System

Abstract

A system is provided.SOLUTION: A system comprising: means for receiving data input by a user via the Internet; means for converting the received data into a text format and performing preprocessing by removing unnecessary information; means for evaluating a risk using a natural language processing model based on the preprocessed data; means for generating a proposed modification based on a risk evaluation result by referring to a past case database; and means for presenting the generated proposed modification and the risk evaluation result to the user.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] With the spread of the internet and social media, there has been an increase in cases where content published by companies and individuals unintentionally sparks outrage. This poses the risk of a company losing credibility and suspending operations. To solve this problem, a system is needed that can diagnose the risks of content before release and propose appropriate corrections and actions. However, current systems have difficulty efficiently analyzing massive amounts of data and providing specific correction suggestions based on the results of risk assessment. Therefore, a new system that can easily help companies and individuals avoid risks is needed. [Means for solving the problem]

[0005] The present invention provides a means for receiving data entered by a user via the Internet, converting the received data into text format, and performing preprocessing. It also includes a means for assessing risk using a natural language processing model based on the preprocessed data. It then provides a means for referencing a database of past cases based on the risk assessment results and generating a remediation plan. It also includes a means for presenting the generated remediation plan and the risk assessment results to the user. This system enables users to grasp risks in advance and take appropriate remediation plans or actions. It also includes a means for generating a notification recommending consultation with the legal department based on the risk assessment results, and a means for converting received image data into text using OCR technology and evaluating whether it contains inappropriate content, thereby achieving even greater accuracy and convenience.

[0006] The "Internet" is a communications infrastructure that interconnects computer networks around the world, enabling the sending and receiving of data.

[0007] "User" refers to any individual or company representative who uses this system.

[0008] "Data" refers to information such as text, images, and video that a user enters or uploads into the system.

[0009] "Text format" refers to a format in which digital data is expressed as a string of characters.

[0010] "Preprocessing" refers to preparatory work such as format conversion, removal of unnecessary information, and normalization that is carried out before analyzing data.

[0011] "Natural language processing models" refer to machine learning and deep learning algorithms used to understand and analyze human language.

[0012] "Risk assessment" refers to the process of assessing the likelihood that elements contained in the data will cause controversy or legal issues.

[0013] A "past case database" refers to a collection of information that accumulates data on past scandals and legal issues.

[0014] "Fixation Suggestion" refers to specific changes or suggestions made to avoid problems based on the results of a risk assessment.

[0015] "Legal department" refers to a specialized department within a company that handles legal issues.

[0016] "OCR technology" refers to optical character recognition technology that converts images and handwritten characters into digital text.

[0017] "Inappropriate content" refers to information that violates the law or ethical standards, or that may be considered socially problematic.

[0018] "Visualization" refers to the visual representation of data or results, such as graphics or charts.

[0019] "Dashboard format" refers to an interface that visualizes information in a centralized manner, allowing users to easily check important data. [Brief explanation of the drawings]

[0020] [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

[0021] 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.

[0022] First, the terms used in the following description will be explained.

[0023] 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).

[0024] 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.

[0025] 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.

[0026] 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.

[0027] 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."

[0028] [First embodiment]

[0029] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0030] 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.

[0031] 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).

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0037] 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.

[0038] 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.

[0039] 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.

[0040] 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."

[0041] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and suggests appropriate corrections and actions. Below, we will generate a program for this system and explain its processing in natural language.

[0042] What the program does

[0043] Receiving and Preprocessing Input Data

[0044] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[0045] Examples:

[0046] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a banner image for the campaign. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into a format that can be analyzed.

[0047] Data analysis and risk assessment

[0048] The server receives the preprocessed data and begins analyzing it. It uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the data with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[0049] Examples:

[0050] When the user types "50% off all products!", the server determines that the expression "all products" poses a risk. A database of past cases records similar campaigns that have caused misunderstandings and sparked outrage, so the server issues a warning based on this.

[0051] Generating results and presenting suggestions

[0052] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[0053] Examples:

[0054] The server generates a correction suggestion for the risky phrase "50% off all products!" to "50% off some products!" and also creates a notification recommending confirmation by the legal department. The user can check this information on the dashboard and make the necessary corrections.

