, and generates and displays a warning message to the user if the input contains offensive words or phrases. This system aims to improve the quality of communication by issuing a warning before the user becomes emotional and sends an offensive message.

[0689] Hardware and software used

[0690] Server: A computer system that receives, analyzes, and generates alert messages.

[0691] Terminal: The device on which the user inputs information (e.g., smartphone, PC)

[0692] Natural language processing module: A software component that performs sentiment analysis and keyword extraction on text data (e.g., Hugging Face's Transformers library, Google Cloud Natural Language API).

[0693] Sentiment analysis engine: A software component that infers emotional states from input speed and patterns (e.g., Apache Kafka)

[0694] Alert Generation Module: A software component that generates an alert message (e.g., AlertManager).

[0695] Specific operation of the system

[0696] First, a user enters a comment using a device such as a chat application or a social networking platform. For example, the user enters "You're an idiot" in a text field. The device temporarily stores the text data entered by the user and metadata such as typing speed and pattern, and then sends the contents to the server. The metadata includes keystroke speed (e.g., 5 characters per second).

[0697] The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis. The received data is in the format of "Text: You're an idiot" and "Speed: 5 characters per second." The server then passes this text data and metadata to the natural language processing module and sentiment analysis engine.

[0698] The natural language processing module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative sentiment scores. At this point, the offensive keyword "idiot" is extracted from the phrase "You're an idiot," and the sentiment score is calculated as "aggressive." The sentiment analysis engine estimates the user's emotional state from their typing speed and patterns. If their typing speed increases suddenly (for example, from 5 characters per second to 8 characters per second), it is assumed that the user is emotional. The server combines the results of the NLP module and the sentiment analysis engine to make a final assessment. For example, the results may be "Text: Aggressive," "Sentiment Score: High," and "Speed: Rapid Increase."

[0699] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, it may generate a warning message such as "Is the text offensive?" with an alternative phrase such as "How about this wording?" The generated warning message is then sent back to the terminal from the server.

[0700] The terminal displays this warning message to the user as a pop-up window. The user can review the warning message and correct the input if necessary. For example, they could correct it to "You seem to be overdoing it, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is not offensive, the server will not generate any further warnings and will allow the posting.

[0701] Examples and prompts

[0702] Specifically, User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and a sentiment analysis engine, detecting that the word "idiot" is offensive and that the user's typing speed has suddenly increased. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard, what's wrong?" The correction is sent to the server and inspected again. If there are no problems, the post is allowed to proceed.

[0703] In this way, by detecting offensive messages in real time and displaying a warning to the user, the quality of communication can be improved and abusive behavior can be reduced.

[0704] Examples of prompts include:

[0705] "Judge whether this message is offensive and issue a warning if necessary: ​​'You're an idiot.'"

[0706] "The user's typing speed has suddenly increased. Please rate whether this typing is emotional."

[0707] Using these prompts, the system can properly detect offensive messages and generate warnings.

[0708] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0709] Program processing steps

[0710] Step 1: Receive user input

[0711] 1. A user types a comment into a chat application or social media platform.

[0712] Example: A user types "You're an idiot" into a text field.

[0713] 2. Input: User's text data (e.g., "You're an idiot")

[0714] 3. Output: Input text data and metadata temporarily saved on the device

[0715] Step 2: The device sends the input

[0716] 1. The device temporarily stores the text data entered by the user and metadata (such as input speed and pattern).

[0717] Example: Temporarily save the text data "You're an idiot" with an input speed of 5 characters per second.

[0718] 2. The device sends the input and metadata to the server.

[0719] An interface API may be used for transmission.

[0720] 3. Input: User text data and metadata

[0721] 4. Output: The input text data and metadata sent to the server.

[0722] Step 3: The server receives the data and stores it in a buffer for analysis.

[0723] 1. The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis.

[0724] Example: Received data "Text: You're an idiot" and "Speed: 5 characters / second" is saved in a buffer.

[0725] 2. Input: Text data and metadata sent from the device

[0726] 3. Output: Data saved in the analysis buffer

[0727] Step 4: Text analysis using the natural language processing module

[0728] 1. The server passes the text data and metadata to a natural language processing (NLP) module.

[0729] Example: Send the text data "You're an idiot" to the NLP module.

[0730] 2. The NLP module analyzes the received text data, calculates sentiment scores, and extracts keywords.

[0731] Example: Extract the keyword "idiot" from the text "You're an idiot" and calculate the emotion score of "aggressive."

[0732] 3. Input: Text data

[0733] 4. Output: Keywords and sentiment scores

[0734] Step 5: Metadata analysis with sentiment analysis engine

[0735] 1. The server passes the metadata to the sentiment analysis engine.

[0736] Example: Sending metadata to a sentiment analysis engine with an input rate of 5 characters per second.

[0737] 2. The sentiment analysis engine infers the user's emotional state from their typing speed and patterns.

[0738] Example: Detecting a sudden increase in typing speed and inferring that the user is becoming "emotional."

[0739] 3. Input: Metadata

[0740] 4. Output: Estimation of emotional state

[0741] Step 6: Synthesis of the analysis results

[0742] 1. The server integrates the results of the NLP module and the sentiment analysis engine and performs the final evaluation.

[0743] Combined results: "Text: Aggressive" "Sentiment score: High" "Speed: Rapid"

[0744] 2. Input: Keywords and sentiment scores from the NLP module, emotional states from the sentiment analysis engine

[0745] 3. Output: Integrated analysis results

[0746] Step 7: Generate a warning message

[0747] 1. The server's warning generation module generates appropriate warning messages based on the integrated analysis results.

[0748] For example, generate a warning message such as "Is this sentence offensive?" with an alternative phrase such as "How about this phrase?"

[0749] 2. Input: Integrated analysis results

[0750] 3. Output: Generated warning messages

[0751] Step 8: Sending a warning message

[0752] 1. The server sends the generated warning message back to the terminal.

[0753] For example: Sending a message saying, "Is the text becoming aggressive?"

[0754] 2. Input: Generated warning message

[0755] 3. Output: Warning message sent to terminal

[0756] Step 9: Your device will display a warning message

[0757] 1. The terminal displays the received warning message to the user as a pop-up window.

[0758] For example, a warning message may appear on the user's screen asking, "Is the text offensive?"

[0759] 2. Input: Received warning message

[0760] 3. Output: The displayed warning message

[0761] Step 10: User corrects input

[0762] 1. The user checks the displayed warning message and corrects the input as necessary.

[0763] For example, correct them with, "You seem like you're overdoing it, what's wrong?"

[0764] 2. Input: Correct the input of the user who saw the warning message.

[0765] 3. Output: Corrected text data

[0766] Step 11: Resubmit your corrections

[0767] 1. The terminal sends the corrected text back to the server.

[0768] Example: Send the modified text "You seem to be overdoing it, what's wrong?" to the server.

[0769] 2. Input: Corrected text data

[0770] 3. Output: The corrected text sent to the server

[0771] Step 12: Reanalysis and final check

[0772] 1. The server analyzes the corrected text again using the NLP module and sentiment analysis engine, and if there are no problems, allows it to be posted.

[0773] Example: The corrected text "You seem to be overdoing it, what's wrong?" is determined to be non-aggressive.

[0774] 2. Input: Corrected text data

[0775] 3. Output: Based on the analysis results, if there are no problems, the submission is permitted.

[0776] In this way, the system can monitor user input in real time and immediately issue a warning if it contains offensive content, allowing users to regain their composure and reduce abusive behavior.

[0777] (Application example 2)

[0778] 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."

[0779] In electronic payment services, there is a need to reduce offensive language and emotional exchanges that can occur during communication between users and support staff. However, conventional systems lack the means to monitor such offensive language in real time and generate and present appropriate warnings or alternative phrases. This poses a challenge in maintaining the dignity of communication and reducing the mental burden on users.

[0780] The specific processing by the specific 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 receiving user input, means for analyzing the received user input, means for generating a warning message in response to the input content, means for displaying the generated warning message to the user, and means for analyzing the user's input pattern and estimating the user's emotional state. This makes it possible to monitor and correct offensive words in real time in communications for electronic payment services and suggest alternative phrases, thereby preventing slander and reducing the mental burden on users.

[0781] "User input" refers to text data sent by a user to support staff or the system in an electronic payment service.

[0782] "Means for receiving" refers to an interface that allows the system to recognize text data entered by the user and then send it to the server.

[0783] "Means for analyzing" refers to the process of analyzing received text data using natural language processing techniques and emotion engines to assess offensive language and the user's emotional state.

[0784] The "means for generating a warning message" refers to a module for creating a message to warn or caution the user based on the results of the analysis.

[0785] "Means for displaying to the user" refers to an interface for displaying the generated warning message as a pop-up or notification on the user's device.

[0786] "User input patterns" refers to metadata such as the speed and timing at which a user types text, and are used to infer emotional states.

[0787] "Means for inferring emotional state" refers to an engine or algorithm that analyzes a user's input pattern and speed to determine whether the user is emotional.

[0788] "Means for suggesting alternative phrases" refers to a module that suggests replacing offensive language with more appropriate and calmer expressions when a user uses it.

[0789] "Natural language processing technology" refers to technology for analyzing text data and understanding and extracting its meaning and emotions.

[0790] "Input speed and pattern metadata" refers to additional information, such as the speed and frequency at which a user actually types text, that can be used to infer a user's emotional state.

[0791] The present invention is a system that monitors communication between users and support staff in electronic payment services in real time, detects offensive words and phrases, and displays a warning message. The system analyzes the user's input data and its metadata (input speed and patterns) to prevent excessive remarks. Specific embodiments for implementing the present invention are described below.

[0792] Hardware / Software used

[0793] Hardware:

[0794] Smartphone

[0795] tablet

[0796] personal computer

[0797] server

[0798] software:

[0799] Natural Language Processing (NLP) modules (e.g., spaCy, NLTK)

[0800] Emotion engine (proprietary or public library)

[0801] Interfaces (e.g. React, Vue.js)

[0802] Means of implementation

[0803] server

[0804] The server receives the text data sent by the user and its metadata (input speed, pattern) and temporarily stores it in a buffer for analysis.

[0805] The server then passes the received data to the NLP module and emotion engine. The NLP module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative emotion scores. The emotion engine also analyzes the user's typing speed and patterns to estimate the user's emotional state.

[0806] The warning generation module combines these analysis results and generates an appropriate warning message. For example, if the user's input is an offensive one such as "You're an idiot," the system generates a warning message saying "Is the sentence offensive?" along with an alternative phrase such as "Let's speak from a more rational perspective."

[0807] Terminal

[0808] The device, such as a smartphone or tablet, collects text data and its metadata entered by the user in real time and sends it to a server, which then displays a warning message and alternative phrases received from the server as a pop-up window.

[0809] User

[0810] The user checks the warning message and corrects the input content as necessary. For example, the user sends the corrected text "You seem to be pushing yourself too hard. What's wrong?" to the server again.

