System

Abstract

A system is provided.SOLUTION: Means for allowing a user to input a new remark content, means for allowing a server to receive the remark content, means for allowing the server to collect past remark data of the user, means for allowing the server to analyze the collected remark data and extract an important keyword or context, and means for allowing the server to compare the new remark content with the past remark data, A system comprising: means for detecting a contradiction; means for the server to check whether the new statement content includes an inappropriate expression or a sensitive topic; means for the server to evaluate a possibility that the new statement content damages a brand image; means for the server to generate a feedback to a user based on an analysis result; and means for the user to receive the feedback and correct the statement content.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 modern society, it is important for certain individuals and organizations, whose individual digital footprints are large and whose statements are likely to have a significant social impact, to be highly conscious of the consistency and appropriateness of their statements. However, it is difficult to memorize or manually check for inconsistencies between all of one's past content and new statements, and self-checking for inappropriate content is easily subjective, posing a risk of damaging brand image. To address these issues, there is a need for a system that automates the consistency of statements, checks for inappropriate language, and evaluates the impact on brand image, providing users with the opportunity to reevaluate their statements before making them public. [Means for solving the problem]

[0005] The present invention solves the above-mentioned problems by providing a system including: a means for a user to input new comments; a means for a server to receive the comments; a means for the server to collect data on past comments from the user; a means for the server to analyze the collected data on past comments and extract important keywords and contexts; a means for the server to compare the new comments with the past data on comments and detect inconsistencies; a means for the server to check whether the new comments contain inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comments will damage a brand image; a means for the server to generate feedback to the user based on the analysis results; and a means for the user to receive the feedback and modify the comments. This system allows users to review and reevaluate their comments before making them public, enabling the creation and protection of a consistent brand image across a digital footprint.

[0006] A "user" is an individual or group that uses the system to input new comments and receive analysis results.

[0007] A "server" is a computer system that receives user comments, collects data, analyzes, compares, evaluates, and generates feedback.

[0008] "Comment content" is text information that the user newly inputs and sends to the system.

[0009] "Past comment data" refers to text information such as social media posts, lecture contents, and books previously published by users.

[0010] "Means of collection" refers to the functions and processes by which the server acquires users' past utterance data.

[0011] "Means of analysis" refers to the function of extracting important keywords and context from collected speech data.

[0012] "Means for detecting contradictions" is a function that compares new statements with past statements and finds contradictions in semantics and content.

[0013] "Inappropriate language" is any word or phrase that, when included in a statement, is offensive or problematic.

[0014] A "sensitive topic" is a socially or politically sensitive subject about which public discourse should be handled with caution.

[0015] "Brand image" is a concept that indicates the trust and reputation that a company or individual has in society.

[0016] "Means of evaluation" is a function that analyzes and scores the impact that statements have on brand image.

[0017] The "means for generating feedback" is a function that creates information that provides specific advice and recommendations to the user based on the analysis results.

[0018] "Means of correction" refers to the process in which the user changes the content of their statement based on the feedback and re-enters it in an appropriate format. [Brief explanation of the drawings]

[0019] [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

[0020] 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.

[0021] First, the terms used in the following description will be explained.

[0022] 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).

[0023] 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.

[0024] 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.

[0025] 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.

[0026] 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."

[0027] [First embodiment]

[0028] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0029] 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.

[0030] 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).

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0036] 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.

[0037] 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.

[0038] 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.

[0039] 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."

[0040] The present invention provides a system for ensuring safe and consistent communication by allowing users to review the content of their statements before they are made public. This system is designed to check whether statements made by individuals or organizations with significant social influence, such as politicians, business leaders, and celebrities, are consistent with past statements, contain inappropriate language, or damage brand image. Specific embodiments of the system are described below.

[0041] System Configuration

[0042] This system consists of the following main modules:

[0043] Data collection module: Collects user's past speech data.

[0044] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[0045] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[0046] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[0047] Brand image evaluation module: Evaluate the impact of statements on brand image.

[0048] Feedback generation module: Generates feedback to the user based on the analysis results.

[0049] Program processing

[0050] 1. User terminal (entering new comments)

[0051] The user inputs a new message and sends it to the system.

[0052] 2. Server (data reception and initial processing)

[0053] The server receives the input speech data and performs initial processing, converting the data into an appropriate format and classifying the type of speech (social media post, lecture content, book, etc.).

[0054] 3. Server (executes data collection modules)

[0055] The server collects past comment data, and retrieves the user's past social media posts, lectures, books, etc. from the database.

[0056] 4. Server (executing text analysis module)

[0057] The server analyzes the collected data to extract important keywords and context, and uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text.

[0058] 5. Server (execution of the inconsistency detection module)

[0059] The server compares the new statement with previous statements to see if there are any inconsistencies. It runs an algorithm to compare previous statements with the new statement and detect semantic inconsistencies.

[0060] 6. Server (execution of profanity detection module)

[0061] The server checks new posts for inappropriate language or sensitive topics by scanning the text against predefined lists and models to detect inappropriate language or sensitive topics.

[0062] 7. Server (Running the brand image evaluation module)

[0063] The server evaluates whether the comment is damaging to the brand image and scores the positive / negative impact of the comment using an evaluation model related to the brand image.

[0064] 8. Server (executing the feedback generation module)

[0065] The server generates feedback to the user based on the analysis results, synthesizing the analysis results and automatically generating a feedback document containing specific advice and recommendations.

[0066] 9. User Device (Receiving and Correcting Feedback)

[0067] The user receives the generated feedback and corrects the statement if necessary. The user checks the feedback, corrects the statement, or re-enters it into the system.

[0068] Specific examples

[0069] Example 1: Posting to social media

[0070] 1. A user enters the text "Excited about the new product" in a social media post.

[0071] 2. The server receives the entered text, and the data collection module collects past posting data.

[0072] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[0073] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[0074] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[0075] 6. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[0076] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[0077] 8. Receive user-generated feedback and revise your post as needed.

[0078] Example 2: Lecture content

[0079] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[0080] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[0081] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[0082] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[0083] 5. The server checks the script for profanity using the profanity detection module.

[0084] 6. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[0085] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[0086] 8. Take the user-generated feedback and modify the script as needed.

[0087] In this way, users can reassess their statements before publishing them to ensure consistent and appropriate communication.

[0088] The processing flow will be explained below.

[0089] Step 1:

[0090] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message on social media saying, "The new product is great!"

[0091] Step 2:

[0092] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[0093] Step 3:

[0094] The server runs a data collection module to collect data on users' past comments. Specifically, it retrieves the user's past social media posts, lecture contents, and book contents from a database.

[0095] Step 4:

[0096] The server runs a text analysis module to analyze the collected speech data. Natural language processing (NLP) techniques are used to extract important keywords, context, entities, and sentiment from the text. For example, keywords such as "new product," "problem," and "excitement" are extracted.

[0097] Step 5:

[0098] The server runs a contradiction detection module, which compares the new statement with past statements. Specifically, it uses an algorithm for detecting semantic contradictions to detect a contradiction between the past statement "The new product has many problems" and the new statement "The new product is great!"

[0099] Step 6:

[0100] The server runs a profanity detection module to check if the new post contains any inappropriate language or sensitive topics, such as inappropriate language or sensitive subjects that should be avoided in public.

[0101] Step 7:

[0102] The server executes the brand image evaluation module to evaluate the impact of the new comment on the brand image, and uses the evaluation model to score the positive / negative impact of the comment.

[0103] Step 8:

[0104] The server runs the feedback generation module and generates feedback to the user based on the analysis results. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[0105] Step 9:

[0106] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[0107] These steps allow users to reassess their content before publishing statements that are consistent and protect their brand image.

[0108] Example 1

[0109] 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."

[0110] Currently, there is a lack of mechanisms for evaluating the appropriateness of user statements before they are made public and for maintaining consistent communication. Furthermore, there are also insufficient means for automatically checking whether statements contradict previous statements, contain inappropriate language, or damage a company's image. These problems are particularly serious for individuals and organizations with significant social influence, such as politicians, corporate leaders, and celebrities. Therefore, the present invention aims to provide a system for reevaluating user statements and ensuring safe and consistent communication.

[0111] 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.

[0112] In this invention, the server includes: a means for a user to input new comments; a means for the server to receive the comments; a means for the server to collect data on past comments from the user; a means for the server to analyze the collected data on comments and extract important keywords and contexts; a means for the server to compare the new comments with the past data on comments and detect semantic inconsistencies; a means for the server to check whether the new comments contain inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comments will damage the company's image; a means for the server to generate feedback to the user based on the analysis results; and a means for the user to receive the feedback and modify the comments. This allows users to reevaluate the content of their comments before making them public, enabling consistent and safe communication.

[0113] "User" refers to a person who uses the system to input comments and receive feedback.

[0114] "Server" refers to a device that includes equipment or software that receives, processes, analyzes, and generates feedback from users.

[0115] "Speech" refers to new text information that a user enters into the system.

[0116] "Past utterance data" refers to the accumulation of text information about utterances made by a user in the past.

[0117] "Important keywords and context" refers to semantically important words and their surrounding context that the text analysis module extracts from speech data.

[0118] A "semantic contradiction" refers to a situation in which a new statement does not semantically match a previous statement.

[0119] "Inappropriate language" refers to language that is socially unacceptable or problematic to use in a particular context.

[0120] "Sensitive topics" refer to subjects or topics that are inappropriate for public disclosure under certain circumstances or social conditions.

[0121] "Corporate image" refers to the overall evaluation and impression of a company from the perspective of general consumers and society.

[0122] "Feedback" refers to information, including advice and recommendations, that the server provides to the user based on the analysis results.

[0123] This invention is a system that allows users to review the content of their statements before they are made public, ensuring safe and consistent communication. In particular, it is designed to check statements made by politicians, business leaders, celebrities, and other influential individuals and organizations to ensure that they are consistent with past statements, contain inappropriate language, and do not damage the company's image.

[0124] The system consists of the following main modules:

[0125] Data collection module: Collects past user comment data. Obtains comment data from users' social media posts, lectures, books, etc.

[0126] Text analysis module: Analyzes collected speech data and extracts important keywords and context. This analysis uses natural language processing (NLP) techniques, specifically tools such as Google's NLP API and SpaCy.

[0127] Inconsistency detection module: Runs algorithms to detect semantic inconsistencies by comparing new statements with previous statements.

[0128] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics by scanning the text using predefined lists and machine learning models.

[0129] Corporate Image Assessment Module: Evaluates whether a statement is damaging to the company's image. Using an assessment model related to corporate image, the module scores the positive / negative impact of the statement.

[0130] Feedback generation module: Generates feedback to users based on the analysis results. Automatically generates specific advice and recommendations.

[0131] The system operates in the following specific steps:

[0132] 1. Entering a new message on the user's terminal: The user enters the new message into the terminal and sends it to the system.

[0133] 2. Data reception and initial processing on the server: The server receives the input speech data, performs initial processing of the data, and classifies the type of speech (SNS post, lecture content, book, etc.).

[0134] 3. Collection of past speech data on the server: The server collects the user's past speech data from the database.

[0135] 4. Text analysis on the server: The server uses NLP techniques to extract important keywords and context.

[0136] 5. Conflict detection on the server: The server compares new statements with previous statements to detect semantic contradictions.

[0137] 6. Server-based profanity detection: The server checks new posts for profanity.

[0138] 7. Corporate image evaluation on the server: The server evaluates the impact of the comments on the corporate image.

[0139] 8. Feedback generation on the server: The server generates feedback to the user based on the analysis results.

[0140] 9. Receiving feedback and correcting content on the user's device: The user receives the generated feedback and corrects the content of the comment if necessary.

[0141] Specific examples

[0142] Example 1: Posting to social media

[0143] 1. A user enters the text "Excited about the new product" in a social media post.

[0144] 2. The server receives the entered text, and the data collection module collects past posting data.

[0145] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[0146] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[0147] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[0148] 6. The server evaluates the impact of this post on the company's image in the company image evaluation module.

[0149] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and will strengthen the company's image."

[0150] 8. Receive user-generated feedback and revise your post as needed.

[0151] Example 2: Lecture content

[0152] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[0153] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[0154] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[0155] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[0156] 5. The server checks the script for profanity using the profanity detection module.

[0157] 6. The server uses the corporate image evaluation module to evaluate the impact of this presentation on the company's image.

[0158] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the company's image, but be careful with certain wording."

[0159] 8. Take the user-generated feedback and modify the script as needed.

[0160] Prompt Sentence Examples

[0161] For social media posts: "Evaluate your social media posts about new products to ensure they are consistent with previous posts, contain no inappropriate language, and fit with the company image."

[0162] For a presentation: "Evaluate the presentation script about our strategy for entering new markets. Check for consistency with previous presentations, inappropriate language, and consistency with our company image."

[0163] These prompts can be used to ask the generative AI model for further evaluation. These examples encourage users to reassess their statements before publishing them, ensuring appropriate and consistent communication.

[0164] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0165] Step 1:

[0166] User enters new comment

[0167] The user inputs new content into the terminal and clicks the "Send" button. For example, the user inputs "I'm looking forward to the release of the new product." The input data is the content of the text message. The output is the text message data sent to the server.

[0168] Step 2:

[0169] The server receives and performs initial processing on the speech data.

[0170] The server receives the text of the comment sent by the user. The data formatter processes this text data, converts it into an appropriate data format, and classifies it by type of comment (social media post, lecture content, book, etc.). The input is the text data sent by the user, and the output is the initially processed data format. Specifically, the data formatter converts the comment content into JSON format.

[0171] Step 3:

[0172] The server collects past utterance data

[0173] The server collects data on users' past comments from a database. For example, it searches the database for social media posts from the past year and retrieves related data. The input is the user's identification information, and the output is collected data on past comments. Specifically, it executes an SQL query to extract the relevant comment data from the database.

[0174] Step 4:

[0175] The server analyzes the collected data

[0176] The server uses natural language processing (NLP) techniques to analyze the collected utterance data. It uses Google's NLP API and SpaCy to extract important keywords and context from the text. The input is the collected utterance data, and the output is the extracted keywords and context. Specifically, the NLP model tokenizes the text and performs named entity recognition (NER).

[0177] Step 5:

[0178] The server compares new and past comments

[0179] The server semantically compares the new utterance with the previous utterance to check for inconsistencies. A comparison algorithm is used to analyze the semantic matches and differences between the new and previous utterances. The input is the new utterance and the previous utterance data, and the output is the result of the inconsistency detection. Specifically, the server calculates context vectors and compares them using cosine similarity.

[0180] Step 6:

[0181] Server detects inappropriate language

[0182] The server checks whether new posts contain inappropriate language or sensitive topics. It scans the text of the post using predefined lists or machine learning models. The input is the new post, and the output is the detection of inappropriate language. Specifically, it compares the text with a list of banned words to see if there are any matches.

[0183] Step 7:

[0184] Server evaluates corporate image

[0185] The server evaluates the impact of new comments on the company's image. It uses an evaluation model related to the company's image to score the positive / negative impact of the comment. The input is the new comment, and the output is an evaluation score. Specifically, the evaluation model performs sentiment analysis on the comment text and scores it.

[0186] Step 8:

[0187] Server generates feedback

[0188] The server generates feedback for the user based on the analysis results. Specific advice and recommendations are automatically generated as a feedback document. The input is the analysis results, and the output is a feedback document. Specifically, the generative AI model generates feedback text based on the input data.

[0189] Step 9:

[0190] Users receive and correct feedback

[0191] The user receives the generated feedback on their device and modifies the utterance as necessary. The input is the feedback document, and the output is the modified utterance. Specifically, the user views the feedback and modifies the utterance in a text editor.

[0192] (Application example 1)

[0193] 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."

[0194] Publicly released comments and statements may lack consistency and contain inappropriate language or legal risks, which could damage the brand image and security policies of companies and celebrities. A system is needed to solve this problem and maintain safe and consistent communication.

[0195] 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.

[0196] In this invention, the server includes: means for a user to input new comments; means for the server to receive the comments; means for the server to collect data on past comments from the user; means for the server to analyze the collected data on comments and extract important keywords and contexts; means for the server to compare the new comments with the past data on comments and detect inconsistencies; means for the server to check whether the new comments contain inappropriate language or sensitive topics; means for the server to evaluate the possibility that the new comments will damage brand image or security policies; means for the server to generate feedback to the user based on the analysis results; and means for the user to receive the feedback and modify the comments. This allows users to evaluate and modify the comments they make public in advance, enabling consistent and appropriate communication.

[0197] A "user" is an individual or group that uses the system to input new comments.

[0198] A "server" is a computer system that analyzes the content of comments received from users and generates feedback based on the analysis results.

[0199] "Comment content" refers to text data such as a comment, script, or post newly entered by a user.

[0200] "Past comment data" is a record of comments and posts made by the user in the past.

[0201] "Important keywords" are words or phrases that characterize the topic and are extracted from the content of the statement or past statement data.

[0202] "Context" refers to the context or background information of what is being said.

[0203] A "contradiction" is a state in which there is a logical inconsistency between the content of a new statement and past statement data.

[0204] "Inappropriate language" is any word or phrase that could pose legal risks or damage a company's reputation.

[0205] A "sensitive topic" is one that is socially, culturally, or legally controversial or carries risk.

[0206] "Brand image" refers to the public reputation or impression of a user or the organization or company to which they belong.

[0207] A "security policy" is a set of guidelines and rules established by an organization or company for the purposes of protecting information and managing risks.

[0208] "Feedback" refers to advice and recommendations to users that are generated based on the analysis results.

[0209] This invention provides a system that allows users to review the content of their comments before publishing them, ensuring consistent and safe communication. This system is particularly useful in the security field, where it can pre-screen comments made by companies and individuals to prevent inappropriate information leaks or comments that violate a company's security policy.

[0210] System Configuration

[0211] This system consists of the following main modules:

[0212] 1. Data Collection Module

[0213] Hardware / Software: Python (version 3.8 or higher), pandas library, SQLAlchemy

[0214] Description: The server collects the user's past speech data from the database and stores it in a data frame format.

[0215] 2. Text Analysis Module

[0216] Hardware / Software: Python (version 3.8 or higher), spaCy library, NLTK

[0217] Description: The server analyzes the speech data and extracts important keywords and entities, as well as analyzing sentiment and semantic relationships.

[0218] 3. Conflict Detection Module

[0219] Hardware / Software: Python (version 3.8 or higher), spaCy, natural language processing model (e.g., BERT)

[0220] Description: The server compares new statements with previous statements to detect logical inconsistencies.

[0221] 4. Profanity Detection Module

[0222] Hardware / Software: Python (version 3.8 or higher), scikit-learn, custom dictionary

[0223] Description: The server checks new posts for legal risks and confidential information.

[0224] 5. Brand Image Evaluation Module

[0225] Hardware / Software: Python (version 3.8 or higher), TensorFlow, scikit-learn

[0226] Description: The server scores new comments on their impact on brand image and security policies.

[0227] 6. Feedback Generation Module

[0228] Hardware / Software: Python (version 3.8 or higher), GPT-3 (OpenAI API)

[0229] Description: The server generates feedback statements from the analysis results and provides specific suggestions to the user.

[0230] Specific processing steps

[0231] 1. The user enters a new comment

[0232] The user inputs new utterances into the terminal.

[0233] 2. The server receives the message

[0234] The server analyzes the content of the message received from the user.

[0235] 3. Data Collection

[0236] The server collects past utterance data from a database.

[0237] 4. Text Analysis

[0238] The server analyzes the speech data and extracts important keywords and context.

[0239] 5. Conflict Detection

[0240] The server compares the new message with previous messages to detect any inconsistencies.

[0241] 6. Profanity Detection

[0242] The server checks new posts for inappropriate language and security risks.

[0243] 7. Brand image evaluation

[0244] The server evaluates the impact of the comments on the brand and security policies.

[0245] 8. Feedback Generation

[0246] The server generates feedback to the user based on the analysis results, prompting them to revise their comments.

[0247] Specific examples

[0248] For example, if a company's security officer types, "I would like to comment on the implementation of a new security system," the server will perform the following process:

[0249] 1. The data collection module collects past speech data (social media posts, news releases, meeting records, etc.).

[0250] 2. The text analysis module analyzes new and past comments and extracts the keywords "security," "system," and "implementation."

[0251] 3. The inconsistency detection module compares the semantic relationships between the new and previous statements to ensure consistency.

[0252] 4. The profanity detection module checks for legal risks and confidential information.

[0253] 5. The brand image evaluation module scores the impact of new statements on security policies and the brand.

[0254] 6. The feedback generation module generates feedback such as "Your comments are appropriate and consistent, but please avoid mentioning specific system names" and displays it to the user.

[0255] Prompt Sentence Examples

[0256] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[0257] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0258] Step 1:

[0259] The user inputs a new statement.

[0260] Input: New statement (e.g., "I would like to comment on the implementation of the new security system")

[0261] Output: What you say is sent to the system.

[0262] Specific operation: The user enters the content of the message into the input form on the terminal and presses the send button.

[0263] Step 2:

[0264] The server receives the message.

[0265] Input: User-submitted message

[0266] Output: The speech is ready for analysis.

[0267] Specific operation: The server receives an HTTP request, converts the content of the comment into an internal data format, and stores it in a database.

[0268] Step 3:

[0269] The server collects past utterance data.