[0055] Specific program operation scenarios

[0056] Users input new campaign ideas and upload campaign banners. The server receives and pre-processes them. It then uses natural language processing and deep learning models to perform a risk assessment and compares it with a database of past cases. If a risk is detected, a notification is generated and presented to the user, suggesting specific remediation measures and consultation with the legal department. Users can review the information through a dashboard and take appropriate remediation or action.

[0057] The above is an embodiment of the present invention and its specific example. This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0058] The processing flow will be explained below.

[0059] Step 1:

[0060] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[0061] Step 2:

[0062] The server receives text data and image data sent from the user via the Internet.

[0063] Step 3:

[0064] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[0065] Step 4:

[0066] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[0067] Step 5:

[0068] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[0069] Step 6:

[0070] The server detects negative elements and the risk of a backlash based on data analyzed by an NLP model.

[0071] Step 7:

[0072] The server refers to a database of past flame war cases, performs pattern matching with the preprocessed data, and checks whether similar cases exist.

[0073] Step 8:

[0074] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[0075] Step 9:

[0076] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction proposal (for example, changing "50% off all products!" to "50% off some products!").

[0077] Step 10:

[0078] The server generates a notification recommending legal consultation if necessary.

[0079] Step 11:

[0080] The server visualizes the generated correction proposals and risk assessment results and sends the results to the terminal in a dashboard format that is easy for the user to understand.

[0081] Step 12:

[0082] The user checks the proposed modifications and risk assessment results presented on the dashboard and makes any necessary modifications.

[0083] Example 1

[0084] 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."

[0085] Content published online by companies and individuals often carries the risk of causing a firestorm. Therefore, there is a need for a system that can perform risk assessments before content is published and propose appropriate revisions and actions. Furthermore, because content is diverse, it is necessary to support multiple data formats, such as text and images. Furthermore, to improve the accuracy of risk assessments, it is important to compare the content with a database of past cases, and a method of presenting information that takes user ease of use into consideration is also required.

[0086] 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.

[0087] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and preprocessing it by removing unnecessary information, and means for assessing risk using a natural language processing model based on the preprocessed text data. This makes it possible to assess risk before content is made public and to propose appropriate modifications and actions.

[0088] The "Internet" is an infrastructure for sending and receiving information through networks around the world.

[0089] "User" means any person or entity that uses the System to input or upload Content.

[0090] "Data" means electronic files in any format that contain information, including text, images, audio, and video.

[0091] "Receiving" is the process by which the server takes in data that a user enters or uploads over the Internet.

[0092] "Text format" is a data format expressed as character information.

[0093] "Unnecessary information" refers to metadata, special characters, HTML tags, etc. that are not necessary for analysis or evaluation, and refers to removing this information.

[0094] "Preprocessing" refers to the preliminary data cleansing operations performed to convert received data into an analyzable format.

[0095] A "natural language processing model" is an artificial intelligence algorithm used to understand and analyze the meaning of text data.

[0096] "Risk assessment" is an analytical process used to determine the likelihood that words or phrases contained in data will spark outrage, misunderstanding, or negative reactions.

[0097] A "fix proposal" is a specific change proposal based on the results of a risk assessment to make the content published by users safer.

[0098] The "dashboard format" is an interface that allows users to visually check risk assessment results and proposed modifications.

[0099] A "legal department" is a department within a company or organization that handles legal matters.

[0100] "Image data" is a digital file containing visual information.

[0101] "OCR technology" is a technology that extracts character information from image data and converts it into text format.

[0102] "Inappropriate content" refers to information or expressions that are considered socially, legally, or morally problematic.

[0103]

[0104] The present invention is an AI-based system that proactively diagnoses the risk of online content being published by companies and individuals causing outrage and proposes appropriate corrections and actions. Detailed embodiments of this system are described below.

[0105] System Overview

[0106] The server receives data (text and images) entered or uploaded by users via the Internet. It then converts the received data into text format and performs preprocessing by removing unnecessary information. Specifically, this includes removing HTML tags and special characters and normalizing the text. For image data, OCR technology is used to extract character information and convert it into an analyzable text format. This completes the preprocessing of the text and image data.