[0811] Specific examples

[0812] If a user types "This system is completely unusable, it's stupid!" into the support section of an electronic payment service, the system analyzes the input data in real time. The server detects the offensive word "stupid" and infers from the metadata that the user is emotional. The warning generation module displays messages on the terminal such as "Is this sentence offensive?" and "Let's speak from a more rational perspective." The user confirms this and corrects their input by saying, "There seems to be a slight problem with this system. How can I fix it?"

[0813] Prompt Sentence Examples

[0814] "Evaluate the offensiveness of messages entered by users in real time and estimate the user's emotional state using an emotion engine. If offensive language is included, generate and display an appropriate warning message."

[0815] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0816] Step 1:

[0817] The device receives the user's text input in real time and collects the input data and metadata such as input speed and timing. The device temporarily stores this information and sends it to the server. Input: User's text data and metadata. Output: Transfer of received text data and metadata to the server.

[0818] Step 2:

[0819] The server receives the text data and metadata sent from the terminal and temporarily stores them in a buffer for analysis. Input: Text data and metadata from the terminal. Output: Data stored in the buffer for analysis.

[0820] Step 3:

[0821] The server analyzes the text data stored in the buffer using a natural language processing (NLP) module to extract sentiment and keywords from the text. The NLP module performs calculations to detect offensive words and negative sentiment scores. Input: Text data in the buffer. Output: Sentiment analysis results and keyword extraction results.

[0822] Step 4:

[0823] In parallel, the server passes the metadata to the emotion engine, which analyzes the user's input pattern and speed to estimate their emotional state. If the input speed suddenly increases, the emotion engine estimates that the user is emotional. Input: Metadata in the buffer. Output: Estimated emotional state.

[0824] Step 5:

[0825] The server integrates the analysis results of the NLP module with the emotional state of the emotion engine and passes them to the warning generation module. The warning generation module generates an appropriate warning message and alternative phrases if the message contains offensive language or has a high negative emotion score. Input: Sentiment analysis results, keyword extraction results, estimated emotional state. Output: Generated warning message and alternative phrases.

[0826] Step 6:

[0827] The server sends the generated warning message and alternative phrase to the terminal. The terminal displays these messages to the user as a popup window. Input: Generated warning message and alternative phrase. Output: Sent warning message to the user terminal.

[0828] Step 7:

[0829] The user checks the displayed warning message and corrects the input as necessary. For example, the user might correct the input by saying, "You seem to be pushing yourself too hard. What's wrong?" Input: The user's reaction after receiving the warning message. Output: The corrected text data.

[0830] Step 8:

[0831] The terminal sends the text data corrected by the user to the server again, and the server repeats the same process to re-analyze it. If there are no offensive words, the server allows the post. Input: Corrected text data. Output: Final decision on whether the post is allowed or not.

[0832] 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.

[0833] 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.

[0834] 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.

[0835] [Third embodiment]

[0836] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0837] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0838] 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).

[0839] 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.

[0840] 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.

[0841] 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).

[0842] 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.

[0843] 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.

[0844] 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.

[0845] 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.

[0846] 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.

[0847] 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."

[0848] The system of the present invention monitors user input in real time, analyzes the content of the input using natural language processing technology, and generates and displays a warning message to the user if the input contains offensive words or phrases. Specific embodiments will be described in detail below.

[0849] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and transmits the content to a server.

[0850] The server receives text data sent from the device, temporarily stores it in a buffer for analysis, and then passes it to a natural language processing (NLP) module.

[0851] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0852] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[0853] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user. The user can check the displayed warning message and correct the input content as necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[0854] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[0855] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content to the server. The server analyzes the received text using an NLP module and detects that "idiot" is an offensive word. The warning generation module generates a warning message and sends it to User A, saying, "Is this sentence offensive?" User A checks the warning and corrects the input, saying, "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[0856] In this way, the system can help users regain their composure even in emotional situations, thereby reducing slander.

[0857] The processing flow will be explained below.

[0858] Step 1:

[0859] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[0860] Step 2:

[0861] The device temporarily stores the user's input and creates an HTTP POST request to send the entered text data to the server.

[0862] Step 3:

[0863] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data is the input text "You're an idiot."

[0864] Step 4:

[0865] The server calls an internal method to pass the received text data to the NLP module, or executes an API call.

[0866] Step 5:

[0867] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[0868] Step 6:

[0869] The NLP module generates the analysis results and sends them back to the server's alert generation module. The resulting data includes a list of offensive keywords and a sentiment score.

[0870] Step 7:

[0871] The server's warning generation module receives the analysis results from the NLP module and generates an appropriate warning message, such as "Is the text offensive?"

[0872] Step 8:

[0873] The interface for sending the generated warning message from the warning generation module to the UI module is called, and the generated warning message is sent to the terminal as a response.

[0874] Step 9:

[0875] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[0876] Step 10:

[0877] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[0878] Step 11:

[0879] The device creates an HTTP POST request to send the corrected text data back to the server.

[0880] Step 12:

[0881] The server receives the corrected text data and checks it again using the same NLP module to ensure it does not contain any offensive content.

[0882] Step 13:

[0883] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[0884] Through these steps, you can respond carefully even when end users are emotional and prevent slander.

[0885] Example 1

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

[0887] In modern online communication, users may unconsciously use offensive language, potentially hurting the other person or a third party. This problem has led to online slander and cyberbullying, becoming a social issue. To prevent this, a system is needed that can monitor user input in real time, automatically detect offensive language, and issue appropriate warnings.

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

[0889] In this invention, the server includes means for acquiring user input, means for transmitting the acquired user input data to the server, means for the server to receive and temporarily store the input data transmitted from the user, means for analyzing the user input data using natural language processing technology, means for generating a warning message based on the input data, means for transmitting the generated warning message to the user, and means for re-analyzing the input data corrected by the user and determining the offensiveness of the input data. This enables the user to receive advice in real time to avoid offensive language without being swayed by emotion.

[0890] "User" refers to an individual or group that uses a terminal to input comments or messages.

[0891] "Input data" refers to text information entered into the system by a user via a terminal.

[0892] "Server" refers to a remote computer system for receiving, storing, and analyzing user-input data.

[0893] "Terminal" refers to a device (e.g., a smartphone, tablet, or personal computer) that a user uses to enter input data.

[0894] "Natural language processing (NLP)" refers to technology that enables computers to understand, interpret, and generate human language.

[0895] "Warning Message" refers to a notification that is generated when text entered by a user is determined to be offensive or inappropriate.

[0896] "Alternative phrases" refers to alternative suggestions for offensive or inappropriate text entered by a user.

[0897] "Aggression assessment" refers to the process of using natural language processing technology to determine whether input data has the potential to harm others.

[0898] "Real-time" refers to data being entered and processed nearly simultaneously, with minimal delay.

[0899] MODE FOR CARRYING OUT THE INVENTION

[0900] The system of the present invention monitors text data entered by users on their terminals in real time and analyzes the data using natural language processing technology. If the analysis detects offensive words or phrases, it generates a warning message and displays it to the user.

[0901] User Input

[0902] Users use devices such as smartphones, tablets, and personal computers to input comments on chat applications and social networking platforms. These inputs are temporarily stored on the device and then sent to a server.

[0903] Data transmission and reception

[0904] The terminal transmits the text data entered by the user. This data is sent to a server via the Internet, and the server temporarily stores the received data in a buffer for analysis. A database management system (e.g., MySQL or PostgreSQL) is used for storage.

[0905] Data analysis using natural language processing (NLP) modules

[0906] The server passes the stored text data to an NLP module. Specifically, it uses natural language processing services such as Google Cloud Natural Language API and IBM Watson Natural Language Understanding. The NLP module analyzes the text data and extracts sentiment scores and offensive keywords. For example, if the word "idiot" is included, it is determined that this word is offensive.

[0907] Generate a warning message

[0908] Once the analysis results are returned to the server, a warning generation module receives them and generates an appropriate warning message. For example, the input "You're an idiot" generates a warning message such as "Is this sentence aggressive?". Additionally, an alternative phrase such as "Let's speak from a more rational perspective" can be included.

[0909] Displaying a warning message

[0910] The generated warning message is sent from the server to the terminal, and the terminal displays the warning message to the user. By displaying the warning message using a pop-up window, the user can check the message in real time.

[0911] Correcting input information

[0912] The user can check the displayed warning message and correct their input if necessary. For example, they could make a correction such as, "You seem to be trying too hard, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is deemed to be non-offensive, the server allows it to be posted.

[0913] Specific scenario example

[0914] 1. User A types "You're an idiot" into a chat app and presses the send button.

[0915] 2. The device sends the input information to the server.

[0916] 3. The server analyzes the text data using an NLP module to detect offensive keywords.

[0917] 4. The warning generation module generates a message saying "Is the text offensive?" and sends it to User A.

[0918] 5. User A sees the warning and corrects his input, saying, "You seem to be pushing yourself too hard. What's wrong?"

[0919] 6. The corrections are sent to the server, inspected again, and if there are no problems, the post is allowed to be posted.

[0920] Prompt Sentence Examples

[0921] Design a system that generates and displays a warning message to users who enter offensive comments, as follows:

[0922] It monitors the text entered by the user in real time, performs sentiment analysis and keyword extraction, and if it detects offensive words such as "idiot" or "die," it displays a message asking, "Is your writing offensive?" and suggests an alternative phrase, such as, "Let's speak from a more calm perspective."

[0923] This system will help curb online slander in real time, allowing users to maintain a calm perspective.

[0924] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0925] Step 1:

[0926] A user inputs a comment on the device. For example, user A inputs "You're an idiot." This input is temporarily stored in the device's memory.

[0927] Input: Text entered by the user (e.g., "You're an idiot")

[0928] Output: Temporarily saved text data

[0929] Step 2:

[0930] The device sends the saved text data to the server using HTTP or WebSocket as the communication protocol.

[0931] Input: Temporarily saved text data

[0932] Output: Text data sent to the server

[0933] Step 3:

[0934] The server stores the received text data in a buffer for analysis, using a database management system (e.g., MySQL or PostgreSQL).

[0935] Input: Text data sent to the server

[0936] Output: Text data saved in the analysis buffer

[0937] Step 4:

[0938] The server passes the stored text data to a natural language processing (NLP) module, which sends the data through an API. The NLP module uses Google Cloud Natural Language API or IBM Watson Natural Language Understanding.

[0939] Input: Text data stored in the parsing buffer

[0940] Output: Text data passed to the NLP module

[0941] Step 5:

[0942] The NLP module analyzes the text data and extracts sentiment scores and offensive keywords, for example, if the word "idiot" is included, it will be determined that this word is offensive.

[0943] Input: Text data passed to the NLP module

[0944] Output: Sentiment score and offensive keywords

[0945] Step 6:

[0946] The server's warning generation module receives the analysis results returned by the NLP module and generates an appropriate warning message. For example, for the input "You're an idiot," it generates the warning message "Is this sentence aggressive?". Additionally, it can include an alternative phrase, "Let's speak from a more rational perspective."

[0947] Input: Sentiment score and offensive keywords

[0948] Output: Warning message and alternative phrase

[0949] Step 7:

[0950] The server then sends the generated alert messages to the terminal, again using HTTP or WebSocket to transmit data in real time.

