System

Abstract

A system is provided.SOLUTION: A system comprising: means for obtaining text data entered by a user; means for analyzing the text data to detect inappropriate expressions; means for generating correction suggestions for the detected inappropriate expressions; means for displaying the correction suggestions to the user; and means for improving the accuracy of the correction suggestions based on user feedback.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] In business, email wording often causes problems. Examples include rude language toward business partners, language that leads to power or sexual harassment toward colleagues or subordinates, and emails that lead to misunderstandings toward superiors. Furthermore, the time it takes to compose emails reduces work efficiency. Furthermore, in the international business environment, the need for multilingual support is increasing. There is a need for a system that can solve these problems and improve the appropriateness and efficiency of business emails. [Means for solving the problem]

[0005] According to the present invention, a system is provided that includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for generating correction suggestions for the detected inappropriate expressions, means for displaying the correction suggestions to the user, and means for improving the accuracy of the correction suggestions based on user feedback. This system enables users to prevent inappropriate expressions when sending business emails and to create efficient and appropriate text. Furthermore, the correction suggestions are compatible with different languages, and settings can be customized based on the user's industry and company characteristics.

[0006] "User" refers to the entity that uses the system to create email text and receives suggested revisions.

[0007] "Text data" refers to the text information entered by the user, and refers to the email text to be analyzed.

[0008] "Analysis" refers to the process of automatically evaluating linguistic expressions in text data and identifying inappropriate expressions and areas for improvement.

[0009] "Inappropriate language" refers to language or phrases in business emails that are considered rude, misleading, or culturally inappropriate.

[0010] The "correction suggestion" refers to an alternative suggestion for presenting a more appropriate expression to the user in response to a detected inappropriate expression.

[0011] "Display" refers to the act of visually showing correction suggestions and feedback to the user on the device.

[0012] "Feedback" refers to the information the system receives based on the user's choice to accept or reject suggested revisions.

[0013] "Accuracy" refers to the degree to which the system's suggested revisions are appropriate to the user's needs and context.

[0014] "Multilingual support" refers to the system's ability to function appropriately and suggest corrections to email text in different languages.

[0015] "Customization" refers to the process of adjusting and optimizing system settings according to the characteristics of the user's industry or company. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0017] 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.

[0018] First, the terms used in the following description will be explained.

[0019] 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).

[0020] 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.

[0021] 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.

[0022] 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.

[0023] 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."

[0024] [First embodiment]

[0025] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0026] 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.

[0027] 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).

[0028] 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.

[0029] 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.

[0030] 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.

[0031] 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.

[0032] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] 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."

[0037] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. Below, the program processing and specific examples of this system will be described.

[0038] Program Processing Overview

[0039] The system mainly consists of three entities: the terminal, the server, and the user. The roles and operations of each entity are as follows:

[0040] 1. User creates email message:

[0041] The user types the text of a business email into the terminal. For example, the user types the following text to a business partner:

[0042] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0043] 2. Sending email content:

[0044] The device sends the entered text data to the server, which includes the entire body of the email.

[0045] 3. Text analysis and profanity detection:

[0046] The server analyzes the received text data using a natural language processing (NLP) engine to detect inappropriate expressions, such as "Thank you for your hard work, John-san."

[0047] 4. Generate correction suggestions:

[0048] The server generates appropriate expressions for the inappropriate expressions it detects. In this case, the suggested correction for "Thank you for your hard work, John-san" would be "John-san."

[0049] 5. View suggested revisions:

[0050] The terminal displays the correction suggestions sent back from the server to the user, who then checks the suggestions and chooses whether to accept them.

[0051] 6. Final User Review:

[0052] The user decides whether to accept the suggested revisions, and if so, the terminal corrects the text to reflect the suggested expressions. The user then confirms the final text and sends the email.

[0053] 7. Gather feedback and learn:

[0054] The server collects user feedback and reflects it in the generative AI algorithm to improve the accuracy of future suggestions, allowing the system to continuously learn and improve.

[0055] Specific examples

[0056] Specific examples are shown below.

[0057] Email text entered by the user:

[0058] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0059] Server analysis results:

[0060] Detected issues:

[0061] "Thank you for your hard work, John" (too friendly)

[0062] "I want to discuss this" (too direct)

[0063] Suggested fix:

[0064] Suggested fix:

[0065] "Thank you for your hard work, John-san" → "John-san"

[0066] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0067] Final confirmation from the user:

[0068] Dear John,

[0069] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[0070] thank you.

[0071] Yamada

[0072] This system allows users to create business emails appropriately and efficiently. Its multilingual capabilities also make it suitable for international business environments, making it suitable for a wide range of uses. Furthermore, the system can be customized to suit the language and cultural characteristics specific to the user's industry or company, allowing it to flexibly meet user needs.

[0073] The processing flow will be explained below.

[0074] Step 1:

[0075] The user composes a business email and types it into the terminal. For example, the user types the following:

[0076] "Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thanks in advance."

[0077] Step 2:

[0078] The device retrieves the text data entered by the user, which includes the entire body of the email.

[0079] Step 3:

[0080] The device sends the acquired text data to the server, which then prepares to begin analyzing the email text.

[0081] Step 4:

[0082] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions.

[0083] Step 5:

[0084] Based on the analysis results, the server lists inappropriate or misleading expressions, such as "Thank you for your hard work, John" or "I'd like to discuss this."

[0085] Step 6:

[0086] Based on the detection results, the server uses a generation AI to generate appropriate correction suggestions. The suggested correction for "Thank you for your hard work, John" is "Dear John," and the suggested correction for "I'd like to discuss this" is "I'd like to take the time to talk to you."

[0087] Step 7:

[0088] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions to the user based on the information received from the server.

[0089] Step 8:

[0090] The user checks the displayed correction suggestions and selects whether to accept them. If the user accepts the corrections, the terminal corrects the email text as suggested.

[0091] Step 9:

[0092] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[0093] Step 10:

[0094] After the final text is confirmed based on the user's input, the device sends the finalized data to the server, which stores the user's feedback and updates the AI's algorithm to improve the accuracy of future suggestions.

[0095] Example 1

[0096] 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."

[0097] When writing business emails, it is difficult for users to use appropriate expressions and avoid inappropriate ones. Furthermore, using expressions appropriate for different cultures and industries requires a great deal of effort and specialized knowledge. Furthermore, systems are required to continuously learn and improve based on user feedback. However, current systems have difficulty effectively resolving all of these issues, and the present invention aims to solve these problems.

[0098] 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.

[0099] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions, thereby enabling the user to efficiently create business email text using appropriate expressions.

[0100] "User" refers to an individual or organization that uses the System to create business emails.

[0101] "Text data" refers to a data format that includes text or sentence information entered by a user.

[0102] "Terminal" refers to the computing device used by a user to enter email text and review suggested revisions.

[0103] "Server" refers to a central computer system that analyzes text data, detects profanity, and generates correction suggestions.

[0104] "Analysis" refers to the process of examining the content of text data using machine learning and natural language processing techniques to identify inappropriate expressions.

[0105] "Inappropriate language" refers to expressions or phrases that are inappropriate to use in a business context.

[0106] "Correction suggestions" refer to alternative suggestions for changing detected inappropriate expressions into appropriate expressions.

[0107] "Feedback" refers to information about whether the user accepted the suggested revisions and whether the suggested revisions were appropriate.

[0108] "Generative AI models" refer to algorithms or machine learning models that generate text data and create correction suggestions.

[0109] A "prompt" refers to a command or question input to a generative AI model.

[0110] "Different languages" refers to languages ​​other than the user's native language.

[0111] "Customization" refers to the process of changing and adjusting system settings to suit the user's industry and corporate characteristics.

[0112] "Sending" refers to the electronic communication act used by a user to deliver the completed email text to the recipient.

[0113] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. This system consists of three entities: a terminal, a server, and a user. The function of each entity is explained below.

[0114] Device Features

[0115] The terminal is a device that allows users to input business emails. Terminals include PCs, tablets, smartphones, etc. Once a user has finished inputting the email, the terminal sends the text data to the server. It also displays any suggested revisions returned by the server.

[0116] Server Features

[0117] The server analyzes text data, detects inappropriate expressions, and generates correction suggestions. Specifically, it uses a natural language processing (NLP) library such as "Hugging Face" and a generative AI model (e.g., "OpenAI GPT"). The server analyzes the text data sent by the user using an NLP engine and detects inappropriate expressions. For detected inappropriate expressions, it uses the generative AI model to generate appropriate correction suggestions and sends them back to the device.

[0118] User operations

[0119] The user types the text of a business email into the terminal. For example, the user types the following text:

[0120] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0121] Once the input is complete, the user presses the send button to send the text data to the server. After that, the user checks the correction suggestions displayed on the terminal and selects whether to accept them. If the correction suggestions are accepted, the user makes a final confirmation and sends the email.

[0122] Specific examples

[0123] Specific examples are shown below.

[0124] Email text entered by the user:

[0125] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0126] Server analysis results:

[0127] Detected issues:

[0128] "Thank you for your hard work, John" (too friendly)

[0129] "I want to discuss this" (too direct)

[0130] Suggested fix:

[0131] Suggested fix:

[0132] "Thank you for your hard work, John-san" → "John-san"

[0133] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0134] Final confirmation from the user:

[0135] Dear John,

[0136] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[0137] thank you.

[0138] Yamada

[0139] Multilingual and customizable

[0140] The system is multilingual and can provide correction suggestions in different languages. It can also be customized to fit the language and cultural characteristics specific to the user's industry or company. This allows for more accurate correction suggestions.

[0141] Gathering feedback and learning

[0142] The server collects user feedback and continuously trains the generative AI model based on it, thereby improving the accuracy of future suggestions.

[0143] As described above, the system of the present invention provides powerful support for users to create business emails quickly and appropriately.

[0144] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0145] Step 1:

[0146] User-generated email content

[0147] A user uses a terminal to type the text of a business email. For example, the user types the following sentence:

[0148] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0149] Once this input is complete, the user presses the send button to prepare to send the text data from the terminal to the server.

[0150] Input: Email text entered by the user

[0151] Output: Text data of the entered email

[0152] Step 2:

[0153] Sending email content

[0154] The device sends the text data entered by the user to the server. At this time, the device sends the email body data to the server using an HTTP request. The request includes the user ID and the email body.

[0155] Input: Text data entered into the terminal

[0156] Output: Text data sent to the server

[0157] Step 3:

[0158] Text analysis and profanity detection

[0159] The server receives the text data and analyzes it using a natural language processing (NLP) engine. This analysis process uses NLP libraries such as "Hugging Face." The server analyzes the content of the email and detects inappropriate expressions. For example, it detects that the expression "Thank you for your hard work, John" is too familiar.

[0160] Input: Text data sent to the server

[0161] Output: Analysis results and a list of inappropriate expressions detected

[0162] Step 4:

[0163] Generate correction suggestions

[0164] The server uses a generative AI model (e.g., "OpenAI GPT") to generate appropriate correction suggestions for the detected inappropriate expressions. Specifically, for the expression "Thank you for your hard work, John-san," the server makes a correction suggestion such as "John-san."

[0165] Input: Profanity detection list

[0166] Output: List of suggested fixes

[0167] Step 5:

[0168] View suggested fixes

[0169] The terminal displays the correction suggestions sent back from the server to the user. The correction suggestions are displayed in a visually intuitive manner, and the user can confirm them and choose whether to accept them.

[0170] Input: A list of suggested revisions sent from the server

[0171] Output: A list of suggested revisions that is displayed to the user

[0172] Step 6:

[0173] Final confirmation by the user

[0174] The user checks the displayed correction suggestions and decides whether to accept them. If they accept, the device corrects the text to the suggested expression. The user then checks the final text and presses the send button to send the email. For example, the corrected text might look like this:

[0175] Dear John,

[0176] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[0177] thank you.

[0178] Yamada

[0179] Input: A list of suggested revisions shown to the user

[0180] Output: The final email message that the user has confirmed and corrected

[0181] Step 7:

[0182] Gathering feedback and learning

[0183] The server collects feedback from users and sends back information such as how the user accepted the correction suggestions and whether the suggestions were appropriate. The server uses this feedback to update the generative AI model and improve the accuracy of future correction suggestions.

[0184] Input: User feedback information

[0185] Output: Update the generative AI model based on feedback

[0186] (Application example 1)

[0187] 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."

[0188] In today's business emails and chats, the text data entered by users often contains inappropriate expressions. Such expressions can lead to misunderstandings and misinterpretations in business communications, negatively impacting a company's credibility and efficiency. Furthermore, in an international business environment, the need for multilingual support makes correcting inappropriate expressions even more complicated. Furthermore, when real-time responses are required, the ability to quickly and accurately correct expressions is essential.

[0189] 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.

[0190] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for analyzing the text data in real time in different response formats and correcting the text data to appropriate expressions, means for generating correction suggestions for the detected inappropriate expressions, means for generating the correction suggestions using a generative AI model, means for displaying the correction suggestions to the user, and means for improving the accuracy of the correction suggestions based on user feedback. This reduces misunderstandings and misinterpretations in business communication, improves corporate reliability and efficiency, and enables multilingual support.

[0191] "User" refers to any individual or corporation that uses the system.

[0192] "Text data" refers to information in the form of a string of characters entered by the user.

[0193] "Analysis" refers to the process of processing and analyzing the content of text data.

[0194] "Inappropriate expressions" refer to expressions that may lead to misunderstandings or misinterpretations in business communications.

[0195] "Correction proposal" refers to an alternative proposal to convert inappropriate language into appropriate language.

[0196] "Display" refers to visually showing the suggested revisions on the user's terminal.

[0197] "Feedback" refers to the opinions and reactions of users to suggested revisions.

[0198] "Accuracy improvement" refers to improvement activities carried out to increase the accuracy and effectiveness of proposals.

[0199] "Different response formats" refers to response methods that are applied in different formats and situations.

[0200] "Real time" refers to the concept of time when data is processed as soon as it is entered.

[0201] "Generative AI model" refers to a model that uses artificial intelligence to generate and modify text data.

[0202] The system of the present invention analyzes text data entered by users, detects inappropriate expressions, and provides appropriate correction suggestions. This system consists of three entities: a terminal, a server, and a user.

[0203] System Configuration

[0204] Hardware and Software Configuration

[0205] Device: The device used by the user, such as a smartphone, tablet, or desktop PC.

[0206] Server: Provides high-performance computing power to analyze text data and generate correction suggestions.

[0207] software:

[0208] OpenAI API: Generates correction suggestions using the generative AI model GPT-3.

[0209] Python: A programming language for implementing the overall system logic.

[0210] Regular expressions (re): Used to process and analyze text data.

[0211] Data processing and calculation

[0212] 1. Acquiring text data:

[0213] A user inputs text data using a terminal, and the text data is transmitted from the terminal to a server.

[0214] 2. Text data analysis:

[0215] The server analyzes the received text data using the OpenAI API. First, it detects inappropriate expressions in the text data.

[0216] 3. Generate correction suggestions:

[0217] For detected inappropriate expressions, a generative AI model is used to generate appropriate correction suggestions.

[0218] 4. View suggested revisions:

[0219] The revision suggestions generated by the server are sent to the terminal and displayed to the user.

[0220] 5. Gathering Feedback:

[0221] The accuracy of suggestions is improved by sending user feedback information to the server and reflecting it in the generative AI model.

[0222] Specific examples

[0223] Example 1: Text analysis of business emails

[0224] A user types the following email on their smartphone:

[0225] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0226] Server analysis results:

[0227] Detected issues:

[0228] "Thank you for your hard work, person in charge" (too friendly)

[0229] "I'd like to discuss this" (too direct)

[0230] Suggested fix:

[0231] Suggested fix:

[0232] "Thank you for your hard work, person in charge" → "Dear person in charge"

[0233] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0234] Final confirmation:

[0235] To the person in charge

[0236] I would like to take the time to speak with you about the meeting we had the other day. I would appreciate it if you could let me know a convenient time and date for us to meet.

[0237] thank you.

[0238] Yamada

[0239] Prompt Sentence Examples

[0240] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[0241] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0242] Fixes:

[0243] This system allows users to quickly correct inappropriate language used in business emails and chats, ensuring appropriate communication. It also supports multiple languages ​​and can be customized based on industry characteristics, making it applicable to a wide range of uses. This is expected to reduce misunderstandings and misinterpretations in business, improving corporate reliability and efficiency.

[0244] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0245] Step 1:

[0246] Users use their devices to input the text of business emails or chat messages. Specifically, users enter text data into an input field on a smartphone, tablet, or desktop PC, and then press the send button. The input data is, for example, in the following format:

[0247] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0248] Input: Text data entered by the user into the terminal.

[0249] Output: Text data ready to be sent to the terminal

[0250] Step 2:

[0251] The terminal sends the entered text data to the server, where the data is transferred securely using protocols such as HTTPS. The entered text is sent in a format that the server can parse.

[0252] Input: Text data sent from the device to the server

[0253] Output: Text data received by the server

[0254] Step 3:

[0255] The server sends the received text data to the OpenAI API for analysis. The server generates a prompt to be used for analysis and passes the text data to the natural language processing engine. At this time, the following prompt is generated:

[0256] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[0257] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0258] Fixes:

[0259] Input: Text data received by the server and prompts

[0260] Output: Analysis results by OpenAI API

[0261] Step 4:

[0262] The OpenAI API uses a generative AI model to detect inappropriate expressions and generate appropriate correction suggestions, such as "Thank you for your hard work, Mr. / Ms. Person in Charge."