[0107] The server then analyzes the preprocessed data and performs a risk assessment using a natural language processing (NLP) model. This involves extracting keywords and phrases from the text and determining whether they pose a negative or potential risk. Specifically, it compares them with a database of past scandals to see if there are any similar patterns. A deep learning model is also applied to image data to assess whether it contains inappropriate content.

[0108] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. These results and suggestions are presented to the user in the form of a visual dashboard.

[0109] Hardware and software used

[0110] Server: The main processing computer system, responsible for receiving, preprocessing, analyzing, and generating recommendations on data.

[0111] Internet: The communications infrastructure used to send and receive data.

[0112] Natural language processing model (NLP model): An AI model used to analyze text data.

[0113] OCR technology: A technology that extracts text information from image data.

[0114] Deep learning model: An AI model used to assess the content of image data.

[0115] Database: A database that stores past incidents of online outrage and is used to check against them when assessing risk.

[0116] Dashboard: An interface for presenting results and recommendations to the user.

[0117] Specific examples

[0118] For example, suppose a user uploads the text "New campaign begins! 50% off all products!" along with a banner image for the campaign. The server receives this data, first removes HTML tags and special characters from the text data, and then uses OCR technology to extract text information from the banner image. Next, it uses a natural language processing model to analyze the phrase "50% off all products!" and determines that the "all products" part poses a risk. It then compares this with a database of past cases and issues a warning based on this.

[0119] The server then generates a suggested change to change the phrase "50% off everything!" to "50% off some items!" and creates a notification recommending that the legal department review the change. This information is presented to the user via a dashboard, allowing them to make the necessary changes.

[0120] Prompt Sentence Examples

[0121] The prompt sentence to be input by the user may have the following format, for example:

[0122] Examples include "New campaign announcement! 50% off all products!" or "This year's new product sale, all items are half price!"

[0123] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0124] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0125] Step 1:

[0126] The server receives data entered or uploaded by users via the Internet. This data is often in text or image format. A specific example of input is the text "New campaign starts! All products 50% off!" and a campaign banner image. The output is the received data itself.

[0127] Step 2:

[0128] The server performs preprocessing on the received text data. Specifically, it removes HTML tags and special characters and normalizes the text. For example, 50% off all products! ", the HTML tags will be removed and the text "50% off all items!" will be output.

[0129] Step 3:

[0130] The server applies OCR technology to the received image data and extracts the text from the image. Specifically, it extracts the text information "50% off all products!" from the campaign banner image and converts it into an analyzable text format. The extracted text is output.

[0131] Step 4:

[0132] The server sends the preprocessed text data to a natural language processing (NLP) model to extract keywords and phrases. Specifically, it extracts the phrase "50% off all items!" and analyzes it. The extracted keywords and phrases are the output.

[0133] Step 5:

[0134] The server uses the extracted keywords and phrases to perform a risk assessment. Specifically, it compares them with a database of past online scandals to see if similar patterns exist. For example, it determines that the phrase "all products" poses a risk. The assessed risk is the output.

[0135] Step 6:

[0136] The server generates a specific correction proposal based on the risk assessment. For example, it generates a correction proposal to change the expression "50% off all products!" to "50% off some products!" The generated correction proposal is the output.

[0137] Step 7:

[0138] The server presents the risk assessment results and suggested modifications to the user in a dashboard format. Specifically, the results are displayed in a format that is easy for the user to understand visually. For example, the dashboard displays suggested modifications such as "50% off all products!" and "50% off some products!" The presented information is the output.

[0139] Step 8:

[0140] The server generates a notification recommending consulting the legal department based on the risk assessment results. Specifically, if high-risk content is detected, it creates a notification such as "We recommend checking with the legal department." The generated notification is the output.

[0141] The above is the processing flow and specific operation of the program for this system.

[0142] (Application example 1)

[0143] 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."

[0144] In recent years, the amount of information published on the Internet has rapidly increased, and the risk of content posted by companies and individuals becoming a hot topic has also increased. However, there is currently a lack of means to identify the risk of a hot topic in advance and respond appropriately. For this reason, there is a need for a system that can efficiently diagnose the risk of a hot topic and propose specific corrections and actions. This system must be realized especially when using mobile devices such as smartphones.

[0145] 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.