[0951] Input: warning message and alternative phrase

[0952] Output: Warning message sent to terminal

[0953] Step 8:

[0954] The terminal displays a warning message to the user, for example, by using a pop-up window.

[0955] Input: The warning message sent to the terminal

[0956] Output: A warning message that is displayed to the user.

[0957] Step 9:

[0958] The user checks the displayed warning message and corrects the input content if necessary. For example, the user might say, "You seem to be pushing yourself too hard. What's wrong?" The corrected text data is then sent to the server again.

[0959] Input: The warning message displayed to the user

[0960] Output: Corrected text data

[0961] Step 10:

[0962] The server then passes the corrected text data back to the NLP module for re-analysis. If the corrected input is deemed non-offensive, the server allows the edited content to be posted as is.

[0963] Input: Modified text data

[0964] Output: Text data that is allowed to be posted

[0965] This series of steps allows users to regain a calm perspective even in emotional situations and suppress or reduce slander.

[0966] (Application example 1)

[0967] 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."

[0968] In recent years, customer service in brick-and-mortar stores has become increasingly important, but there is a challenge in properly managing responses based on human emotions and reactions. In particular, if staff become emotional or use offensive language, it can worsen relationships with customers and damage the credibility of the business. There is a need for a multi-functional system that can prevent such situations and ensure calm and courteous customer service at all times.

[0969] 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.

[0970] In this invention, the server includes means for receiving user input, means for analyzing the received user input, means for generating a warning message for the input content, means for displaying the generated warning message to the user, means for converting the voice input into text data, and means for monitoring emotional words and offensive phrases in real time using the text data. This enables store staff to automatically detect offensive words and content containing negative emotions during conversations with customers, receive appropriate warnings, and encourage calm responses.

[0971] "User" refers to the entity that operates the system or makes input.

[0972] "Input" refers to information or data that a user provides to a system.

[0973] "Voice input" refers to information provided to the system by a user speaking.

[0974] "Text data" refers to written information generated by voice input or other means.

[0975] "Analysis" refers to the process of interpreting input information and understanding its content and intent.

[0976] "Emotional words" refer to words that strongly express the user's emotions.

[0977] "Offensive phrases" refer to words or expressions that can have a negative impact on others.

[0978] A "warning message" refers to a notification that notifies the user that there is a problem with the input content.

[0979] "Voice recognition technology" refers to the technology that converts voice input into text data.

[0980] "Natural language processing technology" refers to technology for understanding and analyzing human language.

[0981] "Smart glasses" refers to glasses-type devices that are worn by the user to display information and receive input.

[0982] "Server" refers to a central processing unit that processes information and stores data.

[0983] "Monitoring" refers to the act of continuously observing specific information or behavior to detect changes or problems.

[0984] The present invention relates to a system for supporting customer service in brick-and-mortar stores, and in particular to a system that detects conversations containing offensive language or negative emotions in real time and generates and displays appropriate warning messages, thereby enabling staff to respond calmly. An embodiment of the present invention will be described in detail below.

[0985] System Configuration

[0986] The system consists of the following main components:

[0987] 1. Smart Glasses

[0988] Staff wear smart glasses and use voice input, which is collected through the smart glasses' microphone.

[0989] 2. Voice recognition function

[0990] Voice input is converted to text in real time using cloud-based or built-in speech recognition technology, such as the Google Speech-to-Text API.

[0991] 3. Server

[0992] The server has the following roles:

[0993] Data reception: Receives text data sent from the smart glasses.

[0994] Natural Language Processing (NLP): Analyzes incoming text data to detect emotional words and offensive phrases. NLP technologies available include spacy and TextBlob.

[0995] Warning generation: When emotive words or offensive phrases are detected, appropriate warning messages and alternative phrases are generated.

[0996] Message sending: The generated warning message is sent to the smart glasses and displayed to staff in real time.

[0997] Data processing and calculation

[0998] 1. Convert voice input to text data:

[0999] The voice data collected by the smart glasses is converted into text data in real time using a voice recognition API. For example, it can be converted into the following format:

[1000] Voice input: "You're an idiot."

[1001] Speech recognition API output: Text data "You're an idiot"

[1002] 2. Text data analysis:

[1003] The server receives the text data and analyzes it using natural language processing technology. Specifically, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is detected as an offensive word.

[1004] 3. Generate a warning message:

[1005] If offensive language or negative sentiment is detected, the server uses a warning generation module to generate an appropriate warning message and alternative phrases. For example, the following warning message may be generated:

[1006] Warning message: "Is the conversation getting aggressive? Try to stay calm."

[1007] 4. Displaying messages:

[1008] The generated warning messages are sent to smart glasses and displayed to staff in real time.

[1009] Specific examples

[1010] For example, this system works if a staff member in a physical store says "You're an idiot" while interacting with a customer. The voice input is recognized by the smart glasses and converted into text data such as "You're an idiot" via the Google Speech-to-Text API. This text data is sent to the server and analyzed by the NLP module. Here, "idiot" is detected as an offensive word, and the server generates a warning message saying, "Is the conversation becoming aggressive? Please try to respond calmly." This message is displayed on the smart glasses, and the staff member can immediately correct their response by asking, "Can you tell me what's wrong?"

[1011] Prompt Sentence Examples

[1012] If a customer types "You're an idiot," create a warning message that this system will display and a suggested calm response.

[1013] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1014] Step 1:

[1015] The user wears the smart glasses and inputs voice data. The microphone in the smart glasses collects the voice data.

[1016] Input: Audio data (e.g., "You're an idiot").

[1017] Output: The audio data is kept intact in the smart glasses and is ready to be sent to the server.

[1018] Step 2:

[1019] The smart glasses send the collected voice data to a server, which then passes the received voice data to a voice recognition API.

[1020] Input: Audio data transmitted from smart glasses.

[1021] Output: A request is sent to the speech recognition API.

[1022] Step 3:

[1023] The server uses a speech recognition API (e.g., Google Speech-to-Text API) to convert the voice data into text data in real time.

[1024] Input: The audio data received by the speech recognition API.

[1025] Output: Text data (e.g., "You're an idiot") is generated.

[1026] Step 4:

[1027] The server passes the generated text data to a natural language processing (NLP) module for analysis, specifically detecting offensive words and negative sentiment scores within the text.

[1028] Input: Text data (e.g., "You're an idiot").

[1029] Output: Analysis results (e.g. "idiot" is detected as an offensive word).

[1030] Step 5:

[1031] Based on the analysis results from the NLP module, the server uses the warning generation module to generate appropriate warning messages and alternative phrases.

[1032] Input: Analysis results (offensive language detection results).

[1033] Output: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.") is generated.

[1034] Step 6:

[1035] The generated warning message is again sent from the server to the smart glasses, which then display the warning message to the user in real time.

[1036] Input: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.").

[1037] Output: A warning message will be displayed on the smart glasses display.

[1038] 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.

[1039] The system according to the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and an emotion engine, and generates and displays a warning message to the user if the input content contains offensive words or phrases. Specific embodiments are described in detail below.

[1040] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and metadata such as input speed and pattern, and then transmits the contents to the server.

[1041] The server receives text data and metadata from the device. This data is temporarily stored in a buffer for analysis. The server then passes the text data to the NLP module and emotion engine.

[1042] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[1043] In parallel, the emotion engine analyzes the user's input speed and patterns to estimate the user's emotional state. For example, a sudden increase in input speed may suggest that the user is emotional. This estimated emotional state is then integrated with the analysis results of the NLP module to make a final assessment.

[1044] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[1045] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user as a pop-up window. The user can check the displayed warning message and correct the input content if necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[1046] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[1047] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and an emotion engine, and detects that "idiot" is an offensive word and that the user's typing speed is increasing. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[1048] In this way, a system that combines an emotion engine can respond carefully even in situations where the user is emotional, thereby reducing slander.

[1049] The processing flow will be explained below.

[1050] Step 1:

[1051] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[1052] Step 2:

[1053] The device temporarily stores the user's input and makes an HTTP POST request to send the entered text data and metadata such as typing speed and pattern to the server.

[1054] Step 3:

[1055] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data consists of the input text "You're an idiot" and metadata such as input speed and pattern.

[1056] Step 4:

[1057] The server calls an internal method to pass the received text data to the NLP module, and the metadata is passed to the emotion engine.

[1058] Step 5:

[1059] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[1060] Step 6:

[1061] The emotion engine analyzes the metadata and analyzes the user's typing speed and patterns. Based on this data, it infers the user's emotional state. For example, if the user's typing speed suddenly increases, it infers that the user is angry.

[1062] Step 7:

[1063] The NLP module and the sentiment engine combine their respective analysis results to generate a comprehensive assessment result, which includes a list of offensive keywords, a sentiment score, and an estimated emotional state.

[1064] Step 8:

[1065] The server's warning generation module receives the integrated evaluation results and generates appropriate warning messages, such as "Is the sentence offensive?" and "How about this wording?" suggestions.

[1066] Step 9:

[1067] The warning generation module calls an interface to send the generated warning message and alternative phrase to the UI module, which then sends the generated warning message and alternative phrase to the terminal as a response.

[1068] Step 10:

[1069] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[1070] Step 11:

[1071] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[1072] Step 12:

[1073] The device creates an HTTP POST request to send the corrected text data back to the server.

[1074] Step 13:

[1075] The server receives the corrected text data and runs it through the same NLP module and emotion engine again to ensure it does not contain any offensive content.

[1076] Step 14:

[1077] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[1078] Through the above steps, the system can carefully respond to even emotional end users and prevent slander, making communication on the Internet healthier and more constructive.

[1079] Example 2

[1080] 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."

[1081] Current online communication platforms have a problem where users become emotional and send offensive messages, which can worsen dialogue within the community. Furthermore, the mental stress and defamation caused by such offensive messages are also serious problems. The problem with existing methods is that they do not adequately implement mechanisms to detect and warn users about offensive messages in advance.

[1082] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: means for receiving a user input; means for temporarily storing the received user input and transmitting it to the server together with metadata such as input speed and pattern; means for saving the text data and metadata received by the server in an analysis buffer and passing them to a natural language processing module and a sentiment analysis engine; means for the natural language processing module to perform sentiment analysis and keyword extraction on the text data and detect offensive words and negative sentiment scores; means for the sentiment analysis engine to estimate the user's emotional state from the user's input speed and pattern and integrate the results; means for generating a warning message based on the integrated analysis results; and means for displaying the generated warning message to the user. This makes it possible to issue a warning before a user becomes emotional and sends an offensive message, thereby improving the quality of communication.

[1083] "User" means a person who enters comments or messages on the Online Platform.

[1084] "Input speed" refers to the speed at which a user types characters, measured in characters per second.

[1085] "Metadata" refers to various data related to user input, including input speed, pattern, timestamp, and the like.

[1086] "Server" refers to a computer system that receives and analyzes user input data.

[1087] The "analysis buffer" refers to a memory area that temporarily stores received data.

[1088] "Natural language processing module" refers to a software component that analyzes received text data and performs sentiment scoring and keyword extraction.