[0270] Input: User ID or metadata related to what was said

[0271] Output: Past speech data (data frame format)

[0272] Specific operation: The server executes a database query to retrieve past comment data, which is then converted into a data frame using the pandas library.

[0273] Step 4:

[0274] The server analyzes the collected speech data and extracts important keywords and contexts.

[0275] Input: Past speech data

[0276] Output: Extracted keywords and context information

[0277] What it does: The server uses the spaCy library to parse text data and extract entities and keywords, as well as perform sentiment analysis and semantic relationships.

[0278] Step 5:

[0279] The server compares new comments with past comments to detect inconsistencies.

[0280] Input: New utterances, extracted keywords and context information

[0281] Output: Conflict detection results

[0282] What happens: The server uses a natural language processing model (e.g., BERT) to compare the semantic relationships between new and previous utterances, detecting any logical inconsistencies.

[0283] Step 6:

[0284] The server checks new posts for inappropriate language or sensitive topics.

[0285] Input: New statement

[0286] Output: Profanity detection results

[0287] What it does: The server uses the scikit-learn library and a custom dictionary to scan posts for profanity and sensitive topics.

[0288] Step 7:

[0289] The server evaluates the likelihood that the comment will damage the brand image or security policy.

[0290] Input: New speech, profanity detection results

[0291] Output: Evaluation results regarding brand image and security policy

[0292] Specific operation: The server uses TensorFlow and scikit-learn to score the impact of statements on brand image and security policies.

[0293] Step 8:

[0294] The server generates feedback to the user based on the analysis results.

[0295] Input: Conflict detection results, inappropriate language detection results, brand image and security policy evaluation results

[0296] Output: Feedback statement

[0297] Specific operation: The server uses GPT-3 (OpenAI API) to generate specific feedback sentences based on the analysis results.

[0298] Step 9:

[0299] The user receives the feedback and corrects the content of the statement.

[0300] Input: Feedback statement

[0301] Output: Corrected statement

[0302] Specific operation: The user checks the feedback text on the terminal, corrects the comment if necessary, and re-enters the corrections into the system.

[0303] Prompt Sentence Examples

[0304] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[0305] 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.

[0306] The present invention is a system that allows users to reevaluate their comments before they are made public, ensuring safe and consistent communication, particularly by combining an emotion engine. This system checks the comments entered by the user for inconsistencies with past comments, inappropriate expressions, and potential damage to the brand image, and also recognizes the user's emotions and adjusts the feedback. A specific embodiment of this system will be described.

[0307] System Configuration

[0308] The system consists of the following main modules:

[0309] Data collection module: Collects user's past speech data.

[0310] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[0311] Emotion engine: Recognizes emotions from user input and past speech data.

[0312] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[0313] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[0314] Brand image evaluation module: Evaluate the impact of statements on brand image.

[0315] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[0316] Program processing

[0317] 1. User terminal (entering new comments)

[0318] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message for posting on social media saying, "The new product is great!"

[0319] 2. Server (data reception and initial processing)

[0320] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[0321] 3. Server (executes data collection modules)

[0322] The server collects past comment data and retrieves the user's past social media posts, lecture contents, and book contents from the database.

[0323] 4. Server (executing text analysis module)

[0324] The server analyzes the collected data and extracts important keywords and context. It uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[0325] 5. Server (execution of emotion engine)

[0326] The server runs an emotion engine to recognize the user's emotions based on the user's input and past utterance data. For example, emotions such as "positive" or "excited" can be recognized from the input.

[0327] 6. Server (execution of the inconsistency detection module)

[0328] The server compares the new statement with previous statements to check for inconsistencies. Specifically, it uses an algorithm to detect semantic inconsistencies and detects a contradiction between the previous statement "The new product has many problems" and the new statement "The new product is great!"

[0329] 7. Server (execution of profanity detection module)

[0330] The server checks new posts for inappropriate language or sensitive topics, for example, detecting inappropriate language or sensitive subjects that should be avoided in public.

[0331] 8. Server (Running the brand image evaluation module)

[0332] The server evaluates whether a comment is damaging to the brand image and uses an evaluation model to score the positive / negative impact of the comment.

[0333] 9. Server (executing the feedback generation module)

[0334] The server generates feedback for the user based on the analysis results and the output of the emotion engine. It automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[0335] 10. User Device (Receiving and Correcting Feedback)

[0336] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[0337] Specific examples

[0338] Example 1: Posting to social media

[0339] 1. A user enters the text "Excited about the new product" in a social media post.

[0340] 2. The server receives the entered text, and the data collection module collects past posting data.

[0341] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[0342] 4. The server uses an emotion engine to recognize emotions such as "positive" and "excited."

[0343] 5. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[0344] 6. The server uses the profanity detection module to check whether the post contains any profanity.

[0345] 7. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[0346] 8. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[0347] 9. Receive user-generated feedback and revise your post as needed.

[0348] Example 2: Lecture content

[0349] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[0350] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[0351] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[0352] 4. The server uses an emotion engine to recognize the emotions "careful" and "thoughtful."

[0353] 5. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[0354] 6. The server checks the script for profanity using the profanity detection module.

[0355] 7. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[0356] 8. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[0357] 9. Take the user-generated feedback and modify the script as needed.

[0358] In this way, users can reassess their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback can be provided in a more user-friendly format.

[0359] The processing flow will be explained below.

[0360] Step 1:

[0361] The user inputs a new message and sends it to the system. For example, the user inputs a message on social media such as "The new product is great!"

[0362] Step 2:

[0363] The server receives the input speech data, saves the received speech content in text format, and converts it into an appropriate format.

[0364] Step 3:

[0365] The server runs a data collection module to collect data on users' past comments. Specifically, the server retrieves information such as users' past social media posts, lectures, and book contents from a database.

[0366] Step 4:

[0367] The server runs a text analysis module to analyze the collected speech data. It uses natural language processing (NLP) techniques to extract important keywords and context from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[0368] Step 5:

[0369] The server runs an emotion engine to recognize emotions from user input. For example, from the input "The new product is great!", emotions such as "positive" and "excited" can be recognized.

[0370] Step 6:

[0371] The server extracts emotional patterns from past speech data and compares them with the content of new speech. It checks for consistency by comparing the emotional patterns in past speech with the emotions of new speech.

[0372] Step 7:

[0373] The server runs a contradiction detection module, which compares the new statement with past statements. It uses an algorithm to detect semantic contradictions, for example, between a past statement "The new product has many problems" and a new statement "The new product is great!"

[0374] Step 8:

[0375] The server runs a profanity detection module to check whether new posts contain inappropriate language or sensitive topics, for example by scanning the text against predefined lists or models to detect inappropriate language or sensitive subjects that should be avoided in public.

[0376] Step 9:

[0377] The server runs a brand image evaluation module to evaluate whether a new comment will damage the brand image, and uses an evaluation model to score the positive / negative impact of the comment.

[0378] Step 10:

[0379] The server runs the feedback generation module and generates feedback to the user based on the analysis results and the output of the emotion engine. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[0380] Step 11:

[0381] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[0382] This allows users to reevaluate their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback content can be provided in a form that is more suited to the user's emotions.

[0383] Example 2

[0384] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0385] In today's digital communications, users' public comments can be inconsistent, contain inappropriate language, or even damage a brand's image. This requires users to carefully evaluate their comments, but this process takes time, effort, and requires specialized knowledge. Furthermore, it is difficult to properly reflect users' emotions in feedback. Technology is needed to solve these issues and enable safe and consistent communication.

[0386] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for collecting past comment data; a means for analyzing the collected comment data and extracting important keywords and context; a means for recognizing emotions from the user's input content and past comment data using an emotion engine; a means for comparing the new comment content with the past comment data and detecting inconsistencies; a means for checking whether the new comment content contains inappropriate expressions or sensitive topics; a means for evaluating the possibility that the new comment content will damage a brand image; a means for generating feedback to the user based on the analysis results and the output of the emotion engine; and a means for the user to receive the feedback and modify the comment content. This allows the user to reevaluate the content of their comment before making it public, thereby achieving consistent and appropriate communication. Furthermore, the user's emotions can be appropriately reflected in the feedback, making it possible to provide more intuitive and adaptable feedback.

[0387] "User" refers to any person or entity that utilizes the system to enter new comments and receive feedback and modify comments.

[0388] "Server" refers to a computer system that processes and analyzes comments received from users and generates feedback.

[0389] "Comment content" refers to the text data that a user inputs into the system.

[0390] "Past utterance data" refers to records of utterances made by the user up to now.

[0391] "Means of collection" refers to the function of retrieving past speech data from databases or other storage media.

[0392] "Means of analysis" refers to the function of analyzing collected speech data using technologies such as natural language processing and extracting important keywords and context.

[0393] "Emotion engine" refers to software or algorithms that recognize emotions from user input and past speech data.

[0394] "Means for detecting contradictions" refers to a function for comparing new statements with past statement data to detect semantic contradictions.

[0395] "Inappropriate language" and "sensitive topics" refer to words or themes that should be avoided in public spaces or expressions that may be legally problematic.

[0396] "Measures to check for inappropriate language" refers to a feature that checks whether new posts contain inappropriate language or sensitive topics.

[0397] "Brand image" refers to consumers' perceptions and evaluations of a company, product, or service.

[0398] "Means for evaluating brand image" refers to the function of evaluating the possibility that new statements will damage brand image.

[0399] "Analysis results" refers to the results of analysis and evaluation of the content of user comments.

[0400] "Means for generating feedback" refers to the function of generating specific advice and recommendations for users based on the analysis results and the output of the emotion engine.

[0401] "Feedback" refers to the analysis results and advice or recommendations based on them that the system provides to the user.

[0402] "Means to correct" refers to the ability for users to receive feedback and adjust or correct what they say based on that feedback.

[0403] MODE FOR CARRYING OUT THE INVENTION

[0404] The present invention is realized by combining an emotion engine with a system that allows users to reevaluate their comments before publishing them and maintain safe and consistent communication. This system is composed of the following specific modules and is executed through data processing between users and a server.

[0405] System configuration details

[0406] 1. User Device

[0407] The user terminal provides an interface for the user to input new comments. This interface includes a text input field and a send button. When the user inputs a comment and clicks the send button, the comment data is sent to the system.

[0408] Examples:

[0409] A user posts a social media post about a new product: "The new product is great!"

[0410] 2. Server Configuration

[0411] The server receives the utterance data sent from the user terminal and performs a series of analysis processes.

[0412] Hardware and software used

[0413] Hardware: High-performance server (general server with CPU, memory, and storage)

[0414] Software: Database Management System (DBMS), Natural Language Processing (NLP) library, Sentiment Analysis Algorithm, Profanity Detection Library, Brand Image Evaluation Model

[0415] Specific Modules

[0416] 1. Data Collection Module:

[0417] The server collects the user's past speech data, which is retrieved from the database and stored for a certain period of time.

[0418] 2. Text Analysis Module:

[0419] The server analyzes the collected speech data and extracts important keywords and context using natural language processing techniques, such as extracting entities and sentiment from text data using NLP libraries (e.g., SpaCy, NLTK).

[0420] 3. Emotion Engine:

[0421] The server recognizes emotions from the user's input and past speech data, and runs a sentiment analysis algorithm to extract positive, negative, excited, and other emotions from the text.

[0422] 4. Conflict Detection Module:

[0423] The server compares new utterances with past utterance data and uses an algorithm to detect semantic inconsistencies.

[0424] 5. Profanity Detection Module:

[0425] The server checks new posts for inappropriate language or sensitive topics. It uses a profanity detection library to scan the text for inappropriate language or topics that should be avoided in public.

[0426] 6. Brand Image Evaluation Module:

[0427] The server evaluates the impact of the comments on the brand image, using a rating model to score the positive / negative impact of the comments.

[0428] 7. Feedback generation module:

[0429] The server generates feedback for the user based on the analysis results and the output of the emotion engine. The feedback includes specific advice and recommendations. For example, it generates a message such as, "The new product is great, but we recommend adding a reference to past issues."

[0430] 3. Receiving feedback and correcting user devices

[0431] The user device receives the feedback generated by the server and presents it to the user, who can then refer to the feedback to modify or adjust the content of their comments and resubmit them to the system.

[0432] Examples:

[0433] Example feedback: "Your new product is great, but I recommend you also mention that this product has caused problems in the past."

[0434] These modules and procedures allow users to reassess their comments before publishing them, ensuring consistent and appropriate communication. Furthermore, incorporating an emotion engine allows for feedback to be provided in a more user-friendly format.

[0435] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0436] Step 1:

[0437] The user enters a new comment

[0438] The user enters the message into the text input field on the terminal and clicks the send button, which sends the message to the system.

[0439] Input: User input (e.g., "Your new product is great!")

[0440] Output: The message is sent to the server.

[0441] Specific behavior:

[0442] The user uses the text input field on the device to enter a new message, and when they click the send button, an HTTP request is sent to the server.

[0443] Step 2:

[0444] The server receives the speech data and performs initial processing.

[0445] The server receives the speech data sent from the user terminal, saves the received data in text format, and converts it into an appropriate format.

[0446] Input: Speech data sent by the user (HTTP request)

[0447] Output: Saved speech data in text format

[0448] Specific behavior:

[0449] The server receives the HTTP request, saves it as text data, and performs the necessary format conversion before storing it in the database.

[0450] Step 3:

[0451] The server collects past speech data

[0452] The server uses the data collection module to obtain the user's past utterance data from the database.

[0453] Input: User ID

[0454] Output: Past speech data (e.g., past social media posts, lecture content)

[0455] Specific behavior:

[0456] The server queries the database to retrieve and search for past comment data based on the user ID, and stores it in temporary storage.

[0457] Step 4:

[0458] The server analyzes the speech data and extracts keywords and context.

[0459] The server uses a text analysis module to analyze past and new utterance data, and uses natural language processing technology to extract important keywords and context.

[0460] Input: Past speech data, new speech data

[0461] Output: Extracted keywords and context (e.g., "new product" or "excitement")

[0462] Specific behavior:

[0463] Analyze text data using a natural language processing (NLP) library (e.g., SpaCy, NLTK), extract entities and sentiment, and convert them into structured data.

[0464] Step 5:

[0465] The server runs the emotion engine to recognize emotions.

[0466] The server executes an emotion engine to recognize the user's emotion from the content of new comments and past comment data.

[0467] Input: New speech data, past speech data

[0468] Output: Recognized emotion data (e.g., "positive" or "excited")

[0469] Specific behavior:

[0470] Apply sentiment analysis algorithms to extract multiple sentiment labels from text, and store the sentiment data in temporary storage.

[0471] Step 6:

[0472] Server performs conflict detection

[0473] The server compares the new utterances with past utterance data to detect semantic inconsistencies.

[0474] Input: New speech data, past speech data

[0475] Output: Flag whether a conflict was detected (e.g. True / False)

[0476] Specific behavior:

[0477] Run a semantic contradiction detection algorithm to check for matches and contradictions with previous statements, and flag any inconsistencies found.

[0478] Step 7:

[0479] Server detects profanity

[0480] The server checks new posts for inappropriate language or sensitive topics using a profanity detection library.

[0481] Input: New speech data

[0482] Output: Profanity flag (e.g. True / False)

[0483] Specific behavior:

[0484] Calls the profanity detection library to scan the content of the post, filters it, and flags any violations.

[0485] Step 8:

[0486] Server evaluates brand image

[0487] The server uses a model to assess the likelihood that new comments will damage the brand image.

[0488] Input: New speech data

[0489] Output: Brand impact score (e.g., positive / negative impact)

[0490] Specific behavior:

[0491] The brand image evaluation algorithm is run to score the content of statements, and the impact of the statements is evaluated based on the score.

[0492] Step 9:

[0493] The server generates feedback

[0494] The server generates specific feedback for the user based on the analysis results and the output of the emotion engine.

[0495] Input: Analysis results, recognized emotion data, contradiction detection results, inappropriate expression detection results, brand image evaluation results

[0496] Output: Specific feedback

[0497] Specific behavior:

[0498] The feedback generation module creates advice based on various analysis results, generates specific recommendations for the comments, and sends them as feedback to the user.

[0499] Step 10:

[0500] Users receive feedback and revise their statements

[0501] The user receives feedback on their device and can correct or adjust what they say as needed.

[0502] Input: Specific feedback

[0503] Output: Corrected statement

[0504] Specific behavior:

[0505] The feedback is displayed on the user's device, and the user can use the feedback to revise their comments and send them back to the system.

[0506] (Application example 2)

[0507] 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."

[0508] Companies need to find ways to prevent inappropriate or brand-damaging remarks from employees when they interact with customers, and maintain consistent and appropriate communication. They also need to consider the impact of what employees say on customer emotions and provide appropriate feedback in real time.

[0509] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for the server to collect the user's past comment data; a means for the server to analyze the collected comment data and extract important keywords and context; a means for the server to compare the new comment content with the past comment data and detect inconsistencies; a means for the server to check whether the new comment content contains inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comment content will damage the brand image; a means for the server to generate feedback based on the analysis results and the user's emotions; a means for the user to receive the feedback and modify the comment content; and a means for employees to use the feedback in real time when dealing with customers. This enables employees to maintain appropriate communication when dealing with customers, providing consistent responses while protecting the brand image. Furthermore, by considering the impact of employee comments on customer emotions in real time and providing appropriate feedback, customer satisfaction can be improved.

[0510] 1. A "user" is someone who uses the system to input comments and receive feedback based on those comments.

[0511] 2. "Utterances" refers to the text or words entered by a user.

[0512] 3. "Server" means a computer system that receives and analyzes input comments and generates appropriate feedback.

[0513] 4. "Past comment data" refers to information about comments made by a user in the past.

[0514] 5. "Keywords" are important words or phrases extracted through text analysis.

[0515] 6. "Context" refers to information that indicates the background and meaning of what is said.

[0516] 7. "Contradiction" refers to a discrepancy between new statements and previous statements.

[0517] 8. "Inappropriate language" refers to words and expressions that should be avoided in public places.

[0518] 9. "Sensitive topics" are those that require special consideration in certain situations or with certain people.

[0519] 10. "Brand image" refers to consumers' impressions and evaluations of a company or product.

[0520] 11. "Means for generating feedback based on emotions" refers to a function that analyzes emotions from the content of a user's comments and provides appropriate feedback based on the results.

[0521] 12. "Real-time means used by employees when interacting with customers" refers to the ability of employees to receive immediate feedback from the system when interacting with customers.

[0522] The system for implementing this invention is designed to receive and analyze user input and generate appropriate feedback. A specific implementation method for this system will be described below.

[0523] System configuration

[0524] The system consists of the following main modules and methods:

[0525] 1. User Device:

[0526] This is a device that allows users to input new comments. This device can be a smartphone or tablet.

[0527] 2. Server:

[0528] Data receiving module: Receives the content of comments sent from the user terminal.

[0529] Data collection module: Collects user's past speech data from the database.

[0530] Text analysis module: Analyzes the collected data and extracts important keywords and context using natural language processing tools (e.g., Spacy and TextBlob).

[0531] Emotion Engine: Recognizes user emotions based on user input and past speech data. Sentiment analysis uses machine learning models such as TextBlob.

[0532] Inconsistency detection module: Compares new statements with past statements to detect inconsistencies.

[0533] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics. Can leverage natural language processing models such as Hugging Face's Transformers.

[0534] Brand image evaluation module: Evaluate the impact of statements on brand image.

[0535] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[0536] User operation procedure

[0537] 1. The user types a new statement, for example, "Can you tell me more about your new product?"

[0538] 2. This statement is sent from the user's device to the server.

[0539] Server Processing

[0540] 1. The server receives the message.

[0541] 2. The server saves the speech content through the data receiving module and converts it into an appropriate format.

[0542] 3. The data collection module collects the user's past utterance data from the database.

[0543] 4. The text analysis module analyzes the collected data and extracts important keywords and context using natural language processing techniques (Spacy, TextBlob).

[0544] 5. The sentiment engine recognizes the emotions expressed by users through their speech. It uses sentiment analysis tools such as TextBlob.

[0545] 6. The contradiction detection module compares the new statement with the previous statement and detects any contradictions.

[0546] 7. The profanity detection module checks new posts for profanity, using tools like Hugging Face Transformers.

[0547] 8. The brand image evaluation module evaluates the impact of the statement on the brand image.

[0548] 9. The feedback generation module generates feedback to the user based on the analysis results and the output of the emotion engine.

[0549] Feedback example:

[0550] In response to a statement such as "Could you please tell us more about your new product?", feedback such as "It would be best to be careful with your choice of words and briefly explain past issues" is generated.

[0551] Prompt Sentence Examples

[0552] Give your users feedback on how they would like to explain the new product to their customers. Evaluate whether it matches your past support record, whether it's inappropriate, and the impact it has on your brand.

[0553] Users (employees) can receive this feedback and modify their statements as necessary to maintain appropriate communication. By using this system, statements made when dealing with customers can be consistent, which will not damage the brand image and improve customer satisfaction.

[0554] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0555] Step 1:

[0556] The user inputs a new statement. This statement is entered in text format and sent from the user's device (smartphone or tablet) to the server. The input is the text entered by the user: "Could you tell me more about the new product?" The output is the data of the statement received by the server.