[0263] Input: Prompt and text data received by the OpenAI API

[0264] Output: Detected profanities and suggested corrections

[0265] Step 5:

[0266] The server receives the correction suggestions returned by the OpenAI API and sends them to the device, where they can be viewed by the user. The correction suggestions are displayed in the following format:

[0267] Suggested fix:

[0268] "Thank you for your hard work, person in charge" → "Dear person in charge"

[0269] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0270] Input: Analysis results returned from the OpenAI API

[0271] Output: Suggested fixes sent to terminal

[0272] Step 6:

[0273] The terminal displays the received correction suggestions to the user, who then checks the displayed correction suggestions and selects whether to accept them. The selection operation is performed via buttons or an interface on the terminal.

[0274] Input: Suggested corrections sent to the terminal

[0275] Output: Correction suggestions and user selections displayed on the device

[0276] Step 7:

[0277] The user decides whether to accept the suggested revisions and confirms the result on the terminal. If the revision is accepted, the terminal corrects the text to the suggested expression and displays the final text.

[0278] Input: User accepts or rejects suggested revisions

[0279] Output: Final corrected text

[0280] Step 8:

[0281] The server collects user feedback and reflects it in the generative AI model algorithm, which improves the accuracy of correction suggestions from the next time onwards.

[0282] Input: User feedback information

[0283] Output: Improved generative AI model algorithm

[0284] This allows users to correct inappropriate expressions in business emails and chat messages to appropriate expressions, enabling efficient and effective communication.

[0285] 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.

[0286] The present invention relates to a system that analyzes text data, detects inappropriate expressions, and provides correction suggestions when a user composes a business email, in addition to combining it with an emotion engine that recognizes the user's emotions and reflects them in the correction suggestions. Below, we will explain the program processing and specific operation examples of this system.

[0287] Program Processing Overview

[0288] The system mainly consists of four entities: the terminal, the server, the user, and the emotion engine. The individual roles and operations of each entity are as follows:

[0289] User-generated email message:

[0290] The user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[0291] example:

[0292] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0293] Send email content:

[0294] The device acquires text data entered by the user. At this time, the device uses an emotion engine to recognize the emotion contained in the text data entered by the user. The acquired text data and emotion information are sent to the server.

[0295] Text analysis and profanity detection:

[0296] The server inputs the received text data into an NLP engine, which analyzes the email text and detects inappropriate expressions. For example, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[0297] Emotion recognition and suggested correction adjustments:

[0298] The server generates correction suggestions based on the emotional information recognized by the emotion engine. In this case, the strong emotional expression "I would like to check this urgently" is adjusted to a softer expression such as "I apologize for bothering you during your busy schedule, but could you please check it?"

[0299] View suggested fixes:

[0300] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions received from the server to the user, who then checks the displayed suggestions and chooses whether to accept them.

[0301] Final user confirmation:

[0302] The user decides whether to accept the suggested revisions, and if so, the terminal will revise the email text as suggested. The user then checks the final text and, if there are no problems, confirms sending the email.

[0303] Gathering feedback and learning:

[0304] The server collects user feedback and reflects it in the generative AI and emotion engine algorithms to improve the accuracy of future suggestions and emotion recognition, allowing the system to continuously learn and improve.

[0305] Specific example (utilizing emotion engine)

[0306] A specific example is given below.

[0307] Email text entered by the user:

[0308] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0309] Emotion recognition results by emotion engine:

[0310] Recognized emotions: impatience, hurry

[0311] Server analysis results:

[0312] Detected issues:

[0313] "I want to check this urgently" (impression of urgency)

[0314] "Please reply as soon as possible" (Pressure)

[0315] Suggested fix:

[0316] Suggested fix:

[0317] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[0318] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[0319] Final confirmation from the user:

[0320] Dear John,

[0321] I know you are busy, but could you please check on the meeting we had the other day? I would appreciate it if you could check it when you have time.

[0322] thank you.

[0323] Yamada

[0324] This system allows users to create emails appropriately and efficiently, and an emotion engine provides revision suggestions that reflect the user's emotions, enabling more human-like communication.The system also supports multiple languages ​​and has customization functions based on industry-specific language usage, making it possible to build a flexible system that meets the diverse needs of users.

[0325] The processing flow will be explained below.

[0326] Step 1:

[0327] The user composes a business email and types it into the terminal. For example, the user types the following:

[0328] "John, I need to confirm something about the meeting we had the other day. Please respond as soon as possible. Thank you."

[0329] Step 2:

[0330] The device retrieves the text data entered by the user, which includes the entire body of the email.

[0331] Step 3:

[0332] The device sends the acquired text data to the emotion engine, which analyzes the user's emotions contained in the text data. For example, the emotion engine recognizes the user's impatience from "I want to check this urgently."

[0333] Step 4:

[0334] The device sends the entire text data along with the emotion analysis results to the server, which then prepares to begin analyzing the email text.

[0335] Step 5:

[0336] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions. In this case, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[0337] Step 6:

[0338] The server generates appropriate correction suggestions for detected inappropriate expressions based on the emotion analysis results from the emotion engine. For example, a strong expression such as "I would like to check this urgently" can be corrected to a milder expression such as "I apologize for bothering you during your busy schedule, but could you please check this?"

[0339] Step 7:

[0340] The server sends the generated revision suggestions together with the emotion analysis results back to the device, which then displays the revision suggestions received from the server to the user.

[0341] Step 8:

[0342] The user checks the displayed correction suggestions and decides whether to accept them. If the user accepts the suggestions, the terminal will correct the email text as suggested.

[0343] Step 9:

[0344] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[0345] Step 10:

[0346] After the user's operation is confirmed, the device will send the confirmation message and feedback information to the server, which will then collect the user's feedback information and use it as learning data to improve the accuracy of the AI ​​algorithm and emotion engine.

[0347] This process allows users to create emails appropriately and efficiently, and by utilizing the emotion engine, it enables more human-like communication that takes into account the user's emotions.

[0348] Example 2

[0349] 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."

[0350] Conventional business email writing support systems focus on detecting inappropriate expressions and suggesting corrections by analyzing text data, but because they are unable to take user emotions into account, they have the problem of difficulty in providing flexible correction suggestions that are in line with emotions.In addition, mechanisms for improving the accuracy of correction suggestions based on user feedback are limited to a few cases, and there is a problem that long-term learning and improvement are not sufficiently carried out.

[0351] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0352] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for recognizing the user's emotions using an emotion recognition engine, means for adjusting revision suggestions based on the recognized emotions, means for generating revision suggestions for the detected inappropriate expressions, means for displaying the revision suggestions to the user, and means for improving the accuracy of the revision suggestions based on user feedback, thereby enabling flexible and appropriate revision suggestions that reflect the user's emotions.

[0353] "User" refers to an individual or organization that uses the system to create business emails.

[0354] "Text data" refers to character string information entered by the user, and includes the email text itself.

[0355] An "emotion recognition engine" refers to software that analyzes and recognizes user emotions contained in text data.

[0356] "Correction suggestions" refer to alternative expressions generated by the system to correct inappropriate expressions in text data.

[0357] "Feedback" refers to the user's evaluation and reaction to the correction suggestions provided by the system.

[0358] A "server" refers to a central device that manages and controls the processes of the entire system and processes and manages various data.

[0359] An "NLP engine" refers to software or algorithms for natural language processing.

[0360] These definitions help clarify the role and function of each element of the system.

[0361] This invention is a system that analyzes text data, detects inappropriate expressions, and provides suggestions for correction when a user composes a business email. Furthermore, by combining it with an emotion engine that recognizes the user's emotions and reflects them in the suggested corrections, it becomes possible to create emails that are more flexible and human-like.

[0362] System configuration

[0363] This system mainly consists of four entities: the terminal, the server, the user, and the emotion engine.

[0364] Terminal: The device (e.g., PC, smartphone) on which the user enters the email text.

[0365] Server: A central processing unit that analyzes text data and generates profanity detection and correction suggestions.

[0366] User: A person or organization that writes business emails.

[0367] Emotion engine: Software for analyzing user emotions (e.g., IBM Watson Tone Analyzer).

[0368] Text data acquisition and sentiment analysis

[0369] A user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[0370] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0371] The device receives text data entered by the user and analyzes the user's emotions contained in the text data using an emotion recognition engine. The analyzed emotion information is then sent to the server together with the text data.

[0372] Text analysis and profanity detection

[0373] The server inputs the received text data into a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to analyze the email text. This is where inappropriate expressions are detected. For example, expressions such as "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong.

[0374] Generate and refine correction suggestions

[0375] The server generates correction suggestions based on the analysis results of the emotion engine. In this case, expressions that are recognized as having strong emotions are adjusted to softer expressions. For example, "I would like to check this urgently" is corrected to "I apologize for bothering you during your busy schedule, but could you please check it?"

[0376] View and finalize suggested revisions

[0377] The server sends the generated revision suggestions to the terminal. The terminal displays the revision suggestions received from the server to the user. The user checks the displayed suggestions and chooses whether to accept them. If accepted, the terminal automatically corrects the email text according to the suggestions. Finally, the user checks the final text and sends it if there are no problems.

[0378] Gathering feedback and training the system

[0379] The server collects user feedback and reflects it in the generative AI model and emotion engine algorithms, which improves the accuracy of future suggestions and emotion recognition.

[0380] Specific examples (prompt sentence examples)

[0381] For example:

[0382] Email text entered by the user

[0383] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0384] Emotion recognition results by emotion engine

[0385] Recognized emotions: impatience, hurry

[0386] Revision suggestions

[0387] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[0388] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[0389] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0390] Program processing flow

[0391] The program processing flow of this system will be explained in detail below, divided into steps.

[0392] Step 1:

[0393] The user enters the text of a business email into the terminal.

[0394] Input: Text data entered by the user into the terminal.

[0395] Action: A user enters text using a keyboard or touchscreen.

[0396] Output: Text data saved in the device.

[0397] Step 2:

[0398] The device acquires text data entered by the user and uses an emotion recognition engine to analyze the user's emotions contained in the text data.

[0399] Input: The text data entered in step 1.

[0400] How it works: The emotion recognition engine extracts emotional information from text (e.g., impatience, urgency).

[0401] Output: User emotion information and text data.

[0402] Step 3:

[0403] The device transmits the acquired text data and emotional information to the server.

[0404] Input: Text data and emotion information.

[0405] Action: The operation of sending data to a server, transferring data over a network.

[0406] Output: Text data and emotion information received by the server.

[0407] Step 4:

[0408] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the text data and detects inappropriate expressions.

[0409] Input: Text data and emotion information sent from the device.

[0410] How it works: An NLP engine analyzes text to detect strong or inappropriate language.

[0411] Output: Text data with profanity flagged.

[0412] Step 5:

[0413] The server generates correction suggestions for detected inappropriate expressions based on the results of the emotion recognition engine.

[0414] Input: Text data with inappropriate language flagged, sentiment information.

[0415] How it works: The revision suggestion algorithm generates revision suggestions that take user sentiment into account.

[0416] Output: Text data with suggested corrections.

[0417] Step 6:

[0418] The server sends the generated revision suggestions to the terminal.

[0419] Input: Text data with suggested corrections.

[0420] Action: The act of sending data to a device, transferring data over a network.

[0421] Output: Correction suggestions received on the device.

[0422] Step 7:

[0423] The terminal displays the revision suggestions received from the server to the user, who then checks the revision suggestions and decides whether to accept them.

[0424] Input: The correction proposal received.

[0425] Behavior: Displays suggested revisions in the user interface.

[0426] Output: User confirmation and choice (accept or not).

[0427] Step 8:

[0428] If the user accepts the suggested revisions, the device automatically corrects the email text as suggested. The user then finally checks the text and, if there are no problems, sends the email.

[0429] Input: The correction suggestion that the user accepted.

[0430] Action: Adjust the email text based on the suggested revisions and ask the user for final confirmation. Send the email.

[0431] Output: The final revised email text and the email sent.

[0432] Step 9:

[0433] The server collects user feedback information and reflects it in the generative AI model and emotion engine algorithms.

[0434] Input: User feedback information.

[0435] How it works: Feedback information is fed into the algorithm and used as training data for the model.

[0436] Output: Improved generative AI models and emotion engines.

[0437] (Application example 2)

[0438] 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."

[0439] When dealing with customers, staff are required to use appropriate language while taking into consideration the customer's feelings, but this can be difficult to judge on the ground, and inappropriate language can be used, especially when dealing with complaints or inquiries. This can lead to a decrease in customer satisfaction and the risk of problems arising. A system to effectively resolve this is needed.

[0440] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0441] In this invention, the server includes means for acquiring text data and voice data input by a user, means for analyzing the text data and voice data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions based on emotion information, thereby enabling support for staff to use appropriate expressions in real time while dealing with customers.

[0442] "Text data" refers to character string information entered by a user. It is generally saved as recorded character string information or a document.

[0443] "Voice Data" means a digital recording of a user's spoken voice, often converted to text using speech recognition technology.

[0444] "Analysis" refers to the act of processing input data using an algorithm to extract specific information or patterns.

[0445] "Inappropriate expressions" refer to expressions that contain inappropriate meanings or emotions in communication and may cause misunderstandings or trouble.

[0446] "Emotional information" refers to emotional elements extracted from the text or voice of a user or customer, and includes emotions such as fear, impatience, and joy.

[0447] "Correction proposals" refer to proposals that suggest appropriate expressions or ways of responding based on the analysis results.

[0448] "Real time" refers to a state in which the time between when data is generated and when it is processed is extremely short, and the data is processed almost simultaneously.

[0449] "Display" refers to the act of visually presenting the analysis results and correction suggestions to the user, outputting them on a display or screen.

[0450] "Feedback" refers to the evaluations and reactions that users provide to a system, which are used to improve and learn from the system.

[0451] "Device" refers to a hardware device with a specific function, including smartphones and robots.

[0452] A "server" refers to a computer system that provides services to clients over a network.

[0453] A "system" refers to a collection of multiple components working together to achieve a specific function.

[0454] The present invention relates to a system that analyzes input text data and voice data when a user is dealing with a customer, and provides appropriate correction suggestions based on emotion information. A specific embodiment of the system includes the following steps.

[0455] The server receives text and voice data sent from the user's device (smartphone or robot). At that time, it uses voice recognition software and an NLP (natural language processing) engine to convert the voice data into text data. Specifically, it uses the "SpeechRecognition" module for voice recognition and the "TextBlob" module for sentiment analysis of the text data.

[0456] The server then uses pre-defined rule-based filters and machine learning algorithms to detect inappropriate content based on the analyzed text data. If an inappropriate content is detected, the server generates appropriate correction suggestions based on the sentiment information. For example, the phrase "I would like to check this urgently" is converted into the suggestion "I apologize for bothering you during your busy schedule, but could you please check it?"

[0457] The generated correction suggestions are displayed in real time on the user's device. The user can review the suggestions and adopt the corrected expressions as needed. In addition, the user's feedback information is sent to the server and used to improve the accuracy of the correction suggestions.

[0458] As a concrete example, the following prompt sentence is used:

[0459] Example prompt sentence:

[0460] Speak specific customer interaction text. Example: "Please check availability of this item immediately."

[0461] It analyzes sentiment and outputs appropriate correction suggestions in the following format: Example correction: "I'm sorry to bother you, but could you please wait a moment while I check this?"

[0462] This system enables staff to use appropriate expressions in real time while interacting with customers, which is expected to improve customer satisfaction. The hybrid system combines voice input and text analysis to enable quick and flexible responses.

[0463] Furthermore, the system continuously learns from feedback and improves the accuracy of correction suggestions, making it possible to adapt to the diverse needs of users.

[0464] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0465] Step 1:

[0466] The user inputs into the device by voice or text.

[0467] Specifically, it uses the smartphone's microphone to capture voice data, and also allows for manual text input.

[0468] Input: Audio or text data

[0469] Output: User input audio or text data

[0470] Step 2:

[0471] The terminal converts the voice data into text data.

[0472] Specifically, the speech data is converted into text data using the SpeechRecognition module.

[0473] Input: Audio data

[0474] Output: Converted text data

[0475] Step 3:

[0476] The terminal transmits the text data to the server.

[0477] As a specific operation, the converted text data is sent to a server via the Internet.

[0478] Input: Text data

[0479] Output: Text data sent to the server

[0480] Step 4:

[0481] The server analyzes the text data and extracts emotional information.

[0482] Specifically, it uses the TextBlob module to analyze the polarity (emotional polarity) and subjectivity of text data.

[0483] Input: Text data

[0484] Output: Emotional information (e.g. polarity, subjectivity)

[0485] Step 5:

[0486] The server detects profanity.

[0487] Specifically, it uses predefined filters and machine learning models to identify inappropriate language.

[0488] Input: Text data

[0489] Output: A list of profanities

[0490] Step 6:

[0491] The server generates revision suggestions based on the emotion information.

[0492] Specifically, it suggests alternative expressions for detected inappropriate expressions and adjusts them taking into account the user's emotions and context.

[0493] Input: Inappropriate language, dishonest behavior

[0494] Output: Sentiment-based revision suggestions

[0495] Step 7:

[0496] The server sends the revision suggestions to the terminal.

[0497] As a specific operation, the generated revision proposal is sent back to the user's terminal via the Internet.

[0498] Input: Proposed correction

[0499] Output: Suggested fixes sent to terminal

[0500] Step 8:

[0501] The terminal displays the suggested revisions to the user.

[0502] Specifically, the proposed corrections are displayed on the screen so that the user can confirm them.

[0503] Input: Proposed correction

[0504] Output: Displayed suggested fixes

[0505] Step 9:

[0506] The user reviews the proposed corrections and selects a response.

[0507] As a specific operation, the user either accepts the suggested revision or creates a new message that reflects the suggested revision.

[0508] Input: Displayed suggested corrections

[0509] Output: User selected or modified text data

[0510] Step 10:

[0511] The server collects user feedback and uses it to improve the accuracy of the system.

[0512] Specifically, user choices and feedback information are stored in a database and used as training data for generative AI models and algorithms.