[1089] "Sentiment analysis engine" refers to a software component that infers a user's emotional state from their typing speed and patterns.

[1090] "Warning Generation Module" means the software component that generates a warning message based on the analysis results of the Natural Language Processing Module and the Sentiment Analysis Engine.

[1091] A "warning message" refers to a message that includes a warning when a user enters emotional and offensive words.

[1092] "Alternative phrases" refer to suggested milder versions of offensive language.

[1093] MODE FOR CARRYING OUT THE INVENTION

[1094] The system of the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and a sentiment analysis engine, and generates and displays a warning message to the user if the input contains offensive words or phrases. This system aims to improve the quality of communication by issuing a warning before the user becomes emotional and sends an offensive message.

[1095] Hardware and software used

[1096] Server: A computer system that receives, analyzes, and generates alert messages.

[1097] Terminal: The device on which the user inputs information (e.g., smartphone, PC)

[1098] Natural language processing module: A software component that performs sentiment analysis and keyword extraction on text data (e.g., Hugging Face's Transformers library, Google Cloud Natural Language API).

[1099] Sentiment analysis engine: A software component that infers emotional states from input speed and patterns (e.g., Apache Kafka)

[1100] Alert Generation Module: A software component that generates an alert message (e.g., AlertManager).

[1101] Specific operation of the system

[1102] First, a user enters a comment using a device such as a chat application or a social networking platform. For example, the user enters "You're an idiot" in a text field. The device temporarily stores the text data entered by the user and metadata such as typing speed and pattern, and then sends the contents to the server. The metadata includes keystroke speed (e.g., 5 characters per second).

[1103] The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis. The received data is in the format of "Text: You're an idiot" and "Speed: 5 characters per second." The server then passes this text data and metadata to the natural language processing module and sentiment analysis engine.

[1104] The natural language processing module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative sentiment scores. At this point, the offensive keyword "idiot" is extracted from the phrase "You're an idiot," and the sentiment score is calculated as "aggressive." The sentiment analysis engine estimates the user's emotional state from their typing speed and patterns. If their typing speed increases suddenly (for example, from 5 characters per second to 8 characters per second), it is assumed that the user is emotional. The server combines the results of the NLP module and the sentiment analysis engine to make a final assessment. For example, the results may be "Text: Aggressive," "Sentiment Score: High," and "Speed: Rapid Increase."

[1105] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, it may generate a warning message such as "Is the text offensive?" with an alternative phrase such as "How about this wording?" The generated warning message is then sent back to the terminal from the server.

[1106] The terminal displays this warning message to the user as a pop-up window. The user can review the warning message and correct the input if necessary. For example, they could correct it to "You seem to be overdoing it, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is not offensive, the server will not generate any further warnings and will allow the posting.

[1107] Examples and prompts

[1108] Specifically, User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and a sentiment analysis engine, detecting that the word "idiot" is offensive and that the user's typing speed has suddenly increased. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard, what's wrong?" The correction is sent to the server and inspected again. If there are no problems, the post is allowed to proceed.

[1109] In this way, by detecting offensive messages in real time and displaying a warning to the user, the quality of communication can be improved and abusive behavior can be reduced.

[1110] Examples of prompts include:

[1111] "Judge whether this message is offensive and issue a warning if necessary: ​​'You're an idiot.'"

[1112] "The user's typing speed has suddenly increased. Please rate whether this typing is emotional."

[1113] Using these prompts, the system can properly detect offensive messages and generate warnings.

[1114] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1115] Program processing steps

[1116] Step 1: Receive user input

[1117] 1. A user types a comment into a chat application or social media platform.

[1118] Example: A user types "You're an idiot" into a text field.

[1119] 2. Input: User's text data (e.g., "You're an idiot")

[1120] 3. Output: Input text data and metadata temporarily saved on the device

[1121] Step 2: The device sends the input

[1122] 1. The device temporarily stores the text data entered by the user and metadata (such as input speed and pattern).

[1123] Example: Temporarily save the text data "You're an idiot" with an input speed of 5 characters per second.

[1124] 2. The device sends the input and metadata to the server.

[1125] An interface API may be used for transmission.

[1126] 3. Input: User text data and metadata

[1127] 4. Output: The input text data and metadata sent to the server.

[1128] Step 3: The server receives the data and stores it in a buffer for analysis.

[1129] 1. The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis.

[1130] Example: Received data "Text: You're an idiot" and "Speed: 5 characters / second" is saved in a buffer.

[1131] 2. Input: Text data and metadata sent from the device

[1132] 3. Output: Data saved in the analysis buffer

[1133] Step 4: Text analysis using the natural language processing module

[1134] 1. The server passes the text data and metadata to a natural language processing (NLP) module.

[1135] Example: Send the text data "You're an idiot" to the NLP module.

[1136] 2. The NLP module analyzes the received text data, calculates sentiment scores, and extracts keywords.

[1137] Example: Extract the keyword "idiot" from the text "You're an idiot" and calculate the emotion score of "aggressive."

[1138] 3. Input: Text data

[1139] 4. Output: Keywords and sentiment scores

[1140] Step 5: Metadata analysis with sentiment analysis engine

[1141] 1. The server passes the metadata to the sentiment analysis engine.

[1142] Example: Sending metadata to a sentiment analysis engine with an input rate of 5 characters per second.

[1143] 2. The sentiment analysis engine infers the user's emotional state from their typing speed and patterns.

[1144] Example: Detecting a sudden increase in typing speed and inferring that the user is becoming "emotional."

[1145] 3. Input: Metadata

[1146] 4. Output: Estimation of emotional state

[1147] Step 6: Synthesis of the analysis results

[1148] 1. The server integrates the results of the NLP module and the sentiment analysis engine and performs the final evaluation.

[1149] Combined results: "Text: Aggressive" "Sentiment score: High" "Speed: Rapid"

[1150] 2. Input: Keywords and sentiment scores from the NLP module, emotional states from the sentiment analysis engine

[1151] 3. Output: Integrated analysis results

[1152] Step 7: Generate a warning message

[1153] 1. The server's warning generation module generates appropriate warning messages based on the integrated analysis results.

[1154] For example, generate a warning message such as "Is this sentence offensive?" with an alternative phrase such as "How about this phrase?"

[1155] 2. Input: Integrated analysis results

[1156] 3. Output: Generated warning messages

[1157] Step 8: Sending a warning message

[1158] 1. The server sends the generated warning message back to the terminal.

[1159] For example: Sending a message saying, "Is the text becoming aggressive?"

[1160] 2. Input: Generated warning message

[1161] 3. Output: Warning message sent to terminal

[1162] Step 9: Your device will display a warning message

[1163] 1. The terminal displays the received warning message to the user as a pop-up window.

[1164] For example, a warning message may appear on the user's screen asking, "Is the text offensive?"

[1165] 2. Input: Received warning message

[1166] 3. Output: The displayed warning message

[1167] Step 10: User corrects input

[1168] 1. The user checks the displayed warning message and corrects the input as necessary.

[1169] For example, correct them with, "You seem like you're overdoing it, what's wrong?"

[1170] 2. Input: Correct the input of the user who saw the warning message.

[1171] 3. Output: Corrected text data

[1172] Step 11: Resubmit your corrections

[1173] 1. The terminal sends the corrected text back to the server.

[1174] Example: Send the modified text "You seem to be overdoing it, what's wrong?" to the server.

[1175] 2. Input: Corrected text data

[1176] 3. Output: The corrected text sent to the server

[1177] Step 12: Reanalysis and final check

[1178] 1. The server analyzes the corrected text again using the NLP module and sentiment analysis engine, and if there are no problems, allows it to be posted.

[1179] Example: The corrected text "You seem to be overdoing it, what's wrong?" is determined to be non-aggressive.

[1180] 2. Input: Corrected text data

[1181] 3. Output: Based on the analysis results, if there are no problems, the submission is permitted.

[1182] In this way, the system can monitor user input in real time and immediately issue a warning if it contains offensive content, allowing users to regain their composure and reduce abusive behavior.

[1183] (Application example 2)

[1184] 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."

[1185] In electronic payment services, there is a need to reduce offensive language and emotional exchanges that can occur during communication between users and support staff. However, conventional systems lack the means to monitor such offensive language in real time and generate and present appropriate warnings or alternative phrases. This poses a challenge in maintaining the dignity of communication and reducing the mental burden on users.

[1186] The specific processing by the specific 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 receiving user input, means for analyzing the received user input, means for generating a warning message in response to the input content, means for displaying the generated warning message to the user, and means for analyzing the user's input pattern and estimating the user's emotional state. This makes it possible to monitor and correct offensive words in real time in communications for electronic payment services and suggest alternative phrases, thereby preventing slander and reducing the mental burden on users.

[1187] "User input" refers to text data sent by a user to support staff or the system in an electronic payment service.

[1188] "Means for receiving" refers to an interface that allows the system to recognize text data entered by the user and then send it to the server.

[1189] "Means for analyzing" refers to the process of analyzing received text data using natural language processing techniques and emotion engines to assess offensive language and the user's emotional state.

[1190] The "means for generating a warning message" refers to a module for creating a message to warn or caution the user based on the results of the analysis.

[1191] "Means for displaying to the user" refers to an interface for displaying the generated warning message as a pop-up or notification on the user's device.

[1192] "User input patterns" refers to metadata such as the speed and timing at which a user types text, and are used to infer emotional states.

[1193] "Means for inferring emotional state" refers to an engine or algorithm that analyzes a user's input pattern and speed to determine whether the user is emotional.

[1194] "Means for suggesting alternative phrases" refers to a module that suggests replacing offensive language with more appropriate and calmer expressions when a user uses it.

[1195] "Natural language processing technology" refers to technology for analyzing text data and understanding and extracting its meaning and emotions.

[1196] "Input speed and pattern metadata" refers to additional information, such as the speed and frequency at which a user actually types text, that can be used to infer a user's emotional state.

[1197] The present invention is a system that monitors communication between users and support staff in electronic payment services in real time, detects offensive words and phrases, and displays a warning message. The system analyzes the user's input data and its metadata (input speed and patterns) to prevent excessive remarks. Specific embodiments for implementing the present invention are described below.

[1198] Hardware / Software used

[1199] Hardware:

[1200] Smartphone

[1201] tablet

[1202] personal computer

[1203] server

[1204] software:

[1205] Natural Language Processing (NLP) modules (e.g., spaCy, NLTK)

[1206] Emotion engine (proprietary or public library)

[1207] Interfaces (e.g. React, Vue.js)

[1208] Means of implementation

[1209] server

[1210] The server receives the text data sent by the user and its metadata (input speed, pattern) and temporarily stores it in a buffer for analysis.

[1211] The server then passes the received data to the NLP module and emotion engine. The NLP module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative emotion scores. The emotion engine also analyzes the user's typing speed and patterns to estimate the user's emotional state.

[1212] The warning generation module combines these analysis results and generates an appropriate warning message. For example, if the user's input is an offensive one such as "You're an idiot," the system generates a warning message saying "Is the sentence offensive?" along with an alternative phrase such as "Let's speak from a more rational perspective."