[0557] Step 2:

[0558] The server receives the message and stores it in an appropriate format, converting it into text for processing by subsequent modules. At this stage, the input is the message sent from the user's terminal, and the output is text data that can be processed within the server.

[0559] Step 3:

[0560] The server executes the data collection module to retrieve the user's past utterance data from the database. At this stage, the input is the user ID, and the output is a set of the user's past utterance data. The server uses this data to perform subsequent analysis.

[0561] Step 4:

[0562] The server runs a text analysis module to extract important keywords and context from past utterance data. It uses natural language processing technologies such as Spacy and TextBlob to analyze the input past utterance data and extract information such as entities, semantic relationships, and sentiment as output.

[0563] Step 5:

[0564] The server runs an emotion engine to recognize the user's emotions based on the user's input and past comment data. It analyzes the input comment using TextBlob or similar tools and outputs an emotion score (positive, negative, neutral, etc.).

[0565] Step 6:

[0566] The server runs a contradiction detection module, which compares the new and previous statements to detect contradictions. Specifically, it uses semantic analysis to compare the meaning of the new and previous statements and detects any contradictions. The input is the new statement and the previous statement data, and the output is the information on the parts where contradictions are detected.

[0567] Step 7:

[0568] The server runs an inappropriate language detection module to check whether new comments contain inappropriate language or sensitive topics. Using tools like Hugging Face's Transformers, the server analyzes the input comments and outputs whether they contain inappropriate language or sensitive topics.

[0569] Step 8:

[0570] The server executes the brand image evaluation module to evaluate the impact of the comment on the brand image. Using the emotion score of the comment, the server scores the impact of the input comment on the brand image as positive or negative, and outputs the score.

[0571] Step 9:

[0572] The server runs the feedback generation module and generates feedback for the user based on the analysis results and the output of the emotion engine. The generated feedback contains specific advice and recommendations. For example, in response to the question, "Could you please tell us more about your new product?", feedback such as, "It would be good to be careful about your choice of words and briefly explain past problems" is generated. The input is the analysis results and emotion score, and the output is the feedback content.

[0573] Step 10:

[0574] The user receives the generated feedback and modifies the utterance as necessary. The feedback is checked on the user's device, and the modified utterance is sent to the server. The input is the feedback sent from the server, and the output is the modified utterance.

[0575] 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.

[0576] 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.

[0577] 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.

[0578] [Second embodiment]

[0579] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0580] 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.

[0581] 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).

[0582] 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.

[0583] 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.

[0584] 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).

[0585] 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.

[0586] 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.

[0587] 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.

[0588] 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.

[0589] 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.

[0590] 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."

[0591] The present invention provides a system for ensuring safe and consistent communication by allowing users to review the content of their statements before they are made public. This system is designed to check whether statements made by individuals or organizations with significant social influence, such as politicians, business leaders, and celebrities, are consistent with past statements, contain inappropriate language, or damage brand image. Specific embodiments of the system are described below.

[0592] System Configuration

[0593] This system consists of the following main modules:

[0594] Data collection module: Collects user's past speech data.

[0595] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[0596] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[0597] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[0598] Brand image evaluation module: Evaluate the impact of statements on brand image.

[0599] Feedback generation module: Generates feedback to the user based on the analysis results.

[0600] Program processing

[0601] 1. User terminal (entering new comments)

[0602] The user inputs a new message and sends it to the system.

[0603] 2. Server (data reception and initial processing)

[0604] The server receives the input speech data and performs initial processing, converting the data into an appropriate format and classifying the type of speech (social media post, lecture content, book, etc.).

[0605] 3. Server (executes data collection modules)

[0606] The server collects past comment data, and retrieves the user's past social media posts, lectures, books, etc. from the database.

[0607] 4. Server (executing text analysis module)

[0608] The server analyzes the collected data to extract important keywords and context, and uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text.

[0609] 5. Server (execution of the inconsistency detection module)

[0610] The server compares the new statement with previous statements to see if there are any inconsistencies. It runs an algorithm to compare previous statements with the new statement and detect semantic inconsistencies.

[0611] 6. Server (execution of profanity detection module)

[0612] The server checks new posts for inappropriate language or sensitive topics by scanning the text against predefined lists and models to detect inappropriate language or sensitive topics.

[0613] 7. Server (Running the brand image evaluation module)

[0614] The server evaluates whether the comment is damaging to the brand image and scores the positive / negative impact of the comment using an evaluation model related to the brand image.

[0615] 8. Server (executing the feedback generation module)

[0616] The server generates feedback to the user based on the analysis results, synthesizing the analysis results and automatically generating a feedback document containing specific advice and recommendations.

[0617] 9. User Device (Receiving and Correcting Feedback)

[0618] The user receives the generated feedback and corrects the statement if necessary. The user checks the feedback, corrects the statement, or re-enters it into the system.

[0619] Specific examples

[0620] Example 1: Posting to social media

[0621] 1. A user enters the text "Excited about the new product" in a social media post.

[0622] 2. The server receives the entered text, and the data collection module collects past posting data.

[0623] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[0624] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[0625] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[0626] 6. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[0627] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[0628] 8. Receive user-generated feedback and revise your post as needed.

[0629] Example 2: Lecture content

[0630] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[0631] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[0632] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[0633] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[0634] 5. The server checks the script for profanity using the profanity detection module.

[0635] 6. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[0636] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[0637] 8. Take the user-generated feedback and modify the script as needed.

[0638] In this way, users can reassess their statements before publishing them to ensure consistent and appropriate communication.

[0639] The processing flow will be explained below.

[0640] Step 1:

[0641] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message on social media saying, "The new product is great!"

[0642] Step 2:

[0643] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[0644] Step 3:

[0645] The server runs a data collection module to collect data on users' past comments. Specifically, it retrieves the user's past social media posts, lecture contents, and book contents from a database.

[0646] Step 4:

[0647] The server runs a text analysis module to analyze the collected speech data. Natural language processing (NLP) techniques are used to extract important keywords, context, entities, and sentiment from the text. For example, keywords such as "new product," "problem," and "excitement" are extracted.

[0648] Step 5:

[0649] The server runs a contradiction detection module, which compares the new statement with past statements. Specifically, it uses an algorithm for detecting semantic contradictions to detect a contradiction between the past statement "The new product has many problems" and the new statement "The new product is great!"

[0650] Step 6:

[0651] The server runs a profanity detection module to check if the new post contains any inappropriate language or sensitive topics, such as inappropriate language or sensitive subjects that should be avoided in public.

[0652] Step 7:

[0653] The server executes the brand image evaluation module to evaluate the impact of the new comment on the brand image, and uses the evaluation model to score the positive / negative impact of the comment.

[0654] Step 8:

[0655] The server runs the feedback generation module and generates feedback to the user based on the analysis results. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[0656] Step 9:

[0657] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[0658] These steps allow users to reassess their content before publishing statements that are consistent and protect their brand image.

[0659] Example 1

[0660] 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."

[0661] Currently, there is a lack of mechanisms for evaluating the appropriateness of user statements before they are made public and for maintaining consistent communication. Furthermore, there are also insufficient means for automatically checking whether statements contradict previous statements, contain inappropriate language, or damage a company's image. These problems are particularly serious for individuals and organizations with significant social influence, such as politicians, corporate leaders, and celebrities. Therefore, the present invention aims to provide a system for reevaluating user statements and ensuring safe and consistent communication.

[0662] 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.

[0663] In this invention, the server includes: a means for a user to input new comments; a means for the server to receive the comments; a means for the server to collect data on past comments from the user; a means for the server to analyze the collected data on comments and extract important keywords and contexts; a means for the server to compare the new comments with the past data on comments and detect semantic inconsistencies; a means for the server to check whether the new comments contain inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comments will damage the company's image; a means for the server to generate feedback to the user based on the analysis results; and a means for the user to receive the feedback and modify the comments. This allows users to reevaluate the content of their comments before making them public, enabling consistent and safe communication.

[0664] "User" refers to a person who uses the system to input comments and receive feedback.

[0665] "Server" refers to a device that includes equipment or software that receives, processes, analyzes, and generates feedback from users.

[0666] "Speech" refers to new text information that a user enters into the system.

[0667] "Past utterance data" refers to the accumulation of text information about utterances made by a user in the past.

[0668] "Important keywords and context" refers to semantically important words and their surrounding context that the text analysis module extracts from speech data.

[0669] A "semantic contradiction" refers to a situation in which a new statement does not semantically match a previous statement.

[0670] "Inappropriate language" refers to language that is socially unacceptable or problematic to use in a particular context.

[0671] "Sensitive topics" refer to subjects or topics that are inappropriate for public disclosure under certain circumstances or social conditions.

[0672] "Corporate image" refers to the overall evaluation and impression of a company from the perspective of general consumers and society.

[0673] "Feedback" refers to information, including advice and recommendations, that the server provides to the user based on the analysis results.

[0674] This invention is a system that allows users to review the content of their statements before they are made public, ensuring safe and consistent communication. In particular, it is designed to check statements made by politicians, business leaders, celebrities, and other influential individuals and organizations to ensure that they are consistent with past statements, contain inappropriate language, and do not damage the company's image.

[0675] The system consists of the following main modules:

[0676] Data collection module: Collects past user comment data. Obtains comment data from users' social media posts, lectures, books, etc.

[0677] Text analysis module: Analyzes collected speech data and extracts important keywords and context. This analysis uses natural language processing (NLP) techniques, specifically tools such as Google's NLP API and SpaCy.

[0678] Inconsistency detection module: Runs algorithms to detect semantic inconsistencies by comparing new statements with previous statements.

[0679] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics by scanning the text using predefined lists and machine learning models.

[0680] Corporate Image Assessment Module: Evaluates whether a statement is damaging to the company's image. Using an assessment model related to corporate image, the module scores the positive / negative impact of the statement.

[0681] Feedback generation module: Generates feedback to users based on the analysis results. Automatically generates specific advice and recommendations.

[0682] The system operates in the following specific steps:

[0683] 1. Entering a new message on the user's terminal: The user enters the new message into the terminal and sends it to the system.

[0684] 2. Data reception and initial processing on the server: The server receives the input speech data, performs initial processing of the data, and classifies the type of speech (SNS post, lecture content, book, etc.).

[0685] 3. Collection of past speech data on the server: The server collects the user's past speech data from the database.

[0686] 4. Text analysis on the server: The server uses NLP techniques to extract important keywords and context.

[0687] 5. Conflict detection on the server: The server compares new statements with previous statements to detect semantic contradictions.

[0688] 6. Server-based profanity detection: The server checks new posts for profanity.

[0689] 7. Corporate image evaluation on the server: The server evaluates the impact of the comments on the corporate image.

[0690] 8. Feedback generation on the server: The server generates feedback to the user based on the analysis results.

[0691] 9. Receiving feedback and correcting content on the user's device: The user receives the generated feedback and corrects the content of the comment if necessary.

[0692] Specific examples

[0693] Example 1: Posting to social media

[0694] 1. A user enters the text "Excited about the new product" in a social media post.

[0695] 2. The server receives the entered text, and the data collection module collects past posting data.

[0696] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[0697] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[0698] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[0699] 6. The server evaluates the impact of this post on the company's image in the company image evaluation module.

[0700] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and will strengthen the company's image."

[0701] 8. Receive user-generated feedback and revise your post as needed.

[0702] Example 2: Lecture content

[0703] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[0704] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[0705] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[0706] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[0707] 5. The server checks the script for profanity using the profanity detection module.

[0708] 6. The server uses the corporate image evaluation module to evaluate the impact of this presentation on the company's image.

[0709] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the company's image, but be careful with certain wording."

[0710] 8. Take the user-generated feedback and modify the script as needed.

[0711] Prompt Sentence Examples

[0712] For social media posts: "Evaluate your social media posts about new products to ensure they are consistent with previous posts, contain no inappropriate language, and fit with the company image."

[0713] For a presentation: "Evaluate the presentation script about our strategy for entering new markets. Check for consistency with previous presentations, inappropriate language, and consistency with our company image."

[0714] These prompts can be used to ask the generative AI model for further evaluation. These examples encourage users to reassess their statements before publishing them, ensuring appropriate and consistent communication.

[0715] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0716] Step 1:

[0717] User enters new comment

[0718] The user inputs new content into the terminal and clicks the "Send" button. For example, the user inputs "I'm looking forward to the release of the new product." The input data is the content of the text message. The output is the text message data sent to the server.

[0719] Step 2:

[0720] The server receives and performs initial processing on the speech data.

[0721] The server receives the text of the comment sent by the user. The data formatter processes this text data, converts it into an appropriate data format, and classifies it by type of comment (social media post, lecture content, book, etc.). The input is the text data sent by the user, and the output is the initially processed data format. Specifically, the data formatter converts the comment content into JSON format.

[0722] Step 3:

[0723] The server collects past utterance data

[0724] The server collects data on users' past comments from a database. For example, it searches the database for social media posts from the past year and retrieves related data. The input is the user's identification information, and the output is collected data on past comments. Specifically, it executes an SQL query to extract the relevant comment data from the database.

[0725] Step 4:

[0726] The server analyzes the collected data

[0727] The server uses natural language processing (NLP) techniques to analyze the collected utterance data. It uses Google's NLP API and SpaCy to extract important keywords and context from the text. The input is the collected utterance data, and the output is the extracted keywords and context. Specifically, the NLP model tokenizes the text and performs named entity recognition (NER).

[0728] Step 5:

[0729] The server compares new and past comments

[0730] The server semantically compares the new utterance with the previous utterance to check for inconsistencies. A comparison algorithm is used to analyze the semantic matches and differences between the new and previous utterances. The input is the new utterance and the previous utterance data, and the output is the result of the inconsistency detection. Specifically, the server calculates context vectors and compares them using cosine similarity.

[0731] Step 6:

[0732] Server detects inappropriate language

[0733] The server checks whether new posts contain inappropriate language or sensitive topics. It scans the text of the post using predefined lists or machine learning models. The input is the new post, and the output is the detection of inappropriate language. Specifically, it compares the text with a list of banned words to see if there are any matches.

[0734] Step 7:

[0735] Server evaluates corporate image

[0736] The server evaluates the impact of new comments on the company's image. It uses an evaluation model related to the company's image to score the positive / negative impact of the comment. The input is the new comment, and the output is an evaluation score. Specifically, the evaluation model performs sentiment analysis on the comment text and scores it.

[0737] Step 8:

[0738] Server generates feedback

[0739] The server generates feedback for the user based on the analysis results. Specific advice and recommendations are automatically generated as a feedback document. The input is the analysis results, and the output is a feedback document. Specifically, the generative AI model generates feedback text based on the input data.

[0740] Step 9:

[0741] Users receive and correct feedback

[0742] The user receives the generated feedback on their device and modifies the utterance as necessary. The input is the feedback document, and the output is the modified utterance. Specifically, the user views the feedback and modifies the utterance in a text editor.

[0743] (Application example 1)

[0744] 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."

[0745] Publicly released comments and statements may lack consistency and contain inappropriate language or legal risks, which could damage the brand image and security policies of companies and celebrities. A system is needed to solve this problem and maintain safe and consistent communication.

[0746] 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.

[0747] In this invention, the server includes: means for a user to input new comments; means for the server to receive the comments; means for the server to collect data on past comments from the user; means for the server to analyze the collected data on comments and extract important keywords and contexts; means for the server to compare the new comments with the past data on comments and detect inconsistencies; means for the server to check whether the new comments contain inappropriate language or sensitive topics; means for the server to evaluate the possibility that the new comments will damage brand image or security policies; means for the server to generate feedback to the user based on the analysis results; and means for the user to receive the feedback and modify the comments. This allows users to evaluate and modify the comments they make public in advance, enabling consistent and appropriate communication.

[0748] A "user" is an individual or group that uses the system to input new comments.

[0749] A "server" is a computer system that analyzes the content of comments received from users and generates feedback based on the analysis results.

[0750] "Comment content" refers to text data such as a comment, script, or post newly entered by a user.

[0751] "Past comment data" is a record of comments and posts made by the user in the past.

[0752] "Important keywords" are words or phrases that characterize the topic and are extracted from the content of the statement or past statement data.

[0753] "Context" refers to the context or background information of what is being said.

[0754] A "contradiction" is a state in which there is a logical inconsistency between the content of a new statement and past statement data.

[0755] "Inappropriate language" is any word or phrase that could pose legal risks or damage a company's reputation.

[0756] A "sensitive topic" is one that is socially, culturally, or legally controversial or carries risk.

[0757] "Brand image" refers to the public reputation or impression of a user or the organization or company to which they belong.

[0758] A "security policy" is a set of guidelines and rules established by an organization or company for the purposes of protecting information and managing risks.

[0759] "Feedback" refers to advice and recommendations to users that are generated based on the analysis results.

[0760] This invention provides a system that allows users to review the content of their comments before publishing them, ensuring consistent and safe communication. This system is particularly useful in the security field, where it can pre-screen comments made by companies and individuals to prevent inappropriate information leaks or comments that violate a company's security policy.

[0761] System Configuration

[0762] This system consists of the following main modules:

[0763] 1. Data Collection Module

[0764] Hardware / Software: Python (version 3.8 or higher), pandas library, SQLAlchemy

[0765] Description: The server collects the user's past speech data from the database and stores it in a data frame format.

[0766] 2. Text Analysis Module

[0767] Hardware / Software: Python (version 3.8 or higher), spaCy library, NLTK

[0768] Description: The server analyzes the speech data and extracts important keywords and entities, as well as analyzing sentiment and semantic relationships.

[0769] 3. Conflict Detection Module

[0770] Hardware / Software: Python (version 3.8 or higher), spaCy, natural language processing model (e.g., BERT)

[0771] Description: The server compares new statements with previous statements to detect logical inconsistencies.

[0772] 4. Profanity Detection Module

[0773] Hardware / Software: Python (version 3.8 or higher), scikit-learn, custom dictionary

[0774] Description: The server checks new posts for legal risks and confidential information.

[0775] 5. Brand Image Evaluation Module

[0776] Hardware / Software: Python (version 3.8 or higher), TensorFlow, scikit-learn

[0777] Description: The server scores new comments on their impact on brand image and security policies.

[0778] 6. Feedback Generation Module

[0779] Hardware / Software: Python (version 3.8 or higher), GPT-3 (OpenAI API)

[0780] Description: The server generates feedback statements from the analysis results and provides specific suggestions to the user.

[0781] Specific processing steps

[0782] 1. The user enters a new comment

[0783] The user inputs new utterances into the terminal.

[0784] 2. The server receives the message

[0785] The server analyzes the content of the message received from the user.

[0786] 3. Data Collection

[0787] The server collects past utterance data from a database.

[0788] 4. Text Analysis

[0789] The server analyzes the speech data and extracts important keywords and context.

[0790] 5. Conflict Detection

[0791] The server compares the new message with previous messages to detect any inconsistencies.

[0792] 6. Profanity Detection

[0793] The server checks new posts for inappropriate language and security risks.

[0794] 7. Brand image evaluation

[0795] The server evaluates the impact of the comments on the brand and security policies.

[0796] 8. Feedback Generation

[0797] The server generates feedback to the user based on the analysis results, prompting them to revise their comments.

[0798] Specific examples

[0799] For example, if a company's security officer types, "I would like to comment on the implementation of a new security system," the server will perform the following process:

[0800] 1. The data collection module collects past speech data (social media posts, news releases, meeting records, etc.).

[0801] 2. The text analysis module analyzes new and past comments and extracts the keywords "security," "system," and "implementation."

[0802] 3. The inconsistency detection module compares the semantic relationships between the new and previous statements to ensure consistency.

[0803] 4. The profanity detection module checks for legal risks and confidential information.

[0804] 5. The brand image evaluation module scores the impact of new statements on security policies and the brand.

[0805] 6. The feedback generation module generates feedback such as "Your comments are appropriate and consistent, but please avoid mentioning specific system names" and displays it to the user.

[0806] Prompt Sentence Examples

[0807] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[0808] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0809] Step 1:

[0810] The user inputs a new statement.

[0811] Input: New statement (e.g., "I would like to comment on the implementation of the new security system")

[0812] Output: What you say is sent to the system.

[0813] Specific operation: The user enters the content of the message into the input form on the terminal and presses the send button.

[0814] Step 2:

[0815] The server receives the message.

[0816] Input: User-submitted message

[0817] Output: The speech is ready for analysis.

[0818] Specific operation: The server receives an HTTP request, converts the content of the comment into an internal data format, and stores it in a database.

[0819] Step 3:

[0820] The server collects past utterance data.

[0821] Input: User ID or metadata related to what was said

[0822] Output: Past speech data (data frame format)

[0823] Specific operation: The server executes a database query to retrieve past comment data, which is then converted into a data frame using the pandas library.

[0824] Step 4:

[0825] The server analyzes the collected speech data and extracts important keywords and contexts.

[0826] Input: Past speech data

[0827] Output: Extracted keywords and context information

[0828] What it does: The server uses the spaCy library to parse text data and extract entities and keywords, as well as perform sentiment analysis and semantic relationships.

[0829] Step 5:

[0830] The server compares new comments with past comments to detect inconsistencies.

[0831] Input: New utterances, extracted keywords and context information

[0832] Output: Conflict detection results

[0833] What happens: The server uses a natural language processing model (e.g., BERT) to compare the semantic relationships between new and previous utterances, detecting any logical inconsistencies.