[0513] Input: User feedback

[0514] Output: Improved correction suggestion algorithm

[0515] 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.

[0516] 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.

[0517] 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.

[0518] [Second embodiment]

[0519] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0520] 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.

[0521] 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).

[0522] 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.

[0523] 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.

[0524] 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).

[0525] 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.

[0526] 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.

[0527] 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.

[0528] 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.

[0529] 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.

[0530] 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."

[0531] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. Below, the program processing and specific examples of this system will be described.

[0532] Program Processing Overview

[0533] The system mainly consists of three entities: the terminal, the server, and the user. The roles and operations of each entity are as follows:

[0534] 1. User creates email message:

[0535] The user types the text of a business email into the terminal. For example, the user types the following text to a business partner:

[0536] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0537] 2. Sending email content:

[0538] The device sends the entered text data to the server, which includes the entire body of the email.

[0539] 3. Text analysis and profanity detection:

[0540] The server analyzes the received text data using a natural language processing (NLP) engine to detect inappropriate expressions, such as "Thank you for your hard work, John-san."

[0541] 4. Generate correction suggestions:

[0542] The server generates appropriate expressions for the inappropriate expressions it detects. In this case, the suggested correction for "Thank you for your hard work, John-san" would be "John-san."

[0543] 5. View suggested revisions:

[0544] The terminal displays the correction suggestions sent back from the server to the user, who then checks the suggestions and chooses whether to accept them.

[0545] 6. Final User Review:

[0546] The user decides whether to accept the suggested revisions, and if so, the terminal corrects the text to reflect the suggested expressions. The user then confirms the final text and sends the email.

[0547] 7. Gather feedback and learn:

[0548] The server collects user feedback and reflects it in the generative AI algorithm to improve the accuracy of future suggestions, allowing the system to continuously learn and improve.

[0549] Specific examples

[0550] Specific examples are shown below.

[0551] Email text entered by the user:

[0552] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0553] Server analysis results:

[0554] Detected issues:

[0555] "Thank you for your hard work, John" (too friendly)

[0556] "I want to discuss this" (too direct)

[0557] Suggested fix:

[0558] Suggested fix:

[0559] "Thank you for your hard work, John-san" → "John-san"

[0560] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0561] Final confirmation from the user:

[0562] Dear John,

[0563] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[0564] thank you.

[0565] Yamada

[0566] This system allows users to create business emails appropriately and efficiently. Its multilingual capabilities also make it suitable for international business environments, making it suitable for a wide range of uses. Furthermore, the system can be customized to suit the language and cultural characteristics specific to the user's industry or company, allowing it to flexibly meet user needs.

[0567] The processing flow will be explained below.

[0568] Step 1:

[0569] The user composes a business email and types it into the terminal. For example, the user types the following:

[0570] "Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thanks in advance."

[0571] Step 2:

[0572] The device retrieves the text data entered by the user, which includes the entire body of the email.

[0573] Step 3:

[0574] The device sends the acquired text data to the server, which then prepares to begin analyzing the email text.

[0575] Step 4:

[0576] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions.

[0577] Step 5:

[0578] Based on the analysis results, the server lists inappropriate or misleading expressions, such as "Thank you for your hard work, John" or "I'd like to discuss this."

[0579] Step 6:

[0580] Based on the detection results, the server uses a generation AI to generate appropriate correction suggestions. The suggested correction for "Thank you for your hard work, John" is "Dear John," and the suggested correction for "I'd like to discuss this" is "I'd like to take the time to talk to you."

[0581] Step 7:

[0582] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions to the user based on the information received from the server.

[0583] Step 8:

[0584] The user checks the displayed correction suggestions and selects whether to accept them. If the user accepts the corrections, the terminal corrects the email text as suggested.

[0585] Step 9:

[0586] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[0587] Step 10:

[0588] After the final text is confirmed based on the user's input, the device sends the finalized data to the server, which stores the user's feedback and updates the AI's algorithm to improve the accuracy of future suggestions.

[0589] Example 1

[0590] 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."

[0591] When writing business emails, it is difficult for users to use appropriate expressions and avoid inappropriate ones. Furthermore, using expressions appropriate for different cultures and industries requires a great deal of effort and specialized knowledge. Furthermore, systems are required to continuously learn and improve based on user feedback. However, current systems have difficulty effectively resolving all of these issues, and the present invention aims to solve these problems.

[0592] 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.

[0593] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions, thereby enabling the user to efficiently create business email text using appropriate expressions.

[0594] "User" refers to an individual or organization that uses the System to create business emails.

[0595] "Text data" refers to a data format that includes text or sentence information entered by a user.

[0596] "Terminal" refers to the computing device used by a user to enter email text and review suggested revisions.

[0597] "Server" refers to a central computer system that analyzes text data, detects profanity, and generates correction suggestions.

[0598] "Analysis" refers to the process of examining the content of text data using machine learning and natural language processing techniques to identify inappropriate expressions.

[0599] "Inappropriate language" refers to expressions or phrases that are inappropriate to use in a business context.

[0600] "Correction suggestions" refer to alternative suggestions for changing detected inappropriate expressions into appropriate expressions.

[0601] "Feedback" refers to information about whether the user accepted the suggested revisions and whether the suggested revisions were appropriate.

[0602] "Generative AI models" refer to algorithms or machine learning models that generate text data and create correction suggestions.

[0603] A "prompt" refers to a command or question input to a generative AI model.

[0604] "Different languages" refers to languages ​​other than the user's native language.

[0605] "Customization" refers to the process of changing and adjusting system settings to suit the user's industry and corporate characteristics.

[0606] "Sending" refers to the electronic communication act used by a user to deliver the completed email text to the recipient.

[0607] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. This system consists of three entities: a terminal, a server, and a user. The function of each entity is explained below.

[0608] Device Features

[0609] The terminal is a device that allows users to input business emails. Terminals include PCs, tablets, smartphones, etc. Once a user has finished inputting the email, the terminal sends the text data to the server. It also displays any suggested revisions returned by the server.

[0610] Server Features

[0611] The server analyzes text data, detects inappropriate expressions, and generates correction suggestions. Specifically, it uses a natural language processing (NLP) library such as "Hugging Face" and a generative AI model (e.g., "OpenAI GPT"). The server analyzes the text data sent by the user using an NLP engine and detects inappropriate expressions. For detected inappropriate expressions, it uses the generative AI model to generate appropriate correction suggestions and sends them back to the device.

[0612] User operations

[0613] The user types the text of a business email into the terminal. For example, the user types the following text:

[0614] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0615] Once the input is complete, the user presses the send button to send the text data to the server. After that, the user checks the correction suggestions displayed on the terminal and selects whether to accept them. If the correction suggestions are accepted, the user makes a final confirmation and sends the email.

[0616] Specific examples

[0617] Specific examples are shown below.

[0618] Email text entered by the user:

[0619] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0620] Server analysis results:

[0621] Detected issues:

[0622] "Thank you for your hard work, John" (too friendly)

[0623] "I want to discuss this" (too direct)

[0624] Suggested fix:

[0625] Suggested fix:

[0626] "Thank you for your hard work, John-san" → "John-san"

[0627] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0628] Final confirmation from the user:

[0629] Dear John,

[0630] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[0631] thank you.

[0632] Yamada

[0633] Multilingual and customizable

[0634] The system is multilingual and can provide correction suggestions in different languages. It can also be customized to fit the language and cultural characteristics specific to the user's industry or company. This allows for more accurate correction suggestions.

[0635] Gathering feedback and learning

[0636] The server collects user feedback and continuously trains the generative AI model based on it, thereby improving the accuracy of future suggestions.

[0637] As described above, the system of the present invention provides powerful support for users to create business emails quickly and appropriately.

[0638] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0639] Step 1:

[0640] User-generated email content

[0641] A user uses a terminal to type the text of a business email. For example, the user types the following sentence:

[0642] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[0643] Once this input is complete, the user presses the send button to prepare to send the text data from the terminal to the server.

[0644] Input: Email text entered by the user

[0645] Output: Text data of the entered email

[0646] Step 2:

[0647] Sending email content

[0648] The device sends the text data entered by the user to the server. At this time, the device sends the email body data to the server using an HTTP request. The request includes the user ID and the email body.

[0649] Input: Text data entered into the terminal

[0650] Output: Text data sent to the server

[0651] Step 3:

[0652] Text analysis and profanity detection

[0653] The server receives the text data and analyzes it using a natural language processing (NLP) engine. This analysis process uses NLP libraries such as "Hugging Face." The server analyzes the content of the email and detects inappropriate expressions. For example, it detects that the expression "Thank you for your hard work, John" is too familiar.

[0654] Input: Text data sent to the server

[0655] Output: Analysis results and a list of inappropriate expressions detected

[0656] Step 4:

[0657] Generate correction suggestions

[0658] The server uses a generative AI model (e.g., "OpenAI GPT") to generate appropriate correction suggestions for the detected inappropriate expressions. Specifically, for the expression "Thank you for your hard work, John-san," the server makes a correction suggestion such as "John-san."

[0659] Input: Profanity detection list

[0660] Output: List of suggested fixes

[0661] Step 5:

[0662] View suggested fixes

[0663] The terminal displays the correction suggestions sent back from the server to the user. The correction suggestions are displayed in a visually intuitive manner, and the user can confirm them and choose whether to accept them.

[0664] Input: A list of suggested revisions sent from the server

[0665] Output: A list of suggested revisions that is displayed to the user

[0666] Step 6:

[0667] Final confirmation by the user

[0668] The user checks the displayed correction suggestions and decides whether to accept them. If they accept, the device corrects the text to the suggested expression. The user then checks the final text and presses the send button to send the email. For example, the corrected text might look like this:

[0669] Dear John,

[0670] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[0671] thank you.

[0672] Yamada

[0673] Input: A list of suggested revisions shown to the user

[0674] Output: The final email message that the user has confirmed and corrected

[0675] Step 7:

[0676] Gathering feedback and learning

[0677] The server collects feedback from users and sends back information such as how the user accepted the correction suggestions and whether the suggestions were appropriate. The server uses this feedback to update the generative AI model and improve the accuracy of future correction suggestions.

[0678] Input: User feedback information

[0679] Output: Update the generative AI model based on feedback

[0680] (Application example 1)

[0681] 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."

[0682] In today's business emails and chats, the text data entered by users often contains inappropriate expressions. Such expressions can lead to misunderstandings and misinterpretations in business communications, negatively impacting a company's credibility and efficiency. Furthermore, in an international business environment, the need for multilingual support makes correcting inappropriate expressions even more complicated. Furthermore, when real-time responses are required, the ability to quickly and accurately correct expressions is essential.

[0683] 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.

[0684] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for analyzing the text data in real time in different response formats and correcting the text data to appropriate expressions, means for generating correction suggestions for the detected inappropriate expressions, means for generating the correction suggestions using a generative AI model, means for displaying the correction suggestions to the user, and means for improving the accuracy of the correction suggestions based on user feedback. This reduces misunderstandings and misinterpretations in business communication, improves corporate reliability and efficiency, and enables multilingual support.

[0685] "User" refers to any individual or corporation that uses the system.

[0686] "Text data" refers to information in the form of a string of characters entered by the user.

[0687] "Analysis" refers to the process of processing and analyzing the content of text data.

[0688] "Inappropriate expressions" refer to expressions that may lead to misunderstandings or misinterpretations in business communications.

[0689] "Correction proposal" refers to an alternative proposal to convert inappropriate language into appropriate language.

[0690] "Display" refers to visually showing the suggested revisions on the user's terminal.

[0691] "Feedback" refers to the opinions and reactions of users to suggested revisions.

[0692] "Accuracy improvement" refers to improvement activities carried out to increase the accuracy and effectiveness of proposals.

[0693] "Different response formats" refers to response methods that are applied in different formats and situations.

[0694] "Real time" refers to the concept of time when data is processed as soon as it is entered.

[0695] "Generative AI model" refers to a model that uses artificial intelligence to generate and modify text data.

[0696] The system of the present invention analyzes text data entered by users, detects inappropriate expressions, and provides appropriate correction suggestions. This system consists of three entities: a terminal, a server, and a user.

[0697] System Configuration

[0698] Hardware and Software Configuration

[0699] Device: The device used by the user, such as a smartphone, tablet, or desktop PC.

[0700] Server: Provides high-performance computing power to analyze text data and generate correction suggestions.

[0701] software:

[0702] OpenAI API: Generates correction suggestions using the generative AI model GPT-3.

[0703] Python: A programming language for implementing the overall system logic.

[0704] Regular expressions (re): Used to process and analyze text data.

[0705] Data processing and calculation

[0706] 1. Acquiring text data:

[0707] A user inputs text data using a terminal, and the text data is transmitted from the terminal to a server.

[0708] 2. Text data analysis:

[0709] The server analyzes the received text data using the OpenAI API. First, it detects inappropriate expressions in the text data.

[0710] 3. Generate correction suggestions:

[0711] For detected inappropriate expressions, a generative AI model is used to generate appropriate correction suggestions.

[0712] 4. View suggested revisions:

[0713] The revision suggestions generated by the server are sent to the terminal and displayed to the user.

[0714] 5. Gathering Feedback:

[0715] The accuracy of suggestions is improved by sending user feedback information to the server and reflecting it in the generative AI model.

[0716] Specific examples

[0717] Example 1: Text analysis of business emails

[0718] A user types the following email on their smartphone:

[0719] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0720] Server analysis results:

[0721] Detected issues:

[0722] "Thank you for your hard work, person in charge" (too friendly)

[0723] "I'd like to discuss this" (too direct)

[0724] Suggested fix:

[0725] Suggested fix:

[0726] "Thank you for your hard work, person in charge" → "Dear person in charge"

[0727] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0728] Final confirmation:

[0729] To the person in charge

[0730] I would like to take the time to speak with you about the meeting we had the other day. I would appreciate it if you could let me know a convenient time and date for us to meet.

[0731] thank you.

[0732] Yamada

[0733] Prompt Sentence Examples

[0734] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[0735] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0736] Fixes:

[0737] This system allows users to quickly correct inappropriate language used in business emails and chats, ensuring appropriate communication. It also supports multiple languages ​​and can be customized based on industry characteristics, making it applicable to a wide range of uses. This is expected to reduce misunderstandings and misinterpretations in business, improving corporate reliability and efficiency.

[0738] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0739] Step 1:

[0740] Users use their devices to input the text of business emails or chat messages. Specifically, users enter text data into an input field on a smartphone, tablet, or desktop PC, and then press the send button. The input data is, for example, in the following format:

[0741] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0742] Input: Text data entered by the user into the terminal.

[0743] Output: Text data ready to be sent to the terminal

[0744] Step 2:

[0745] The terminal sends the entered text data to the server, where the data is transferred securely using protocols such as HTTPS. The entered text is sent in a format that the server can parse.

[0746] Input: Text data sent from the device to the server

[0747] Output: Text data received by the server

[0748] Step 3:

[0749] The server sends the received text data to the OpenAI API for analysis. The server generates a prompt to be used for analysis and passes the text data to the natural language processing engine. At this time, the following prompt is generated:

[0750] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[0751] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[0752] Fixes:

[0753] Input: Text data received by the server and prompts

[0754] Output: Analysis results by OpenAI API

[0755] Step 4:

[0756] The OpenAI API uses a generative AI model to detect inappropriate expressions and generate appropriate correction suggestions, such as "Thank you for your hard work, Mr. / Ms. Person in Charge."

[0757] Input: Prompt and text data received by the OpenAI API

[0758] Output: Detected profanities and suggested corrections

[0759] Step 5:

[0760] The server receives the correction suggestions returned by the OpenAI API and sends them to the device, where they can be viewed by the user. The correction suggestions are displayed in the following format:

[0761] Suggested fix:

[0762] "Thank you for your hard work, person in charge" → "Dear person in charge"

[0763] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[0764] Input: Analysis results returned from the OpenAI API

[0765] Output: Suggested fixes sent to terminal

[0766] Step 6:

[0767] The terminal displays the received correction suggestions to the user, who then checks the displayed correction suggestions and selects whether to accept them. The selection operation is performed via buttons or an interface on the terminal.

[0768] Input: Suggested corrections sent to the terminal

[0769] Output: Correction suggestions and user selections displayed on the device

[0770] Step 7:

[0771] The user decides whether to accept the suggested revisions and confirms the result on the terminal. If the revision is accepted, the terminal corrects the text to the suggested expression and displays the final text.

[0772] Input: User accepts or rejects suggested revisions

[0773] Output: Final corrected text

[0774] Step 8:

[0775] The server collects user feedback and reflects it in the generative AI model algorithm, which improves the accuracy of correction suggestions from the next time onwards.

[0776] Input: User feedback information

[0777] Output: Improved generative AI model algorithm

[0778] This allows users to correct inappropriate expressions in business emails and chat messages to appropriate expressions, enabling efficient and effective communication.

[0779] 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.

[0780] The present invention relates to a system that analyzes text data, detects inappropriate expressions, and provides correction suggestions when a user composes a business email, in addition to combining it with an emotion engine that recognizes the user's emotions and reflects them in the correction suggestions. Below, we will explain the program processing and specific operation examples of this system.

[0781] Program Processing Overview

[0782] The system mainly consists of four entities: the terminal, the server, the user, and the emotion engine. The individual roles and operations of each entity are as follows:

[0783] User-generated email message:

[0784] The user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[0785] example:

[0786] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0787] Send email content:

[0788] The device acquires text data entered by the user. At this time, the device uses an emotion engine to recognize the emotion contained in the text data entered by the user. The acquired text data and emotion information are sent to the server.