[0146] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for referencing a database of past cases and generating a revision plan based on the risk assessment result, means for presenting the generated revision plan and the risk assessment result to the user, and means for suggesting specific revision plans and actions when a risk is detected. This enables users to take measures to quickly and accurately avoid the risk of a flame war using a mobile device such as a smartphone.

[0147] "User" refers to any individual or company that uses the System.

[0148] "Data" refers to text, images, or other information entered or uploaded by Users.

[0149] "Means for receiving" refers to the function of taking data entered or uploaded by the user into the server via the Internet.

[0150] "Means for converting to text format and preprocessing by removing unnecessary information" refers to the process of converting received data into a format that can be analyzed and removing irrelevant information.

[0151] A "natural language processing model" refers to algorithms and techniques for analyzing text data, understanding its content, and extracting its intent.

[0152] "Means for assessing risk" refers to a function that assesses the risk of content causing a backlash based on preprocessed data.

[0153] The "past case database" refers to a database that accumulates past cases of online outrage.

[0154] "Means of generation" refers to the function of proposing appropriate corrections or actions based on the results of risk assessment.

[0155] "Means of presentation" refers to the function of notifying the user of generated correction proposals and risk assessment results.

[0156] "Means to suggest specific corrections and actions" refers to a function that shows the user specific countermeasures and corrections for detected risks.

[0157] "Dashboard function" refers to an interface that displays risk assessment results and correction suggestions in a format that makes it easy for users to visually confirm them.

[0158] "OCR technology" refers to the technology that optically recognizes text in an image and extracts it as electronic data.

[0159] The system of the present invention receives data (text and images) entered or uploaded by users via the Internet, analyzes the received data, and evaluates the risk of a social media outcry. If a risk is detected, the system proposes specific corrections and actions. The system has the following processing flow:

[0160] Hardware and Software

[0161] This system mainly uses the following hardware and software.

[0162] 1. Server: The central device that receives, preprocesses, analyzes data, and generates and presents results.

[0163] 2. Smartphone or computer: This is the device where users enter and upload data and check the results.

[0164] 3. Natural language processing model (spacy): A library for analyzing text data.

[0165] 4. OCR technology (pytesseract and PIL): Technology for extracting text from images and converting it into an analyzable format.

[0166] Data reception and preprocessing

[0167] The server receives text data entered by users or image data uploaded via the Internet. The received text data is preprocessed by removing HTML tags and special characters and normalizing it. For image data, OCR technology is used to extract the text within the image and normalize it.

[0168] Data analysis and risk assessment

[0169] The preprocessed text and image data are analyzed using a natural language processing model (SPACY). This analysis extracts keywords and phrases from the text and evaluates negative elements and potential risks. The results are then compared with a database of past online outrage cases to assess risk.

[0170] Suggested fixes and actions

[0171] The server generates specific correction suggestions based on the risk assessment results. For example, it determines that the expression "50% off all products!" is risky and suggests changing it to "50% off some products!" It also generates a notification recommending consulting the legal department if necessary. These results are presented to the user in the form of a visual dashboard.

[0172] Specific operation scenario

[0173] 1. User: Enter a new campaign idea and upload a campaign banner. Example: "New campaign starts! 50% off all products!"

[0174] 2. Server: Receives this and performs preprocessing.

[0175] 3. Server: Performs risk assessment using natural language processing and deep learning models and compares the results with a database of past cases.

[0176] 4. Server: If a risk is detected, a notification is generated and presented to the user with specific remediation suggestions and recommendations for consulting the legal department.

[0177] 5. User: Check the information through the dashboard and take appropriate corrections or actions.

[0178] Example prompt sentence:

[0179] content = "New campaign starts! 50% off all products!"

[0180] image_path = "path / to / banner_image.jpg"

[0181] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0182] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0183] Step 1:

[0184] The server receives text data entered by the user via the Internet and image data uploaded by the user.

[0185] Input: Text data entered by the user and image data uploaded by the user.

[0186] Output: Received data.

[0187] Specific operation: The user uploads the text data "New campaign starts! All products 50% off!" and the campaign banner image. The server receives this data via the Internet.

[0188] Step 2:

[0189] The server pre-processes the received text data, i.e., removes HTML tags and special characters, and normalizes it.