[1213] Terminal

[1214] The device, such as a smartphone or tablet, collects text data and its metadata entered by the user in real time and sends it to a server, which then displays a warning message and alternative phrases received from the server as a pop-up window.

[1215] User

[1216] The user checks the warning message and corrects the input content as necessary. For example, the user sends the corrected text "You seem to be pushing yourself too hard. What's wrong?" to the server again.

[1217] Specific examples

[1218] If a user types "This system is completely unusable, it's stupid!" into the support section of an electronic payment service, the system analyzes the input data in real time. The server detects the offensive word "stupid" and infers from the metadata that the user is emotional. The warning generation module displays messages on the terminal such as "Is this sentence offensive?" and "Let's speak from a more rational perspective." The user confirms this and corrects their input by saying, "There seems to be a slight problem with this system. How can I fix it?"

[1219] Prompt Sentence Examples

[1220] "Evaluate the offensiveness of messages entered by users in real time and estimate the user's emotional state using an emotion engine. If offensive language is included, generate and display an appropriate warning message."

[1221] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1222] Step 1:

[1223] The device receives the user's text input in real time and collects the input data and metadata such as input speed and timing. The device temporarily stores this information and sends it to the server. Input: User's text data and metadata. Output: Transfer of received text data and metadata to the server.

[1224] Step 2:

[1225] The server receives the text data and metadata sent from the terminal and temporarily stores them in a buffer for analysis. Input: Text data and metadata from the terminal. Output: Data stored in the buffer for analysis.

[1226] Step 3:

[1227] The server analyzes the text data stored in the buffer using a natural language processing (NLP) module to extract sentiment and keywords from the text. The NLP module performs calculations to detect offensive words and negative sentiment scores. Input: Text data in the buffer. Output: Sentiment analysis results and keyword extraction results.

[1228] Step 4:

[1229] In parallel, the server passes the metadata to the emotion engine, which analyzes the user's input pattern and speed to estimate their emotional state. If the input speed suddenly increases, the emotion engine estimates that the user is emotional. Input: Metadata in the buffer. Output: Estimated emotional state.

[1230] Step 5:

[1231] The server integrates the analysis results of the NLP module with the emotional state of the emotion engine and passes them to the warning generation module. The warning generation module generates an appropriate warning message and alternative phrases if the message contains offensive language or has a high negative emotion score. Input: Sentiment analysis results, keyword extraction results, estimated emotional state. Output: Generated warning message and alternative phrases.

[1232] Step 6:

[1233] The server sends the generated warning message and alternative phrase to the terminal. The terminal displays these messages to the user as a popup window. Input: Generated warning message and alternative phrase. Output: Sent warning message to the user terminal.

[1234] Step 7:

[1235] The user checks the displayed warning message and corrects the input as necessary. For example, the user might correct the input by saying, "You seem to be pushing yourself too hard. What's wrong?" Input: The user's reaction after receiving the warning message. Output: The corrected text data.

[1236] Step 8:

[1237] The terminal sends the text data corrected by the user to the server again, and the server repeats the same process to re-analyze it. If there are no offensive words, the server allows the post. Input: Corrected text data. Output: Final decision on whether the post is allowed or not.

[1238] 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.

[1239] 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.

[1240] 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.

[1241] [Fourth embodiment]

[1242] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1243] 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.

[1244] 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).

[1245] 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.

[1246] 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.

[1247] 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).

[1248] 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.

[1249] 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.

[1250] 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.

[1251] 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.

[1252] 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.

[1253] 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.

[1254] 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."

[1255] The system of the present invention monitors user input in real time, analyzes the content of the input using natural language processing technology, and generates and displays a warning message to the user if the input contains offensive words or phrases. Specific embodiments will be described in detail below.

[1256] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and transmits the content to a server.

[1257] The server receives text data sent from the device, temporarily stores it in a buffer for analysis, and then passes it to a natural language processing (NLP) module.

[1258] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[1259] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[1260] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user. The user can check the displayed warning message and correct the input content as necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[1261] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[1262] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content to the server. The server analyzes the received text using an NLP module and detects that "idiot" is an offensive word. The warning generation module generates a warning message and sends it to User A, saying, "Is this sentence offensive?" User A checks the warning and corrects the input, saying, "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[1263] In this way, the system can help users regain their composure even in emotional situations, thereby reducing slander.

[1264] The processing flow will be explained below.

[1265] Step 1:

[1266] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[1267] Step 2:

[1268] The device temporarily stores the user's input and creates an HTTP POST request to send the entered text data to the server.

[1269] Step 3:

[1270] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data is the input text "You're an idiot."

[1271] Step 4:

[1272] The server calls an internal method to pass the received text data to the NLP module, or executes an API call.

[1273] Step 5:

[1274] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[1275] Step 6:

[1276] The NLP module generates the analysis results and sends them back to the server's alert generation module. The resulting data includes a list of offensive keywords and a sentiment score.

[1277] Step 7:

[1278] The server's warning generation module receives the analysis results from the NLP module and generates an appropriate warning message, such as "Is the text offensive?"

[1279] Step 8:

[1280] The interface for sending the generated warning message from the warning generation module to the UI module is called, and the generated warning message is sent to the terminal as a response.

[1281] Step 9:

[1282] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[1283] Step 10:

[1284] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[1285] Step 11:

[1286] The device creates an HTTP POST request to send the corrected text data back to the server.

[1287] Step 12:

[1288] The server receives the corrected text data and checks it again using the same NLP module to ensure it does not contain any offensive content.

[1289] Step 13:

[1290] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[1291] Through these steps, you can respond carefully even when end users are emotional and prevent slander.

[1292] Example 1

[1293] 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."

[1294] In modern online communication, users may unconsciously use offensive language, potentially hurting the other person or a third party. This problem has led to online slander and cyberbullying, becoming a social issue. To prevent this, a system is needed that can monitor user input in real time, automatically detect offensive language, and issue appropriate warnings.

[1295] 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.

[1296] In this invention, the server includes means for acquiring user input, means for transmitting the acquired user input data to the server, means for the server to receive and temporarily store the input data transmitted from the user, means for analyzing the user input data using natural language processing technology, means for generating a warning message based on the input data, means for transmitting the generated warning message to the user, and means for re-analyzing the input data corrected by the user and determining the offensiveness of the input data. This enables the user to receive advice in real time to avoid offensive language without being swayed by emotion.

[1297] "User" refers to an individual or group that uses a terminal to input comments or messages.

[1298] "Input data" refers to text information entered into the system by a user via a terminal.

[1299] "Server" refers to a remote computer system for receiving, storing, and analyzing user-input data.

[1300] "Terminal" refers to a device (e.g., a smartphone, tablet, or personal computer) that a user uses to enter input data.

[1301] "Natural language processing (NLP)" refers to technology that enables computers to understand, interpret, and generate human language.

[1302] "Warning Message" refers to a notification that is generated when text entered by a user is determined to be offensive or inappropriate.

[1303] "Alternative phrases" refers to alternative suggestions for offensive or inappropriate text entered by a user.

[1304] "Aggression assessment" refers to the process of using natural language processing technology to determine whether input data has the potential to harm others.

[1305] "Real-time" refers to data being entered and processed nearly simultaneously, with minimal delay.

[1306] MODE FOR CARRYING OUT THE INVENTION

[1307] The system of the present invention monitors text data entered by users on their terminals in real time and analyzes the data using natural language processing technology. If the analysis detects offensive words or phrases, it generates a warning message and displays it to the user.

[1308] User Input

[1309] Users use devices such as smartphones, tablets, and personal computers to input comments on chat applications and social networking platforms. These inputs are temporarily stored on the device and then sent to a server.

[1310] Data transmission and reception

[1311] The terminal transmits the text data entered by the user. This data is sent to a server via the Internet, and the server temporarily stores the received data in a buffer for analysis. A database management system (e.g., MySQL or PostgreSQL) is used for storage.

[1312] Data analysis using natural language processing (NLP) modules

[1313] The server passes the stored text data to an NLP module. Specifically, it uses natural language processing services such as Google Cloud Natural Language API and IBM Watson Natural Language Understanding. The NLP module analyzes the text data and extracts sentiment scores and offensive keywords. For example, if the word "idiot" is included, it is determined that this word is offensive.

[1314] Generate a warning message

[1315] Once the analysis results are returned to the server, a warning generation module receives them and generates an appropriate warning message. For example, the input "You're an idiot" generates a warning message such as "Is this sentence aggressive?". Additionally, an alternative phrase such as "Let's speak from a more rational perspective" can be included.

[1316] Displaying a warning message

[1317] The generated warning message is sent from the server to the terminal, and the terminal displays the warning message to the user. By displaying the warning message using a pop-up window, the user can check the message in real time.

[1318] Correcting input information

[1319] The user can check the displayed warning message and correct their input if necessary. For example, they could make a correction such as, "You seem to be trying too hard, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is deemed to be non-offensive, the server allows it to be posted.

[1320] Specific scenario example

[1321] 1. User A types "You're an idiot" into a chat app and presses the send button.

[1322] 2. The device sends the input information to the server.

[1323] 3. The server analyzes the text data using an NLP module to detect offensive keywords.

[1324] 4. The warning generation module generates a message saying "Is the text offensive?" and sends it to User A.

[1325] 5. User A sees the warning and corrects his input, saying, "You seem to be pushing yourself too hard. What's wrong?"

[1326] 6. The corrections are sent to the server, inspected again, and if there are no problems, the post is allowed to be posted.

[1327] Prompt Sentence Examples

[1328] Design a system that generates and displays a warning message to users who enter offensive comments, as follows:

[1329] It monitors the text entered by the user in real time, performs sentiment analysis and keyword extraction, and if it detects offensive words such as "idiot" or "die," it displays a message asking, "Is your writing offensive?" and suggests an alternative phrase, such as, "Let's speak from a more calm perspective."

[1330] This system will help curb online slander in real time, allowing users to maintain a calm perspective.

[1331] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1332] Step 1:

[1333] A user inputs a comment on the device. For example, user A inputs "You're an idiot." This input is temporarily stored in the device's memory.

[1334] Input: Text entered by the user (e.g., "You're an idiot")

[1335] Output: Temporarily saved text data

[1336] Step 2:

[1337] The device sends the saved text data to the server using HTTP or WebSocket as the communication protocol.

[1338] Input: Temporarily saved text data

[1339] Output: Text data sent to the server

[1340] Step 3:

[1341] The server stores the received text data in a buffer for analysis, using a database management system (e.g., MySQL or PostgreSQL).

[1342] Input: Text data sent to the server

[1343] Output: Text data saved in the analysis buffer

[1344] Step 4:

[1345] The server passes the stored text data to a natural language processing (NLP) module, which sends the data through an API. The NLP module uses Google Cloud Natural Language API or IBM Watson Natural Language Understanding.

[1346] Input: Text data stored in the parsing buffer

[1347] Output: Text data passed to the NLP module

[1348] Step 5:

[1349] The NLP module analyzes the text data and extracts sentiment scores and offensive keywords, for example, if the word "idiot" is included, it will be determined that this word is offensive.