[0789] Text analysis and profanity detection:

[0790] The server inputs the received text data into an NLP engine, which analyzes the email text and detects inappropriate expressions. For example, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[0791] Emotion recognition and suggested correction adjustments:

[0792] The server generates correction suggestions based on the emotional information recognized by the emotion engine. In this case, the strong emotional expression "I would like to check this urgently" is adjusted to a softer expression such as "I apologize for bothering you during your busy schedule, but could you please check it?"

[0793] View suggested fixes:

[0794] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions received from the server to the user, who then checks the displayed suggestions and chooses whether to accept them.

[0795] Final user confirmation:

[0796] The user decides whether to accept the suggested revisions, and if so, the terminal will revise the email text as suggested. The user then checks the final text and, if there are no problems, confirms sending the email.

[0797] Gathering feedback and learning:

[0798] The server collects user feedback and reflects it in the generative AI and emotion engine algorithms to improve the accuracy of future suggestions and emotion recognition, allowing the system to continuously learn and improve.

[0799] Specific example (utilizing emotion engine)

[0800] A specific example is given below.

[0801] Email text entered by the user:

[0802] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0803] Emotion recognition results by emotion engine:

[0804] Recognized emotions: impatience, hurry

[0805] Server analysis results:

[0806] Detected issues:

[0807] "I want to check this urgently" (impression of urgency)

[0808] "Please reply as soon as possible" (Pressure)

[0809] Suggested fix:

[0810] Suggested fix:

[0811] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[0812] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[0813] Final confirmation from the user:

[0814] Dear John,

[0815] I know you are busy, but could you please check on the meeting we had the other day? I would appreciate it if you could check it when you have time.

[0816] thank you.

[0817] Yamada

[0818] This system allows users to create emails appropriately and efficiently, and an emotion engine provides revision suggestions that reflect the user's emotions, enabling more human-like communication.The system also supports multiple languages ​​and has customization functions based on industry-specific language usage, making it possible to build a flexible system that meets the diverse needs of users.

[0819] The processing flow will be explained below.

[0820] Step 1:

[0821] The user composes a business email and types it into the terminal. For example, the user types the following:

[0822] "John, I need to confirm something about the meeting we had the other day. Please respond as soon as possible. Thank you."

[0823] Step 2:

[0824] The device retrieves the text data entered by the user, which includes the entire body of the email.

[0825] Step 3:

[0826] The device sends the acquired text data to the emotion engine, which analyzes the user's emotions contained in the text data. For example, the emotion engine recognizes the user's impatience from "I want to check this urgently."

[0827] Step 4:

[0828] The device sends the entire text data along with the emotion analysis results to the server, which then prepares to begin analyzing the email text.

[0829] Step 5:

[0830] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions. In this case, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[0831] Step 6:

[0832] The server generates appropriate correction suggestions for detected inappropriate expressions based on the emotion analysis results from the emotion engine. For example, a strong expression such as "I would like to check this urgently" can be corrected to a milder expression such as "I apologize for bothering you during your busy schedule, but could you please check this?"

[0833] Step 7:

[0834] The server sends the generated revision suggestions together with the emotion analysis results back to the device, which then displays the revision suggestions received from the server to the user.

[0835] Step 8:

[0836] The user checks the displayed correction suggestions and decides whether to accept them. If the user accepts the suggestions, the terminal will correct the email text as suggested.

[0837] Step 9:

[0838] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[0839] Step 10:

[0840] After the user's operation is confirmed, the device will send the confirmation message and feedback information to the server, which will then collect the user's feedback information and use it as learning data to improve the accuracy of the AI ​​algorithm and emotion engine.

[0841] This process allows users to create emails appropriately and efficiently, and by utilizing the emotion engine, it enables more human-like communication that takes into account the user's emotions.

[0842] Example 2

[0843] 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."

[0844] Conventional business email writing support systems focus on detecting inappropriate expressions and suggesting corrections by analyzing text data, but because they are unable to take user emotions into account, they have the problem of difficulty in providing flexible correction suggestions that are in line with emotions.In addition, mechanisms for improving the accuracy of correction suggestions based on user feedback are limited to a few cases, and there is a problem that long-term learning and improvement are not sufficiently carried out.

[0845] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0846] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for recognizing the user's emotions using an emotion recognition engine, means for adjusting revision suggestions based on the recognized emotions, means for generating revision suggestions for the detected inappropriate expressions, means for displaying the revision suggestions to the user, and means for improving the accuracy of the revision suggestions based on user feedback, thereby enabling flexible and appropriate revision suggestions that reflect the user's emotions.

[0847] "User" refers to an individual or organization that uses the system to create business emails.

[0848] "Text data" refers to character string information entered by the user, and includes the email text itself.

[0849] An "emotion recognition engine" refers to software that analyzes and recognizes user emotions contained in text data.

[0850] "Correction suggestions" refer to alternative expressions generated by the system to correct inappropriate expressions in text data.

[0851] "Feedback" refers to the user's evaluation and reaction to the correction suggestions provided by the system.

[0852] A "server" refers to a central device that manages and controls the processes of the entire system and processes and manages various data.

[0853] An "NLP engine" refers to software or algorithms for natural language processing.

[0854] These definitions help clarify the role and function of each element of the system.

[0855] This invention is a system that analyzes text data, detects inappropriate expressions, and provides suggestions for correction when a user composes a business email. Furthermore, by combining it with an emotion engine that recognizes the user's emotions and reflects them in the suggested corrections, it becomes possible to create emails that are more flexible and human-like.

[0856] System configuration

[0857] This system mainly consists of four entities: the terminal, the server, the user, and the emotion engine.

[0858] Terminal: The device (e.g., PC, smartphone) on which the user enters the email text.

[0859] Server: A central processing unit that analyzes text data and generates profanity detection and correction suggestions.

[0860] User: A person or organization that writes business emails.

[0861] Emotion engine: Software for analyzing user emotions (e.g., IBM Watson Tone Analyzer).

[0862] Text data acquisition and sentiment analysis

[0863] A user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[0864] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0865] The device receives text data entered by the user and analyzes the user's emotions contained in the text data using an emotion recognition engine. The analyzed emotion information is then sent to the server together with the text data.

[0866] Text analysis and profanity detection

[0867] The server inputs the received text data into a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to analyze the email text. This is where inappropriate expressions are detected. For example, expressions such as "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong.

[0868] Generate and refine correction suggestions

[0869] The server generates correction suggestions based on the analysis results of the emotion engine. In this case, expressions that are recognized as having strong emotions are adjusted to softer expressions. For example, "I would like to check this urgently" is corrected to "I apologize for bothering you during your busy schedule, but could you please check it?"

[0870] View and finalize suggested revisions

[0871] The server sends the generated revision suggestions to the terminal. The terminal displays the revision suggestions received from the server to the user. The user checks the displayed suggestions and chooses whether to accept them. If accepted, the terminal automatically corrects the email text according to the suggestions. Finally, the user checks the final text and sends it if there are no problems.

[0872] Gathering feedback and training the system

[0873] The server collects user feedback and reflects it in the generative AI model and emotion engine algorithms, which improves the accuracy of future suggestions and emotion recognition.

[0874] Specific examples (prompt sentence examples)

[0875] For example:

[0876] Email text entered by the user

[0877] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[0878] Emotion recognition results by emotion engine

[0879] Recognized emotions: impatience, hurry

[0880] Revision suggestions

[0881] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[0882] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[0883] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0884] Program processing flow

[0885] The program processing flow of this system will be explained in detail below, divided into steps.

[0886] Step 1:

[0887] The user enters the text of a business email into the terminal.

[0888] Input: Text data entered by the user into the terminal.

[0889] Action: A user enters text using a keyboard or touchscreen.

[0890] Output: Text data saved in the device.

[0891] Step 2:

[0892] The device acquires text data entered by the user and uses an emotion recognition engine to analyze the user's emotions contained in the text data.

[0893] Input: The text data entered in step 1.

[0894] How it works: The emotion recognition engine extracts emotional information from text (e.g., impatience, urgency).

[0895] Output: User emotion information and text data.

[0896] Step 3:

[0897] The device transmits the acquired text data and emotional information to the server.

[0898] Input: Text data and emotion information.

[0899] Action: The operation of sending data to a server, transferring data over a network.

[0900] Output: Text data and emotion information received by the server.

[0901] Step 4:

[0902] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the text data and detects inappropriate expressions.

[0903] Input: Text data and emotion information sent from the device.

[0904] How it works: An NLP engine analyzes text to detect strong or inappropriate language.

[0905] Output: Text data with profanity flagged.

[0906] Step 5:

[0907] The server generates correction suggestions for detected inappropriate expressions based on the results of the emotion recognition engine.

[0908] Input: Text data with inappropriate language flagged, sentiment information.

[0909] How it works: The revision suggestion algorithm generates revision suggestions that take user sentiment into account.

[0910] Output: Text data with suggested corrections.

[0911] Step 6:

[0912] The server sends the generated revision suggestions to the terminal.

[0913] Input: Text data with suggested corrections.

[0914] Action: The act of sending data to a device, transferring data over a network.

[0915] Output: Correction suggestions received on the device.

[0916] Step 7:

[0917] The terminal displays the revision suggestions received from the server to the user, who then checks the revision suggestions and decides whether to accept them.

[0918] Input: The correction proposal received.

[0919] Behavior: Displays suggested revisions in the user interface.

[0920] Output: User confirmation and choice (accept or not).

[0921] Step 8:

[0922] If the user accepts the suggested revisions, the device automatically corrects the email text as suggested. The user then finally checks the text and, if there are no problems, sends the email.

[0923] Input: The correction suggestion that the user accepted.

[0924] Action: Adjust the email text based on the suggested revisions and ask the user for final confirmation. Send the email.

[0925] Output: The final revised email text and the email sent.

[0926] Step 9:

[0927] The server collects user feedback information and reflects it in the generative AI model and emotion engine algorithms.

[0928] Input: User feedback information.

[0929] How it works: Feedback information is fed into the algorithm and used as training data for the model.

[0930] Output: Improved generative AI models and emotion engines.

[0931] (Application example 2)

[0932] 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."

[0933] When dealing with customers, staff are required to use appropriate language while taking into consideration the customer's feelings, but this can be difficult to judge on the ground, and inappropriate language can be used, especially when dealing with complaints or inquiries. This can lead to a decrease in customer satisfaction and the risk of problems arising. A system to effectively resolve this is needed.

[0934] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0935] In this invention, the server includes means for acquiring text data and voice data input by a user, means for analyzing the text data and voice data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions based on emotion information, thereby enabling support for staff to use appropriate expressions in real time while dealing with customers.

[0936] "Text data" refers to character string information entered by a user. It is generally saved as recorded character string information or a document.

[0937] "Voice Data" means a digital recording of a user's spoken voice, often converted to text using speech recognition technology.

[0938] "Analysis" refers to the act of processing input data using an algorithm to extract specific information or patterns.

[0939] "Inappropriate expressions" refer to expressions that contain inappropriate meanings or emotions in communication and may cause misunderstandings or trouble.

[0940] "Emotional information" refers to emotional elements extracted from the text or voice of a user or customer, and includes emotions such as fear, impatience, and joy.

[0941] "Correction proposals" refer to proposals that suggest appropriate expressions or ways of responding based on the analysis results.

[0942] "Real time" refers to a state in which the time between when data is generated and when it is processed is extremely short, and the data is processed almost simultaneously.

[0943] "Display" refers to the act of visually presenting the analysis results and correction suggestions to the user, outputting them on a display or screen.

[0944] "Feedback" refers to the evaluations and reactions that users provide to a system, which are used to improve and learn from the system.

[0945] "Device" refers to a hardware device with a specific function, including smartphones and robots.

[0946] A "server" refers to a computer system that provides services to clients over a network.

[0947] A "system" refers to a collection of multiple components working together to achieve a specific function.

[0948] The present invention relates to a system that analyzes input text data and voice data when a user is dealing with a customer, and provides appropriate correction suggestions based on emotion information. A specific embodiment of the system includes the following steps.

[0949] The server receives text and voice data sent from the user's device (smartphone or robot). At that time, it uses voice recognition software and an NLP (natural language processing) engine to convert the voice data into text data. Specifically, it uses the "SpeechRecognition" module for voice recognition and the "TextBlob" module for sentiment analysis of the text data.

[0950] The server then uses pre-defined rule-based filters and machine learning algorithms to detect inappropriate content based on the analyzed text data. If an inappropriate content is detected, the server generates appropriate correction suggestions based on the sentiment information. For example, the phrase "I would like to check this urgently" is converted into the suggestion "I apologize for bothering you during your busy schedule, but could you please check it?"

[0951] The generated correction suggestions are displayed in real time on the user's device. The user can review the suggestions and adopt the corrected expressions as needed. In addition, the user's feedback information is sent to the server and used to improve the accuracy of the correction suggestions.

[0952] As a concrete example, the following prompt sentence is used:

[0953] Example prompt sentence:

[0954] Speak specific customer interaction text. Example: "Please check availability of this item immediately."

[0955] It analyzes sentiment and outputs appropriate correction suggestions in the following format: Example correction: "I'm sorry to bother you, but could you please wait a moment while I check this?"

[0956] This system enables staff to use appropriate expressions in real time while interacting with customers, which is expected to improve customer satisfaction. The hybrid system combines voice input and text analysis to enable quick and flexible responses.

[0957] Furthermore, the system continuously learns from feedback and improves the accuracy of correction suggestions, making it possible to adapt to the diverse needs of users.

[0958] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0959] Step 1:

[0960] The user inputs into the device by voice or text.

[0961] Specifically, it uses the smartphone's microphone to capture voice data, and also allows for manual text input.

[0962] Input: Audio or text data

[0963] Output: User input audio or text data

[0964] Step 2:

[0965] The terminal converts the voice data into text data.

[0966] Specifically, the speech data is converted into text data using the SpeechRecognition module.

[0967] Input: Audio data

[0968] Output: Converted text data

[0969] Step 3:

[0970] The terminal transmits the text data to the server.

[0971] As a specific operation, the converted text data is sent to a server via the Internet.

[0972] Input: Text data

[0973] Output: Text data sent to the server

[0974] Step 4:

[0975] The server analyzes the text data and extracts emotional information.

[0976] Specifically, it uses the TextBlob module to analyze the polarity (emotional polarity) and subjectivity of text data.

[0977] Input: Text data

[0978] Output: Emotional information (e.g. polarity, subjectivity)

[0979] Step 5:

[0980] The server detects profanity.

[0981] Specifically, it uses predefined filters and machine learning models to identify inappropriate language.

[0982] Input: Text data

[0983] Output: A list of profanities

[0984] Step 6:

[0985] The server generates revision suggestions based on the emotion information.

[0986] Specifically, it suggests alternative expressions for detected inappropriate expressions and adjusts them taking into account the user's emotions and context.

[0987] Input: Inappropriate language, dishonest behavior

[0988] Output: Sentiment-based revision suggestions

[0989] Step 7:

[0990] The server sends the revision suggestions to the terminal.

[0991] As a specific operation, the generated revision proposal is sent back to the user's terminal via the Internet.

[0992] Input: Proposed correction

[0993] Output: Suggested fixes sent to terminal

[0994] Step 8:

[0995] The terminal displays the suggested revisions to the user.

[0996] Specifically, the proposed corrections are displayed on the screen so that the user can confirm them.

[0997] Input: Proposed correction

[0998] Output: Displayed suggested fixes

[0999] Step 9:

[1000] The user reviews the proposed corrections and selects a response.

[1001] As a specific operation, the user either accepts the suggested revision or creates a new message that reflects the suggested revision.

[1002] Input: Displayed suggested corrections

[1003] Output: User selected or modified text data

[1004] Step 10:

[1005] The server collects user feedback and uses it to improve the accuracy of the system.

[1006] Specifically, user choices and feedback information are stored in a database and used as training data for generative AI models and algorithms.

[1007] Input: User feedback

[1008] Output: Improved correction suggestion algorithm

[1009] 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.

[1010] 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.

[1011] 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.

[1012] [Third embodiment]

[1013] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1014] 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.

[1015] 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).

[1016] 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.

[1017] 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.

[1018] 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).

[1019] 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.

[1020] 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.

[1021] 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.

[1022] 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.

[1023] 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.

[1024] 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."

[1025] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. Below, the program processing and specific examples of this system will be described.

[1026] Program Processing Overview

[1027] The system mainly consists of three entities: the terminal, the server, and the user. The roles and operations of each entity are as follows:

[1028] 1. User creates email message:

[1029] The user types the text of a business email into the terminal. For example, the user types the following text to a business partner:

[1030] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1031] 2. Sending email content:

[1032] The device sends the entered text data to the server, which includes the entire body of the email.

[1033] 3. Text analysis and profanity detection:

[1034] The server analyzes the received text data using a natural language processing (NLP) engine to detect inappropriate expressions, such as "Thank you for your hard work, John-san."

[1035] 4. Generate correction suggestions:

[1036] The server generates appropriate expressions for the inappropriate expressions it detects. In this case, the suggested correction for "Thank you for your hard work, John-san" would be "John-san."

[1037] 5. View suggested revisions:

[1038] The terminal displays the correction suggestions sent back from the server to the user, who then checks the suggestions and chooses whether to accept them.

[1039] 6. Final User Review:

[1040] The user decides whether to accept the suggested revisions, and if so, the terminal corrects the text to reflect the suggested expressions. The user then confirms the final text and sends the email.

[1041] 7. Gather feedback and learn:

[1042] The server collects user feedback and reflects it in the generative AI algorithm to improve the accuracy of future suggestions, allowing the system to continuously learn and improve.

[1043] Specific examples

[1044] Specific examples are shown below.