[0834] Step 6:

[0835] The server checks new posts for inappropriate language or sensitive topics.

[0836] Input: New statement

[0837] Output: Profanity detection results

[0838] What it does: The server uses the scikit-learn library and a custom dictionary to scan posts for profanity and sensitive topics.

[0839] Step 7:

[0840] The server evaluates the likelihood that the comment will damage the brand image or security policy.

[0841] Input: New speech, profanity detection results

[0842] Output: Evaluation results regarding brand image and security policy

[0843] Specific operation: The server uses TensorFlow and scikit-learn to score the impact of statements on brand image and security policies.

[0844] Step 8:

[0845] The server generates feedback to the user based on the analysis results.

[0846] Input: Conflict detection results, inappropriate language detection results, brand image and security policy evaluation results

[0847] Output: Feedback statement

[0848] Specific operation: The server uses GPT-3 (OpenAI API) to generate specific feedback sentences based on the analysis results.

[0849] Step 9:

[0850] The user receives the feedback and corrects the content of the statement.

[0851] Input: Feedback statement

[0852] Output: Corrected statement

[0853] Specific operation: The user checks the feedback text on the terminal, corrects the comment if necessary, and re-enters the corrections into the system.

[0854] Prompt Sentence Examples

[0855] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[0856] 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.

[0857] The present invention is a system that allows users to reevaluate their comments before they are made public, ensuring safe and consistent communication, particularly by combining an emotion engine. This system checks the comments entered by the user for inconsistencies with past comments, inappropriate expressions, and potential damage to the brand image, and also recognizes the user's emotions and adjusts the feedback. A specific embodiment of this system will be described.

[0858] System Configuration

[0859] The system consists of the following main modules:

[0860] Data collection module: Collects user's past speech data.

[0861] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[0862] Emotion engine: Recognizes emotions from user input and past speech data.

[0863] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[0864] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[0865] Brand image evaluation module: Evaluate the impact of statements on brand image.

[0866] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[0867] Program processing

[0868] 1. User terminal (entering new comments)

[0869] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message for posting on social media saying, "The new product is great!"

[0870] 2. Server (data reception and initial processing)

[0871] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[0872] 3. Server (executes data collection modules)

[0873] The server collects past comment data and retrieves the user's past social media posts, lecture contents, and book contents from the database.

[0874] 4. Server (executing text analysis module)

[0875] The server analyzes the collected data and extracts important keywords and context. It uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[0876] 5. Server (execution of emotion engine)

[0877] The server runs an emotion engine to recognize the user's emotions based on the user's input and past utterance data. For example, emotions such as "positive" or "excited" can be recognized from the input.

[0878] 6. Server (execution of the inconsistency detection module)

[0879] The server compares the new statement with previous statements to check for inconsistencies. Specifically, it uses an algorithm to detect semantic inconsistencies and detects a contradiction between the previous statement "The new product has many problems" and the new statement "The new product is great!"

[0880] 7. Server (execution of profanity detection module)

[0881] The server checks new posts for inappropriate language or sensitive topics, for example, detecting inappropriate language or sensitive subjects that should be avoided in public.

[0882] 8. Server (Running the brand image evaluation module)

[0883] The server evaluates whether a comment is damaging to the brand image and uses an evaluation model to score the positive / negative impact of the comment.

[0884] 9. Server (executing the feedback generation module)

[0885] The server generates feedback for the user based on the analysis results and the output of the emotion engine. It automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[0886] 10. User Device (Receiving and Correcting Feedback)

[0887] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[0888] Specific examples

[0889] Example 1: Posting to social media

[0890] 1. A user enters the text "Excited about the new product" in a social media post.

[0891] 2. The server receives the entered text, and the data collection module collects past posting data.

[0892] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[0893] 4. The server uses an emotion engine to recognize emotions such as "positive" and "excited."

[0894] 5. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[0895] 6. The server uses the profanity detection module to check whether the post contains any profanity.

[0896] 7. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[0897] 8. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[0898] 9. Receive user-generated feedback and revise your post as needed.

[0899] Example 2: Lecture content

[0900] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[0901] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[0902] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[0903] 4. The server uses an emotion engine to recognize the emotions "careful" and "thoughtful."

[0904] 5. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[0905] 6. The server checks the script for profanity using the profanity detection module.

[0906] 7. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[0907] 8. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[0908] 9. Take the user-generated feedback and modify the script as needed.

[0909] In this way, users can reassess their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback can be provided in a more user-friendly format.

[0910] The processing flow will be explained below.

[0911] Step 1:

[0912] The user inputs a new message and sends it to the system. For example, the user inputs a message on social media such as "The new product is great!"

[0913] Step 2:

[0914] The server receives the input speech data, saves the received speech content in text format, and converts it into an appropriate format.

[0915] Step 3:

[0916] The server runs a data collection module to collect data on users' past comments. Specifically, the server retrieves information such as users' past social media posts, lectures, and book contents from a database.

[0917] Step 4:

[0918] The server runs a text analysis module to analyze the collected speech data. It uses natural language processing (NLP) techniques to extract important keywords and context from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[0919] Step 5:

[0920] The server runs an emotion engine to recognize emotions from user input. For example, from the input "The new product is great!", emotions such as "positive" and "excited" can be recognized.

[0921] Step 6:

[0922] The server extracts emotional patterns from past speech data and compares them with the content of new speech. It checks for consistency by comparing the emotional patterns in past speech with the emotions of new speech.

[0923] Step 7:

[0924] The server runs a contradiction detection module, which compares the new statement with past statements. It uses an algorithm to detect semantic contradictions, for example, between a past statement "The new product has many problems" and a new statement "The new product is great!"

[0925] Step 8:

[0926] The server runs a profanity detection module to check whether new posts contain inappropriate language or sensitive topics, for example by scanning the text against predefined lists or models to detect inappropriate language or sensitive subjects that should be avoided in public.

[0927] Step 9:

[0928] The server runs a brand image evaluation module to evaluate whether a new comment will damage the brand image, and uses an evaluation model to score the positive / negative impact of the comment.

[0929] Step 10:

[0930] The server runs the feedback generation module and generates feedback to the user based on the analysis results and the output of the emotion engine. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[0931] Step 11:

[0932] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[0933] This allows users to reevaluate their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback content can be provided in a form that is more suited to the user's emotions.

[0934] Example 2

[0935] 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."

[0936] In today's digital communications, users' public comments can be inconsistent, contain inappropriate language, or even damage a brand's image. This requires users to carefully evaluate their comments, but this process takes time, effort, and requires specialized knowledge. Furthermore, it is difficult to properly reflect users' emotions in feedback. Technology is needed to solve these issues and enable safe and consistent communication.

[0937] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for collecting past comment data; a means for analyzing the collected comment data and extracting important keywords and context; a means for recognizing emotions from the user's input content and past comment data using an emotion engine; a means for comparing the new comment content with the past comment data and detecting inconsistencies; a means for checking whether the new comment content contains inappropriate expressions or sensitive topics; a means for evaluating the possibility that the new comment content will damage a brand image; a means for generating feedback to the user based on the analysis results and the output of the emotion engine; and a means for the user to receive the feedback and modify the comment content. This allows the user to reevaluate the content of their comment before making it public, thereby achieving consistent and appropriate communication. Furthermore, the user's emotions can be appropriately reflected in the feedback, making it possible to provide more intuitive and adaptable feedback.

[0938] "User" refers to any person or entity that utilizes the system to enter new comments and receive feedback and modify comments.

[0939] "Server" refers to a computer system that processes and analyzes comments received from users and generates feedback.

[0940] "Comment content" refers to the text data that a user inputs into the system.

[0941] "Past utterance data" refers to records of utterances made by the user up to now.

[0942] "Means of collection" refers to the function of retrieving past speech data from databases or other storage media.

[0943] "Means of analysis" refers to the function of analyzing collected speech data using technologies such as natural language processing and extracting important keywords and context.

[0944] "Emotion engine" refers to software or algorithms that recognize emotions from user input and past speech data.

[0945] "Means for detecting contradictions" refers to a function for comparing new statements with past statement data to detect semantic contradictions.

[0946] "Inappropriate language" and "sensitive topics" refer to words or themes that should be avoided in public spaces or expressions that may be legally problematic.

[0947] "Measures to check for inappropriate language" refers to a feature that checks whether new posts contain inappropriate language or sensitive topics.

[0948] "Brand image" refers to consumers' perceptions and evaluations of a company, product, or service.

[0949] "Means for evaluating brand image" refers to the function of evaluating the possibility that new statements will damage brand image.

[0950] "Analysis results" refers to the results of analysis and evaluation of the content of user comments.

[0951] "Means for generating feedback" refers to the function of generating specific advice and recommendations for users based on the analysis results and the output of the emotion engine.

[0952] "Feedback" refers to the analysis results and advice or recommendations based on them that the system provides to the user.

[0953] "Means to correct" refers to the ability for users to receive feedback and adjust or correct what they say based on that feedback.

[0954] MODE FOR CARRYING OUT THE INVENTION

[0955] The present invention is realized by combining an emotion engine with a system that allows users to reevaluate their comments before publishing them and maintain safe and consistent communication. This system is composed of the following specific modules and is executed through data processing between users and a server.

[0956] System configuration details

[0957] 1. User Device

[0958] The user terminal provides an interface for the user to input new comments. This interface includes a text input field and a send button. When the user inputs a comment and clicks the send button, the comment data is sent to the system.

[0959] Examples:

[0960] A user posts a social media post about a new product: "The new product is great!"

[0961] 2. Server Configuration

[0962] The server receives the utterance data sent from the user terminal and performs a series of analysis processes.

[0963] Hardware and software used

[0964] Hardware: High-performance server (general server with CPU, memory, and storage)

[0965] Software: Database Management System (DBMS), Natural Language Processing (NLP) library, Sentiment Analysis Algorithm, Profanity Detection Library, Brand Image Evaluation Model

[0966] Specific Modules

[0967] 1. Data Collection Module:

[0968] The server collects the user's past speech data, which is retrieved from the database and stored for a certain period of time.

[0969] 2. Text Analysis Module:

[0970] The server analyzes the collected speech data and extracts important keywords and context using natural language processing techniques, such as extracting entities and sentiment from text data using NLP libraries (e.g., SpaCy, NLTK).

[0971] 3. Emotion Engine:

[0972] The server recognizes emotions from the user's input and past speech data, and runs a sentiment analysis algorithm to extract positive, negative, excited, and other emotions from the text.

[0973] 4. Conflict Detection Module:

[0974] The server compares new utterances with past utterance data and uses an algorithm to detect semantic inconsistencies.

[0975] 5. Profanity Detection Module:

[0976] The server checks new posts for inappropriate language or sensitive topics. It uses a profanity detection library to scan the text for inappropriate language or topics that should be avoided in public.

[0977] 6. Brand Image Evaluation Module:

[0978] The server evaluates the impact of the comments on the brand image, using a rating model to score the positive / negative impact of the comments.

[0979] 7. Feedback generation module:

[0980] The server generates feedback for the user based on the analysis results and the output of the emotion engine. The feedback includes specific advice and recommendations. For example, it generates a message such as, "The new product is great, but we recommend adding a reference to past issues."

[0981] 3. Receiving feedback and correcting user devices

[0982] The user device receives the feedback generated by the server and presents it to the user, who can then refer to the feedback to modify or adjust the content of their comments and resubmit them to the system.

[0983] Examples:

[0984] Example feedback: "Your new product is great, but I recommend you also mention that this product has caused problems in the past."

[0985] These modules and procedures allow users to reassess their comments before publishing them, ensuring consistent and appropriate communication. Furthermore, incorporating an emotion engine allows for feedback to be provided in a more user-friendly format.

[0986] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0987] Step 1:

[0988] The user enters a new comment

[0989] The user enters the message into the text input field on the terminal and clicks the send button, which sends the message to the system.

[0990] Input: User input (e.g., "Your new product is great!")

[0991] Output: The message is sent to the server.

[0992] Specific behavior:

[0993] The user uses the text input field on the device to enter a new message, and when they click the send button, an HTTP request is sent to the server.

[0994] Step 2:

[0995] The server receives the speech data and performs initial processing.

[0996] The server receives the speech data sent from the user terminal, saves the received data in text format, and converts it into an appropriate format.

[0997] Input: Speech data sent by the user (HTTP request)

[0998] Output: Saved speech data in text format

[0999] Specific behavior:

[1000] The server receives the HTTP request, saves it as text data, and performs the necessary format conversion before storing it in the database.

[1001] Step 3:

[1002] The server collects past speech data

[1003] The server uses the data collection module to obtain the user's past utterance data from the database.

[1004] Input: User ID

[1005] Output: Past speech data (e.g., past social media posts, lecture content)

[1006] Specific behavior:

[1007] The server queries the database to retrieve and search for past comment data based on the user ID, and stores it in temporary storage.

[1008] Step 4:

[1009] The server analyzes the speech data and extracts keywords and context.

[1010] The server uses a text analysis module to analyze past and new utterance data, and uses natural language processing technology to extract important keywords and context.

[1011] Input: Past speech data, new speech data

[1012] Output: Extracted keywords and context (e.g., "new product" or "excitement")

[1013] Specific behavior:

[1014] Analyze text data using a natural language processing (NLP) library (e.g., SpaCy, NLTK), extract entities and sentiment, and convert them into structured data.

[1015] Step 5:

[1016] The server runs the emotion engine to recognize emotions.

[1017] The server executes an emotion engine to recognize the user's emotion from the content of new comments and past comment data.

[1018] Input: New speech data, past speech data

[1019] Output: Recognized emotion data (e.g., "positive" or "excited")

[1020] Specific behavior:

[1021] Apply sentiment analysis algorithms to extract multiple sentiment labels from text, and store the sentiment data in temporary storage.

[1022] Step 6:

[1023] Server performs conflict detection

[1024] The server compares the new utterances with past utterance data to detect semantic inconsistencies.

[1025] Input: New speech data, past speech data

[1026] Output: Flag whether a conflict was detected (e.g. True / False)

[1027] Specific behavior:

[1028] Run a semantic contradiction detection algorithm to check for matches and contradictions with previous statements, and flag any inconsistencies found.

[1029] Step 7:

[1030] Server detects profanity

[1031] The server checks new posts for inappropriate language or sensitive topics using a profanity detection library.

[1032] Input: New speech data

[1033] Output: Profanity flag (e.g. True / False)

[1034] Specific behavior:

[1035] Calls the profanity detection library to scan the content of the post, filters it, and flags any violations.

[1036] Step 8:

[1037] Server evaluates brand image

[1038] The server uses a model to assess the likelihood that new comments will damage the brand image.

[1039] Input: New speech data

[1040] Output: Brand impact score (e.g., positive / negative impact)

[1041] Specific behavior:

[1042] The brand image evaluation algorithm is run to score the content of statements, and the impact of the statements is evaluated based on the score.

[1043] Step 9:

[1044] The server generates feedback

[1045] The server generates specific feedback for the user based on the analysis results and the output of the emotion engine.

[1046] Input: Analysis results, recognized emotion data, contradiction detection results, inappropriate expression detection results, brand image evaluation results

[1047] Output: Specific feedback

[1048] Specific behavior:

[1049] The feedback generation module creates advice based on various analysis results, generates specific recommendations for the comments, and sends them as feedback to the user.

[1050] Step 10:

[1051] Users receive feedback and revise their statements

[1052] The user receives feedback on their device and can correct or adjust what they say as needed.

[1053] Input: Specific feedback

[1054] Output: Corrected statement

[1055] Specific behavior:

[1056] The feedback is displayed on the user's device, and the user can use the feedback to revise their comments and send them back to the system.

[1057] (Application example 2)

[1058] 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."

[1059] Companies need to find ways to prevent inappropriate or brand-damaging remarks from employees when they interact with customers, and maintain consistent and appropriate communication. They also need to consider the impact of what employees say on customer emotions and provide appropriate feedback in real time.

[1060] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for the server to collect the user's past comment data; a means for the server to analyze the collected comment data and extract important keywords and context; a means for the server to compare the new comment content with the past comment data and detect inconsistencies; a means for the server to check whether the new comment content contains inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comment content will damage the brand image; a means for the server to generate feedback based on the analysis results and the user's emotions; a means for the user to receive the feedback and modify the comment content; and a means for employees to use the feedback in real time when dealing with customers. This enables employees to maintain appropriate communication when dealing with customers, providing consistent responses while protecting the brand image. Furthermore, by considering the impact of employee comments on customer emotions in real time and providing appropriate feedback, customer satisfaction can be improved.

[1061] 1. A "user" is someone who uses the system to input comments and receive feedback based on those comments.

[1062] 2. "Utterances" refers to the text or words entered by a user.

[1063] 3. "Server" means a computer system that receives and analyzes input comments and generates appropriate feedback.

[1064] 4. "Past comment data" refers to information about comments made by a user in the past.

[1065] 5. "Keywords" are important words or phrases extracted through text analysis.

[1066] 6. "Context" refers to information that indicates the background and meaning of what is said.

[1067] 7. "Contradiction" refers to a discrepancy between new statements and previous statements.

[1068] 8. "Inappropriate language" refers to words and expressions that should be avoided in public places.

[1069] 9. "Sensitive topics" are those that require special consideration in certain situations or with certain people.

[1070] 10. "Brand image" refers to consumers' impressions and evaluations of a company or product.

[1071] 11. "Means for generating feedback based on emotions" refers to a function that analyzes emotions from the content of a user's comments and provides appropriate feedback based on the results.

[1072] 12. "Real-time means used by employees when interacting with customers" refers to the ability of employees to receive immediate feedback from the system when interacting with customers.

[1073] The system for implementing this invention is designed to receive and analyze user input and generate appropriate feedback. A specific implementation method for this system will be described below.

[1074] System configuration

[1075] The system consists of the following main modules and methods:

[1076] 1. User Device:

[1077] This is a device that allows users to input new comments. This device can be a smartphone or tablet.

[1078] 2. Server:

[1079] Data receiving module: Receives the content of comments sent from the user terminal.

[1080] Data collection module: Collects user's past speech data from the database.

[1081] Text analysis module: Analyzes the collected data and extracts important keywords and context using natural language processing tools (e.g., Spacy and TextBlob).

[1082] Emotion Engine: Recognizes user emotions based on user input and past speech data. Sentiment analysis uses machine learning models such as TextBlob.

[1083] Inconsistency detection module: Compares new statements with past statements to detect inconsistencies.

[1084] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics. Can leverage natural language processing models such as Hugging Face's Transformers.

[1085] Brand image evaluation module: Evaluate the impact of statements on brand image.

[1086] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[1087] User operation procedure

[1088] 1. The user types a new statement, for example, "Can you tell me more about your new product?"

[1089] 2. This statement is sent from the user's device to the server.

[1090] Server Processing

[1091] 1. The server receives the message.

[1092] 2. The server saves the speech content through the data receiving module and converts it into an appropriate format.

[1093] 3. The data collection module collects the user's past utterance data from the database.

[1094] 4. The text analysis module analyzes the collected data and extracts important keywords and context using natural language processing techniques (Spacy, TextBlob).

[1095] 5. The sentiment engine recognizes the emotions expressed by users through their speech. It uses sentiment analysis tools such as TextBlob.

[1096] 6. The contradiction detection module compares the new statement with the previous statement and detects any contradictions.

[1097] 7. The profanity detection module checks new posts for profanity, using tools like Hugging Face Transformers.

[1098] 8. The brand image evaluation module evaluates the impact of the statement on the brand image.

[1099] 9. The feedback generation module generates feedback to the user based on the analysis results and the output of the emotion engine.

[1100] Feedback example:

[1101] In response to a statement such as "Could you please tell us more about your new product?", feedback such as "It would be best to be careful with your choice of words and briefly explain past issues" is generated.

[1102] Prompt Sentence Examples

[1103] Give your users feedback on how they would like to explain the new product to their customers. Evaluate whether it matches your past support record, whether it's inappropriate, and the impact it has on your brand.

[1104] Users (employees) can receive this feedback and modify their statements as necessary to maintain appropriate communication. By using this system, statements made when dealing with customers can be consistent, which will not damage the brand image and improve customer satisfaction.

[1105] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1106] Step 1:

[1107] The user inputs a new statement. This statement is entered in text format and sent from the user's device (smartphone or tablet) to the server. The input is the text entered by the user: "Could you tell me more about the new product?" The output is the data of the statement received by the server.

[1108] Step 2:

[1109] The server receives the message and stores it in an appropriate format, converting it into text for processing by subsequent modules. At this stage, the input is the message sent from the user's terminal, and the output is text data that can be processed within the server.

[1110] Step 3:

[1111] The server executes the data collection module to retrieve the user's past utterance data from the database. At this stage, the input is the user ID, and the output is a set of the user's past utterance data. The server uses this data to perform subsequent analysis.

[1112] Step 4:

[1113] The server runs a text analysis module to extract important keywords and context from past utterance data. It uses natural language processing technologies such as Spacy and TextBlob to analyze the input past utterance data and extract information such as entities, semantic relationships, and sentiment as output.

[1114] Step 5:

[1115] The server runs an emotion engine to recognize the user's emotions based on the user's input and past comment data. It analyzes the input comment using TextBlob or similar tools and outputs an emotion score (positive, negative, neutral, etc.).