[1350] Input: Text data passed to the NLP module

[1351] Output: Sentiment score and offensive keywords

[1352] Step 6:

[1353] The server's warning generation module receives the analysis results returned by the NLP module and generates an appropriate warning message. For example, for the input "You're an idiot," it generates the warning message "Is this sentence aggressive?". Additionally, it can include an alternative phrase, "Let's speak from a more rational perspective."

[1354] Input: Sentiment score and offensive keywords

[1355] Output: Warning message and alternative phrase

[1356] Step 7:

[1357] The server then sends the generated alert messages to the terminal, again using HTTP or WebSocket to transmit data in real time.

[1358] Input: warning message and alternative phrase

[1359] Output: Warning message sent to terminal

[1360] Step 8:

[1361] The terminal displays a warning message to the user, for example, by using a pop-up window.

[1362] Input: The warning message sent to the terminal

[1363] Output: A warning message that is displayed to the user.

[1364] Step 9:

[1365] The user checks the displayed warning message and corrects the input content if necessary. For example, the user might say, "You seem to be pushing yourself too hard. What's wrong?" The corrected text data is then sent to the server again.

[1366] Input: The warning message displayed to the user

[1367] Output: Corrected text data

[1368] Step 10:

[1369] The server then passes the corrected text data back to the NLP module for re-analysis. If the corrected input is deemed non-offensive, the server allows the edited content to be posted as is.

[1370] Input: Modified text data

[1371] Output: Text data that is allowed to be posted

[1372] This series of steps allows users to regain a calm perspective even in emotional situations and suppress or reduce slander.

[1373] (Application example 1)

[1374] 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."

[1375] In recent years, customer service in brick-and-mortar stores has become increasingly important, but there is a challenge in properly managing responses based on human emotions and reactions. In particular, if staff become emotional or use offensive language, it can worsen relationships with customers and damage the credibility of the business. There is a need for a multi-functional system that can prevent such situations and ensure calm and courteous customer service at all times.

[1376] 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.

[1377] In this invention, the server includes means for receiving user input, means for analyzing the received user input, means for generating a warning message for the input content, means for displaying the generated warning message to the user, means for converting the voice input into text data, and means for monitoring emotional words and offensive phrases in real time using the text data. This enables store staff to automatically detect offensive words and content containing negative emotions during conversations with customers, receive appropriate warnings, and encourage calm responses.

[1378] "User" refers to the entity that operates the system or makes input.

[1379] "Input" refers to information or data that a user provides to a system.

[1380] "Voice input" refers to information provided to the system by a user speaking.

[1381] "Text data" refers to written information generated by voice input or other means.

[1382] "Analysis" refers to the process of interpreting input information and understanding its content and intent.

[1383] "Emotional words" refer to words that strongly express the user's emotions.

[1384] "Offensive phrases" refer to words or expressions that can have a negative impact on others.

[1385] A "warning message" refers to a notification that notifies the user that there is a problem with the input content.

[1386] "Voice recognition technology" refers to the technology that converts voice input into text data.

[1387] "Natural language processing technology" refers to technology for understanding and analyzing human language.

[1388] "Smart glasses" refers to glasses-type devices that are worn by the user to display information and receive input.

[1389] "Server" refers to a central processing unit that processes information and stores data.

[1390] "Monitoring" refers to the act of continuously observing specific information or behavior to detect changes or problems.

[1391] The present invention relates to a system for supporting customer service in brick-and-mortar stores, and in particular to a system that detects conversations containing offensive language or negative emotions in real time and generates and displays appropriate warning messages, thereby enabling staff to respond calmly. An embodiment of the present invention will be described in detail below.

[1392] System Configuration

[1393] The system consists of the following main components:

[1394] 1. Smart Glasses

[1395] Staff wear smart glasses and use voice input, which is collected through the smart glasses' microphone.

[1396] 2. Voice recognition function

[1397] Voice input is converted to text in real time using cloud-based or built-in speech recognition technology, such as the Google Speech-to-Text API.

[1398] 3. Server

[1399] The server has the following roles:

[1400] Data reception: Receives text data sent from the smart glasses.

[1401] Natural Language Processing (NLP): Analyzes incoming text data to detect emotional words and offensive phrases. NLP technologies available include spacy and TextBlob.

[1402] Warning generation: When emotive words or offensive phrases are detected, appropriate warning messages and alternative phrases are generated.

[1403] Message sending: The generated warning message is sent to the smart glasses and displayed to staff in real time.

[1404] Data processing and calculation

[1405] 1. Convert voice input to text data:

[1406] The voice data collected by the smart glasses is converted into text data in real time using a voice recognition API. For example, it can be converted into the following format:

[1407] Voice input: "You're an idiot."

[1408] Speech recognition API output: Text data "You're an idiot"

[1409] 2. Text data analysis:

[1410] The server receives the text data and analyzes it using natural language processing technology. Specifically, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is detected as an offensive word.

[1411] 3. Generate a warning message:

[1412] If offensive language or negative sentiment is detected, the server uses a warning generation module to generate an appropriate warning message and alternative phrases. For example, the following warning message may be generated:

[1413] Warning message: "Is the conversation getting aggressive? Try to stay calm."

[1414] 4. Displaying messages:

[1415] The generated warning messages are sent to smart glasses and displayed to staff in real time.

[1416] Specific examples

[1417] For example, this system works if a staff member in a physical store says "You're an idiot" while interacting with a customer. The voice input is recognized by the smart glasses and converted into text data such as "You're an idiot" via the Google Speech-to-Text API. This text data is sent to the server and analyzed by the NLP module. Here, "idiot" is detected as an offensive word, and the server generates a warning message saying, "Is the conversation becoming aggressive? Please try to respond calmly." This message is displayed on the smart glasses, and the staff member can immediately correct their response by asking, "Can you tell me what's wrong?"

[1418] Prompt Sentence Examples

[1419] If a customer types "You're an idiot," create a warning message that this system will display and a suggested calm response.

[1420] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1421] Step 1:

[1422] The user wears the smart glasses and inputs voice data. The microphone in the smart glasses collects the voice data.

[1423] Input: Audio data (e.g., "You're an idiot").

[1424] Output: The audio data is kept intact in the smart glasses and is ready to be sent to the server.

[1425] Step 2:

[1426] The smart glasses send the collected voice data to a server, which then passes the received voice data to a voice recognition API.

[1427] Input: Audio data transmitted from smart glasses.

[1428] Output: A request is sent to the speech recognition API.

[1429] Step 3:

[1430] The server uses a speech recognition API (e.g., Google Speech-to-Text API) to convert the voice data into text data in real time.

[1431] Input: The audio data received by the speech recognition API.

[1432] Output: Text data (e.g., "You're an idiot") is generated.

[1433] Step 4:

[1434] The server passes the generated text data to a natural language processing (NLP) module for analysis, specifically detecting offensive words and negative sentiment scores within the text.

[1435] Input: Text data (e.g., "You're an idiot").

[1436] Output: Analysis results (e.g. "idiot" is detected as an offensive word).

[1437] Step 5:

[1438] Based on the analysis results from the NLP module, the server uses the warning generation module to generate appropriate warning messages and alternative phrases.

[1439] Input: Analysis results (offensive language detection results).

[1440] Output: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.") is generated.

[1441] Step 6:

[1442] The generated warning message is again sent from the server to the smart glasses, which then display the warning message to the user in real time.

[1443] Input: A warning message (e.g., "Is the conversation becoming aggressive? Please try to stay calm.").

[1444] Output: A warning message will be displayed on the smart glasses display.

[1445] 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.

[1446] The system according to the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and an emotion engine, and generates and displays a warning message to the user if the input content contains offensive words or phrases. Specific embodiments are described in detail below.

[1447] First, a user inputs a comment using a device such as a chat application or a social networking platform. The device temporarily stores the text data entered by the user and metadata such as input speed and pattern, and then transmits the contents to the server.

[1448] The server receives text data and metadata from the device. This data is temporarily stored in a buffer for analysis. The server then passes the text data to the NLP module and emotion engine.

[1449] The NLP module performs sentiment analysis and keyword extraction on the text data. At this point, it detects offensive words and negative sentiment scores in the text. For example, if the word "idiot" is included, it is determined that this word is offensive.

[1450] In parallel, the emotion engine analyzes the user's input speed and patterns to estimate the user's emotional state. For example, a sudden increase in input speed may suggest that the user is emotional. This estimated emotional state is then integrated with the analysis results of the NLP module to make a final assessment.

[1451] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, if the text entered by the user is "You're an idiot," the warning generation module will generate a warning message saying "Is this sentence offensive?" It can also include a suggestion for an alternative phrase, such as "Let's speak from a more rational perspective."

[1452] The generated warning message is sent again from the server to the terminal. The terminal displays this warning message to the user as a pop-up window. The user can check the displayed warning message and correct the input content if necessary. For example, the user can correct it by saying, "It looks like you're pushing yourself too hard. What's wrong?"

[1453] The corrected text is then sent back to the server, and the same process is repeated. However, if the corrected text is not offensive, the server will not generate any further warnings and will allow the post to be posted. This process allows users to regain a calm perspective even in emotional situations, and to suppress and reduce abusive comments.

[1454] Specifically, consider the following scenario: User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and an emotion engine, and detects that "idiot" is an offensive word and that the user's typing speed is increasing. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard. What's wrong?" The correction is sent to the server, where it is inspected again, and if there are no problems, the post is allowed to proceed.

[1455] In this way, a system that combines an emotion engine can respond carefully even in situations where the user is emotional, thereby reducing slander.

[1456] The processing flow will be explained below.

[1457] Step 1:

[1458] A user operates a device and enters text into an input field in a chat application or social networking platform, for example, typing "You're an idiot."

[1459] Step 2:

[1460] The device temporarily stores the user's input and makes an HTTP POST request to send the entered text data and metadata such as typing speed and pattern to the server.

[1461] Step 3:

[1462] The server receives the HTTP POST request sent from the terminal and stores its contents in a buffer for analysis. The received data consists of the input text "You're an idiot" and metadata such as input speed and pattern.

[1463] Step 4:

[1464] The server calls an internal method to pass the received text data to the NLP module, and the metadata is passed to the emotion engine.

[1465] Step 5:

[1466] The NLP module analyzes the text data, performs sentiment analysis and keyword extraction, detects the word "idiot" as an offensive word, and calculates a sentiment score.

[1467] Step 6:

[1468] The emotion engine analyzes the metadata and analyzes the user's typing speed and patterns. Based on this data, it infers the user's emotional state. For example, if the user's typing speed suddenly increases, it infers that the user is angry.

[1469] Step 7:

[1470] The NLP module and the sentiment engine combine their respective analysis results to generate a comprehensive assessment result, which includes a list of offensive keywords, a sentiment score, and an estimated emotional state.

[1471] Step 8:

[1472] The server's warning generation module receives the integrated evaluation results and generates appropriate warning messages, such as "Is the sentence offensive?" and "How about this wording?" suggestions.