[1045] Email text entered by the user:

[1046] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1047] Server analysis results:

[1048] Detected issues:

[1049] "Thank you for your hard work, John" (too friendly)

[1050] "I want to discuss this" (too direct)

[1051] Suggested fix:

[1052] Suggested fix:

[1053] "Thank you for your hard work, John-san" → "John-san"

[1054] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1055] Final confirmation from the user:

[1056] Dear John,

[1057] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[1058] thank you.

[1059] Yamada

[1060] This system allows users to create business emails appropriately and efficiently. Its multilingual capabilities also make it suitable for international business environments, making it suitable for a wide range of uses. Furthermore, the system can be customized to suit the language and cultural characteristics specific to the user's industry or company, allowing it to flexibly meet user needs.

[1061] The processing flow will be explained below.

[1062] Step 1:

[1063] The user composes a business email and types it into the terminal. For example, the user types the following:

[1064] "Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thanks in advance."

[1065] Step 2:

[1066] The device retrieves the text data entered by the user, which includes the entire body of the email.

[1067] Step 3:

[1068] The device sends the acquired text data to the server, which then prepares to begin analyzing the email text.

[1069] Step 4:

[1070] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions.

[1071] Step 5:

[1072] Based on the analysis results, the server lists inappropriate or misleading expressions, such as "Thank you for your hard work, John" or "I'd like to discuss this."

[1073] Step 6:

[1074] Based on the detection results, the server uses a generation AI to generate appropriate correction suggestions. The suggested correction for "Thank you for your hard work, John" is "Dear John," and the suggested correction for "I'd like to discuss this" is "I'd like to take the time to talk to you."

[1075] Step 7:

[1076] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions to the user based on the information received from the server.

[1077] Step 8:

[1078] The user checks the displayed correction suggestions and selects whether to accept them. If the user accepts the corrections, the terminal corrects the email text as suggested.

[1079] Step 9:

[1080] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[1081] Step 10:

[1082] After the final text is confirmed based on the user's input, the device sends the finalized data to the server, which stores the user's feedback and updates the AI's algorithm to improve the accuracy of future suggestions.

[1083] Example 1

[1084] 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."

[1085] When writing business emails, it is difficult for users to use appropriate expressions and avoid inappropriate ones. Furthermore, using expressions appropriate for different cultures and industries requires a great deal of effort and specialized knowledge. Furthermore, systems are required to continuously learn and improve based on user feedback. However, current systems have difficulty effectively resolving all of these issues, and the present invention aims to solve these problems.

[1086] 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.

[1087] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions, thereby enabling the user to efficiently create business email text using appropriate expressions.

[1088] "User" refers to an individual or organization that uses the System to create business emails.

[1089] "Text data" refers to a data format that includes text or sentence information entered by a user.

[1090] "Terminal" refers to the computing device used by a user to enter email text and review suggested revisions.

[1091] "Server" refers to a central computer system that analyzes text data, detects profanity, and generates correction suggestions.

[1092] "Analysis" refers to the process of examining the content of text data using machine learning and natural language processing techniques to identify inappropriate expressions.

[1093] "Inappropriate language" refers to expressions or phrases that are inappropriate to use in a business context.

[1094] "Correction suggestions" refer to alternative suggestions for changing detected inappropriate expressions into appropriate expressions.

[1095] "Feedback" refers to information about whether the user accepted the suggested revisions and whether the suggested revisions were appropriate.

[1096] "Generative AI models" refer to algorithms or machine learning models that generate text data and create correction suggestions.

[1097] A "prompt" refers to a command or question input to a generative AI model.

[1098] "Different languages" refers to languages ​​other than the user's native language.

[1099] "Customization" refers to the process of changing and adjusting system settings to suit the user's industry and corporate characteristics.

[1100] "Sending" refers to the electronic communication act used by a user to deliver the completed email text to the recipient.

[1101] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. This system consists of three entities: a terminal, a server, and a user. The function of each entity is explained below.

[1102] Device Features

[1103] The terminal is a device that allows users to input business emails. Terminals include PCs, tablets, smartphones, etc. Once a user has finished inputting the email, the terminal sends the text data to the server. It also displays any suggested revisions returned by the server.

[1104] Server Features

[1105] The server analyzes text data, detects inappropriate expressions, and generates correction suggestions. Specifically, it uses a natural language processing (NLP) library such as "Hugging Face" and a generative AI model (e.g., "OpenAI GPT"). The server analyzes the text data sent by the user using an NLP engine and detects inappropriate expressions. For detected inappropriate expressions, it uses the generative AI model to generate appropriate correction suggestions and sends them back to the device.

[1106] User operations

[1107] The user types the text of a business email into the terminal. For example, the user types the following text:

[1108] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1109] Once the input is complete, the user presses the send button to send the text data to the server. After that, the user checks the correction suggestions displayed on the terminal and selects whether to accept them. If the correction suggestions are accepted, the user makes a final confirmation and sends the email.

[1110] Specific examples

[1111] Specific examples are shown below.

[1112] Email text entered by the user:

[1113] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1114] Server analysis results:

[1115] Detected issues:

[1116] "Thank you for your hard work, John" (too friendly)

[1117] "I want to discuss this" (too direct)

[1118] Suggested fix:

[1119] Suggested fix:

[1120] "Thank you for your hard work, John-san" → "John-san"

[1121] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1122] Final confirmation from the user:

[1123] Dear John,

[1124] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[1125] thank you.

[1126] Yamada

[1127] Multilingual and customizable

[1128] The system is multilingual and can provide correction suggestions in different languages. It can also be customized to fit the language and cultural characteristics specific to the user's industry or company. This allows for more accurate correction suggestions.

[1129] Gathering feedback and learning

[1130] The server collects user feedback and continuously trains the generative AI model based on it, thereby improving the accuracy of future suggestions.

[1131] As described above, the system of the present invention provides powerful support for users to create business emails quickly and appropriately.

[1132] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1133] Step 1:

[1134] User-generated email content

[1135] A user uses a terminal to type the text of a business email. For example, the user types the following sentence:

[1136] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1137] Once this input is complete, the user presses the send button to prepare to send the text data from the terminal to the server.

[1138] Input: Email text entered by the user

[1139] Output: Text data of the entered email

[1140] Step 2:

[1141] Sending email content

[1142] The device sends the text data entered by the user to the server. At this time, the device sends the email body data to the server using an HTTP request. The request includes the user ID and the email body.

[1143] Input: Text data entered into the terminal

[1144] Output: Text data sent to the server

[1145] Step 3:

[1146] Text analysis and profanity detection

[1147] The server receives the text data and analyzes it using a natural language processing (NLP) engine. This analysis process uses NLP libraries such as "Hugging Face." The server analyzes the content of the email and detects inappropriate expressions. For example, it detects that the expression "Thank you for your hard work, John" is too familiar.

[1148] Input: Text data sent to the server

[1149] Output: Analysis results and a list of inappropriate expressions detected

[1150] Step 4:

[1151] Generate correction suggestions

[1152] The server uses a generative AI model (e.g., "OpenAI GPT") to generate appropriate correction suggestions for the detected inappropriate expressions. Specifically, for the expression "Thank you for your hard work, John-san," the server makes a correction suggestion such as "John-san."

[1153] Input: Profanity detection list

[1154] Output: List of suggested fixes

[1155] Step 5:

[1156] View suggested fixes

[1157] The terminal displays the correction suggestions sent back from the server to the user. The correction suggestions are displayed in a visually intuitive manner, and the user can confirm them and choose whether to accept them.

[1158] Input: A list of suggested revisions sent from the server

[1159] Output: A list of suggested revisions that is displayed to the user

[1160] Step 6:

[1161] Final confirmation by the user

[1162] The user checks the displayed correction suggestions and decides whether to accept them. If they accept, the device corrects the text to the suggested expression. The user then checks the final text and presses the send button to send the email. For example, the corrected text might look like this:

[1163] Dear John,

[1164] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[1165] thank you.

[1166] Yamada

[1167] Input: A list of suggested revisions shown to the user

[1168] Output: The final email message that the user has confirmed and corrected

[1169] Step 7:

[1170] Gathering feedback and learning

[1171] The server collects feedback from users and sends back information such as how the user accepted the correction suggestions and whether the suggestions were appropriate. The server uses this feedback to update the generative AI model and improve the accuracy of future correction suggestions.

[1172] Input: User feedback information

[1173] Output: Update the generative AI model based on feedback

[1174] (Application example 1)

[1175] 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."

[1176] In today's business emails and chats, the text data entered by users often contains inappropriate expressions. Such expressions can lead to misunderstandings and misinterpretations in business communications, negatively impacting a company's credibility and efficiency. Furthermore, in an international business environment, the need for multilingual support makes correcting inappropriate expressions even more complicated. Furthermore, when real-time responses are required, the ability to quickly and accurately correct expressions is essential.

[1177] 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.

[1178] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for analyzing the text data in real time in different response formats and correcting the text data to appropriate expressions, means for generating correction suggestions for the detected inappropriate expressions, means for generating the correction suggestions using a generative AI model, means for displaying the correction suggestions to the user, and means for improving the accuracy of the correction suggestions based on user feedback. This reduces misunderstandings and misinterpretations in business communication, improves corporate reliability and efficiency, and enables multilingual support.

[1179] "User" refers to any individual or corporation that uses the system.

[1180] "Text data" refers to information in the form of a string of characters entered by the user.

[1181] "Analysis" refers to the process of processing and analyzing the content of text data.

[1182] "Inappropriate expressions" refer to expressions that may lead to misunderstandings or misinterpretations in business communications.

[1183] "Correction proposal" refers to an alternative proposal to convert inappropriate language into appropriate language.

[1184] "Display" refers to visually showing the suggested revisions on the user's terminal.

[1185] "Feedback" refers to the opinions and reactions of users to suggested revisions.

[1186] "Accuracy improvement" refers to improvement activities carried out to increase the accuracy and effectiveness of proposals.

[1187] "Different response formats" refers to response methods that are applied in different formats and situations.

[1188] "Real time" refers to the concept of time when data is processed as soon as it is entered.

[1189] "Generative AI model" refers to a model that uses artificial intelligence to generate and modify text data.

[1190] The system of the present invention analyzes text data entered by users, detects inappropriate expressions, and provides appropriate correction suggestions. This system consists of three entities: a terminal, a server, and a user.

[1191] System Configuration

[1192] Hardware and Software Configuration

[1193] Device: The device used by the user, such as a smartphone, tablet, or desktop PC.

[1194] Server: Provides high-performance computing power to analyze text data and generate correction suggestions.

[1195] software:

[1196] OpenAI API: Generates correction suggestions using the generative AI model GPT-3.

[1197] Python: A programming language for implementing the overall system logic.

[1198] Regular expressions (re): Used to process and analyze text data.

[1199] Data processing and calculation

[1200] 1. Acquiring text data:

[1201] A user inputs text data using a terminal, and the text data is transmitted from the terminal to a server.

[1202] 2. Text data analysis:

[1203] The server analyzes the received text data using the OpenAI API. First, it detects inappropriate expressions in the text data.

[1204] 3. Generate correction suggestions:

[1205] For detected inappropriate expressions, a generative AI model is used to generate appropriate correction suggestions.

[1206] 4. View suggested revisions:

[1207] The revision suggestions generated by the server are sent to the terminal and displayed to the user.

[1208] 5. Gathering Feedback:

[1209] The accuracy of suggestions is improved by sending user feedback information to the server and reflecting it in the generative AI model.

[1210] Specific examples

[1211] Example 1: Text analysis of business emails

[1212] A user types the following email on their smartphone:

[1213] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1214] Server analysis results:

[1215] Detected issues:

[1216] "Thank you for your hard work, person in charge" (too friendly)

[1217] "I'd like to discuss this" (too direct)

[1218] Suggested fix:

[1219] Suggested fix:

[1220] "Thank you for your hard work, person in charge" → "Dear person in charge"

[1221] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1222] Final confirmation:

[1223] To the person in charge

[1224] I would like to take the time to speak with you about the meeting we had the other day. I would appreciate it if you could let me know a convenient time and date for us to meet.

[1225] thank you.

[1226] Yamada

[1227] Prompt Sentence Examples

[1228] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[1229] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1230] Fixes:

[1231] This system allows users to quickly correct inappropriate language used in business emails and chats, ensuring appropriate communication. It also supports multiple languages ​​and can be customized based on industry characteristics, making it applicable to a wide range of uses. This is expected to reduce misunderstandings and misinterpretations in business, improving corporate reliability and efficiency.

[1232] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1233] Step 1:

[1234] Users use their devices to input the text of business emails or chat messages. Specifically, users enter text data into an input field on a smartphone, tablet, or desktop PC, and then press the send button. The input data is, for example, in the following format:

[1235] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1236] Input: Text data entered by the user into the terminal.

[1237] Output: Text data ready to be sent to the terminal

[1238] Step 2:

[1239] The terminal sends the entered text data to the server, where the data is transferred securely using protocols such as HTTPS. The entered text is sent in a format that the server can parse.

[1240] Input: Text data sent from the device to the server

[1241] Output: Text data received by the server

[1242] Step 3:

[1243] The server sends the received text data to the OpenAI API for analysis. The server generates a prompt to be used for analysis and passes the text data to the natural language processing engine. At this time, the following prompt is generated:

[1244] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[1245] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1246] Fixes:

[1247] Input: Text data received by the server and prompts

[1248] Output: Analysis results by OpenAI API

[1249] Step 4:

[1250] The OpenAI API uses a generative AI model to detect inappropriate expressions and generate appropriate correction suggestions, such as "Thank you for your hard work, Mr. / Ms. Person in Charge."

[1251] Input: Prompt and text data received by the OpenAI API

[1252] Output: Detected profanities and suggested corrections

[1253] Step 5:

[1254] The server receives the correction suggestions returned by the OpenAI API and sends them to the device, where they can be viewed by the user. The correction suggestions are displayed in the following format:

[1255] Suggested fix:

[1256] "Thank you for your hard work, person in charge" → "Dear person in charge"

[1257] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1258] Input: Analysis results returned from the OpenAI API

[1259] Output: Suggested fixes sent to terminal

[1260] Step 6:

[1261] The terminal displays the received correction suggestions to the user, who then checks the displayed correction suggestions and selects whether to accept them. The selection operation is performed via buttons or an interface on the terminal.

[1262] Input: Suggested corrections sent to the terminal

[1263] Output: Correction suggestions and user selections displayed on the device

[1264] Step 7:

[1265] The user decides whether to accept the suggested revisions and confirms the result on the terminal. If the revision is accepted, the terminal corrects the text to the suggested expression and displays the final text.

[1266] Input: User accepts or rejects suggested revisions

[1267] Output: Final corrected text

[1268] Step 8:

[1269] The server collects user feedback and reflects it in the generative AI model algorithm, which improves the accuracy of correction suggestions from the next time onwards.

[1270] Input: User feedback information

[1271] Output: Improved generative AI model algorithm

[1272] This allows users to correct inappropriate expressions in business emails and chat messages to appropriate expressions, enabling efficient and effective communication.

[1273] 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.

[1274] The present invention relates to a system that analyzes text data, detects inappropriate expressions, and provides correction suggestions when a user composes a business email, in addition to combining it with an emotion engine that recognizes the user's emotions and reflects them in the correction suggestions. Below, we will explain the program processing and specific operation examples of this system.

[1275] Program Processing Overview

[1276] The system mainly consists of four entities: the terminal, the server, the user, and the emotion engine. The individual roles and operations of each entity are as follows:

[1277] User-generated email message:

[1278] The user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[1279] example:

[1280] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1281] Send email content:

[1282] The device acquires text data entered by the user. At this time, the device uses an emotion engine to recognize the emotion contained in the text data entered by the user. The acquired text data and emotion information are sent to the server.

[1283] Text analysis and profanity detection:

[1284] The server inputs the received text data into an NLP engine, which analyzes the email text and detects inappropriate expressions. For example, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[1285] Emotion recognition and suggested correction adjustments:

[1286] The server generates correction suggestions based on the emotional information recognized by the emotion engine. In this case, the strong emotional expression "I would like to check this urgently" is adjusted to a softer expression such as "I apologize for bothering you during your busy schedule, but could you please check it?"

[1287] View suggested fixes:

[1288] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions received from the server to the user, who then checks the displayed suggestions and chooses whether to accept them.

[1289] Final user confirmation:

[1290] The user decides whether to accept the suggested revisions, and if so, the terminal will revise the email text as suggested. The user then checks the final text and, if there are no problems, confirms sending the email.

[1291] Gathering feedback and learning:

[1292] The server collects user feedback and reflects it in the generative AI and emotion engine algorithms to improve the accuracy of future suggestions and emotion recognition, allowing the system to continuously learn and improve.

[1293] Specific example (utilizing emotion engine)

[1294] A specific example is given below.

[1295] Email text entered by the user:

[1296] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1297] Emotion recognition results by emotion engine:

[1298] Recognized emotions: impatience, hurry

[1299] Server analysis results:

[1300] Detected issues:

[1301] "I want to check this urgently" (impression of urgency)

[1302] "Please reply as soon as possible" (Pressure)

[1303] Suggested fix:

[1304] Suggested fix:

[1305] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[1306] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[1307] Final confirmation from the user:

[1308] Dear John,

[1309] I know you are busy, but could you please check on the meeting we had the other day? I would appreciate it if you could check it when you have time.

[1310] thank you.

[1311] Yamada

[1312] This system allows users to create emails appropriately and efficiently, and an emotion engine provides revision suggestions that reflect the user's emotions, enabling more human-like communication.The system also supports multiple languages ​​and has customization functions based on industry-specific language usage, making it possible to build a flexible system that meets the diverse needs of users.

[1313] The processing flow will be explained below.

[1314] Step 1:

[1315] The user composes a business email and types it into the terminal. For example, the user types the following:

[1316] "John, I need to confirm something about the meeting we had the other day. Please respond as soon as possible. Thank you."