[0190] Input: Received data (text data).

[0191] Output: Preprocessed text data.

[0192] Specific operation: Removes HTML tags and special characters from the text data "New campaign starts! All products 50% off!" and performs normalization such as lowercasing.

[0193] Step 3:

[0194] The server uses OCR technology on the received image data to extract and normalize the text within the image.

[0195] Input: Received data (image data).

[0196] Output: Preprocessed text data.

[0197] Specific operation: Extract text from campaign banner images using pytesseract and PIL, and normalize the text by lowercasing it, etc.

[0198] Step 4:

[0199] The server analyzes the preprocessed text data using a natural language processing model (spacy), which extracts keywords and phrases and evaluates negative elements and risks.

[0200] Input: Preprocessed text data.

[0201] Output: Analysis results (keywords, phrases, risk assessment).

[0202] Specific operation: The spacy model analyzes text data containing the phrase "50% off all products!", extracts keywords and phrases, and compares them with a past database to assess risk.

[0203] Step 5:

[0204] Based on the risk assessment results, the server refers to a database of past controversy cases and generates appropriate correction proposals.

[0205] Input: Analysis results, database of past flame war cases.

[0206] Output: Revision proposal.

[0207] Specific behavior: Generate a suggestion to correct the phrase "50% off all items!" to "50% off some items!"

[0208] Step 6:

[0209] The server presents the generated remediation suggestions and risk assessment results to the user, and if necessary, generates a notification recommending consultation with the legal department.

[0210] Input: Correction proposal, risk assessment results.

[0211] Output: Presentation information, notifications.

[0212] Specific behavior: The user is presented with a visual dashboard showing suggested fixes such as "50% off selected items!" and the results of the risk assessment, and a notification is generated recommending that the user contact the legal department.

[0213] Step 7:

[0214] Users can view the information through the dashboard and take appropriate corrective action.

[0215] Input: Presented information, notifications.

[0216] Output: Modified content, behavior.

[0217] Specific actions: The user checks the proposed revisions and risk assessment results on the dashboard, changes "50% off all items!" to "50% off some items!", and consults with the legal department if necessary.

[0218] 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.

[0219] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and proposes appropriate corrections and actions. Furthermore, by combining it with an emotion engine that recognizes user emotions, the accuracy of risk assessment is improved.

[0220] What the program does

[0221] Receiving and Preprocessing Input Data

[0222] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[0223] Examples:

[0224] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a campaign banner image. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into an analyzable format.

[0225] Data analysis and risk assessment

[0226] The server receives the preprocessed data and first analyzes user sentiment using an emotion engine. It then uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the results with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[0227] Examples:

[0228] When the user inputs "50% off all products!", the server uses an emotion engine to recognize the emotion of "excitement, anticipation." The NLP model then determines that the expression "all products" poses a risk, and issues a warning because a database of past cases records similar campaigns that have caused misunderstandings and sparked outrage.

[0229] Generating results and presenting suggestions

[0230] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[0231] Examples:

[0232] The server generates a suggested correction for the risky phrase "50% off all items!" to "50% off some items!" and also creates a notification recommending confirmation by the legal department. Taking into account the results of sentiment analysis, the server reevaluates whether this suggested correction is appropriate for the user. The user can check this information on the dashboard and make any necessary corrections.

[0233] Specific program operation scenarios

[0234] Users input new campaign ideas and upload campaign banners. The server receives and preprocesses them. It then uses an emotion engine, natural language processing models, and deep learning models to perform a risk assessment. It compares the results with a database of past cases, and if a risk is detected, it generates a notification to the user suggesting specific corrections and consultation with the legal department. Users can view the information through a dashboard and take appropriate corrections or actions.

[0235] The above is an embodiment of the present invention and its specific example. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and maintain credibility.

[0236] The processing flow will be explained below.

[0237] What the program does

[0238] Processing steps of a system that combines emotion engines

[0239] Step 1:

[0240] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[0241] Step 2:

[0242] The server receives text data and image data sent from the user via the Internet.

[0243] Step 3:

[0244] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[0245] Step 4:

[0246] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[0247] Step 5:

[0248] The server applies an emotion engine to the preprocessed text data to recognize the user's emotions, which are then recorded in a database for subsequent processing.