[1473] Step 9:

[1474] The warning generation module calls an interface to send the generated warning message and alternative phrase to the UI module, which then sends the generated warning message and alternative phrase to the terminal as a response.

[1475] Step 10:

[1476] The device's UI module receives the warning message from the server and displays it to the user as a pop-up window, which reads "Is the text offensive?"

[1477] Step 11:

[1478] The user checks the pop-up window and corrects their input if necessary, for example changing their comment to "You seem to be pushing yourself, what's wrong?"

[1479] Step 12:

[1480] The device creates an HTTP POST request to send the corrected text data back to the server.

[1481] Step 13:

[1482] The server receives the corrected text data and runs it through the same NLP module and emotion engine again to ensure it does not contain any offensive content.

[1483] Step 14:

[1484] After the server makes a final check, if there are no problems with the revised text, it will allow the comment to be posted as is. The user's comment will be posted normally.

[1485] Through the above steps, the system can carefully respond to even emotional end users and prevent slander, making communication on the Internet healthier and more constructive.

[1486] Example 2

[1487] 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."

[1488] Current online communication platforms have a problem where users become emotional and send offensive messages, which can worsen dialogue within the community. Furthermore, the mental stress and defamation caused by such offensive messages are also serious problems. The problem with existing methods is that they do not adequately implement mechanisms to detect and warn users about offensive messages in advance.

[1489] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes: means for receiving a user input; means for temporarily storing the received user input and transmitting it to the server together with metadata such as input speed and pattern; means for saving the text data and metadata received by the server in an analysis buffer and passing them to a natural language processing module and a sentiment analysis engine; means for the natural language processing module to perform sentiment analysis and keyword extraction on the text data and detect offensive words and negative sentiment scores; means for the sentiment analysis engine to estimate the user's emotional state from the user's input speed and pattern and integrate the results; means for generating a warning message based on the integrated analysis results; and means for displaying the generated warning message to the user. This makes it possible to issue a warning before a user becomes emotional and sends an offensive message, thereby improving the quality of communication.

[1490] "User" means a person who enters comments or messages on the Online Platform.

[1491] "Input speed" refers to the speed at which a user types characters, measured in characters per second.

[1492] "Metadata" refers to various data related to user input, including input speed, pattern, timestamp, and the like.

[1493] "Server" refers to a computer system that receives and analyzes user input data.

[1494] The "analysis buffer" refers to a memory area that temporarily stores received data.

[1495] "Natural language processing module" refers to a software component that analyzes received text data and performs sentiment scoring and keyword extraction.

[1496] "Sentiment analysis engine" refers to a software component that infers a user's emotional state from their typing speed and patterns.

[1497] "Warning Generation Module" means the software component that generates a warning message based on the analysis results of the Natural Language Processing Module and the Sentiment Analysis Engine.

[1498] A "warning message" refers to a message that includes a warning when a user enters emotional and offensive words.

[1499] "Alternative phrases" refer to suggested milder versions of offensive language.

[1500] MODE FOR CARRYING OUT THE INVENTION

[1501] The system of the present invention monitors user input in real time, analyzes the input content using natural language processing (NLP) technology and a sentiment analysis engine, and generates and displays a warning message to the user if the input contains offensive words or phrases. This system aims to improve the quality of communication by issuing a warning before the user becomes emotional and sends an offensive message.

[1502] Hardware and software used

[1503] Server: A computer system that receives, analyzes, and generates alert messages.

[1504] Terminal: The device on which the user inputs information (e.g., smartphone, PC)

[1505] Natural language processing module: A software component that performs sentiment analysis and keyword extraction on text data (e.g., Hugging Face's Transformers library, Google Cloud Natural Language API).

[1506] Sentiment analysis engine: A software component that infers emotional states from input speed and patterns (e.g., Apache Kafka)

[1507] Alert Generation Module: A software component that generates an alert message (e.g., AlertManager).

[1508] Specific operation of the system

[1509] First, a user enters a comment using a device such as a chat application or a social networking platform. For example, the user enters "You're an idiot" in a text field. The device temporarily stores the text data entered by the user and metadata such as typing speed and pattern, and then sends the contents to the server. The metadata includes keystroke speed (e.g., 5 characters per second).

[1510] The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis. The received data is in the format of "Text: You're an idiot" and "Speed: 5 characters per second." The server then passes this text data and metadata to the natural language processing module and sentiment analysis engine.

[1511] The natural language processing module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative sentiment scores. At this point, the offensive keyword "idiot" is extracted from the phrase "You're an idiot," and the sentiment score is calculated as "aggressive." The sentiment analysis engine estimates the user's emotional state from their typing speed and patterns. If their typing speed increases suddenly (for example, from 5 characters per second to 8 characters per second), it is assumed that the user is emotional. The server combines the results of the NLP module and the sentiment analysis engine to make a final assessment. For example, the results may be "Text: Aggressive," "Sentiment Score: High," and "Speed: Rapid Increase."

[1512] Once the analysis is complete, the results are transferred to the server's warning generation module, which generates an appropriate warning message based on the analysis results. For example, it may generate a warning message such as "Is the text offensive?" with an alternative phrase such as "How about this wording?" The generated warning message is then sent back to the terminal from the server.

[1513] The terminal displays this warning message to the user as a pop-up window. The user can review the warning message and correct the input if necessary. For example, they could correct it to "You seem to be overdoing it, what's wrong?" The corrected text is then sent to the server again, and the same process is repeated. If the corrected input is not offensive, the server will not generate any further warnings and will allow the posting.

[1514] Examples and prompts

[1515] Specifically, User A types "You're an idiot" into a chat app and presses the send button. At this point, the device sends the input content and input patterns to the server. The server analyzes the received text and metadata using an NLP module and a sentiment analysis engine, detecting that the word "idiot" is offensive and that the user's typing speed has suddenly increased. The warning generation module sends User A a warning message saying "Is this sentence becoming offensive?" along with an alternative phrase such as "How about this wording?" User A checks the warning and corrects the input, saying "It looks like you're trying too hard, what's wrong?" The correction is sent to the server and inspected again. If there are no problems, the post is allowed to proceed.

[1516] In this way, by detecting offensive messages in real time and displaying a warning to the user, the quality of communication can be improved and abusive behavior can be reduced.

[1517] Examples of prompts include:

[1518] "Judge whether this message is offensive and issue a warning if necessary: ​​'You're an idiot.'"

[1519] "The user's typing speed has suddenly increased. Please rate whether this typing is emotional."

[1520] Using these prompts, the system can properly detect offensive messages and generate warnings.

[1521] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1522] Program processing steps

[1523] Step 1: Receive user input

[1524] 1. A user types a comment into a chat application or social media platform.

[1525] Example: A user types "You're an idiot" into a text field.

[1526] 2. Input: User's text data (e.g., "You're an idiot")

[1527] 3. Output: Input text data and metadata temporarily saved on the device

[1528] Step 2: The device sends the input

[1529] 1. The device temporarily stores the text data entered by the user and metadata (such as input speed and pattern).

[1530] Example: Temporarily save the text data "You're an idiot" with an input speed of 5 characters per second.

[1531] 2. The device sends the input and metadata to the server.

[1532] An interface API may be used for transmission.

[1533] 3. Input: User text data and metadata

[1534] 4. Output: The input text data and metadata sent to the server.

[1535] Step 3: The server receives the data and stores it in a buffer for analysis.

[1536] 1. The server receives the text data and metadata sent from the device and temporarily stores them in a buffer for analysis.

[1537] Example: Received data "Text: You're an idiot" and "Speed: 5 characters / second" is saved in a buffer.

[1538] 2. Input: Text data and metadata sent from the device

[1539] 3. Output: Data saved in the analysis buffer

[1540] Step 4: Text analysis using the natural language processing module

[1541] 1. The server passes the text data and metadata to a natural language processing (NLP) module.

[1542] Example: Send the text data "You're an idiot" to the NLP module.

[1543] 2. The NLP module analyzes the received text data, calculates sentiment scores, and extracts keywords.

[1544] Example: Extract the keyword "idiot" from the text "You're an idiot" and calculate the emotion score of "aggressive."

[1545] 3. Input: Text data

[1546] 4. Output: Keywords and sentiment scores

[1547] Step 5: Metadata analysis with sentiment analysis engine

[1548] 1. The server passes the metadata to the sentiment analysis engine.

[1549] Example: Sending metadata to a sentiment analysis engine with an input rate of 5 characters per second.

[1550] 2. The sentiment analysis engine infers the user's emotional state from their typing speed and patterns.

[1551] Example: Detecting a sudden increase in typing speed and inferring that the user is becoming "emotional."

[1552] 3. Input: Metadata

[1553] 4. Output: Estimation of emotional state

[1554] Step 6: Synthesis of the analysis results

[1555] 1. The server integrates the results of the NLP module and the sentiment analysis engine and performs the final evaluation.

[1556] Combined results: "Text: Aggressive" "Sentiment score: High" "Speed: Rapid"

[1557] 2. Input: Keywords and sentiment scores from the NLP module, emotional states from the sentiment analysis engine

[1558] 3. Output: Integrated analysis results

[1559] Step 7: Generate a warning message

[1560] 1. The server's warning generation module generates appropriate warning messages based on the integrated analysis results.

[1561] For example, generate a warning message such as "Is this sentence offensive?" with an alternative phrase such as "How about this phrase?"

[1562] 2. Input: Integrated analysis results

[1563] 3. Output: Generated warning messages

[1564] Step 8: Sending a warning message

[1565] 1. The server sends the generated warning message back to the terminal.

[1566] For example: Sending a message saying, "Is the text becoming aggressive?"

[1567] 2. Input: Generated warning message

[1568] 3. Output: Warning message sent to terminal

[1569] Step 9: Your device will display a warning message

[1570] 1. The terminal displays the received warning message to the user as a pop-up window.

[1571] For example, a warning message may appear on the user's screen asking, "Is the text offensive?"

[1572] 2. Input: Received warning message

[1573] 3. Output: The displayed warning message

[1574] Step 10: User corrects input

[1575] 1. The user checks the displayed warning message and corrects the input as necessary.

[1576] For example, correct them with, "You seem like you're overdoing it, what's wrong?"

[1577] 2. Input: Correct the input of the user who saw the warning message.

[1578] 3. Output: Corrected text data

[1579] Step 11: Resubmit your corrections

[1580] 1. The terminal sends the corrected text back to the server.

[1581] Example: Send the modified text "You seem to be overdoing it, what's wrong?" to the server.

[1582] 2. Input: Corrected text data

[1583] 3. Output: The corrected text sent to the server

[1584] Step 12: Reanalysis and final check

[1585] 1. The server analyzes the corrected text again using the NLP module and sentiment analysis engine, and if there are no problems, allows it to be posted.

[1586] Example: The corrected text "You seem to be overdoing it, what's wrong?" is determined to be non-aggressive.

[1587] 2. Input: Corrected text data

[1588] 3. Output: Based on the analysis results, if there are no problems, the submission is permitted.

[1589] In this way, the system can monitor user input in real time and immediately issue a warning if it contains offensive content, allowing users to regain their composure and reduce abusive behavior.