[1317] Step 2:

[1318] The device retrieves the text data entered by the user, which includes the entire body of the email.

[1319] Step 3:

[1320] The device sends the acquired text data to the emotion engine, which analyzes the user's emotions contained in the text data. For example, the emotion engine recognizes the user's impatience from "I want to check this urgently."

[1321] Step 4:

[1322] The device sends the entire text data along with the emotion analysis results to the server, which then prepares to begin analyzing the email text.

[1323] Step 5:

[1324] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions. In this case, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[1325] Step 6:

[1326] The server generates appropriate correction suggestions for detected inappropriate expressions based on the emotion analysis results from the emotion engine. For example, a strong expression such as "I would like to check this urgently" can be corrected to a milder expression such as "I apologize for bothering you during your busy schedule, but could you please check this?"

[1327] Step 7:

[1328] The server sends the generated revision suggestions together with the emotion analysis results back to the device, which then displays the revision suggestions received from the server to the user.

[1329] Step 8:

[1330] The user checks the displayed correction suggestions and decides whether to accept them. If the user accepts the suggestions, the terminal will correct the email text as suggested.

[1331] Step 9:

[1332] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[1333] Step 10:

[1334] After the user's operation is confirmed, the device will send the confirmation message and feedback information to the server, which will then collect the user's feedback information and use it as learning data to improve the accuracy of the AI ​​algorithm and emotion engine.

[1335] This process allows users to create emails appropriately and efficiently, and by utilizing the emotion engine, it enables more human-like communication that takes into account the user's emotions.

[1336] Example 2

[1337] 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."

[1338] Conventional business email writing support systems focus on detecting inappropriate expressions and suggesting corrections by analyzing text data, but because they are unable to take user emotions into account, they have the problem of difficulty in providing flexible correction suggestions that are in line with emotions.In addition, mechanisms for improving the accuracy of correction suggestions based on user feedback are limited to a few cases, and there is a problem that long-term learning and improvement are not sufficiently carried out.

[1339] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1340] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for recognizing the user's emotions using an emotion recognition engine, means for adjusting revision suggestions based on the recognized emotions, means for generating revision suggestions for the detected inappropriate expressions, means for displaying the revision suggestions to the user, and means for improving the accuracy of the revision suggestions based on user feedback, thereby enabling flexible and appropriate revision suggestions that reflect the user's emotions.

[1341] "User" refers to an individual or organization that uses the system to create business emails.

[1342] "Text data" refers to character string information entered by the user, and includes the email text itself.

[1343] An "emotion recognition engine" refers to software that analyzes and recognizes user emotions contained in text data.

[1344] "Correction suggestions" refer to alternative expressions generated by the system to correct inappropriate expressions in text data.

[1345] "Feedback" refers to the user's evaluation and reaction to the correction suggestions provided by the system.

[1346] A "server" refers to a central device that manages and controls the processes of the entire system and processes and manages various data.

[1347] An "NLP engine" refers to software or algorithms for natural language processing.

[1348] These definitions help clarify the role and function of each element of the system.

[1349] This invention is a system that analyzes text data, detects inappropriate expressions, and provides suggestions for correction when a user composes a business email. Furthermore, by combining it with an emotion engine that recognizes the user's emotions and reflects them in the suggested corrections, it becomes possible to create emails that are more flexible and human-like.

[1350] System configuration

[1351] This system mainly consists of four entities: the terminal, the server, the user, and the emotion engine.

[1352] Terminal: The device (e.g., PC, smartphone) on which the user enters the email text.

[1353] Server: A central processing unit that analyzes text data and generates profanity detection and correction suggestions.

[1354] User: A person or organization that writes business emails.

[1355] Emotion engine: Software for analyzing user emotions (e.g., IBM Watson Tone Analyzer).

[1356] Text data acquisition and sentiment analysis

[1357] A user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[1358] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1359] The device receives text data entered by the user and analyzes the user's emotions contained in the text data using an emotion recognition engine. The analyzed emotion information is then sent to the server together with the text data.

[1360] Text analysis and profanity detection

[1361] The server inputs the received text data into a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to analyze the email text. This is where inappropriate expressions are detected. For example, expressions such as "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong.

[1362] Generate and refine correction suggestions

[1363] The server generates correction suggestions based on the analysis results of the emotion engine. In this case, expressions that are recognized as having strong emotions are adjusted to softer expressions. For example, "I would like to check this urgently" is corrected to "I apologize for bothering you during your busy schedule, but could you please check it?"

[1364] View and finalize suggested revisions

[1365] The server sends the generated revision suggestions to the terminal. The terminal displays the revision suggestions received from the server to the user. The user checks the displayed suggestions and chooses whether to accept them. If accepted, the terminal automatically corrects the email text according to the suggestions. Finally, the user checks the final text and sends it if there are no problems.

[1366] Gathering feedback and training the system

[1367] The server collects user feedback and reflects it in the generative AI model and emotion engine algorithms, which improves the accuracy of future suggestions and emotion recognition.

[1368] Specific examples (prompt sentence examples)

[1369] For example:

[1370] Email text entered by the user

[1371] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1372] Emotion recognition results by emotion engine

[1373] Recognized emotions: impatience, hurry

[1374] Revision suggestions

[1375] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[1376] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[1377] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1378] Program processing flow

[1379] The program processing flow of this system will be explained in detail below, divided into steps.

[1380] Step 1:

[1381] The user enters the text of a business email into the terminal.

[1382] Input: Text data entered by the user into the terminal.

[1383] Action: A user enters text using a keyboard or touchscreen.

[1384] Output: Text data saved in the device.

[1385] Step 2:

[1386] The device acquires text data entered by the user and uses an emotion recognition engine to analyze the user's emotions contained in the text data.

[1387] Input: The text data entered in step 1.

[1388] How it works: The emotion recognition engine extracts emotional information from text (e.g., impatience, urgency).

[1389] Output: User emotion information and text data.

[1390] Step 3:

[1391] The device transmits the acquired text data and emotional information to the server.

[1392] Input: Text data and emotion information.

[1393] Action: The operation of sending data to a server, transferring data over a network.

[1394] Output: Text data and emotion information received by the server.

[1395] Step 4:

[1396] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the text data and detects inappropriate expressions.

[1397] Input: Text data and emotion information sent from the device.

[1398] How it works: An NLP engine analyzes text to detect strong or inappropriate language.

[1399] Output: Text data with profanity flagged.

[1400] Step 5:

[1401] The server generates correction suggestions for detected inappropriate expressions based on the results of the emotion recognition engine.

[1402] Input: Text data with inappropriate language flagged, sentiment information.

[1403] How it works: The revision suggestion algorithm generates revision suggestions that take user sentiment into account.

[1404] Output: Text data with suggested corrections.

[1405] Step 6:

[1406] The server sends the generated revision suggestions to the terminal.

[1407] Input: Text data with suggested corrections.

[1408] Action: The act of sending data to a device, transferring data over a network.

[1409] Output: Correction suggestions received on the device.

[1410] Step 7:

[1411] The terminal displays the revision suggestions received from the server to the user, who then checks the revision suggestions and decides whether to accept them.

[1412] Input: The correction proposal received.

[1413] Behavior: Displays suggested revisions in the user interface.

[1414] Output: User confirmation and choice (accept or not).

[1415] Step 8:

[1416] If the user accepts the suggested revisions, the device automatically corrects the email text as suggested. The user then finally checks the text and, if there are no problems, sends the email.

[1417] Input: The correction suggestion that the user accepted.

[1418] Action: Adjust the email text based on the suggested revisions and ask the user for final confirmation. Send the email.

[1419] Output: The final revised email text and the email sent.

[1420] Step 9:

[1421] The server collects user feedback information and reflects it in the generative AI model and emotion engine algorithms.

[1422] Input: User feedback information.

[1423] How it works: Feedback information is fed into the algorithm and used as training data for the model.

[1424] Output: Improved generative AI models and emotion engines.

[1425] (Application example 2)

[1426] 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."

[1427] When dealing with customers, staff are required to use appropriate language while taking into consideration the customer's feelings, but this can be difficult to judge on the ground, and inappropriate language can be used, especially when dealing with complaints or inquiries. This can lead to a decrease in customer satisfaction and the risk of problems arising. A system to effectively resolve this is needed.

[1428] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1429] In this invention, the server includes means for acquiring text data and voice data input by a user, means for analyzing the text data and voice data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions based on emotion information, thereby enabling support for staff to use appropriate expressions in real time while dealing with customers.

[1430] "Text data" refers to character string information entered by a user. It is generally saved as recorded character string information or a document.

[1431] "Voice Data" means a digital recording of a user's spoken voice, often converted to text using speech recognition technology.

[1432] "Analysis" refers to the act of processing input data using an algorithm to extract specific information or patterns.

[1433] "Inappropriate expressions" refer to expressions that contain inappropriate meanings or emotions in communication and may cause misunderstandings or trouble.

[1434] "Emotional information" refers to emotional elements extracted from the text or voice of a user or customer, and includes emotions such as fear, impatience, and joy.

[1435] "Correction proposals" refer to proposals that suggest appropriate expressions or ways of responding based on the analysis results.

[1436] "Real time" refers to a state in which the time between when data is generated and when it is processed is extremely short, and the data is processed almost simultaneously.

[1437] "Display" refers to the act of visually presenting the analysis results and correction suggestions to the user, outputting them on a display or screen.

[1438] "Feedback" refers to the evaluations and reactions that users provide to a system, which are used to improve and learn from the system.

[1439] "Device" refers to a hardware device with a specific function, including smartphones and robots.

[1440] A "server" refers to a computer system that provides services to clients over a network.

[1441] A "system" refers to a collection of multiple components working together to achieve a specific function.

[1442] The present invention relates to a system that analyzes input text data and voice data when a user is dealing with a customer, and provides appropriate correction suggestions based on emotion information. A specific embodiment of the system includes the following steps.

[1443] The server receives text and voice data sent from the user's device (smartphone or robot). At that time, it uses voice recognition software and an NLP (natural language processing) engine to convert the voice data into text data. Specifically, it uses the "SpeechRecognition" module for voice recognition and the "TextBlob" module for sentiment analysis of the text data.

[1444] The server then uses pre-defined rule-based filters and machine learning algorithms to detect inappropriate content based on the analyzed text data. If an inappropriate content is detected, the server generates appropriate correction suggestions based on the sentiment information. For example, the phrase "I would like to check this urgently" is converted into the suggestion "I apologize for bothering you during your busy schedule, but could you please check it?"

[1445] The generated correction suggestions are displayed in real time on the user's device. The user can review the suggestions and adopt the corrected expressions as needed. In addition, the user's feedback information is sent to the server and used to improve the accuracy of the correction suggestions.

[1446] As a concrete example, the following prompt sentence is used:

[1447] Example prompt sentence:

[1448] Speak specific customer interaction text. Example: "Please check availability of this item immediately."

[1449] It analyzes sentiment and outputs appropriate correction suggestions in the following format: Example correction: "I'm sorry to bother you, but could you please wait a moment while I check this?"

[1450] This system enables staff to use appropriate expressions in real time while interacting with customers, which is expected to improve customer satisfaction. The hybrid system combines voice input and text analysis to enable quick and flexible responses.

[1451] Furthermore, the system continuously learns from feedback and improves the accuracy of correction suggestions, making it possible to adapt to the diverse needs of users.

[1452] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1453] Step 1:

[1454] The user inputs into the device by voice or text.

[1455] Specifically, it uses the smartphone's microphone to capture voice data, and also allows for manual text input.

[1456] Input: Audio or text data

[1457] Output: User input audio or text data

[1458] Step 2:

[1459] The terminal converts the voice data into text data.

[1460] Specifically, the speech data is converted into text data using the SpeechRecognition module.

[1461] Input: Audio data

[1462] Output: Converted text data

[1463] Step 3:

[1464] The terminal transmits the text data to the server.

[1465] As a specific operation, the converted text data is sent to a server via the Internet.

[1466] Input: Text data

[1467] Output: Text data sent to the server

[1468] Step 4:

[1469] The server analyzes the text data and extracts emotional information.

[1470] Specifically, it uses the TextBlob module to analyze the polarity (emotional polarity) and subjectivity of text data.

[1471] Input: Text data

[1472] Output: Emotional information (e.g. polarity, subjectivity)

[1473] Step 5:

[1474] The server detects profanity.

[1475] Specifically, it uses predefined filters and machine learning models to identify inappropriate language.

[1476] Input: Text data

[1477] Output: A list of profanities

[1478] Step 6:

[1479] The server generates revision suggestions based on the emotion information.

[1480] Specifically, it suggests alternative expressions for detected inappropriate expressions and adjusts them taking into account the user's emotions and context.

[1481] Input: Inappropriate language, dishonest behavior

[1482] Output: Sentiment-based revision suggestions

[1483] Step 7:

[1484] The server sends the revision suggestions to the terminal.

[1485] As a specific operation, the generated revision proposal is sent back to the user's terminal via the Internet.

[1486] Input: Proposed correction

[1487] Output: Suggested fixes sent to terminal

[1488] Step 8:

[1489] The terminal displays the suggested revisions to the user.

[1490] Specifically, the proposed corrections are displayed on the screen so that the user can confirm them.

[1491] Input: Proposed correction

[1492] Output: Displayed suggested fixes

[1493] Step 9:

[1494] The user reviews the proposed corrections and selects a response.

[1495] As a specific operation, the user either accepts the suggested revision or creates a new message that reflects the suggested revision.

[1496] Input: Displayed suggested corrections

[1497] Output: User selected or modified text data

[1498] Step 10:

[1499] The server collects user feedback and uses it to improve the accuracy of the system.

[1500] Specifically, user choices and feedback information are stored in a database and used as training data for generative AI models and algorithms.

[1501] Input: User feedback

[1502] Output: Improved correction suggestion algorithm

[1503] 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.

[1504] 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.

[1505] 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.

[1506] [Fourth embodiment]

[1507] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1508] 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.

[1509] 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).

[1510] 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.

[1511] 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.

[1512] 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).

[1513] 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.

[1514] 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.

[1515] 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.

[1516] 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.

[1517] 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.

[1518] 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.

[1519] 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."

[1520] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. Below, the program processing and specific examples of this system will be described.

[1521] Program Processing Overview

[1522] The system mainly consists of three entities: the terminal, the server, and the user. The roles and operations of each entity are as follows:

[1523] 1. User creates email message:

[1524] The user types the text of a business email into the terminal. For example, the user types the following text to a business partner:

[1525] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1526] 2. Sending email content:

[1527] The device sends the entered text data to the server, which includes the entire body of the email.

[1528] 3. Text analysis and profanity detection:

[1529] The server analyzes the received text data using a natural language processing (NLP) engine to detect inappropriate expressions, such as "Thank you for your hard work, John-san."

[1530] 4. Generate correction suggestions:

[1531] The server generates appropriate expressions for the inappropriate expressions it detects. In this case, the suggested correction for "Thank you for your hard work, John-san" would be "John-san."

[1532] 5. View suggested revisions:

[1533] The terminal displays the correction suggestions sent back from the server to the user, who then checks the suggestions and chooses whether to accept them.

[1534] 6. Final User Review:

[1535] The user decides whether to accept the suggested revisions, and if so, the terminal corrects the text to reflect the suggested expressions. The user then confirms the final text and sends the email.

[1536] 7. Gather feedback and learn:

[1537] The server collects user feedback and reflects it in the generative AI algorithm to improve the accuracy of future suggestions, allowing the system to continuously learn and improve.

[1538] Specific examples

[1539] Specific examples are shown below.

[1540] Email text entered by the user:

[1541] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1542] Server analysis results:

[1543] Detected issues:

[1544] "Thank you for your hard work, John" (too friendly)

[1545] "I want to discuss this" (too direct)

[1546] Suggested fix:

[1547] Suggested fix:

[1548] "Thank you for your hard work, John-san" → "John-san"

[1549] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1550] Final confirmation from the user:

[1551] Dear John,

[1552] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[1553] thank you.

[1554] Yamada

[1555] This system allows users to create business emails appropriately and efficiently. Its multilingual capabilities also make it suitable for international business environments, making it suitable for a wide range of uses. Furthermore, the system can be customized to suit the language and cultural characteristics specific to the user's industry or company, allowing it to flexibly meet user needs.

[1556] The processing flow will be explained below.

[1557] Step 1:

[1558] The user composes a business email and types it into the terminal. For example, the user types the following:

[1559] "Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thanks in advance."

[1560] Step 2:

[1561] The device retrieves the text data entered by the user, which includes the entire body of the email.

[1562] Step 3:

[1563] The device sends the acquired text data to the server, which then prepares to begin analyzing the email text.

[1564] Step 4:

[1565] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions.

[1566] Step 5:

[1567] Based on the analysis results, the server lists inappropriate or misleading expressions, such as "Thank you for your hard work, John" or "I'd like to discuss this."

[1568] Step 6:

[1569] Based on the detection results, the server uses a generation AI to generate appropriate correction suggestions. The suggested correction for "Thank you for your hard work, John" is "Dear John," and the suggested correction for "I'd like to discuss this" is "I'd like to take the time to talk to you."

[1570] Step 7:

[1571] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions to the user based on the information received from the server.

[1572] Step 8:

[1573] The user checks the displayed correction suggestions and selects whether to accept them. If the user accepts the corrections, the terminal corrects the email text as suggested.

[1574] Step 9:

[1575] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[1576] Step 10:

[1577] After the final text is confirmed based on the user's input, the device sends the finalized data to the server, which stores the user's feedback and updates the AI's algorithm to improve the accuracy of future suggestions.

[1578] Example 1

[1579] 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."