[1116] Step 6:

[1117] The server runs a contradiction detection module, which compares the new and previous statements to detect contradictions. Specifically, it uses semantic analysis to compare the meaning of the new and previous statements and detects any contradictions. The input is the new statement and the previous statement data, and the output is the information on the parts where contradictions are detected.

[1118] Step 7:

[1119] The server runs an inappropriate language detection module to check whether new comments contain inappropriate language or sensitive topics. Using tools like Hugging Face's Transformers, the server analyzes the input comments and outputs whether they contain inappropriate language or sensitive topics.

[1120] Step 8:

[1121] The server executes the brand image evaluation module to evaluate the impact of the comment on the brand image. Using the emotion score of the comment, the server scores the impact of the input comment on the brand image as positive or negative, and outputs the score.

[1122] Step 9:

[1123] The server runs the feedback generation module and generates feedback for the user based on the analysis results and the output of the emotion engine. The generated feedback contains specific advice and recommendations. For example, in response to the question, "Could you please tell us more about your new product?", feedback such as, "It would be good to be careful about your choice of words and briefly explain past problems" is generated. The input is the analysis results and emotion score, and the output is the feedback content.

[1124] Step 10:

[1125] The user receives the generated feedback and modifies the utterance as necessary. The feedback is checked on the user's device, and the modified utterance is sent to the server. The input is the feedback sent from the server, and the output is the modified utterance.

[1126] 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.

[1127] 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.

[1128] 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.

[1129] [Third embodiment]

[1130] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1131] 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.

[1132] 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).

[1133] 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.

[1134] 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.

[1135] 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).

[1136] 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.

[1137] 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.

[1138] 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.

[1139] 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.

[1140] 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.

[1141] 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."

[1142] The present invention provides a system for ensuring safe and consistent communication by allowing users to review the content of their statements before they are made public. This system is designed to check whether statements made by individuals or organizations with significant social influence, such as politicians, business leaders, and celebrities, are consistent with past statements, contain inappropriate language, or damage brand image. Specific embodiments of the system are described below.

[1143] System Configuration

[1144] This system consists of the following main modules:

[1145] Data collection module: Collects user's past speech data.

[1146] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[1147] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[1148] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[1149] Brand image evaluation module: Evaluate the impact of statements on brand image.

[1150] Feedback generation module: Generates feedback to the user based on the analysis results.

[1151] Program processing

[1152] 1. User terminal (entering new comments)

[1153] The user inputs a new message and sends it to the system.

[1154] 2. Server (data reception and initial processing)

[1155] The server receives the input speech data and performs initial processing, converting the data into an appropriate format and classifying the type of speech (social media post, lecture content, book, etc.).

[1156] 3. Server (executes data collection modules)

[1157] The server collects past comment data, and retrieves the user's past social media posts, lectures, books, etc. from the database.

[1158] 4. Server (executing text analysis module)

[1159] The server analyzes the collected data to extract important keywords and context, and uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text.

[1160] 5. Server (execution of the inconsistency detection module)

[1161] The server compares the new statement with previous statements to see if there are any inconsistencies. It runs an algorithm to compare previous statements with the new statement and detect semantic inconsistencies.

[1162] 6. Server (execution of profanity detection module)

[1163] The server checks new posts for inappropriate language or sensitive topics by scanning the text against predefined lists and models to detect inappropriate language or sensitive topics.

[1164] 7. Server (Running the brand image evaluation module)

[1165] The server evaluates whether the comment is damaging to the brand image and scores the positive / negative impact of the comment using an evaluation model related to the brand image.

[1166] 8. Server (executing the feedback generation module)

[1167] The server generates feedback to the user based on the analysis results, synthesizing the analysis results and automatically generating a feedback document containing specific advice and recommendations.

[1168] 9. User Device (Receiving and Correcting Feedback)

[1169] The user receives the generated feedback and corrects the statement if necessary. The user checks the feedback, corrects the statement, or re-enters it into the system.

[1170] Specific examples

[1171] Example 1: Posting to social media

[1172] 1. A user enters the text "Excited about the new product" in a social media post.

[1173] 2. The server receives the entered text, and the data collection module collects past posting data.

[1174] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[1175] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[1176] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[1177] 6. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[1178] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[1179] 8. Receive user-generated feedback and revise your post as needed.

[1180] Example 2: Lecture content

[1181] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[1182] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[1183] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[1184] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[1185] 5. The server checks the script for profanity using the profanity detection module.

[1186] 6. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[1187] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[1188] 8. Take the user-generated feedback and modify the script as needed.

[1189] In this way, users can reassess their statements before publishing them to ensure consistent and appropriate communication.

[1190] The processing flow will be explained below.

[1191] Step 1:

[1192] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message on social media saying, "The new product is great!"

[1193] Step 2:

[1194] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[1195] Step 3:

[1196] The server runs a data collection module to collect data on users' past comments. Specifically, it retrieves the user's past social media posts, lecture contents, and book contents from a database.

[1197] Step 4:

[1198] The server runs a text analysis module to analyze the collected speech data. Natural language processing (NLP) techniques are used to extract important keywords, context, entities, and sentiment from the text. For example, keywords such as "new product," "problem," and "excitement" are extracted.

[1199] Step 5:

[1200] The server runs a contradiction detection module, which compares the new statement with past statements. Specifically, it uses an algorithm for detecting semantic contradictions to detect a contradiction between the past statement "The new product has many problems" and the new statement "The new product is great!"

[1201] Step 6:

[1202] The server runs a profanity detection module to check if the new post contains any inappropriate language or sensitive topics, such as inappropriate language or sensitive subjects that should be avoided in public.

[1203] Step 7:

[1204] The server executes the brand image evaluation module to evaluate the impact of the new comment on the brand image, and uses the evaluation model to score the positive / negative impact of the comment.

[1205] Step 8:

[1206] The server runs the feedback generation module and generates feedback to the user based on the analysis results. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[1207] Step 9:

[1208] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[1209] These steps allow users to reassess their content before publishing statements that are consistent and protect their brand image.

[1210] Example 1

[1211] 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."

[1212] Currently, there is a lack of mechanisms for evaluating the appropriateness of user statements before they are made public and for maintaining consistent communication. Furthermore, there are also insufficient means for automatically checking whether statements contradict previous statements, contain inappropriate language, or damage a company's image. These problems are particularly serious for individuals and organizations with significant social influence, such as politicians, corporate leaders, and celebrities. Therefore, the present invention aims to provide a system for reevaluating user statements and ensuring safe and consistent communication.

[1213] 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.

[1214] In this invention, the server includes: a means for a user to input new comments; a means for the server to receive the comments; a means for the server to collect data on past comments from the user; a means for the server to analyze the collected data on comments and extract important keywords and contexts; a means for the server to compare the new comments with the past data on comments and detect semantic inconsistencies; a means for the server to check whether the new comments contain inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comments will damage the company's image; a means for the server to generate feedback to the user based on the analysis results; and a means for the user to receive the feedback and modify the comments. This allows users to reevaluate the content of their comments before making them public, enabling consistent and safe communication.

[1215] "User" refers to a person who uses the system to input comments and receive feedback.

[1216] "Server" refers to a device that includes equipment or software that receives, processes, analyzes, and generates feedback from users.

[1217] "Speech" refers to new text information that a user enters into the system.

[1218] "Past utterance data" refers to the accumulation of text information about utterances made by a user in the past.

[1219] "Important keywords and context" refers to semantically important words and their surrounding context that the text analysis module extracts from speech data.

[1220] A "semantic contradiction" refers to a situation in which a new statement does not semantically match a previous statement.

[1221] "Inappropriate language" refers to language that is socially unacceptable or problematic to use in a particular context.

[1222] "Sensitive topics" refer to subjects or topics that are inappropriate for public disclosure under certain circumstances or social conditions.

[1223] "Corporate image" refers to the overall evaluation and impression of a company from the perspective of general consumers and society.

[1224] "Feedback" refers to information, including advice and recommendations, that the server provides to the user based on the analysis results.

[1225] This invention is a system that allows users to review the content of their statements before they are made public, ensuring safe and consistent communication. In particular, it is designed to check statements made by politicians, business leaders, celebrities, and other influential individuals and organizations to ensure that they are consistent with past statements, contain inappropriate language, and do not damage the company's image.

[1226] The system consists of the following main modules:

[1227] Data collection module: Collects past user comment data. Obtains comment data from users' social media posts, lectures, books, etc.

[1228] Text analysis module: Analyzes collected speech data and extracts important keywords and context. This analysis uses natural language processing (NLP) techniques, specifically tools such as Google's NLP API and SpaCy.

[1229] Inconsistency detection module: Runs algorithms to detect semantic inconsistencies by comparing new statements with previous statements.

[1230] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics by scanning the text using predefined lists and machine learning models.

[1231] Corporate Image Assessment Module: Evaluates whether a statement is damaging to the company's image. Using an assessment model related to corporate image, the module scores the positive / negative impact of the statement.

[1232] Feedback generation module: Generates feedback to users based on the analysis results. Automatically generates specific advice and recommendations.

[1233] The system operates in the following specific steps:

[1234] 1. Entering a new message on the user's terminal: The user enters the new message into the terminal and sends it to the system.

[1235] 2. Data reception and initial processing on the server: The server receives the input speech data, performs initial processing of the data, and classifies the type of speech (SNS post, lecture content, book, etc.).

[1236] 3. Collection of past speech data on the server: The server collects the user's past speech data from the database.

[1237] 4. Text analysis on the server: The server uses NLP techniques to extract important keywords and context.

[1238] 5. Conflict detection on the server: The server compares new statements with previous statements to detect semantic contradictions.

[1239] 6. Server-based profanity detection: The server checks new posts for profanity.

[1240] 7. Corporate image evaluation on the server: The server evaluates the impact of the comments on the corporate image.

[1241] 8. Feedback generation on the server: The server generates feedback to the user based on the analysis results.

[1242] 9. Receiving feedback and correcting content on the user's device: The user receives the generated feedback and corrects the content of the comment if necessary.

[1243] Specific examples

[1244] Example 1: Posting to social media

[1245] 1. A user enters the text "Excited about the new product" in a social media post.

[1246] 2. The server receives the entered text, and the data collection module collects past posting data.

[1247] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[1248] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[1249] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[1250] 6. The server evaluates the impact of this post on the company's image in the company image evaluation module.

[1251] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and will strengthen the company's image."

[1252] 8. Receive user-generated feedback and revise your post as needed.

[1253] Example 2: Lecture content

[1254] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[1255] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[1256] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[1257] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[1258] 5. The server checks the script for profanity using the profanity detection module.

[1259] 6. The server uses the corporate image evaluation module to evaluate the impact of this presentation on the company's image.

[1260] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the company's image, but be careful with certain wording."

[1261] 8. Take the user-generated feedback and modify the script as needed.

[1262] Prompt Sentence Examples

[1263] For social media posts: "Evaluate your social media posts about new products to ensure they are consistent with previous posts, contain no inappropriate language, and fit with the company image."

[1264] For a presentation: "Evaluate the presentation script about our strategy for entering new markets. Check for consistency with previous presentations, inappropriate language, and consistency with our company image."

[1265] These prompts can be used to ask the generative AI model for further evaluation. These examples encourage users to reassess their statements before publishing them, ensuring appropriate and consistent communication.

[1266] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1267] Step 1:

[1268] User enters new comment

[1269] The user inputs new content into the terminal and clicks the "Send" button. For example, the user inputs "I'm looking forward to the release of the new product." The input data is the content of the text message. The output is the text message data sent to the server.

[1270] Step 2:

[1271] The server receives and performs initial processing on the speech data.

[1272] The server receives the text of the comment sent by the user. The data formatter processes this text data, converts it into an appropriate data format, and classifies it by type of comment (social media post, lecture content, book, etc.). The input is the text data sent by the user, and the output is the initially processed data format. Specifically, the data formatter converts the comment content into JSON format.

[1273] Step 3:

[1274] The server collects past utterance data

[1275] The server collects data on users' past comments from a database. For example, it searches the database for social media posts from the past year and retrieves related data. The input is the user's identification information, and the output is collected data on past comments. Specifically, it executes an SQL query to extract the relevant comment data from the database.

[1276] Step 4:

[1277] The server analyzes the collected data

[1278] The server uses natural language processing (NLP) techniques to analyze the collected utterance data. It uses Google's NLP API and SpaCy to extract important keywords and context from the text. The input is the collected utterance data, and the output is the extracted keywords and context. Specifically, the NLP model tokenizes the text and performs named entity recognition (NER).

[1279] Step 5:

[1280] The server compares new and past comments

[1281] The server semantically compares the new utterance with the previous utterance to check for inconsistencies. A comparison algorithm is used to analyze the semantic matches and differences between the new and previous utterances. The input is the new utterance and the previous utterance data, and the output is the result of the inconsistency detection. Specifically, the server calculates context vectors and compares them using cosine similarity.

[1282] Step 6:

[1283] Server detects inappropriate language

[1284] The server checks whether new posts contain inappropriate language or sensitive topics. It scans the text of the post using predefined lists or machine learning models. The input is the new post, and the output is the detection of inappropriate language. Specifically, it compares the text with a list of banned words to see if there are any matches.

[1285] Step 7:

[1286] Server evaluates corporate image

[1287] The server evaluates the impact of new comments on the company's image. It uses an evaluation model related to the company's image to score the positive / negative impact of the comment. The input is the new comment, and the output is an evaluation score. Specifically, the evaluation model performs sentiment analysis on the comment text and scores it.

[1288] Step 8:

[1289] Server generates feedback

[1290] The server generates feedback for the user based on the analysis results. Specific advice and recommendations are automatically generated as a feedback document. The input is the analysis results, and the output is a feedback document. Specifically, the generative AI model generates feedback text based on the input data.

[1291] Step 9:

[1292] Users receive and correct feedback

[1293] The user receives the generated feedback on their device and modifies the utterance as necessary. The input is the feedback document, and the output is the modified utterance. Specifically, the user views the feedback and modifies the utterance in a text editor.

[1294] (Application example 1)

[1295] 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."

[1296] Publicly released comments and statements may lack consistency and contain inappropriate language or legal risks, which could damage the brand image and security policies of companies and celebrities. A system is needed to solve this problem and maintain safe and consistent communication.

[1297] 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.

[1298] In this invention, the server includes: means for a user to input new comments; means for the server to receive the comments; means for the server to collect data on past comments from the user; means for the server to analyze the collected data on comments and extract important keywords and contexts; means for the server to compare the new comments with the past data on comments and detect inconsistencies; means for the server to check whether the new comments contain inappropriate language or sensitive topics; means for the server to evaluate the possibility that the new comments will damage brand image or security policies; means for the server to generate feedback to the user based on the analysis results; and means for the user to receive the feedback and modify the comments. This allows users to evaluate and modify the comments they make public in advance, enabling consistent and appropriate communication.

[1299] A "user" is an individual or group that uses the system to input new comments.

[1300] A "server" is a computer system that analyzes the content of comments received from users and generates feedback based on the analysis results.

[1301] "Comment content" refers to text data such as a comment, script, or post newly entered by a user.

[1302] "Past comment data" is a record of comments and posts made by the user in the past.

[1303] "Important keywords" are words or phrases that characterize the topic and are extracted from the content of the statement or past statement data.

[1304] "Context" refers to the context or background information of what is being said.

[1305] A "contradiction" is a state in which there is a logical inconsistency between the content of a new statement and past statement data.

[1306] "Inappropriate language" is any word or phrase that could pose legal risks or damage a company's reputation.

[1307] A "sensitive topic" is one that is socially, culturally, or legally controversial or carries risk.

[1308] "Brand image" refers to the public reputation or impression of a user or the organization or company to which they belong.

[1309] A "security policy" is a set of guidelines and rules established by an organization or company for the purposes of protecting information and managing risks.

[1310] "Feedback" refers to advice and recommendations to users that are generated based on the analysis results.

[1311] This invention provides a system that allows users to review the content of their comments before publishing them, ensuring consistent and safe communication. This system is particularly useful in the security field, where it can pre-screen comments made by companies and individuals to prevent inappropriate information leaks or comments that violate a company's security policy.

[1312] System Configuration

[1313] This system consists of the following main modules:

[1314] 1. Data Collection Module

[1315] Hardware / Software: Python (version 3.8 or higher), pandas library, SQLAlchemy

[1316] Description: The server collects the user's past speech data from the database and stores it in a data frame format.

[1317] 2. Text Analysis Module

[1318] Hardware / Software: Python (version 3.8 or higher), spaCy library, NLTK

[1319] Description: The server analyzes the speech data and extracts important keywords and entities, as well as analyzing sentiment and semantic relationships.

[1320] 3. Conflict Detection Module

[1321] Hardware / Software: Python (version 3.8 or higher), spaCy, natural language processing model (e.g., BERT)

[1322] Description: The server compares new statements with previous statements to detect logical inconsistencies.

[1323] 4. Profanity Detection Module

[1324] Hardware / Software: Python (version 3.8 or higher), scikit-learn, custom dictionary

[1325] Description: The server checks new posts for legal risks and confidential information.

[1326] 5. Brand Image Evaluation Module

[1327] Hardware / Software: Python (version 3.8 or higher), TensorFlow, scikit-learn

[1328] Description: The server scores new comments on their impact on brand image and security policies.

[1329] 6. Feedback Generation Module

[1330] Hardware / Software: Python (version 3.8 or higher), GPT-3 (OpenAI API)

[1331] Description: The server generates feedback statements from the analysis results and provides specific suggestions to the user.

[1332] Specific processing steps

[1333] 1. The user enters a new comment

[1334] The user inputs new utterances into the terminal.

[1335] 2. The server receives the message

[1336] The server analyzes the content of the message received from the user.

[1337] 3. Data Collection

[1338] The server collects past utterance data from a database.

[1339] 4. Text Analysis

[1340] The server analyzes the speech data and extracts important keywords and context.

[1341] 5. Conflict Detection

[1342] The server compares the new message with previous messages to detect any inconsistencies.

[1343] 6. Profanity Detection

[1344] The server checks new posts for inappropriate language and security risks.

[1345] 7. Brand image evaluation

[1346] The server evaluates the impact of the comments on the brand and security policies.

[1347] 8. Feedback Generation

[1348] The server generates feedback to the user based on the analysis results, prompting them to revise their comments.

[1349] Specific examples

[1350] For example, if a company's security officer types, "I would like to comment on the implementation of a new security system," the server will perform the following process:

[1351] 1. The data collection module collects past speech data (social media posts, news releases, meeting records, etc.).

[1352] 2. The text analysis module analyzes new and past comments and extracts the keywords "security," "system," and "implementation."

[1353] 3. The inconsistency detection module compares the semantic relationships between the new and previous statements to ensure consistency.

[1354] 4. The profanity detection module checks for legal risks and confidential information.

[1355] 5. The brand image evaluation module scores the impact of new statements on security policies and the brand.

[1356] 6. The feedback generation module generates feedback such as "Your comments are appropriate and consistent, but please avoid mentioning specific system names" and displays it to the user.

[1357] Prompt Sentence Examples

[1358] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[1359] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1360] Step 1:

[1361] The user inputs a new statement.

[1362] Input: New statement (e.g., "I would like to comment on the implementation of the new security system")

[1363] Output: What you say is sent to the system.

[1364] Specific operation: The user enters the content of the message into the input form on the terminal and presses the send button.

[1365] Step 2:

[1366] The server receives the message.

[1367] Input: User-submitted message

[1368] Output: The speech is ready for analysis.

[1369] Specific operation: The server receives an HTTP request, converts the content of the comment into an internal data format, and stores it in a database.

[1370] Step 3:

[1371] The server collects past utterance data.

[1372] Input: User ID or metadata related to what was said

[1373] Output: Past speech data (data frame format)

[1374] Specific operation: The server executes a database query to retrieve past comment data, which is then converted into a data frame using the pandas library.

[1375] Step 4:

[1376] The server analyzes the collected speech data and extracts important keywords and contexts.

[1377] Input: Past speech data

[1378] Output: Extracted keywords and context information

[1379] What it does: The server uses the spaCy library to parse text data and extract entities and keywords, as well as perform sentiment analysis and semantic relationships.

[1380] Step 5:

[1381] The server compares new comments with past comments to detect inconsistencies.

[1382] Input: New utterances, extracted keywords and context information

[1383] Output: Conflict detection results

[1384] What happens: The server uses a natural language processing model (e.g., BERT) to compare the semantic relationships between new and previous utterances, detecting any logical inconsistencies.

[1385] Step 6:

[1386] The server checks new posts for inappropriate language or sensitive topics.

[1387] Input: New statement

[1388] Output: Profanity detection results

[1389] What it does: The server uses the scikit-learn library and a custom dictionary to scan posts for profanity and sensitive topics.

[1390] Step 7:

[1391] The server evaluates the likelihood that the comment will damage the brand image or security policy.

[1392] Input: New speech, profanity detection results

[1393] Output: Evaluation results regarding brand image and security policy

[1394] Specific operation: The server uses TensorFlow and scikit-learn to score the impact of statements on brand image and security policies.

[1395] Step 8:

[1396] The server generates feedback to the user based on the analysis results.

[1397] Input: Conflict detection results, inappropriate language detection results, brand image and security policy evaluation results

[1398] Output: Feedback statement

[1399] Specific operation: The server uses GPT-3 (OpenAI API) to generate specific feedback sentences based on the analysis results.