[0249] Step 6:

[0250] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[0251] Step 7:

[0252] The server detects negative elements and potential risks based on the data analyzed by the NLP model, taking into account the results of the sentiment analysis recognized in step 5.

[0253] Step 8:

[0254] The server compares the data with a database of past flame war cases, performs pattern matching with the given data, and checks whether similar cases exist.

[0255] Step 9:

[0256] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[0257] Step 10:

[0258] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction suggestion. For example, it suggests correcting the expression "50% off all products!" to "50% off some products!"

[0259] Step 11:

[0260] The server generates notifications recommending legal consultation as needed, with specific justification based on past experience and current risk assessment.

[0261] Step 12:

[0262] The server visualizes the generated correction proposals and risk assessment results and sends the results to the user's terminal in a dashboard format that is easy for the user to understand.

[0263] Step 13:

[0264] Users can check the proposed modifications and risk assessment results presented on the dashboard and make any necessary modifications. Once the modifications are complete, they can also perform a reassessment.

[0265] Example 2

[0266] 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."

[0267] There is a lack of methods to proactively diagnose the risk of online content becoming controversial and propose appropriate revisions and actions. In particular, there is a need for a system that can assess the risk of not only text data but also image data, and that also takes user emotions into account.

[0268] 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 means for receiving data entered by a user via the Internet, means for performing preprocessing of the received data by removing HTML tags and special characters and normalizing the text, means for extracting text from image data using OCR technology, means for performing sentiment analysis using an emotion engine based on the preprocessed data, means for assessing risk using a natural language processing model, means for confirming the risk assessment result by comparing it with a database of similar past cases, means for generating a revision proposal based on the risk assessment result, and means for presenting the generated revision proposal and the risk assessment result to the user. This makes it possible to prevent the risk of a controversy in advance and increase the reliability of content.

[0269] "User" means any person or entity that uses the System to input or upload data.

[0270] "Server" refers to the hardware or software that receives, processes, and analyzes data entered or uploaded by users.

[0271] "Data" refers to information entered or uploaded by a user into the system, in the form of text, images, etc.

[0272] "Internet" refers to the global information network that enables the communication of data between users and servers.

[0273] "Preprocessing" refers to the processing performed on received data, specifically tasks such as removing HTML tags and special characters, and normalizing text.

[0274] "OCR technology" refers to optical character recognition technology for extracting character information from image data and converting it into text format.

[0275] An "emotion engine" refers to software or algorithms that analyze and determine user emotions from text data.

[0276] A "natural language processing model" refers to an algorithm or software that analyzes text data, extracts keywords and phrases, and assesses risk.

[0277] "Deep learning model" refers to a machine learning algorithm that analyzes image data and automatically evaluates inappropriate content or problematic elements.

[0278] A "database" refers to an information resource that stores past cases and information and is referenced when conducting risk assessments.

[0279] "Proposed fixes" refer to proposals that are generated based on the results of risk assessment and indicate specific fixes or countermeasures that the user should take.

[0280] A "dashboard" refers to an interface that visually displays risk assessment results and suggested corrections to users.

[0281] This invention is an AI-based system that proactively diagnoses the risk of online content published by companies and individuals becoming controversial and proposes appropriate corrections and actions. The system receives data (text and images) entered or uploaded by users, and performs a comprehensive process from data preprocessing, sentiment analysis, risk assessment, generation of correction suggestions, and presentation of the results to the user.

[0282] Receiving and Preprocessing Input Data

[0283] The server receives data entered or uploaded by users via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. This converts the data into analyzable data. For image data, OCR (Optical Character Recognition) technology is used to extract text from the image and convert it into an analyzable text format.

[0284] sentiment analysis

[0285] The server performs sentiment analysis on the preprocessed text data using an emotion engine, which is software that uses natural language processing (NLP) technology to determine the emotions contained in the text, such as "excitement" or "anticipation."

[0286] Risk Assessment

[0287] The server not only evaluates content based on sentiment analysis, but also uses natural language processing models to extract keywords and phrases from the text to detect negative elements and potential risks. It also uses deep learning models to analyze preprocessed image data to assess whether it contains inappropriate content or problematic elements. These processes also involve cross-checking the data with a database of past scandals.