[1580] When writing business emails, it is difficult for users to use appropriate expressions and avoid inappropriate ones. Furthermore, using expressions appropriate for different cultures and industries requires a great deal of effort and specialized knowledge. Furthermore, systems are required to continuously learn and improve based on user feedback. However, current systems have difficulty effectively resolving all of these issues, and the present invention aims to solve these problems.

[1581] 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.

[1582] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions, thereby enabling the user to efficiently create business email text using appropriate expressions.

[1583] "User" refers to an individual or organization that uses the System to create business emails.

[1584] "Text data" refers to a data format that includes text or sentence information entered by a user.

[1585] "Terminal" refers to the computing device used by a user to enter email text and review suggested revisions.

[1586] "Server" refers to a central computer system that analyzes text data, detects profanity, and generates correction suggestions.

[1587] "Analysis" refers to the process of examining the content of text data using machine learning and natural language processing techniques to identify inappropriate expressions.

[1588] "Inappropriate language" refers to expressions or phrases that are inappropriate to use in a business context.

[1589] "Correction suggestions" refer to alternative suggestions for changing detected inappropriate expressions into appropriate expressions.

[1590] "Feedback" refers to information about whether the user accepted the suggested revisions and whether the suggested revisions were appropriate.

[1591] "Generative AI models" refer to algorithms or machine learning models that generate text data and create correction suggestions.

[1592] A "prompt" refers to a command or question input to a generative AI model.

[1593] "Different languages" refers to languages ​​other than the user's native language.

[1594] "Customization" refers to the process of changing and adjusting system settings to suit the user's industry and corporate characteristics.

[1595] "Sending" refers to the electronic communication act used by a user to deliver the completed email text to the recipient.

[1596] The system of the present invention analyzes text data when a user composes a business email, detects inappropriate expressions, and provides suggestions for correction. This system consists of three entities: a terminal, a server, and a user. The function of each entity is explained below.

[1597] Device Features

[1598] The terminal is a device that allows users to input business emails. Terminals include PCs, tablets, smartphones, etc. Once a user has finished inputting the email, the terminal sends the text data to the server. It also displays any suggested revisions returned by the server.

[1599] Server Features

[1600] The server analyzes text data, detects inappropriate expressions, and generates correction suggestions. Specifically, it uses a natural language processing (NLP) library such as "Hugging Face" and a generative AI model (e.g., "OpenAI GPT"). The server analyzes the text data sent by the user using an NLP engine and detects inappropriate expressions. For detected inappropriate expressions, it uses the generative AI model to generate appropriate correction suggestions and sends them back to the device.

[1601] User operations

[1602] The user types the text of a business email into the terminal. For example, the user types the following text:

[1603] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1604] Once the input is complete, the user presses the send button to send the text data to the server. After that, the user checks the correction suggestions displayed on the terminal and selects whether to accept them. If the correction suggestions are accepted, the user makes a final confirmation and sends the email.

[1605] Specific examples

[1606] Specific examples are shown below.

[1607] Email text entered by the user:

[1608] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1609] Server analysis results:

[1610] Detected issues:

[1611] "Thank you for your hard work, John" (too friendly)

[1612] "I want to discuss this" (too direct)

[1613] Suggested fix:

[1614] Suggested fix:

[1615] "Thank you for your hard work, John-san" → "John-san"

[1616] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1617] Final confirmation from the user:

[1618] Dear John,

[1619] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[1620] thank you.

[1621] Yamada

[1622] Multilingual and customizable

[1623] The system is multilingual and can provide correction suggestions in different languages. It can also be customized to fit the language and cultural characteristics specific to the user's industry or company. This allows for more accurate correction suggestions.

[1624] Gathering feedback and learning

[1625] The server collects user feedback and continuously trains the generative AI model based on it, thereby improving the accuracy of future suggestions.

[1626] As described above, the system of the present invention provides powerful support for users to create business emails quickly and appropriately.

[1627] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1628] Step 1:

[1629] User-generated email content

[1630] A user uses a terminal to type the text of a business email. For example, the user types the following sentence:

[1631] Thank you for your hard work, John. I'd like to discuss our meeting the other day. Please let me know a time that's convenient for you. Thank you.

[1632] Once this input is complete, the user presses the send button to prepare to send the text data from the terminal to the server.

[1633] Input: Email text entered by the user

[1634] Output: Text data of the entered email

[1635] Step 2:

[1636] Sending email content

[1637] The device sends the text data entered by the user to the server. At this time, the device sends the email body data to the server using an HTTP request. The request includes the user ID and the email body.

[1638] Input: Text data entered into the terminal

[1639] Output: Text data sent to the server

[1640] Step 3:

[1641] Text analysis and profanity detection

[1642] The server receives the text data and analyzes it using a natural language processing (NLP) engine. This analysis process uses NLP libraries such as "Hugging Face." The server analyzes the content of the email and detects inappropriate expressions. For example, it detects that the expression "Thank you for your hard work, John" is too familiar.

[1643] Input: Text data sent to the server

[1644] Output: Analysis results and a list of inappropriate expressions detected

[1645] Step 4:

[1646] Generate correction suggestions

[1647] The server uses a generative AI model (e.g., "OpenAI GPT") to generate appropriate correction suggestions for the detected inappropriate expressions. Specifically, for the expression "Thank you for your hard work, John-san," the server makes a correction suggestion such as "John-san."

[1648] Input: Profanity detection list

[1649] Output: List of suggested fixes

[1650] Step 5:

[1651] View suggested fixes

[1652] The terminal displays the correction suggestions sent back from the server to the user. The correction suggestions are displayed in a visually intuitive manner, and the user can confirm them and choose whether to accept them.

[1653] Input: A list of suggested revisions sent from the server

[1654] Output: A list of suggested revisions that is displayed to the user

[1655] Step 6:

[1656] Final confirmation by the user

[1657] The user checks the displayed correction suggestions and decides whether to accept them. If they accept, the device corrects the text to the suggested expression. The user then checks the final text and presses the send button to send the email. For example, the corrected text might look like this:

[1658] Dear John,

[1659] I would like to take the time to speak with you about our recent meeting. I would appreciate it if you could let me know a time that would be convenient for you.

[1660] thank you.

[1661] Yamada

[1662] Input: A list of suggested revisions shown to the user

[1663] Output: The final email message that the user has confirmed and corrected

[1664] Step 7:

[1665] Gathering feedback and learning

[1666] The server collects feedback from users and sends back information such as how the user accepted the correction suggestions and whether the suggestions were appropriate. The server uses this feedback to update the generative AI model and improve the accuracy of future correction suggestions.

[1667] Input: User feedback information

[1668] Output: Update the generative AI model based on feedback

[1669] (Application example 1)

[1670] 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."

[1671] In today's business emails and chats, the text data entered by users often contains inappropriate expressions. Such expressions can lead to misunderstandings and misinterpretations in business communications, negatively impacting a company's credibility and efficiency. Furthermore, in an international business environment, the need for multilingual support makes correcting inappropriate expressions even more complicated. Furthermore, when real-time responses are required, the ability to quickly and accurately correct expressions is essential.

[1672] 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.

[1673] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for analyzing the text data in real time in different response formats and correcting the text data to appropriate expressions, means for generating correction suggestions for the detected inappropriate expressions, means for generating the correction suggestions using a generative AI model, means for displaying the correction suggestions to the user, and means for improving the accuracy of the correction suggestions based on user feedback. This reduces misunderstandings and misinterpretations in business communication, improves corporate reliability and efficiency, and enables multilingual support.

[1674] "User" refers to any individual or corporation that uses the system.

[1675] "Text data" refers to information in the form of a string of characters entered by the user.

[1676] "Analysis" refers to the process of processing and analyzing the content of text data.

[1677] "Inappropriate expressions" refer to expressions that may lead to misunderstandings or misinterpretations in business communications.

[1678] "Correction proposal" refers to an alternative proposal to convert inappropriate language into appropriate language.

[1679] "Display" refers to visually showing the suggested revisions on the user's terminal.

[1680] "Feedback" refers to the opinions and reactions of users to suggested revisions.

[1681] "Accuracy improvement" refers to improvement activities carried out to increase the accuracy and effectiveness of proposals.

[1682] "Different response formats" refers to response methods that are applied in different formats and situations.

[1683] "Real time" refers to the concept of time when data is processed as soon as it is entered.

[1684] "Generative AI model" refers to a model that uses artificial intelligence to generate and modify text data.

[1685] The system of the present invention analyzes text data entered by users, detects inappropriate expressions, and provides appropriate correction suggestions. This system consists of three entities: a terminal, a server, and a user.

[1686] System Configuration

[1687] Hardware and Software Configuration

[1688] Device: The device used by the user, such as a smartphone, tablet, or desktop PC.

[1689] Server: Provides high-performance computing power to analyze text data and generate correction suggestions.

[1690] software:

[1691] OpenAI API: Generates correction suggestions using the generative AI model GPT-3.

[1692] Python: A programming language for implementing the overall system logic.

[1693] Regular expressions (re): Used to process and analyze text data.

[1694] Data processing and calculation

[1695] 1. Acquiring text data:

[1696] A user inputs text data using a terminal, and the text data is transmitted from the terminal to a server.

[1697] 2. Text data analysis:

[1698] The server analyzes the received text data using the OpenAI API. First, it detects inappropriate expressions in the text data.

[1699] 3. Generate correction suggestions:

[1700] For detected inappropriate expressions, a generative AI model is used to generate appropriate correction suggestions.

[1701] 4. View suggested revisions:

[1702] The revision suggestions generated by the server are sent to the terminal and displayed to the user.

[1703] 5. Gathering Feedback:

[1704] The accuracy of suggestions is improved by sending user feedback information to the server and reflecting it in the generative AI model.

[1705] Specific examples

[1706] Example 1: Text analysis of business emails

[1707] A user types the following email on their smartphone:

[1708] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1709] Server analysis results:

[1710] Detected issues:

[1711] "Thank you for your hard work, person in charge" (too friendly)

[1712] "I'd like to discuss this" (too direct)

[1713] Suggested fix:

[1714] Suggested fix:

[1715] "Thank you for your hard work, person in charge" → "Dear person in charge"

[1716] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1717] Final confirmation:

[1718] To the person in charge

[1719] I would like to take the time to speak with you about the meeting we had the other day. I would appreciate it if you could let me know a convenient time and date for us to meet.

[1720] thank you.

[1721] Yamada

[1722] Prompt Sentence Examples

[1723] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[1724] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1725] Fixes:

[1726] This system allows users to quickly correct inappropriate language used in business emails and chats, ensuring appropriate communication. It also supports multiple languages ​​and can be customized based on industry characteristics, making it applicable to a wide range of uses. This is expected to reduce misunderstandings and misinterpretations in business, improving corporate reliability and efficiency.

[1727] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1728] Step 1:

[1729] Users use their devices to input the text of business emails or chat messages. Specifically, users enter text data into an input field on a smartphone, tablet, or desktop PC, and then press the send button. The input data is, for example, in the following format:

[1730] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1731] Input: Text data entered by the user into the terminal.

[1732] Output: Text data ready to be sent to the terminal

[1733] Step 2:

[1734] The terminal sends the entered text data to the server, where the data is transferred securely using protocols such as HTTPS. The entered text is sent in a format that the server can parse.

[1735] Input: Text data sent from the device to the server

[1736] Output: Text data received by the server

[1737] Step 3:

[1738] The server sends the received text data to the OpenAI API for analysis. The server generates a prompt to be used for analysis and passes the text data to the natural language processing engine. At this time, the following prompt is generated:

[1739] Analyze the following business chat message and provide suggestions for correcting inappropriate language:

[1740] Thank you for your hard work, Mr. / Ms. Person in Charge. I would like to discuss the meeting we had the other day. Please let me know a convenient time and date. Thank you in advance.

[1741] Fixes:

[1742] Input: Text data received by the server and prompts

[1743] Output: Analysis results by OpenAI API

[1744] Step 4:

[1745] The OpenAI API uses a generative AI model to detect inappropriate expressions and generate appropriate correction suggestions, such as "Thank you for your hard work, Mr. / Ms. Person in Charge."

[1746] Input: Prompt and text data received by the OpenAI API

[1747] Output: Detected profanities and suggested corrections

[1748] Step 5:

[1749] The server receives the correction suggestions returned by the OpenAI API and sends them to the device, where they can be viewed by the user. The correction suggestions are displayed in the following format:

[1750] Suggested fix:

[1751] "Thank you for your hard work, person in charge" → "Dear person in charge"

[1752] "I'd like to discuss this" → "I'd like to take the time to talk to you."

[1753] Input: Analysis results returned from the OpenAI API

[1754] Output: Suggested fixes sent to terminal

[1755] Step 6:

[1756] The terminal displays the received correction suggestions to the user, who then checks the displayed correction suggestions and selects whether to accept them. The selection operation is performed via buttons or an interface on the terminal.

[1757] Input: Suggested corrections sent to the terminal

[1758] Output: Correction suggestions and user selections displayed on the device

[1759] Step 7:

[1760] The user decides whether to accept the suggested revisions and confirms the result on the terminal. If the revision is accepted, the terminal corrects the text to the suggested expression and displays the final text.

[1761] Input: User accepts or rejects suggested revisions

[1762] Output: Final corrected text

[1763] Step 8:

[1764] The server collects user feedback and reflects it in the generative AI model algorithm, which improves the accuracy of correction suggestions from the next time onwards.

[1765] Input: User feedback information

[1766] Output: Improved generative AI model algorithm

[1767] This allows users to correct inappropriate expressions in business emails and chat messages to appropriate expressions, enabling efficient and effective communication.

[1768] 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.

[1769] The present invention relates to a system that analyzes text data, detects inappropriate expressions, and provides correction suggestions when a user composes a business email, in addition to combining it with an emotion engine that recognizes the user's emotions and reflects them in the correction suggestions. Below, we will explain the program processing and specific operation examples of this system.

[1770] Program Processing Overview

[1771] The system mainly consists of four entities: the terminal, the server, the user, and the emotion engine. The individual roles and operations of each entity are as follows:

[1772] User-generated email message:

[1773] The user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[1774] example:

[1775] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1776] Send email content:

[1777] The device acquires text data entered by the user. At this time, the device uses an emotion engine to recognize the emotion contained in the text data entered by the user. The acquired text data and emotion information are sent to the server.

[1778] Text analysis and profanity detection:

[1779] The server inputs the received text data into an NLP engine, which analyzes the email text and detects inappropriate expressions. For example, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[1780] Emotion recognition and suggested correction adjustments:

[1781] The server generates correction suggestions based on the emotional information recognized by the emotion engine. In this case, the strong emotional expression "I would like to check this urgently" is adjusted to a softer expression such as "I apologize for bothering you during your busy schedule, but could you please check it?"

[1782] View suggested fixes:

[1783] The server sends the generated revision suggestions back to the terminal, which then displays the revision suggestions received from the server to the user, who then checks the displayed suggestions and chooses whether to accept them.

[1784] Final user confirmation:

[1785] The user decides whether to accept the suggested revisions, and if so, the terminal will revise the email text as suggested. The user then checks the final text and, if there are no problems, confirms sending the email.

[1786] Gathering feedback and learning:

[1787] The server collects user feedback and reflects it in the generative AI and emotion engine algorithms to improve the accuracy of future suggestions and emotion recognition, allowing the system to continuously learn and improve.

[1788] Specific example (utilizing emotion engine)

[1789] A specific example is given below.

[1790] Email text entered by the user:

[1791] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1792] Emotion recognition results by emotion engine:

[1793] Recognized emotions: impatience, hurry

[1794] Server analysis results:

[1795] Detected issues:

[1796] "I want to check this urgently" (impression of urgency)

[1797] "Please reply as soon as possible" (Pressure)

[1798] Suggested fix:

[1799] Suggested fix:

[1800] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[1801] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[1802] Final confirmation from the user:

[1803] Dear John,

[1804] I know you are busy, but could you please check on the meeting we had the other day? I would appreciate it if you could check it when you have time.

[1805] thank you.

[1806] Yamada

[1807] This system allows users to create emails appropriately and efficiently, and an emotion engine provides revision suggestions that reflect the user's emotions, enabling more human-like communication.The system also supports multiple languages ​​and has customization functions based on industry-specific language usage, making it possible to build a flexible system that meets the diverse needs of users.

[1808] The processing flow will be explained below.

[1809] Step 1:

[1810] The user composes a business email and types it into the terminal. For example, the user types the following:

[1811] "John, I need to confirm something about the meeting we had the other day. Please respond as soon as possible. Thank you."

[1812] Step 2:

[1813] The device retrieves the text data entered by the user, which includes the entire body of the email.

[1814] Step 3:

[1815] The device sends the acquired text data to the emotion engine, which analyzes the user's emotions contained in the text data. For example, the emotion engine recognizes the user's impatience from "I want to check this urgently."

[1816] Step 4:

[1817] The device sends the entire text data along with the emotion analysis results to the server, which then prepares to begin analyzing the email text.

[1818] Step 5:

[1819] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the email text and detects inappropriate expressions. In this case, "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong expressions.

[1820] Step 6:

[1821] The server generates appropriate correction suggestions for detected inappropriate expressions based on the emotion analysis results from the emotion engine. For example, a strong expression such as "I would like to check this urgently" can be corrected to a milder expression such as "I apologize for bothering you during your busy schedule, but could you please check this?"

[1822] Step 7:

[1823] The server sends the generated revision suggestions together with the emotion analysis results back to the device, which then displays the revision suggestions received from the server to the user.

[1824] Step 8:

[1825] The user checks the displayed correction suggestions and decides whether to accept them. If the user accepts the suggestions, the terminal will correct the email text as suggested.

[1826] Step 9:

[1827] The terminal displays the final, corrected message to the user, who then checks the message and confirms sending the email if there are no problems.