[1400] Step 9:

[1401] The user receives the feedback and corrects the content of the statement.

[1402] Input: Feedback statement

[1403] Output: Corrected statement

[1404] Specific operation: The user checks the feedback text on the terminal, corrects the comment if necessary, and re-enters the corrections into the system.

[1405] Prompt Sentence Examples

[1406] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[1407] 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.

[1408] The present invention is a system that allows users to reevaluate their comments before they are made public, ensuring safe and consistent communication, particularly by combining an emotion engine. This system checks the comments entered by the user for inconsistencies with past comments, inappropriate expressions, and potential damage to the brand image, and also recognizes the user's emotions and adjusts the feedback. A specific embodiment of this system will be described.

[1409] System Configuration

[1410] The system consists of the following main modules:

[1411] Data collection module: Collects user's past speech data.

[1412] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[1413] Emotion engine: Recognizes emotions from user input and past speech data.

[1414] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[1415] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[1416] Brand image evaluation module: Evaluate the impact of statements on brand image.

[1417] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[1418] Program processing

[1419] 1. User terminal (entering new comments)

[1420] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message for posting on social media saying, "The new product is great!"

[1421] 2. Server (data reception and initial processing)

[1422] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[1423] 3. Server (executes data collection modules)

[1424] The server collects past comment data and retrieves the user's past social media posts, lecture contents, and book contents from the database.

[1425] 4. Server (executing text analysis module)

[1426] The server analyzes the collected data and extracts important keywords and context. It uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[1427] 5. Server (execution of emotion engine)

[1428] The server runs an emotion engine to recognize the user's emotions based on the user's input and past utterance data. For example, emotions such as "positive" or "excited" can be recognized from the input.

[1429] 6. Server (execution of the inconsistency detection module)

[1430] The server compares the new statement with previous statements to check for inconsistencies. Specifically, it uses an algorithm to detect semantic inconsistencies and detects a contradiction between the previous statement "The new product has many problems" and the new statement "The new product is great!"

[1431] 7. Server (execution of profanity detection module)

[1432] The server checks new posts for inappropriate language or sensitive topics, for example, detecting inappropriate language or sensitive subjects that should be avoided in public.

[1433] 8. Server (Running the brand image evaluation module)

[1434] The server evaluates whether a comment is damaging to the brand image and uses an evaluation model to score the positive / negative impact of the comment.

[1435] 9. Server (executing the feedback generation module)

[1436] The server generates feedback for the user based on the analysis results and the output of the emotion engine. It automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[1437] 10. User Device (Receiving and Correcting Feedback)

[1438] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[1439] Specific examples

[1440] Example 1: Posting to social media

[1441] 1. A user enters the text "Excited about the new product" in a social media post.

[1442] 2. The server receives the entered text, and the data collection module collects past posting data.

[1443] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[1444] 4. The server uses an emotion engine to recognize emotions such as "positive" and "excited."

[1445] 5. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[1446] 6. The server uses the profanity detection module to check whether the post contains any profanity.

[1447] 7. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[1448] 8. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[1449] 9. Receive user-generated feedback and revise your post as needed.

[1450] Example 2: Lecture content

[1451] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[1452] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[1453] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[1454] 4. The server uses an emotion engine to recognize the emotions "careful" and "thoughtful."

[1455] 5. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[1456] 6. The server checks the script for profanity using the profanity detection module.

[1457] 7. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[1458] 8. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[1459] 9. Take the user-generated feedback and modify the script as needed.

[1460] In this way, users can reassess their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback can be provided in a more user-friendly format.

[1461] The processing flow will be explained below.

[1462] Step 1:

[1463] The user inputs a new message and sends it to the system. For example, the user inputs a message on social media such as "The new product is great!"

[1464] Step 2:

[1465] The server receives the input speech data, saves the received speech content in text format, and converts it into an appropriate format.

[1466] Step 3:

[1467] The server runs a data collection module to collect data on users' past comments. Specifically, the server retrieves information such as users' past social media posts, lectures, and book contents from a database.

[1468] Step 4:

[1469] The server runs a text analysis module to analyze the collected speech data. It uses natural language processing (NLP) techniques to extract important keywords and context from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[1470] Step 5:

[1471] The server runs an emotion engine to recognize emotions from user input. For example, from the input "The new product is great!", emotions such as "positive" and "excited" can be recognized.

[1472] Step 6:

[1473] The server extracts emotional patterns from past speech data and compares them with the content of new speech. It checks for consistency by comparing the emotional patterns in past speech with the emotions of new speech.

[1474] Step 7:

[1475] The server runs a contradiction detection module, which compares the new statement with past statements. It uses an algorithm to detect semantic contradictions, for example, between a past statement "The new product has many problems" and a new statement "The new product is great!"

[1476] Step 8:

[1477] The server runs a profanity detection module to check whether new posts contain inappropriate language or sensitive topics, for example by scanning the text against predefined lists or models to detect inappropriate language or sensitive subjects that should be avoided in public.

[1478] Step 9:

[1479] The server runs a brand image evaluation module to evaluate whether a new comment will damage the brand image, and uses an evaluation model to score the positive / negative impact of the comment.

[1480] Step 10:

[1481] The server runs the feedback generation module and generates feedback to the user based on the analysis results and the output of the emotion engine. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[1482] Step 11:

[1483] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[1484] This allows users to reevaluate their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback content can be provided in a form that is more suited to the user's emotions.

[1485] Example 2

[1486] 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."

[1487] In today's digital communications, users' public comments can be inconsistent, contain inappropriate language, or even damage a brand's image. This requires users to carefully evaluate their comments, but this process takes time, effort, and requires specialized knowledge. Furthermore, it is difficult to properly reflect users' emotions in feedback. Technology is needed to solve these issues and enable safe and consistent communication.

[1488] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for collecting past comment data; a means for analyzing the collected comment data and extracting important keywords and context; a means for recognizing emotions from the user's input content and past comment data using an emotion engine; a means for comparing the new comment content with the past comment data and detecting inconsistencies; a means for checking whether the new comment content contains inappropriate expressions or sensitive topics; a means for evaluating the possibility that the new comment content will damage a brand image; a means for generating feedback to the user based on the analysis results and the output of the emotion engine; and a means for the user to receive the feedback and modify the comment content. This allows the user to reevaluate the content of their comment before making it public, thereby achieving consistent and appropriate communication. Furthermore, the user's emotions can be appropriately reflected in the feedback, making it possible to provide more intuitive and adaptable feedback.

[1489] "User" refers to any person or entity that utilizes the system to enter new comments and receive feedback and modify comments.

[1490] "Server" refers to a computer system that processes and analyzes comments received from users and generates feedback.

[1491] "Comment content" refers to the text data that a user inputs into the system.

[1492] "Past utterance data" refers to records of utterances made by the user up to now.

[1493] "Means of collection" refers to the function of retrieving past speech data from databases or other storage media.

[1494] "Means of analysis" refers to the function of analyzing collected speech data using technologies such as natural language processing and extracting important keywords and context.

[1495] "Emotion engine" refers to software or algorithms that recognize emotions from user input and past speech data.

[1496] "Means for detecting contradictions" refers to a function for comparing new statements with past statement data to detect semantic contradictions.

[1497] "Inappropriate language" and "sensitive topics" refer to words or themes that should be avoided in public spaces or expressions that may be legally problematic.

[1498] "Measures to check for inappropriate language" refers to a feature that checks whether new posts contain inappropriate language or sensitive topics.

[1499] "Brand image" refers to consumers' perceptions and evaluations of a company, product, or service.

[1500] "Means for evaluating brand image" refers to the function of evaluating the possibility that new statements will damage brand image.

[1501] "Analysis results" refers to the results of analysis and evaluation of the content of user comments.

[1502] "Means for generating feedback" refers to the function of generating specific advice and recommendations for users based on the analysis results and the output of the emotion engine.

[1503] "Feedback" refers to the analysis results and advice or recommendations based on them that the system provides to the user.

[1504] "Means to correct" refers to the ability for users to receive feedback and adjust or correct what they say based on that feedback.

[1505] MODE FOR CARRYING OUT THE INVENTION

[1506] The present invention is realized by combining an emotion engine with a system that allows users to reevaluate their comments before publishing them and maintain safe and consistent communication. This system is composed of the following specific modules and is executed through data processing between users and a server.

[1507] System configuration details

[1508] 1. User Device

[1509] The user terminal provides an interface for the user to input new comments. This interface includes a text input field and a send button. When the user inputs a comment and clicks the send button, the comment data is sent to the system.

[1510] Examples:

[1511] A user posts a social media post about a new product: "The new product is great!"

[1512] 2. Server Configuration

[1513] The server receives the utterance data sent from the user terminal and performs a series of analysis processes.

[1514] Hardware and software used

[1515] Hardware: High-performance server (general server with CPU, memory, and storage)

[1516] Software: Database Management System (DBMS), Natural Language Processing (NLP) library, Sentiment Analysis Algorithm, Profanity Detection Library, Brand Image Evaluation Model

[1517] Specific Modules

[1518] 1. Data Collection Module:

[1519] The server collects the user's past speech data, which is retrieved from the database and stored for a certain period of time.

[1520] 2. Text Analysis Module:

[1521] The server analyzes the collected speech data and extracts important keywords and context using natural language processing techniques, such as extracting entities and sentiment from text data using NLP libraries (e.g., SpaCy, NLTK).

[1522] 3. Emotion Engine:

[1523] The server recognizes emotions from the user's input and past speech data, and runs a sentiment analysis algorithm to extract positive, negative, excited, and other emotions from the text.

[1524] 4. Conflict Detection Module:

[1525] The server compares new utterances with past utterance data and uses an algorithm to detect semantic inconsistencies.

[1526] 5. Profanity Detection Module:

[1527] The server checks new posts for inappropriate language or sensitive topics. It uses a profanity detection library to scan the text for inappropriate language or topics that should be avoided in public.

[1528] 6. Brand Image Evaluation Module:

[1529] The server evaluates the impact of the comments on the brand image, using a rating model to score the positive / negative impact of the comments.

[1530] 7. Feedback generation module:

[1531] The server generates feedback for the user based on the analysis results and the output of the emotion engine. The feedback includes specific advice and recommendations. For example, it generates a message such as, "The new product is great, but we recommend adding a reference to past issues."

[1532] 3. Receiving feedback and correcting user devices

[1533] The user device receives the feedback generated by the server and presents it to the user, who can then refer to the feedback to modify or adjust the content of their comments and resubmit them to the system.

[1534] Examples:

[1535] Example feedback: "Your new product is great, but I recommend you also mention that this product has caused problems in the past."

[1536] These modules and procedures allow users to reassess their comments before publishing them, ensuring consistent and appropriate communication. Furthermore, incorporating an emotion engine allows for feedback to be provided in a more user-friendly format.

[1537] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1538] Step 1:

[1539] The user enters a new comment

[1540] The user enters the message into the text input field on the terminal and clicks the send button, which sends the message to the system.

[1541] Input: User input (e.g., "Your new product is great!")

[1542] Output: The message is sent to the server.

[1543] Specific behavior:

[1544] The user uses the text input field on the device to enter a new message, and when they click the send button, an HTTP request is sent to the server.

[1545] Step 2:

[1546] The server receives the speech data and performs initial processing.

[1547] The server receives the speech data sent from the user terminal, saves the received data in text format, and converts it into an appropriate format.

[1548] Input: Speech data sent by the user (HTTP request)

[1549] Output: Saved speech data in text format

[1550] Specific behavior:

[1551] The server receives the HTTP request, saves it as text data, and performs the necessary format conversion before storing it in the database.

[1552] Step 3:

[1553] The server collects past speech data

[1554] The server uses the data collection module to obtain the user's past utterance data from the database.

[1555] Input: User ID

[1556] Output: Past speech data (e.g., past social media posts, lecture content)

[1557] Specific behavior:

[1558] The server queries the database to retrieve and search for past comment data based on the user ID, and stores it in temporary storage.

[1559] Step 4:

[1560] The server analyzes the speech data and extracts keywords and context.

[1561] The server uses a text analysis module to analyze past and new utterance data, and uses natural language processing technology to extract important keywords and context.

[1562] Input: Past speech data, new speech data

[1563] Output: Extracted keywords and context (e.g., "new product" or "excitement")

[1564] Specific behavior:

[1565] Analyze text data using a natural language processing (NLP) library (e.g., SpaCy, NLTK), extract entities and sentiment, and convert them into structured data.

[1566] Step 5:

[1567] The server runs the emotion engine to recognize emotions.

[1568] The server executes an emotion engine to recognize the user's emotion from the content of new comments and past comment data.

[1569] Input: New speech data, past speech data

[1570] Output: Recognized emotion data (e.g., "positive" or "excited")

[1571] Specific behavior:

[1572] Apply sentiment analysis algorithms to extract multiple sentiment labels from text, and store the sentiment data in temporary storage.

[1573] Step 6:

[1574] Server performs conflict detection

[1575] The server compares the new utterances with past utterance data to detect semantic inconsistencies.

[1576] Input: New speech data, past speech data

[1577] Output: Flag whether a conflict was detected (e.g. True / False)

[1578] Specific behavior:

[1579] Run a semantic contradiction detection algorithm to check for matches and contradictions with previous statements, and flag any inconsistencies found.

[1580] Step 7:

[1581] Server detects profanity

[1582] The server checks new posts for inappropriate language or sensitive topics using a profanity detection library.

[1583] Input: New speech data

[1584] Output: Profanity flag (e.g. True / False)

[1585] Specific behavior:

[1586] Calls the profanity detection library to scan the content of the post, filters it, and flags any violations.

[1587] Step 8:

[1588] Server evaluates brand image

[1589] The server uses a model to assess the likelihood that new comments will damage the brand image.

[1590] Input: New speech data

[1591] Output: Brand impact score (e.g., positive / negative impact)

[1592] Specific behavior:

[1593] The brand image evaluation algorithm is run to score the content of statements, and the impact of the statements is evaluated based on the score.

[1594] Step 9:

[1595] The server generates feedback

[1596] The server generates specific feedback for the user based on the analysis results and the output of the emotion engine.

[1597] Input: Analysis results, recognized emotion data, contradiction detection results, inappropriate expression detection results, brand image evaluation results

[1598] Output: Specific feedback

[1599] Specific behavior:

[1600] The feedback generation module creates advice based on various analysis results, generates specific recommendations for the comments, and sends them as feedback to the user.

[1601] Step 10:

[1602] Users receive feedback and revise their statements

[1603] The user receives feedback on their device and can correct or adjust what they say as needed.

[1604] Input: Specific feedback

[1605] Output: Corrected statement

[1606] Specific behavior:

[1607] The feedback is displayed on the user's device, and the user can use the feedback to revise their comments and send them back to the system.

[1608] (Application example 2)

[1609] 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."

[1610] Companies need to find ways to prevent inappropriate or brand-damaging remarks from employees when they interact with customers, and maintain consistent and appropriate communication. They also need to consider the impact of what employees say on customer emotions and provide appropriate feedback in real time.

[1611] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for the server to collect the user's past comment data; a means for the server to analyze the collected comment data and extract important keywords and context; a means for the server to compare the new comment content with the past comment data and detect inconsistencies; a means for the server to check whether the new comment content contains inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comment content will damage the brand image; a means for the server to generate feedback based on the analysis results and the user's emotions; a means for the user to receive the feedback and modify the comment content; and a means for employees to use the feedback in real time when dealing with customers. This enables employees to maintain appropriate communication when dealing with customers, providing consistent responses while protecting the brand image. Furthermore, by considering the impact of employee comments on customer emotions in real time and providing appropriate feedback, customer satisfaction can be improved.

[1612] 1. A "user" is someone who uses the system to input comments and receive feedback based on those comments.

[1613] 2. "Utterances" refers to the text or words entered by a user.

[1614] 3. "Server" means a computer system that receives and analyzes input comments and generates appropriate feedback.

[1615] 4. "Past comment data" refers to information about comments made by a user in the past.

[1616] 5. "Keywords" are important words or phrases extracted through text analysis.

[1617] 6. "Context" refers to information that indicates the background and meaning of what is said.

[1618] 7. "Contradiction" refers to a discrepancy between new statements and previous statements.

[1619] 8. "Inappropriate language" refers to words and expressions that should be avoided in public places.

[1620] 9. "Sensitive topics" are those that require special consideration in certain situations or with certain people.

[1621] 10. "Brand image" refers to consumers' impressions and evaluations of a company or product.

[1622] 11. "Means for generating feedback based on emotions" refers to a function that analyzes emotions from the content of a user's comments and provides appropriate feedback based on the results.

[1623] 12. "Real-time means used by employees when interacting with customers" refers to the ability of employees to receive immediate feedback from the system when interacting with customers.

[1624] The system for implementing this invention is designed to receive and analyze user input and generate appropriate feedback. A specific implementation method for this system will be described below.

[1625] System configuration

[1626] The system consists of the following main modules and methods:

[1627] 1. User Device:

[1628] This is a device that allows users to input new comments. This device can be a smartphone or tablet.

[1629] 2. Server:

[1630] Data receiving module: Receives the content of comments sent from the user terminal.

[1631] Data collection module: Collects user's past speech data from the database.

[1632] Text analysis module: Analyzes the collected data and extracts important keywords and context using natural language processing tools (e.g., Spacy and TextBlob).

[1633] Emotion Engine: Recognizes user emotions based on user input and past speech data. Sentiment analysis uses machine learning models such as TextBlob.

[1634] Inconsistency detection module: Compares new statements with past statements to detect inconsistencies.

[1635] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics. Can leverage natural language processing models such as Hugging Face's Transformers.

[1636] Brand image evaluation module: Evaluate the impact of statements on brand image.

[1637] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[1638] User operation procedure

[1639] 1. The user types a new statement, for example, "Can you tell me more about your new product?"

[1640] 2. This statement is sent from the user's device to the server.

[1641] Server Processing

[1642] 1. The server receives the message.

[1643] 2. The server saves the speech content through the data receiving module and converts it into an appropriate format.

[1644] 3. The data collection module collects the user's past utterance data from the database.

[1645] 4. The text analysis module analyzes the collected data and extracts important keywords and context using natural language processing techniques (Spacy, TextBlob).

[1646] 5. The sentiment engine recognizes the emotions expressed by users through their speech. It uses sentiment analysis tools such as TextBlob.

[1647] 6. The contradiction detection module compares the new statement with the previous statement and detects any contradictions.

[1648] 7. The profanity detection module checks new posts for profanity, using tools like Hugging Face Transformers.

[1649] 8. The brand image evaluation module evaluates the impact of the statement on the brand image.

[1650] 9. The feedback generation module generates feedback to the user based on the analysis results and the output of the emotion engine.

[1651] Feedback example:

[1652] In response to a statement such as "Could you please tell us more about your new product?", feedback such as "It would be best to be careful with your choice of words and briefly explain past issues" is generated.

[1653] Prompt Sentence Examples

[1654] Give your users feedback on how they would like to explain the new product to their customers. Evaluate whether it matches your past support record, whether it's inappropriate, and the impact it has on your brand.

[1655] Users (employees) can receive this feedback and modify their statements as necessary to maintain appropriate communication. By using this system, statements made when dealing with customers can be consistent, which will not damage the brand image and improve customer satisfaction.

[1656] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1657] Step 1:

[1658] The user inputs a new statement. This statement is entered in text format and sent from the user's device (smartphone or tablet) to the server. The input is the text entered by the user: "Could you tell me more about the new product?" The output is the data of the statement received by the server.

[1659] Step 2:

[1660] The server receives the message and stores it in an appropriate format, converting it into text for processing by subsequent modules. At this stage, the input is the message sent from the user's terminal, and the output is text data that can be processed within the server.

[1661] Step 3:

[1662] The server executes the data collection module to retrieve the user's past utterance data from the database. At this stage, the input is the user ID, and the output is a set of the user's past utterance data. The server uses this data to perform subsequent analysis.

[1663] Step 4:

[1664] The server runs a text analysis module to extract important keywords and context from past utterance data. It uses natural language processing technologies such as Spacy and TextBlob to analyze the input past utterance data and extract information such as entities, semantic relationships, and sentiment as output.

[1665] Step 5:

[1666] The server runs an emotion engine to recognize the user's emotions based on the user's input and past comment data. It analyzes the input comment using TextBlob or similar tools and outputs an emotion score (positive, negative, neutral, etc.).

[1667] Step 6:

[1668] The server runs a contradiction detection module, which compares the new and previous statements to detect contradictions. Specifically, it uses semantic analysis to compare the meaning of the new and previous statements and detects any contradictions. The input is the new statement and the previous statement data, and the output is the information on the parts where contradictions are detected.

[1669] Step 7:

[1670] The server runs an inappropriate language detection module to check whether new comments contain inappropriate language or sensitive topics. Using tools like Hugging Face's Transformers, the server analyzes the input comments and outputs whether they contain inappropriate language or sensitive topics.

[1671] Step 8:

[1672] The server executes the brand image evaluation module to evaluate the impact of the comment on the brand image. Using the emotion score of the comment, the server scores the impact of the input comment on the brand image as positive or negative, and outputs the score.