[0288] Generating and Presenting Results

[0289] The server generates specific correction suggestions based on the risk assessment results. For example, it suggests changing the phrase "50% off all products!" to "50% off some products!" It also generates a notification recommending consultation with the legal department if necessary. These results and correction suggestions are presented to the user in the form of a visual dashboard.

[0290] An example of a specific prompt sentence is the action of entering the text "New campaign starts! All products 50% off!" and uploading a campaign banner image. In this case, the server receives the data corresponding to the prompt, preprocesses it, analyzes sentiment, and evaluates risks, and presents optimal correction suggestions to the user. The user can check this information on the dashboard and make any necessary corrections.

[0291] This completes the description of the embodiment of the present invention. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and increase the reliability of their content.

[0292] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0293] Step 1: Receive the database

[0294] The server supports users. (PC, 스마트폰) HTTP

[0295] Specific behavior:

[0296] User 가 "New campaign starts! 50% off all products!" Campaign Yes, the server is private.

[0297] Input: Data entered or uploaded by Yuzaga (text and images)

[0298] Output: Received data requiring preprocessing

[0299] ステップ2:Preprocessing of データ

[0300] Server performs preprocessing on the received data. For text data, it removes HTML tags and special characters and normalizes the text. For image data, it uses OCR technology to extract the text within the image and converts it into an analyzable text format.

[0301] Artistic work:

[0302] HTML tags and special characters are removed from the text "New campaign opening! All products 50% off!" entered by Yuza, and the text "All products 50% off!" is extracted from the deployed campaign banner image using OCR technology.

[0303] Input: Received data (text and images)

[0304] Output: Preprocessed data, normalized and converted to text.

[0305] Step 3: Emotion Analysis

[0306] サーバ inputs preprocessed text data into the emotion engine to analyze Yuza's emotions. This emotion engine is software that uses natural language processing (NLP) technology to determine the emotions of the text.

[0307] Artistic work:

[0308] The emotion engine analyzes the text "New campaign opening! 50% off all products!" and detects emotions such as "excitement" and "anticipation."

[0309] Input: Normalized text data

[0310] Output: Detected emotion result

[0311] Step 4: Risk assessment

[0312] The server analyzes the pre - processed data by combining the NLP model and the deep - learning model. In text data, keywords and phrases are extracted to detect negative elements and potential risks. Similarly, for image data, the deep - learning model is applied to evaluate whether there is inappropriate content or problem elements. In this evaluation, the risk assessment result is reviewed by querying the database of similar cases that occurred before.

[0313] Specific operation:

[0314] The NLP model determines that the keyword "all products" has a risk. There is a record in the previous similar case database where a campaign was themed with a similar expression, so a warning is issued.

[0315] Input: Emotion analysis result, pre - processed text and image data

[0316] Output: Risk assessment result and detected risks

[0317] Step 5: Result generation

[0318] When risks are detected, the server generates specific modification suggestions. For example, for the expression "50% off all products!", it is proposed to change it to "50% off some products!". Also, a notice recommending consultation with the legal department is generated if necessary. <​​​​​The server creates a notice recommending verification by the legal department by drafting a revision to "50% off all products!" to "50% off some products!" for the expression "50% off all products!".

[0321] Input: Risk assessment result

[0322] Output: Specific revision and notice of recommendation

[0323] Step 6: Presenting the result

[0324] The server presents the generated revision or notice to the user in the form of a visualized dashboard. The user can view the information through the dashboard and make necessary revisions.

[0325] Specific operations:

[0326] The user opens the dashboard on the terminal to view the risk assessment result of "50% off all products!" and the revision of "50% off some products!". A notice of verification to the legal department is also displayed.

[0327] Input: Specific revision and notice of recommendation

[0328] Output: Visualized assessment result and revision for the user

[0329] The above are the specific processing steps performed by the system of the present invention.

[0330] [[ID=***]]

[0331] Next, Application Example 2 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart device 14 is referred to as the "terminal".

[0332] ​There is a need for a system that can proactively diagnose the risk of online content becoming a hot topic and propose appropriate revisions and actions. It is also necessary to provide a more accurate risk assessment that can handle a variety of data formats, including images and audio, and that takes user emotions into account.