[1828] Step 10:

[1829] After the user's operation is confirmed, the device will send the confirmation message and feedback information to the server, which will then collect the user's feedback information and use it as learning data to improve the accuracy of the AI ​​algorithm and emotion engine.

[1830] This process allows users to create emails appropriately and efficiently, and by utilizing the emotion engine, it enables more human-like communication that takes into account the user's emotions.

[1831] Example 2

[1832] 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."

[1833] Conventional business email writing support systems focus on detecting inappropriate expressions and suggesting corrections by analyzing text data, but because they are unable to take user emotions into account, they have the problem of difficulty in providing flexible correction suggestions that are in line with emotions.In addition, mechanisms for improving the accuracy of correction suggestions based on user feedback are limited to a few cases, and there is a problem that long-term learning and improvement are not sufficiently carried out.

[1834] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1835] In this invention, the server includes means for acquiring text data entered by a user, means for analyzing the text data to detect inappropriate expressions, means for recognizing the user's emotions using an emotion recognition engine, means for adjusting revision suggestions based on the recognized emotions, means for generating revision suggestions for the detected inappropriate expressions, means for displaying the revision suggestions to the user, and means for improving the accuracy of the revision suggestions based on user feedback, thereby enabling flexible and appropriate revision suggestions that reflect the user's emotions.

[1836] "User" refers to an individual or organization that uses the system to create business emails.

[1837] "Text data" refers to character string information entered by the user, and includes the email text itself.

[1838] An "emotion recognition engine" refers to software that analyzes and recognizes user emotions contained in text data.

[1839] "Correction suggestions" refer to alternative expressions generated by the system to correct inappropriate expressions in text data.

[1840] "Feedback" refers to the user's evaluation and reaction to the correction suggestions provided by the system.

[1841] A "server" refers to a central device that manages and controls the processes of the entire system and processes and manages various data.

[1842] An "NLP engine" refers to software or algorithms for natural language processing.

[1843] These definitions help clarify the role and function of each element of the system.

[1844] This invention is a system that analyzes text data, detects inappropriate expressions, and provides suggestions for correction when a user composes a business email. Furthermore, by combining it with an emotion engine that recognizes the user's emotions and reflects them in the suggested corrections, it becomes possible to create emails that are more flexible and human-like.

[1845] System configuration

[1846] This system mainly consists of four entities: the terminal, the server, the user, and the emotion engine.

[1847] Terminal: The device (e.g., PC, smartphone) on which the user enters the email text.

[1848] Server: A central processing unit that analyzes text data and generates profanity detection and correction suggestions.

[1849] User: A person or organization that writes business emails.

[1850] Emotion engine: Software for analyzing user emotions (e.g., IBM Watson Tone Analyzer).

[1851] Text data acquisition and sentiment analysis

[1852] A user types the text of a business email into the terminal. For example, the user types the text of an email to be sent to a business partner as follows:

[1853] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1854] The device receives text data entered by the user and analyzes the user's emotions contained in the text data using an emotion recognition engine. The analyzed emotion information is then sent to the server together with the text data.

[1855] Text analysis and profanity detection

[1856] The server inputs the received text data into a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to analyze the email text. This is where inappropriate expressions are detected. For example, expressions such as "I would like to check this as soon as possible" and "Please reply as soon as possible" are detected as being too strong.

[1857] Generate and refine correction suggestions

[1858] The server generates correction suggestions based on the analysis results of the emotion engine. In this case, expressions that are recognized as having strong emotions are adjusted to softer expressions. For example, "I would like to check this urgently" is corrected to "I apologize for bothering you during your busy schedule, but could you please check it?"

[1859] View and finalize suggested revisions

[1860] The server sends the generated revision suggestions to the terminal. The terminal displays the revision suggestions received from the server to the user. The user checks the displayed suggestions and chooses whether to accept them. If accepted, the terminal automatically corrects the email text according to the suggestions. Finally, the user checks the final text and sends it if there are no problems.

[1861] Gathering feedback and training the system

[1862] The server collects user feedback and reflects it in the generative AI model and emotion engine algorithms, which improves the accuracy of future suggestions and emotion recognition.

[1863] Specific examples (prompt sentence examples)

[1864] For example:

[1865] Email text entered by the user

[1866] John, I'd like to confirm the details of the meeting we had the other day as soon as possible. Please reply as soon as possible. Thank you.

[1867] Emotion recognition results by emotion engine

[1868] Recognized emotions: impatience, hurry

[1869] Revision suggestions

[1870] "I need to check this urgently" → "I know you're busy, but could you please check this for me?"

[1871] "Please reply as soon as possible" → "I would appreciate it if you could check it when you have time"

[1872] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1873] Program processing flow

[1874] The program processing flow of this system will be explained in detail below, divided into steps.

[1875] Step 1:

[1876] The user enters the text of a business email into the terminal.

[1877] Input: Text data entered by the user into the terminal.

[1878] Action: A user enters text using a keyboard or touchscreen.

[1879] Output: Text data saved in the device.

[1880] Step 2:

[1881] The device acquires text data entered by the user and uses an emotion recognition engine to analyze the user's emotions contained in the text data.

[1882] Input: The text data entered in step 1.

[1883] How it works: The emotion recognition engine extracts emotional information from text (e.g., impatience, urgency).

[1884] Output: User emotion information and text data.

[1885] Step 3:

[1886] The device transmits the acquired text data and emotional information to the server.

[1887] Input: Text data and emotion information.

[1888] Action: The operation of sending data to a server, transferring data over a network.

[1889] Output: Text data and emotion information received by the server.

[1890] Step 4:

[1891] The server inputs the received text data into a natural language processing (NLP) engine, which analyzes the text data and detects inappropriate expressions.

[1892] Input: Text data and emotion information sent from the device.

[1893] How it works: An NLP engine analyzes text to detect strong or inappropriate language.

[1894] Output: Text data with profanity flagged.

[1895] Step 5:

[1896] The server generates correction suggestions for detected inappropriate expressions based on the results of the emotion recognition engine.

[1897] Input: Text data with inappropriate language flagged, sentiment information.

[1898] How it works: The revision suggestion algorithm generates revision suggestions that take user sentiment into account.

[1899] Output: Text data with suggested corrections.

[1900] Step 6:

[1901] The server sends the generated revision suggestions to the terminal.

[1902] Input: Text data with suggested corrections.

[1903] Action: The act of sending data to a device, transferring data over a network.

[1904] Output: Correction suggestions received on the device.

[1905] Step 7:

[1906] The terminal displays the revision suggestions received from the server to the user, who then checks the revision suggestions and decides whether to accept them.

[1907] Input: The correction proposal received.

[1908] Behavior: Displays suggested revisions in the user interface.

[1909] Output: User confirmation and choice (accept or not).

[1910] Step 8:

[1911] If the user accepts the suggested revisions, the device automatically corrects the email text as suggested. The user then finally checks the text and, if there are no problems, sends the email.

[1912] Input: The correction suggestion that the user accepted.

[1913] Action: Adjust the email text based on the suggested revisions and ask the user for final confirmation. Send the email.

[1914] Output: The final revised email text and the email sent.

[1915] Step 9:

[1916] The server collects user feedback information and reflects it in the generative AI model and emotion engine algorithms.

[1917] Input: User feedback information.

[1918] How it works: Feedback information is fed into the algorithm and used as training data for the model.

[1919] Output: Improved generative AI models and emotion engines.

[1920] (Application example 2)

[1921] 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."

[1922] When dealing with customers, staff are required to use appropriate language while taking into consideration the customer's feelings, but this can be difficult to judge on the ground, and inappropriate language can be used, especially when dealing with complaints or inquiries. This can lead to a decrease in customer satisfaction and the risk of problems arising. A system to effectively resolve this is needed.

[1923] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1924] In this invention, the server includes means for acquiring text data and voice data input by a user, means for analyzing the text data and voice data to detect inappropriate expressions, and means for generating correction suggestions for the detected inappropriate expressions based on emotion information, thereby enabling support for staff to use appropriate expressions in real time while dealing with customers.

[1925] "Text data" refers to character string information entered by a user. It is generally saved as recorded character string information or a document.

[1926] "Voice Data" means a digital recording of a user's spoken voice, often converted to text using speech recognition technology.

[1927] "Analysis" refers to the act of processing input data using an algorithm to extract specific information or patterns.

[1928] "Inappropriate expressions" refer to expressions that contain inappropriate meanings or emotions in communication and may cause misunderstandings or trouble.

[1929] "Emotional information" refers to emotional elements extracted from the text or voice of a user or customer, and includes emotions such as fear, impatience, and joy.

[1930] "Correction proposals" refer to proposals that suggest appropriate expressions or ways of responding based on the analysis results.

[1931] "Real time" refers to a state in which the time between when data is generated and when it is processed is extremely short, and the data is processed almost simultaneously.

[1932] "Display" refers to the act of visually presenting the analysis results and correction suggestions to the user, outputting them on a display or screen.

[1933] "Feedback" refers to the evaluations and reactions that users provide to a system, which are used to improve and learn from the system.

[1934] "Device" refers to a hardware device with a specific function, including smartphones and robots.

[1935] A "server" refers to a computer system that provides services to clients over a network.

[1936] A "system" refers to a collection of multiple components working together to achieve a specific function.

[1937] The present invention relates to a system that analyzes input text data and voice data when a user is dealing with a customer, and provides appropriate correction suggestions based on emotion information. A specific embodiment of the system includes the following steps.

[1938] The server receives text and voice data sent from the user's device (smartphone or robot). At that time, it uses voice recognition software and an NLP (natural language processing) engine to convert the voice data into text data. Specifically, it uses the "SpeechRecognition" module for voice recognition and the "TextBlob" module for sentiment analysis of the text data.

[1939] The server then uses pre-defined rule-based filters and machine learning algorithms to detect inappropriate content based on the analyzed text data. If an inappropriate content is detected, the server generates appropriate correction suggestions based on the sentiment information. For example, the phrase "I would like to check this urgently" is converted into the suggestion "I apologize for bothering you during your busy schedule, but could you please check it?"

[1940] The generated correction suggestions are displayed in real time on the user's device. The user can review the suggestions and adopt the corrected expressions as needed. In addition, the user's feedback information is sent to the server and used to improve the accuracy of the correction suggestions.

[1941] As a concrete example, the following prompt sentence is used:

[1942] Example prompt sentence:

[1943] Speak specific customer interaction text. Example: "Please check availability of this item immediately."

[1944] It analyzes sentiment and outputs appropriate correction suggestions in the following format: Example correction: "I'm sorry to bother you, but could you please wait a moment while I check this?"

[1945] This system enables staff to use appropriate expressions in real time while interacting with customers, which is expected to improve customer satisfaction. The hybrid system combines voice input and text analysis to enable quick and flexible responses.

[1946] Furthermore, the system continuously learns from feedback and improves the accuracy of correction suggestions, making it possible to adapt to the diverse needs of users.

[1947] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1948] Step 1:

[1949] The user inputs into the device by voice or text.

[1950] Specifically, it uses the smartphone's microphone to capture voice data, and also allows for manual text input.

[1951] Input: Audio or text data

[1952] Output: User input audio or text data

[1953] Step 2:

[1954] The terminal converts the voice data into text data.

[1955] Specifically, the speech data is converted into text data using the SpeechRecognition module.

[1956] Input: Audio data

[1957] Output: Converted text data

[1958] Step 3:

[1959] The terminal transmits the text data to the server.

[1960] As a specific operation, the converted text data is sent to a server via the Internet.

[1961] Input: Text data

[1962] Output: Text data sent to the server

[1963] Step 4:

[1964] The server analyzes the text data and extracts emotional information.

[1965] Specifically, it uses the TextBlob module to analyze the polarity (emotional polarity) and subjectivity of text data.

[1966] Input: Text data

[1967] Output: Emotional information (e.g. polarity, subjectivity)

[1968] Step 5:

[1969] The server detects profanity.

[1970] Specifically, it uses predefined filters and machine learning models to identify inappropriate language.

[1971] Input: Text data

[1972] Output: A list of profanities

[1973] Step 6:

[1974] The server generates revision suggestions based on the emotion information.

[1975] Specifically, it suggests alternative expressions for detected inappropriate expressions and adjusts them taking into account the user's emotions and context.

[1976] Input: Inappropriate language, dishonest behavior

[1977] Output: Sentiment-based revision suggestions

[1978] Step 7:

[1979] The server sends the revision suggestions to the terminal.

[1980] As a specific operation, the generated revision proposal is sent back to the user's terminal via the Internet.

[1981] Input: Proposed correction

[1982] Output: Suggested fixes sent to terminal

[1983] Step 8:

[1984] The terminal displays the suggested revisions to the user.

[1985] Specifically, the proposed corrections are displayed on the screen so that the user can confirm them.

[1986] Input: Proposed correction

[1987] Output: Displayed suggested fixes

[1988] Step 9:

[1989] The user reviews the proposed corrections and selects a response.

[1990] As a specific operation, the user either accepts the suggested revision or creates a new message that reflects the suggested revision.

[1991] Input: Displayed suggested corrections

[1992] Output: User selected or modified text data

[1993] Step 10:

[1994] The server collects user feedback and uses it to improve the accuracy of the system.

[1995] Specifically, user choices and feedback information are stored in a database and used as training data for generative AI models and algorithms.

[1996] Input: User feedback

[1997] Output: Improved correction suggestion algorithm

[1998] 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.

[1999] 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.

[2000] 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.

[2001] 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.

[2002] 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.

[2003] 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.

[2004] 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).

[2005] 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.

[2006] 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."

[2007] 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.

[2008] 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).

[2009] 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.

[2010] 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.

[2011] 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.

[2012] 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.

[2013] 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.

[2014] 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.

[2015] 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.

[2016] 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.

[2017] 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.

[2018] 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.

[2019] The following is further disclosed regarding the above embodiment.

[2020] (Claim 1)

[2021] A means for obtaining text data entered by a user;

[2022] means for analyzing the text data to detect inappropriate expressions;

[2023] means for generating correction suggestions for the detected profanity;

[2024] means for displaying said revision suggestions to a user;

[2025] A means to improve the accuracy of correction suggestions based on user feedback; and

[2026] A system including:

[2027] (Claim 2)

[2028] The system of claim 1 , wherein the revision suggestions correspond to different languages.

[2029] (Claim 3)

[2030] 10. The system of claim 1, further comprising means for customizing settings based on the user's industry and business characteristics.

[2031] "Example 1"

[2032] (Claim 1)

[2033] A means for obtaining text data entered by a user;

[2034] means for analyzing the text data to detect inappropriate expressions;

[2035] means for generating correction suggestions for the detected profanity;

[2036] means for displaying said revision suggestions to a user;

[2037] A means for the user to select a suggested revision and revise the email text;

[2038] a means for sending the email after final confirmation by the user;

[2039] A means to improve the accuracy of correction suggestions based on user feedback; and

[2040] A system including:

[2041] (Claim 2)

[2042] The system of claim 1 , wherein the revision suggestions correspond to different languages.

[2043] (Claim 3)

[2044] 10. The system of claim 1, further comprising means for customizing settings based on the user's industry and business characteristics.

[2045] "Application Example 1"

[2046] (Claim 1)

[2047] A means for obtaining text data entered by a user;

[2048] means for analyzing the text data to detect inappropriate expressions;

[2049] means for generating correction suggestions for the detected profanity;

[2050] means for displaying said revision suggestions to a user;

[2051] A means to improve the accuracy of correction suggestions based on user feedback; and

[2052] A means for analyzing the text data in real time and correcting it to an appropriate expression in different response formats;

[2053] means for generating said revision suggestions utilizing a generative AI model;

[2054] A system including:

[2055] (Claim 2)

[2056] The system of claim 1 , wherein the revision suggestions correspond to different languages.

[2057] (Claim 3)

[2058] 10. The system of claim 1, further comprising means for customizing settings based on the user's industry and business characteristics.

[2059] "Example 2: Combining Emotion Engines"

[2060] (Claim 1)

[2061] A means for obtaining text data entered by a user;

[2062] means for analyzing the text data to detect inappropriate expressions;

[2063] means for recognizing a user's emotion using an emotion recognition engine;

[2064] a means for adjusting the revision suggestions based on the perceived sentiment;

[2065] means for generating correction suggestions for the detected profanity;

[2066] means for displaying said revision suggestions to a user;

[2067] A means to improve the accuracy of correction suggestions based on user feedback; and

[2068] A system including:

[2069] (Claim 2)

[2070] The system of claim 1 , wherein the revision suggestions correspond to different languages.

[2071] (Claim 3)

[2072] 10. The system of claim 1, further comprising means for customizing settings based on the user's industry and business characteristics.

[2073] "Application example 2 when combining emotion engines"

[2074] (Claim 1)

[2075] means for acquiring text data and voice data input by a user;

[2076] means for analyzing the text data and the voice data to detect inappropriate expressions;

[2077] means for generating emotion-based correction suggestions for the detected inappropriate expressions;

[2078] means for displaying said revision suggestions to a user;

[2079] A means to improve the accuracy of correction suggestions based on user feedback; and

[2080] a means for providing real-time revision suggestions across multiple devices;

[2081] A system including:

[2082] (Claim 2)

[2083] The system of claim 1 , wherein the revision suggestions correspond to different languages.

[2084] (Claim 3)

[2085] 10. The system of claim 1, further comprising means for customizing settings based on the user's industry and business characteristics. [Explanation of symbols]

[2086] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for obtaining text data entered by a user; means for analyzing the text data to detect inappropriate expressions; means for generating correction suggestions for the detected profanity; means for displaying said revision suggestions to a user; A means to improve the accuracy of correction suggestions based on user feedback; and A system including:

2. The system of claim 1 , wherein the revision suggestions correspond to different languages.

3. 10. The system of claim 1, further comprising means for customizing settings based on the user's industry or business characteristics.

Images