[1673] Step 9:

[1674] The server runs the feedback generation module and generates feedback for the user based on the analysis results and the output of the emotion engine. The generated feedback contains specific advice and recommendations. For example, in response to the question, "Could you please tell us more about your new product?", feedback such as, "It would be good to be careful about your choice of words and briefly explain past problems" is generated. The input is the analysis results and emotion score, and the output is the feedback content.

[1675] Step 10:

[1676] The user receives the generated feedback and modifies the utterance as necessary. The feedback is checked on the user's device, and the modified utterance is sent to the server. The input is the feedback sent from the server, and the output is the modified utterance.

[1677] 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.

[1678] 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.

[1679] 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.

[1680] [Fourth embodiment]

[1681] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1682] 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.

[1683] 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).

[1684] 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.

[1685] 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.

[1686] 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).

[1687] 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.

[1688] 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.

[1689] 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.

[1690] 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.

[1691] 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.

[1692] 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.

[1693] 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."

[1694] The present invention provides a system for ensuring safe and consistent communication by allowing users to review the content of their statements before they are made public. This system is designed to check whether statements made by individuals or organizations with significant social influence, such as politicians, business leaders, and celebrities, are consistent with past statements, contain inappropriate language, or damage brand image. Specific embodiments of the system are described below.

[1695] System Configuration

[1696] This system consists of the following main modules:

[1697] Data collection module: Collects user's past speech data.

[1698] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[1699] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[1700] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[1701] Brand image evaluation module: Evaluate the impact of statements on brand image.

[1702] Feedback generation module: Generates feedback to the user based on the analysis results.

[1703] Program processing

[1704] 1. User terminal (entering new comments)

[1705] The user inputs a new message and sends it to the system.

[1706] 2. Server (data reception and initial processing)

[1707] The server receives the input speech data and performs initial processing, converting the data into an appropriate format and classifying the type of speech (social media post, lecture content, book, etc.).

[1708] 3. Server (executes data collection modules)

[1709] The server collects past comment data, and retrieves the user's past social media posts, lectures, books, etc. from the database.

[1710] 4. Server (executing text analysis module)

[1711] The server analyzes the collected data to extract important keywords and context, and uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text.

[1712] 5. Server (execution of the inconsistency detection module)

[1713] The server compares the new statement with previous statements to see if there are any inconsistencies. It runs an algorithm to compare previous statements with the new statement and detect semantic inconsistencies.

[1714] 6. Server (execution of profanity detection module)

[1715] The server checks new posts for inappropriate language or sensitive topics by scanning the text against predefined lists and models to detect inappropriate language or sensitive topics.

[1716] 7. Server (Running the brand image evaluation module)

[1717] The server evaluates whether the comment is damaging to the brand image and scores the positive / negative impact of the comment using an evaluation model related to the brand image.

[1718] 8. Server (executing the feedback generation module)

[1719] The server generates feedback to the user based on the analysis results, synthesizing the analysis results and automatically generating a feedback document containing specific advice and recommendations.

[1720] 9. User Device (Receiving and Correcting Feedback)

[1721] The user receives the generated feedback and corrects the statement if necessary. The user checks the feedback, corrects the statement, or re-enters it into the system.

[1722] Specific examples

[1723] Example 1: Posting to social media

[1724] 1. A user enters the text "Excited about the new product" in a social media post.

[1725] 2. The server receives the entered text, and the data collection module collects past posting data.

[1726] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[1727] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[1728] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[1729] 6. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[1730] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[1731] 8. Receive user-generated feedback and revise your post as needed.

[1732] Example 2: Lecture content

[1733] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[1734] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[1735] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[1736] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[1737] 5. The server checks the script for profanity using the profanity detection module.

[1738] 6. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[1739] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[1740] 8. Take the user-generated feedback and modify the script as needed.

[1741] In this way, users can reassess their statements before publishing them to ensure consistent and appropriate communication.

[1742] The processing flow will be explained below.

[1743] Step 1:

[1744] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message on social media saying, "The new product is great!"

[1745] Step 2:

[1746] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[1747] Step 3:

[1748] The server runs a data collection module to collect data on users' past comments. Specifically, it retrieves the user's past social media posts, lecture contents, and book contents from a database.

[1749] Step 4:

[1750] The server runs a text analysis module to analyze the collected speech data. Natural language processing (NLP) techniques are used to extract important keywords, context, entities, and sentiment from the text. For example, keywords such as "new product," "problem," and "excitement" are extracted.

[1751] Step 5:

[1752] The server runs a contradiction detection module, which compares the new statement with past statements. Specifically, it uses an algorithm for detecting semantic contradictions to detect a contradiction between the past statement "The new product has many problems" and the new statement "The new product is great!"

[1753] Step 6:

[1754] The server runs a profanity detection module to check if the new post contains any inappropriate language or sensitive topics, such as inappropriate language or sensitive subjects that should be avoided in public.

[1755] Step 7:

[1756] The server executes the brand image evaluation module to evaluate the impact of the new comment on the brand image, and uses the evaluation model to score the positive / negative impact of the comment.

[1757] Step 8:

[1758] The server runs the feedback generation module and generates feedback to the user based on the analysis results. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[1759] Step 9:

[1760] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[1761] These steps allow users to reassess their content before publishing statements that are consistent and protect their brand image.

[1762] Example 1

[1763] 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."

[1764] Currently, there is a lack of mechanisms for evaluating the appropriateness of user statements before they are made public and for maintaining consistent communication. Furthermore, there are also insufficient means for automatically checking whether statements contradict previous statements, contain inappropriate language, or damage a company's image. These problems are particularly serious for individuals and organizations with significant social influence, such as politicians, corporate leaders, and celebrities. Therefore, the present invention aims to provide a system for reevaluating user statements and ensuring safe and consistent communication.

[1765] 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.

[1766] In this invention, the server includes: a means for a user to input new comments; a means for the server to receive the comments; a means for the server to collect data on past comments from the user; a means for the server to analyze the collected data on comments and extract important keywords and contexts; a means for the server to compare the new comments with the past data on comments and detect semantic inconsistencies; a means for the server to check whether the new comments contain inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comments will damage the company's image; a means for the server to generate feedback to the user based on the analysis results; and a means for the user to receive the feedback and modify the comments. This allows users to reevaluate the content of their comments before making them public, enabling consistent and safe communication.

[1767] "User" refers to a person who uses the system to input comments and receive feedback.

[1768] "Server" refers to a device that includes equipment or software that receives, processes, analyzes, and generates feedback from users.

[1769] "Speech" refers to new text information that a user enters into the system.

[1770] "Past utterance data" refers to the accumulation of text information about utterances made by a user in the past.

[1771] "Important keywords and context" refers to semantically important words and their surrounding context that the text analysis module extracts from speech data.

[1772] A "semantic contradiction" refers to a situation in which a new statement does not semantically match a previous statement.

[1773] "Inappropriate language" refers to language that is socially unacceptable or problematic to use in a particular context.

[1774] "Sensitive topics" refer to subjects or topics that are inappropriate for public disclosure under certain circumstances or social conditions.

[1775] "Corporate image" refers to the overall evaluation and impression of a company from the perspective of general consumers and society.

[1776] "Feedback" refers to information, including advice and recommendations, that the server provides to the user based on the analysis results.

[1777] This invention is a system that allows users to review the content of their statements before they are made public, ensuring safe and consistent communication. In particular, it is designed to check statements made by politicians, business leaders, celebrities, and other influential individuals and organizations to ensure that they are consistent with past statements, contain inappropriate language, and do not damage the company's image.

[1778] The system consists of the following main modules:

[1779] Data collection module: Collects past user comment data. Obtains comment data from users' social media posts, lectures, books, etc.

[1780] Text analysis module: Analyzes collected speech data and extracts important keywords and context. This analysis uses natural language processing (NLP) techniques, specifically tools such as Google's NLP API and SpaCy.

[1781] Inconsistency detection module: Runs algorithms to detect semantic inconsistencies by comparing new statements with previous statements.

[1782] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics by scanning the text using predefined lists and machine learning models.

[1783] Corporate Image Assessment Module: Evaluates whether a statement is damaging to the company's image. Using an assessment model related to corporate image, the module scores the positive / negative impact of the statement.

[1784] Feedback generation module: Generates feedback to users based on the analysis results. Automatically generates specific advice and recommendations.

[1785] The system operates in the following specific steps:

[1786] 1. Entering a new message on the user's terminal: The user enters the new message into the terminal and sends it to the system.

[1787] 2. Data reception and initial processing on the server: The server receives the input speech data, performs initial processing of the data, and classifies the type of speech (SNS post, lecture content, book, etc.).

[1788] 3. Collection of past speech data on the server: The server collects the user's past speech data from the database.

[1789] 4. Text analysis on the server: The server uses NLP techniques to extract important keywords and context.

[1790] 5. Conflict detection on the server: The server compares new statements with previous statements to detect semantic contradictions.

[1791] 6. Server-based profanity detection: The server checks new posts for profanity.

[1792] 7. Corporate image evaluation on the server: The server evaluates the impact of the comments on the corporate image.

[1793] 8. Feedback generation on the server: The server generates feedback to the user based on the analysis results.

[1794] 9. Receiving feedback and correcting content on the user's device: The user receives the generated feedback and corrects the content of the comment if necessary.

[1795] Specific examples

[1796] Example 1: Posting to social media

[1797] 1. A user enters the text "Excited about the new product" in a social media post.

[1798] 2. The server receives the entered text, and the data collection module collects past posting data.

[1799] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[1800] 4. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[1801] 5. The server uses the profanity detection module to check whether the post contains any profanity.

[1802] 6. The server evaluates the impact of this post on the company's image in the company image evaluation module.

[1803] 7. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and will strengthen the company's image."

[1804] 8. Receive user-generated feedback and revise your post as needed.

[1805] Example 2: Lecture content

[1806] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[1807] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[1808] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[1809] 4. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[1810] 5. The server checks the script for profanity using the profanity detection module.

[1811] 6. The server uses the corporate image evaluation module to evaluate the impact of this presentation on the company's image.

[1812] 7. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the company's image, but be careful with certain wording."

[1813] 8. Take the user-generated feedback and modify the script as needed.

[1814] Prompt Sentence Examples

[1815] For social media posts: "Evaluate your social media posts about new products to ensure they are consistent with previous posts, contain no inappropriate language, and fit with the company image."

[1816] For a presentation: "Evaluate the presentation script about our strategy for entering new markets. Check for consistency with previous presentations, inappropriate language, and consistency with our company image."

[1817] These prompts can be used to ask the generative AI model for further evaluation. These examples encourage users to reassess their statements before publishing them, ensuring appropriate and consistent communication.

[1818] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1819] Step 1:

[1820] User enters new comment

[1821] The user inputs new content into the terminal and clicks the "Send" button. For example, the user inputs "I'm looking forward to the release of the new product." The input data is the content of the text message. The output is the text message data sent to the server.

[1822] Step 2:

[1823] The server receives and performs initial processing on the speech data.

[1824] The server receives the text of the comment sent by the user. The data formatter processes this text data, converts it into an appropriate data format, and classifies it by type of comment (social media post, lecture content, book, etc.). The input is the text data sent by the user, and the output is the initially processed data format. Specifically, the data formatter converts the comment content into JSON format.

[1825] Step 3:

[1826] The server collects past utterance data

[1827] The server collects data on users' past comments from a database. For example, it searches the database for social media posts from the past year and retrieves related data. The input is the user's identification information, and the output is collected data on past comments. Specifically, it executes an SQL query to extract the relevant comment data from the database.

[1828] Step 4:

[1829] The server analyzes the collected data

[1830] The server uses natural language processing (NLP) techniques to analyze the collected utterance data. It uses Google's NLP API and SpaCy to extract important keywords and context from the text. The input is the collected utterance data, and the output is the extracted keywords and context. Specifically, the NLP model tokenizes the text and performs named entity recognition (NER).

[1831] Step 5:

[1832] The server compares new and past comments

[1833] The server semantically compares the new utterance with the previous utterance to check for inconsistencies. A comparison algorithm is used to analyze the semantic matches and differences between the new and previous utterances. The input is the new utterance and the previous utterance data, and the output is the result of the inconsistency detection. Specifically, the server calculates context vectors and compares them using cosine similarity.

[1834] Step 6:

[1835] Server detects inappropriate language

[1836] The server checks whether new posts contain inappropriate language or sensitive topics. It scans the text of the post using predefined lists or machine learning models. The input is the new post, and the output is the detection of inappropriate language. Specifically, it compares the text with a list of banned words to see if there are any matches.

[1837] Step 7:

[1838] Server evaluates corporate image

[1839] The server evaluates the impact of new comments on the company's image. It uses an evaluation model related to the company's image to score the positive / negative impact of the comment. The input is the new comment, and the output is an evaluation score. Specifically, the evaluation model performs sentiment analysis on the comment text and scores it.

[1840] Step 8:

[1841] Server generates feedback

[1842] The server generates feedback for the user based on the analysis results. Specific advice and recommendations are automatically generated as a feedback document. The input is the analysis results, and the output is a feedback document. Specifically, the generative AI model generates feedback text based on the input data.

[1843] Step 9:

[1844] Users receive and correct feedback

[1845] The user receives the generated feedback on their device and modifies the utterance as necessary. The input is the feedback document, and the output is the modified utterance. Specifically, the user views the feedback and modifies the utterance in a text editor.

[1846] (Application example 1)

[1847] 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."

[1848] Publicly released comments and statements may lack consistency and contain inappropriate language or legal risks, which could damage the brand image and security policies of companies and celebrities. A system is needed to solve this problem and maintain safe and consistent communication.

[1849] 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.

[1850] In this invention, the server includes: means for a user to input new comments; means for the server to receive the comments; means for the server to collect data on past comments from the user; means for the server to analyze the collected data on comments and extract important keywords and contexts; means for the server to compare the new comments with the past data on comments and detect inconsistencies; means for the server to check whether the new comments contain inappropriate language or sensitive topics; means for the server to evaluate the possibility that the new comments will damage brand image or security policies; means for the server to generate feedback to the user based on the analysis results; and means for the user to receive the feedback and modify the comments. This allows users to evaluate and modify the comments they make public in advance, enabling consistent and appropriate communication.

[1851] A "user" is an individual or group that uses the system to input new comments.

[1852] A "server" is a computer system that analyzes the content of comments received from users and generates feedback based on the analysis results.

[1853] "Comment content" refers to text data such as a comment, script, or post newly entered by a user.

[1854] "Past comment data" is a record of comments and posts made by the user in the past.

[1855] "Important keywords" are words or phrases that characterize the topic and are extracted from the content of the statement or past statement data.

[1856] "Context" refers to the context or background information of what is being said.

[1857] A "contradiction" is a state in which there is a logical inconsistency between the content of a new statement and past statement data.

[1858] "Inappropriate language" is any word or phrase that could pose legal risks or damage a company's reputation.

[1859] A "sensitive topic" is one that is socially, culturally, or legally controversial or carries risk.

[1860] "Brand image" refers to the public reputation or impression of a user or the organization or company to which they belong.

[1861] A "security policy" is a set of guidelines and rules established by an organization or company for the purposes of protecting information and managing risks.

[1862] "Feedback" refers to advice and recommendations to users that are generated based on the analysis results.

[1863] This invention provides a system that allows users to review the content of their comments before publishing them, ensuring consistent and safe communication. This system is particularly useful in the security field, where it can pre-screen comments made by companies and individuals to prevent inappropriate information leaks or comments that violate a company's security policy.

[1864] System Configuration

[1865] This system consists of the following main modules:

[1866] 1. Data Collection Module

[1867] Hardware / Software: Python (version 3.8 or higher), pandas library, SQLAlchemy

[1868] Description: The server collects the user's past speech data from the database and stores it in a data frame format.

[1869] 2. Text Analysis Module

[1870] Hardware / Software: Python (version 3.8 or higher), spaCy library, NLTK

[1871] Description: The server analyzes the speech data and extracts important keywords and entities, as well as analyzing sentiment and semantic relationships.

[1872] 3. Conflict Detection Module

[1873] Hardware / Software: Python (version 3.8 or higher), spaCy, natural language processing model (e.g., BERT)

[1874] Description: The server compares new statements with previous statements to detect logical inconsistencies.

[1875] 4. Profanity Detection Module

[1876] Hardware / Software: Python (version 3.8 or higher), scikit-learn, custom dictionary

[1877] Description: The server checks new posts for legal risks and confidential information.

[1878] 5. Brand Image Evaluation Module

[1879] Hardware / Software: Python (version 3.8 or higher), TensorFlow, scikit-learn

[1880] Description: The server scores new comments on their impact on brand image and security policies.

[1881] 6. Feedback Generation Module

[1882] Hardware / Software: Python (version 3.8 or higher), GPT-3 (OpenAI API)

[1883] Description: The server generates feedback statements from the analysis results and provides specific suggestions to the user.

[1884] Specific processing steps

[1885] 1. The user enters a new comment

[1886] The user inputs new utterances into the terminal.

[1887] 2. The server receives the message

[1888] The server analyzes the content of the message received from the user.

[1889] 3. Data Collection

[1890] The server collects past utterance data from a database.

[1891] 4. Text Analysis

[1892] The server analyzes the speech data and extracts important keywords and context.

[1893] 5. Conflict Detection

[1894] The server compares the new message with previous messages to detect any inconsistencies.

[1895] 6. Profanity Detection

[1896] The server checks new posts for inappropriate language and security risks.

[1897] 7. Brand image evaluation

[1898] The server evaluates the impact of the comments on the brand and security policies.

[1899] 8. Feedback Generation

[1900] The server generates feedback to the user based on the analysis results, prompting them to revise their comments.

[1901] Specific examples

[1902] For example, if a company's security officer types, "I would like to comment on the implementation of a new security system," the server will perform the following process:

[1903] 1. The data collection module collects past speech data (social media posts, news releases, meeting records, etc.).

[1904] 2. The text analysis module analyzes new and past comments and extracts the keywords "security," "system," and "implementation."

[1905] 3. The inconsistency detection module compares the semantic relationships between the new and previous statements to ensure consistency.

[1906] 4. The profanity detection module checks for legal risks and confidential information.

[1907] 5. The brand image evaluation module scores the impact of new statements on security policies and the brand.

[1908] 6. The feedback generation module generates feedback such as "Your comments are appropriate and consistent, but please avoid mentioning specific system names" and displays it to the user.

[1909] Prompt Sentence Examples

[1910] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[1911] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1912] Step 1:

[1913] The user inputs a new statement.

[1914] Input: New statement (e.g., "I would like to comment on the implementation of the new security system")

[1915] Output: What you say is sent to the system.

[1916] Specific operation: The user enters the content of the message into the input form on the terminal and presses the send button.

[1917] Step 2:

[1918] The server receives the message.

[1919] Input: User-submitted message

[1920] Output: The speech is ready for analysis.

[1921] Specific operation: The server receives an HTTP request, converts the content of the comment into an internal data format, and stores it in a database.

[1922] Step 3:

[1923] The server collects past utterance data.

[1924] Input: User ID or metadata related to what was said

[1925] Output: Past speech data (data frame format)

[1926] Specific operation: The server executes a database query to retrieve past comment data, which is then converted into a data frame using the pandas library.

[1927] Step 4:

[1928] The server analyzes the collected speech data and extracts important keywords and contexts.

[1929] Input: Past speech data

[1930] Output: Extracted keywords and context information

[1931] What it does: The server uses the spaCy library to parse text data and extract entities and keywords, as well as perform sentiment analysis and semantic relationships.

[1932] Step 5:

[1933] The server compares new comments with past comments to detect inconsistencies.

[1934] Input: New utterances, extracted keywords and context information

[1935] Output: Conflict detection results

[1936] What happens: The server uses a natural language processing model (e.g., BERT) to compare the semantic relationships between new and previous utterances, detecting any logical inconsistencies.

[1937] Step 6:

[1938] The server checks new posts for inappropriate language or sensitive topics.

[1939] Input: New statement

[1940] Output: Profanity detection results

[1941] What it does: The server uses the scikit-learn library and a custom dictionary to scan posts for profanity and sensitive topics.

[1942] Step 7:

[1943] The server evaluates the likelihood that the comment will damage the brand image or security policy.

[1944] Input: New speech, profanity detection results

[1945] Output: Evaluation results regarding brand image and security policy

[1946] Specific operation: The server uses TensorFlow and scikit-learn to score the impact of statements on brand image and security policies.

[1947] Step 8:

[1948] The server generates feedback to the user based on the analysis results.

[1949] Input: Conflict detection results, inappropriate language detection results, brand image and security policy evaluation results

[1950] Output: Feedback statement

[1951] Specific operation: The server uses GPT-3 (OpenAI API) to generate specific feedback sentences based on the analysis results.

[1952] Step 9:

[1953] The user receives the feedback and corrects the content of the statement.

[1954] Input: Feedback statement

[1955] Output: Corrected statement

[1956] Specific operation: The user checks the feedback text on the terminal, corrects the comment if necessary, and re-enters the corrections into the system.

[1957] Prompt Sentence Examples

[1958] "Please review your comments about the implementation of new security systems to ensure they are appropriate. Conduct a risk assessment to ensure they are consistent with past statements, contain inappropriate language, and address the impact on your brand image."

[1959] 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.