[1590] (Application example 2)

[1591] 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."

[1592] In electronic payment services, there is a need to reduce offensive language and emotional exchanges that can occur during communication between users and support staff. However, conventional systems lack the means to monitor such offensive language in real time and generate and present appropriate warnings or alternative phrases. This poses a challenge in maintaining the dignity of communication and reducing the mental burden on users.

[1593] The specific processing by the specific 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 receiving user input, means for analyzing the received user input, means for generating a warning message in response to the input content, means for displaying the generated warning message to the user, and means for analyzing the user's input pattern and estimating the user's emotional state. This makes it possible to monitor and correct offensive words in real time in communications for electronic payment services and suggest alternative phrases, thereby preventing slander and reducing the mental burden on users.

[1594] "User input" refers to text data sent by a user to support staff or the system in an electronic payment service.

[1595] "Means for receiving" refers to an interface that allows the system to recognize text data entered by the user and then send it to the server.

[1596] "Means for analyzing" refers to the process of analyzing received text data using natural language processing techniques and emotion engines to assess offensive language and the user's emotional state.

[1597] The "means for generating a warning message" refers to a module for creating a message to warn or caution the user based on the results of the analysis.

[1598] "Means for displaying to the user" refers to an interface for displaying the generated warning message as a pop-up or notification on the user's device.

[1599] "User input patterns" refers to metadata such as the speed and timing at which a user types text, and are used to infer emotional states.

[1600] "Means for inferring emotional state" refers to an engine or algorithm that analyzes a user's input pattern and speed to determine whether the user is emotional.

[1601] "Means for suggesting alternative phrases" refers to a module that suggests replacing offensive language with more appropriate and calmer expressions when a user uses it.

[1602] "Natural language processing technology" refers to technology for analyzing text data and understanding and extracting its meaning and emotions.

[1603] "Input speed and pattern metadata" refers to additional information, such as the speed and frequency at which a user actually types text, that can be used to infer a user's emotional state.

[1604] The present invention is a system that monitors communication between users and support staff in electronic payment services in real time, detects offensive words and phrases, and displays a warning message. The system analyzes the user's input data and its metadata (input speed and patterns) to prevent excessive remarks. Specific embodiments for implementing the present invention are described below.

[1605] Hardware / Software used

[1606] Hardware:

[1607] Smartphone

[1608] tablet

[1609] personal computer

[1610] server

[1611] software:

[1612] Natural Language Processing (NLP) modules (e.g., spaCy, NLTK)

[1613] Emotion engine (proprietary or public library)

[1614] Interfaces (e.g. React, Vue.js)

[1615] Means of implementation

[1616] server

[1617] The server receives the text data sent by the user and its metadata (input speed, pattern) and temporarily stores it in a buffer for analysis.

[1618] The server then passes the received data to the NLP module and emotion engine. The NLP module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative emotion scores. The emotion engine also analyzes the user's typing speed and patterns to estimate the user's emotional state.

[1619] The warning generation module combines these analysis results and generates an appropriate warning message. For example, if the user's input is an offensive one such as "You're an idiot," the system generates a warning message saying "Is the sentence offensive?" along with an alternative phrase such as "Let's speak from a more rational perspective."

[1620] Terminal

[1621] The device, such as a smartphone or tablet, collects text data and its metadata entered by the user in real time and sends it to a server, which then displays a warning message and alternative phrases received from the server as a pop-up window.

[1622] User

[1623] The user checks the warning message and corrects the input content as necessary. For example, the user sends the corrected text "You seem to be pushing yourself too hard. What's wrong?" to the server again.

[1624] Specific examples

[1625] If a user types "This system is completely unusable, it's stupid!" into the support section of an electronic payment service, the system analyzes the input data in real time. The server detects the offensive word "stupid" and infers from the metadata that the user is emotional. The warning generation module displays messages on the terminal such as "Is this sentence offensive?" and "Let's speak from a more rational perspective." The user confirms this and corrects their input by saying, "There seems to be a slight problem with this system. How can I fix it?"

[1626] Prompt Sentence Examples

[1627] "Evaluate the offensiveness of messages entered by users in real time and estimate the user's emotional state using an emotion engine. If offensive language is included, generate and display an appropriate warning message."

[1628] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1629] Step 1:

[1630] The device receives the user's text input in real time and collects the input data and metadata such as input speed and timing. The device temporarily stores this information and sends it to the server. Input: User's text data and metadata. Output: Transfer of received text data and metadata to the server.

[1631] Step 2:

[1632] The server receives the text data and metadata sent from the terminal and temporarily stores them in a buffer for analysis. Input: Text data and metadata from the terminal. Output: Data stored in the buffer for analysis.

[1633] Step 3:

[1634] The server analyzes the text data stored in the buffer using a natural language processing (NLP) module to extract sentiment and keywords from the text. The NLP module performs calculations to detect offensive words and negative sentiment scores. Input: Text data in the buffer. Output: Sentiment analysis results and keyword extraction results.

[1635] Step 4:

[1636] In parallel, the server passes the metadata to the emotion engine, which analyzes the user's input pattern and speed to estimate their emotional state. If the input speed suddenly increases, the emotion engine estimates that the user is emotional. Input: Metadata in the buffer. Output: Estimated emotional state.

[1637] Step 5:

[1638] The server integrates the analysis results of the NLP module with the emotional state of the emotion engine and passes them to the warning generation module. The warning generation module generates an appropriate warning message and alternative phrases if the message contains offensive language or has a high negative emotion score. Input: Sentiment analysis results, keyword extraction results, estimated emotional state. Output: Generated warning message and alternative phrases.

[1639] Step 6:

[1640] The server sends the generated warning message and alternative phrase to the terminal. The terminal displays these messages to the user as a popup window. Input: Generated warning message and alternative phrase. Output: Sent warning message to the user terminal.

[1641] Step 7:

[1642] The user checks the displayed warning message and corrects the input as necessary. For example, the user might correct the input by saying, "You seem to be pushing yourself too hard. What's wrong?" Input: The user's reaction after receiving the warning message. Output: The corrected text data.

[1643] Step 8:

[1644] The terminal sends the text data corrected by the user to the server again, and the server repeats the same process to re-analyze it. If there are no offensive words, the server allows the post. Input: Corrected text data. Output: Final decision on whether the post is allowed or not.

[1645] 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.

[1646] 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.

[1647] 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.

[1648] 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.

[1649] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1650] 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.

[1651] 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).

[1652] 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.

[1653] 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."

[1654] 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.

[1655] 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).

[1656] 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.

[1657] 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.

[1658] 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.

[1659] 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.

[1660] 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.

[1661] 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.

[1662] 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.

[1663] 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.

[1664] 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.

[1665] 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.

[1666] The following is further disclosed regarding the above embodiment.

[1667] (Claim 1)

[1668] means for receiving user input;

[1669] means for analyzing received user input;

[1670] means for generating a warning message in response to the input;

[1671] means for displaying the generated warning message to a user;

[1672] A system including:

[1673] (Claim 2)

[1674] a means of suggesting alternative phrases when a warning message contains potentially offensive language;

[1675] The system of claim 1 further comprising:

[1676] (Claim 3)

[1677] an analysis means for analyzing the user's input using natural language processing technology;

[1678] 10. The system of claim 1, comprising:

[1679] "Example 1"

[1680] (Claim 1)

[1681] a means for obtaining user input;

[1682] means for transmitting the acquired user input data to a server;

[1683] A means for the server to receive and temporarily store input data sent by the user;

[1684] means for analyzing user input data using natural language processing techniques;

[1685] means for generating a warning message based on input data;

[1686] means for transmitting the generated warning message to a user;

[1687] a means for re-analyzing the input data modified by the user to determine whether it is offensive;

[1688] A system including:

[1689] (Claim 2)

[1690] Further includes a means to suggest alternative phrases when warning messages contain potentially offensive language.

[1691] 10. The system of claim 1.

[1692] (Claim 3)

[1693] A means for displaying the generated warning message to the user on the terminal, and for the user to confirm the displayed warning message and correct the input contents, is included.

[1694] 10. The system of claim 1.

[1695] "Application Example 1"

[1696] (Claim 1)

[1697] means for receiving user input;

[1698] means for analyzing received user input;

[1699] means for generating a warning message in response to the input;

[1700] means for displaying the generated warning message to a user;

[1701] means for converting voice input into text data;

[1702] A means of monitoring text data in real time for emotional language and offensive phrases;

[1703] A system including:

[1704] (Claim 2)

[1705] a means of suggesting alternative phrases when a warning message contains potentially offensive language;

[1706] A means for converting a voice input by a user into text data using a voice recognition technology;

[1707] The system of claim 1 further comprising:

[1708] (Claim 3)

[1709] an analysis means for analyzing the user's input using natural language processing technology;

[1710] a means for displaying a warning message to a user using the smart glasses based on the analysis result;

[1711] 10. The system of claim 1, comprising:

[1712] "Example 2: Combining Emotion Engines"

[1713] (Claim 1)

[1714] means for receiving user input;

[1715] a means for temporarily storing received user input and transmitting it to a server along with metadata such as input speed and pattern;

[1716] A means for storing the text data and metadata received by the server in a buffer for analysis and passing them to a natural language processing module and a sentiment analysis engine;

[1717] A natural language processing module performs sentiment analysis and keyword extraction on the text data to detect offensive words and negative sentiment scores;

[1718] A means for the sentiment analysis engine to infer the user's emotional state from their input speed and patterns and integrate the results;

[1719] means for generating a warning message based on the combined analysis results;

[1720] means for displaying the generated warning message to a user;

[1721] A system including:

[1722] (Claim 2)

[1723] 10. The system of claim 1, further comprising means for suggesting an alternative phrase if the warning message contains potentially offensive language.

[1724] (Claim 3)

[1725] 10. The system of claim 1, wherein the analyzing means uses natural language processing techniques to analyze the user's input, and the sentiment analysis engine includes means for inferring the user's emotional state.

[1726] "Application example 2 when combining emotion engines"

[1727] (Claim 1)

[1728] means for receiving user input;

[1729] means for analyzing received user input;

[1730] means for generating a warning message in response to the input;

[1731] means for displaying the generated warning message to a user;

[1732] means for estimating an emotional state of a user by analyzing the user's input pattern;

[1733] A system including:

[1734] (Claim 2)

[1735] a means of suggesting alternative phrases when a warning message contains potentially offensive language;

[1736] means for suggesting alternative phrases related to electronic payment services actually used by the user;

[1737] The system of claim 1 further comprising:

[1738] (Claim 3)

[1739] an analysis means for analyzing the user's input using natural language processing technology;

[1740] a means for using typing speed and pattern metadata to infer emotional state;

[1741] 10. The system of claim 1, comprising: [Explanation of symbols]

[1742] 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. means for receiving user input; means for analyzing received user input; means for generating a warning message in response to the input; means for displaying the generated warning message to a user; A system including:

2. a means of suggesting alternative phrases when a warning message contains potentially offensive language; The system of claim 1 further comprising:

3. an analysis means for analyzing the user's input using natural language processing technology; The system of claim 1 , comprising:

Images