[1960] The present invention is a system that allows users to reevaluate their comments before they are made public, ensuring safe and consistent communication, particularly by combining an emotion engine. This system checks the comments entered by the user for inconsistencies with past comments, inappropriate expressions, and potential damage to the brand image, and also recognizes the user's emotions and adjusts the feedback. A specific embodiment of this system will be described.

[1961] System Configuration

[1962] The system consists of the following main modules:

[1963] Data collection module: Collects user's past speech data.

[1964] Text analysis module: Analyzes collected speech data and extracts important keywords and context.

[1965] Emotion engine: Recognizes emotions from user input and past speech data.

[1966] Inconsistency detection module: Compares new statements with past statements to check for inconsistencies.

[1967] Profanity detection module: Checks new posts for inappropriate language or sensitive topics.

[1968] Brand image evaluation module: Evaluate the impact of statements on brand image.

[1969] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[1970] Program processing

[1971] 1. User terminal (entering new comments)

[1972] The user inputs a new message and sends it to the system. For example, consider the case where a user inputs a message for posting on social media saying, "The new product is great!"

[1973] 2. Server (data reception and initial processing)

[1974] The server receives the input speech data, saves it in text format, and converts it into an appropriate format.

[1975] 3. Server (executes data collection modules)

[1976] The server collects past comment data and retrieves the user's past social media posts, lecture contents, and book contents from the database.

[1977] 4. Server (executing text analysis module)

[1978] The server analyzes the collected data and extracts important keywords and context. It uses natural language processing (NLP) techniques to extract entities, sentiment, and semantic relationships from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[1979] 5. Server (execution of emotion engine)

[1980] The server runs an emotion engine to recognize the user's emotions based on the user's input and past utterance data. For example, emotions such as "positive" or "excited" can be recognized from the input.

[1981] 6. Server (execution of the inconsistency detection module)

[1982] The server compares the new statement with previous statements to check for inconsistencies. Specifically, it uses an algorithm to detect semantic inconsistencies and detects a contradiction between the previous statement "The new product has many problems" and the new statement "The new product is great!"

[1983] 7. Server (execution of profanity detection module)

[1984] The server checks new posts for inappropriate language or sensitive topics, for example, detecting inappropriate language or sensitive subjects that should be avoided in public.

[1985] 8. Server (Running the brand image evaluation module)

[1986] The server evaluates whether a comment is damaging to the brand image and uses an evaluation model to score the positive / negative impact of the comment.

[1987] 9. Server (executing the feedback generation module)

[1988] The server generates feedback for the user based on the analysis results and the output of the emotion engine. It automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[1989] 10. User Device (Receiving and Correcting Feedback)

[1990] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[1991] Specific examples

[1992] Example 1: Posting to social media

[1993] 1. A user enters the text "Excited about the new product" in a social media post.

[1994] 2. The server receives the entered text, and the data collection module collects past posting data.

[1995] 3. The server uses the text analysis module to extract the keywords "new product" and "excitement."

[1996] 4. The server uses an emotion engine to recognize emotions such as "positive" and "excited."

[1997] 5. The server uses a contradiction detection module to check for inconsistencies with previous posts (e.g., whether previous posts criticize the same product).

[1998] 6. The server uses the profanity detection module to check whether the post contains any profanity.

[1999] 7. The server evaluates the impact of this post on brand image in the brand image evaluation module.

[2000] 8. The server uses a feedback generation module to generate feedback such as "This expression is appropriate and strengthens the brand image."

[2001] 9. Receive user-generated feedback and revise your post as needed.

[2002] Example 2: Lecture content

[2003] 1. The user inputs the lecture script "Strategy for Entering New Markets."

[2004] 2. The server receives the input script, and the data collection module collects past lecture scripts.

[2005] 3. The server uses the text analysis module to extract the keywords "new market" and "entry strategy."

[2006] 4. The server uses an emotion engine to recognize the emotions "careful" and "thoughtful."

[2007] 5. The server uses a contradiction detection module to check for any contradictions with previous talks (e.g., whether a different strategy was proposed in a previous talk).

[2008] 6. The server checks the script for profanity using the profanity detection module.

[2009] 7. The server uses the brand image evaluation module to evaluate the impact of the presentation on the brand image.

[2010] 8. The server uses a feedback generation module to generate feedback such as, "This content will strengthen the brand image, but be careful with certain wording."

[2011] 9. Take the user-generated feedback and modify the script as needed.

[2012] In this way, users can reassess their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback can be provided in a more user-friendly format.

[2013] The processing flow will be explained below.

[2014] Step 1:

[2015] The user inputs a new message and sends it to the system. For example, the user inputs a message on social media such as "The new product is great!"

[2016] Step 2:

[2017] The server receives the input speech data, saves the received speech content in text format, and converts it into an appropriate format.

[2018] Step 3:

[2019] The server runs a data collection module to collect data on users' past comments. Specifically, the server retrieves information such as users' past social media posts, lectures, and book contents from a database.

[2020] Step 4:

[2021] The server runs a text analysis module to analyze the collected speech data. It uses natural language processing (NLP) techniques to extract important keywords and context from the text. For example, it extracts keywords such as "new product," "problem," and "excitement."

[2022] Step 5:

[2023] The server runs an emotion engine to recognize emotions from user input. For example, from the input "The new product is great!", emotions such as "positive" and "excited" can be recognized.

[2024] Step 6:

[2025] The server extracts emotional patterns from past speech data and compares them with the content of new speech. It checks for consistency by comparing the emotional patterns in past speech with the emotions of new speech.

[2026] Step 7:

[2027] The server runs a contradiction detection module, which compares the new statement with past statements. It uses an algorithm to detect semantic contradictions, for example, between a past statement "The new product has many problems" and a new statement "The new product is great!"

[2028] Step 8:

[2029] The server runs a profanity detection module to check whether new posts contain inappropriate language or sensitive topics, for example by scanning the text against predefined lists or models to detect inappropriate language or sensitive subjects that should be avoided in public.

[2030] Step 9:

[2031] The server runs a brand image evaluation module to evaluate whether a new comment will damage the brand image, and uses an evaluation model to score the positive / negative impact of the comment.

[2032] Step 10:

[2033] The server runs the feedback generation module and generates feedback to the user based on the analysis results and the output of the emotion engine. The module automatically generates feedback containing specific advice and recommendations, such as "The new product is great, but we recommend adding a reference to past issues."

[2034] Step 11:

[2035] The user receives the generated feedback and corrects the utterance as necessary. The user then refers to the feedback, adjusts the utterance, and submits it back to the system.

[2036] This allows users to reevaluate their comments before making them public, ensuring consistent and appropriate communication. By incorporating an emotion engine, feedback content can be provided in a form that is more suited to the user's emotions.

[2037] Example 2

[2038] 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."

[2039] In today's digital communications, users' public comments can be inconsistent, contain inappropriate language, or even damage a brand's image. This requires users to carefully evaluate their comments, but this process takes time, effort, and requires specialized knowledge. Furthermore, it is difficult to properly reflect users' emotions in feedback. Technology is needed to solve these issues and enable safe and consistent communication.

[2040] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for collecting past comment data; a means for analyzing the collected comment data and extracting important keywords and context; a means for recognizing emotions from the user's input content and past comment data using an emotion engine; a means for comparing the new comment content with the past comment data and detecting inconsistencies; a means for checking whether the new comment content contains inappropriate expressions or sensitive topics; a means for evaluating the possibility that the new comment content will damage a brand image; a means for generating feedback to the user based on the analysis results and the output of the emotion engine; and a means for the user to receive the feedback and modify the comment content. This allows the user to reevaluate the content of their comment before making it public, thereby achieving consistent and appropriate communication. Furthermore, the user's emotions can be appropriately reflected in the feedback, making it possible to provide more intuitive and adaptable feedback.

[2041] "User" refers to any person or entity that utilizes the system to enter new comments and receive feedback and modify comments.

[2042] "Server" refers to a computer system that processes and analyzes comments received from users and generates feedback.

[2043] "Comment content" refers to the text data that a user inputs into the system.

[2044] "Past utterance data" refers to records of utterances made by the user up to now.

[2045] "Means of collection" refers to the function of retrieving past speech data from databases or other storage media.

[2046] "Means of analysis" refers to the function of analyzing collected speech data using technologies such as natural language processing and extracting important keywords and context.

[2047] "Emotion engine" refers to software or algorithms that recognize emotions from user input and past speech data.

[2048] "Means for detecting contradictions" refers to a function for comparing new statements with past statement data to detect semantic contradictions.

[2049] "Inappropriate language" and "sensitive topics" refer to words or themes that should be avoided in public spaces or expressions that may be legally problematic.

[2050] "Measures to check for inappropriate language" refers to a feature that checks whether new posts contain inappropriate language or sensitive topics.

[2051] "Brand image" refers to consumers' perceptions and evaluations of a company, product, or service.

[2052] "Means for evaluating brand image" refers to the function of evaluating the possibility that new statements will damage brand image.

[2053] "Analysis results" refers to the results of analysis and evaluation of the content of user comments.

[2054] "Means for generating feedback" refers to the function of generating specific advice and recommendations for users based on the analysis results and the output of the emotion engine.

[2055] "Feedback" refers to the analysis results and advice or recommendations based on them that the system provides to the user.

[2056] "Means to correct" refers to the ability for users to receive feedback and adjust or correct what they say based on that feedback.

[2057] MODE FOR CARRYING OUT THE INVENTION

[2058] The present invention is realized by combining an emotion engine with a system that allows users to reevaluate their comments before publishing them and maintain safe and consistent communication. This system is composed of the following specific modules and is executed through data processing between users and a server.

[2059] System configuration details

[2060] 1. User Device

[2061] The user terminal provides an interface for the user to input new comments. This interface includes a text input field and a send button. When the user inputs a comment and clicks the send button, the comment data is sent to the system.

[2062] Examples:

[2063] A user posts a social media post about a new product: "The new product is great!"

[2064] 2. Server Configuration

[2065] The server receives the utterance data sent from the user terminal and performs a series of analysis processes.

[2066] Hardware and software used

[2067] Hardware: High-performance server (general server with CPU, memory, and storage)

[2068] Software: Database Management System (DBMS), Natural Language Processing (NLP) library, Sentiment Analysis Algorithm, Profanity Detection Library, Brand Image Evaluation Model

[2069] Specific Modules

[2070] 1. Data Collection Module:

[2071] The server collects the user's past speech data, which is retrieved from the database and stored for a certain period of time.

[2072] 2. Text Analysis Module:

[2073] The server analyzes the collected speech data and extracts important keywords and context using natural language processing techniques, such as extracting entities and sentiment from text data using NLP libraries (e.g., SpaCy, NLTK).

[2074] 3. Emotion Engine:

[2075] The server recognizes emotions from the user's input and past speech data, and runs a sentiment analysis algorithm to extract positive, negative, excited, and other emotions from the text.

[2076] 4. Conflict Detection Module:

[2077] The server compares new utterances with past utterance data and uses an algorithm to detect semantic inconsistencies.

[2078] 5. Profanity Detection Module:

[2079] The server checks new posts for inappropriate language or sensitive topics. It uses a profanity detection library to scan the text for inappropriate language or topics that should be avoided in public.

[2080] 6. Brand Image Evaluation Module:

[2081] The server evaluates the impact of the comments on the brand image, using a rating model to score the positive / negative impact of the comments.

[2082] 7. Feedback generation module:

[2083] The server generates feedback for the user based on the analysis results and the output of the emotion engine. The feedback includes specific advice and recommendations. For example, it generates a message such as, "The new product is great, but we recommend adding a reference to past issues."

[2084] 3. Receiving feedback and correcting user devices

[2085] The user device receives the feedback generated by the server and presents it to the user, who can then refer to the feedback to modify or adjust the content of their comments and resubmit them to the system.

[2086] Examples:

[2087] Example feedback: "Your new product is great, but I recommend you also mention that this product has caused problems in the past."

[2088] These modules and procedures allow users to reassess their comments before publishing them, ensuring consistent and appropriate communication. Furthermore, incorporating an emotion engine allows for feedback to be provided in a more user-friendly format.

[2089] The flow of the identification process in the second embodiment will be described with reference to FIG.

[2090] Step 1:

[2091] The user enters a new comment

[2092] The user enters the message into the text input field on the terminal and clicks the send button, which sends the message to the system.

[2093] Input: User input (e.g., "Your new product is great!")

[2094] Output: The message is sent to the server.

[2095] Specific behavior:

[2096] The user uses the text input field on the device to enter a new message, and when they click the send button, an HTTP request is sent to the server.

[2097] Step 2:

[2098] The server receives the speech data and performs initial processing.

[2099] The server receives the speech data sent from the user terminal, saves the received data in text format, and converts it into an appropriate format.

[2100] Input: Speech data sent by the user (HTTP request)

[2101] Output: Saved speech data in text format

[2102] Specific behavior:

[2103] The server receives the HTTP request, saves it as text data, and performs the necessary format conversion before storing it in the database.

[2104] Step 3:

[2105] The server collects past speech data

[2106] The server uses the data collection module to obtain the user's past utterance data from the database.

[2107] Input: User ID

[2108] Output: Past speech data (e.g., past social media posts, lecture content)

[2109] Specific behavior:

[2110] The server queries the database to retrieve and search for past comment data based on the user ID, and stores it in temporary storage.

[2111] Step 4:

[2112] The server analyzes the speech data and extracts keywords and context.

[2113] The server uses a text analysis module to analyze past and new utterance data, and uses natural language processing technology to extract important keywords and context.

[2114] Input: Past speech data, new speech data

[2115] Output: Extracted keywords and context (e.g., "new product" or "excitement")

[2116] Specific behavior:

[2117] Analyze text data using a natural language processing (NLP) library (e.g., SpaCy, NLTK), extract entities and sentiment, and convert them into structured data.

[2118] Step 5:

[2119] The server runs the emotion engine to recognize emotions.

[2120] The server executes an emotion engine to recognize the user's emotion from the content of new comments and past comment data.

[2121] Input: New speech data, past speech data

[2122] Output: Recognized emotion data (e.g., "positive" or "excited")

[2123] Specific behavior:

[2124] Apply sentiment analysis algorithms to extract multiple sentiment labels from text, and store the sentiment data in temporary storage.

[2125] Step 6:

[2126] Server performs conflict detection

[2127] The server compares the new utterances with past utterance data to detect semantic inconsistencies.

[2128] Input: New speech data, past speech data

[2129] Output: Flag whether a conflict was detected (e.g. True / False)

[2130] Specific behavior:

[2131] Run a semantic contradiction detection algorithm to check for matches and contradictions with previous statements, and flag any inconsistencies found.

[2132] Step 7:

[2133] Server detects profanity

[2134] The server checks new posts for inappropriate language or sensitive topics using a profanity detection library.

[2135] Input: New speech data

[2136] Output: Profanity flag (e.g. True / False)

[2137] Specific behavior:

[2138] Calls the profanity detection library to scan the content of the post, filters it, and flags any violations.

[2139] Step 8:

[2140] Server evaluates brand image

[2141] The server uses a model to assess the likelihood that new comments will damage the brand image.

[2142] Input: New speech data

[2143] Output: Brand impact score (e.g., positive / negative impact)

[2144] Specific behavior:

[2145] The brand image evaluation algorithm is run to score the content of statements, and the impact of the statements is evaluated based on the score.

[2146] Step 9:

[2147] The server generates feedback

[2148] The server generates specific feedback for the user based on the analysis results and the output of the emotion engine.

[2149] Input: Analysis results, recognized emotion data, contradiction detection results, inappropriate expression detection results, brand image evaluation results

[2150] Output: Specific feedback

[2151] Specific behavior:

[2152] The feedback generation module creates advice based on various analysis results, generates specific recommendations for the comments, and sends them as feedback to the user.

[2153] Step 10:

[2154] Users receive feedback and revise their statements

[2155] The user receives feedback on their device and can correct or adjust what they say as needed.

[2156] Input: Specific feedback

[2157] Output: Corrected statement

[2158] Specific behavior:

[2159] The feedback is displayed on the user's device, and the user can use the feedback to revise their comments and send them back to the system.

[2160] (Application example 2)

[2161] 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."

[2162] Companies need to find ways to prevent inappropriate or brand-damaging remarks from employees when they interact with customers, and maintain consistent and appropriate communication. They also need to consider the impact of what employees say on customer emotions and provide appropriate feedback in real time.

[2163] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a means for a user to input new comment content; a means for the server to receive the comment content; a means for the server to collect the user's past comment data; a means for the server to analyze the collected comment data and extract important keywords and context; a means for the server to compare the new comment content with the past comment data and detect inconsistencies; a means for the server to check whether the new comment content contains inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comment content will damage the brand image; a means for the server to generate feedback based on the analysis results and the user's emotions; a means for the user to receive the feedback and modify the comment content; and a means for employees to use the feedback in real time when dealing with customers. This enables employees to maintain appropriate communication when dealing with customers, providing consistent responses while protecting the brand image. Furthermore, by considering the impact of employee comments on customer emotions in real time and providing appropriate feedback, customer satisfaction can be improved.

[2164] 1. A "user" is someone who uses the system to input comments and receive feedback based on those comments.

[2165] 2. "Utterances" refers to the text or words entered by a user.

[2166] 3. "Server" means a computer system that receives and analyzes input comments and generates appropriate feedback.

[2167] 4. "Past comment data" refers to information about comments made by a user in the past.

[2168] 5. "Keywords" are important words or phrases extracted through text analysis.

[2169] 6. "Context" refers to information that indicates the background and meaning of what is said.

[2170] 7. "Contradiction" refers to a discrepancy between new statements and previous statements.

[2171] 8. "Inappropriate language" refers to words and expressions that should be avoided in public places.

[2172] 9. "Sensitive topics" are those that require special consideration in certain situations or with certain people.

[2173] 10. "Brand image" refers to consumers' impressions and evaluations of a company or product.

[2174] 11. "Means for generating feedback based on emotions" refers to a function that analyzes emotions from the content of a user's comments and provides appropriate feedback based on the results.

[2175] 12. "Real-time means used by employees when interacting with customers" refers to the ability of employees to receive immediate feedback from the system when interacting with customers.

[2176] The system for implementing this invention is designed to receive and analyze user input and generate appropriate feedback. A specific implementation method for this system will be described below.

[2177] System configuration

[2178] The system consists of the following main modules and methods:

[2179] 1. User Device:

[2180] This is a device that allows users to input new comments. This device can be a smartphone or tablet.

[2181] 2. Server:

[2182] Data receiving module: Receives the content of comments sent from the user terminal.

[2183] Data collection module: Collects user's past speech data from the database.

[2184] Text analysis module: Analyzes the collected data and extracts important keywords and context using natural language processing tools (e.g., Spacy and TextBlob).

[2185] Emotion Engine: Recognizes user emotions based on user input and past speech data. Sentiment analysis uses machine learning models such as TextBlob.

[2186] Inconsistency detection module: Compares new statements with past statements to detect inconsistencies.

[2187] Profanity Detection Module: Checks new posts for inappropriate language or sensitive topics. Can leverage natural language processing models such as Hugging Face's Transformers.

[2188] Brand image evaluation module: Evaluate the impact of statements on brand image.

[2189] Feedback generation module: Generates feedback to the user based on the analysis results and the output of the emotion engine.

[2190] User operation procedure

[2191] 1. The user types a new statement, for example, "Can you tell me more about your new product?"

[2192] 2. This statement is sent from the user's device to the server.

[2193] Server Processing

[2194] 1. The server receives the message.

[2195] 2. The server saves the speech content through the data receiving module and converts it into an appropriate format.

[2196] 3. The data collection module collects the user's past utterance data from the database.

[2197] 4. The text analysis module analyzes the collected data and extracts important keywords and context using natural language processing techniques (Spacy, TextBlob).

[2198] 5. The sentiment engine recognizes the emotions expressed by users through their speech. It uses sentiment analysis tools such as TextBlob.

[2199] 6. The contradiction detection module compares the new statement with the previous statement and detects any contradictions.

[2200] 7. The profanity detection module checks new posts for profanity, using tools like Hugging Face Transformers.

[2201] 8. The brand image evaluation module evaluates the impact of the statement on the brand image.

[2202] 9. The feedback generation module generates feedback to the user based on the analysis results and the output of the emotion engine.

[2203] Feedback example:

[2204] In response to a statement such as "Could you please tell us more about your new product?", feedback such as "It would be best to be careful with your choice of words and briefly explain past issues" is generated.

[2205] Prompt Sentence Examples

[2206] Give your users feedback on how they would like to explain the new product to their customers. Evaluate whether it matches your past support record, whether it's inappropriate, and the i...

Claims

1. a means for a user to input new comments; A means for the server to receive the content of the message; a means for the server to collect past utterance data of the user; A means for the server to analyze the collected utterance data and extract important keywords and contexts; means for the server to compare the new message content with the past message data and detect any inconsistencies; means for the server to check whether the new comments contain inappropriate language or sensitive topics; a means for the server to evaluate the possibility that the new comment content will damage a brand image; a means for the server to generate feedback to the user based on the analysis result; A means for the user to receive the feedback and modify the content of the comment; A system including:

2. 10. The system of claim 1, wherein the server includes means for checking text against predefined lists or models to detect profanity or sensitive topics.

3. 2. The system of claim 1, wherein the server includes means for scoring the positive / negative impact of a statement using an evaluation model related to brand image.

Images