[0333] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for generating a revision plan based on the risk assessment result by referring to a database of past cases, means for presenting the generated revision plan and the risk assessment result to the user, means for recognizing the user's emotions using an emotion engine to improve the accuracy of the risk assessment, and means for collecting and analyzing audio and video data via a smart device. This enables advance diagnosis of the risk of a controversy in content that a user intends to distribute and appropriate measures to be taken.

[0334] The "Internet" is a global information and communications network that interconnects computer networks around the world.

[0335] "User" means any person or entity that uses a system or device.

[0336] "Data" is a collection of information and includes various formats such as text, images, and audio.

[0337] A "server" is a computer system used to receive, process, and transmit data.

[0338] "Preprocessing" is the initial processing of data to convert it into an analyzable format.

[0339] "Natural language processing model" is a general term for algorithms and technologies for understanding and manipulating human language.

[0340] "Risk assessment" is the process of evaluating the potential dangers or problems of a particular activity or situation.

[0341] A "case database" is a database that systematically collects past cases and data.

[0342] "Fixation Suggestions" refers to specific improvements or proposals for addressing detected issues or risks.

[0343] "Emotion engine" is an artificial intelligence technology for analyzing and recognizing user emotions.

[0344] A "smart device" is an advanced device that has internet connectivity and is capable of collecting and analyzing data.

[0345] "Audio data" refers to audio information recorded in digital format.

[0346] "Video data" refers to visual information such as images and videos recorded in digital format.

[0347] "Analysis" is the process of analyzing data in detail and extracting its meaning and value.

[0348] This invention is a system that proactively diagnoses the risk of online content becoming controversial and suggests appropriate modifications and actions. The system receives data entered by users and performs risk assessment using an emotion engine and natural language processing model. Furthermore, the accuracy of the assessment is improved by collecting and analyzing audio and video data via smart devices.

[0349] The server first receives data entered by the user via the Internet. The data may be in a variety of formats, including text and image formats, but after receiving it, it undergoes preprocessing to make it analyzable. For example, text data is normalized by removing HTML tags and special characters. For image data, OCR technology (pytesseract library) is used to extract text from the image, and further preprocessing is performed.

[0350] Next, we use the preprocessed data to run the emotion engine, which performs sentiment analysis on the content users are about to post. For example, we can recognize emotions such as "excitement," "expectation," and "dissatisfaction." We then use the TextBlob library to evaluate the polarity of the emotions.

[0351] Once the sentiment analysis is complete, a natural language processing model (Keras and a text tokenizer) is used to assess the risk of the received text data. During this process, the data is compared with a database of past scandals to determine whether there is any risk. Specific revisions are then proposed based on the generated risk assessment results. For example, if "50% off all products!" is judged to be a risky expression, a suggestion is made to change it to "50% off some products!"

[0352] Finally, the server presents the risk assessment results and suggested remediation to the user in a visual dashboard, allowing the user to take concrete corrective action. Depending on the level of risk, a notification may also be generated recommending that the user consult with the legal department.

[0353] These processes are carried out via smart devices, making it easy to collect audio and video data. For example, content can be evaluated on the spot using a smartphone or tablet, and suggestions for revisions can be received immediately.

[0354] Specific examples

[0355] The user inputs the text "New campaign starts! All products 50% off!" as a new campaign proposal and uploads a campaign banner image. The server receives this and preprocesses the text and image. It then uses an emotion engine and natural language processing model to perform sentiment analysis and risk assessment, generating a revised proposal: "50% off some products!"

[0356] Example prompts to input to a generative AI model:

[0357] test_text = "New campaign starts! 50% off all items!"

[0358] test_image_path = "path_to_campaign_banner.jpg"

[0359] main(test_text, test_image_path)

[0360] By entering this prompt, users can check the risk assessment results and suggested corrections on the dashboard, enabling safer content delivery.

[0361] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0362] Step 1:

[0363] The user sends text data or image data from their terminal to the server via the Internet. The user's input data is content information such as campaign proposals. In this step, it is confirmed whether the user's input reaches the server.

[0364] Input: Text data and image data entered by the user

[0365] Output: Raw data sent to the server

[0366] Step 2:

[0367] The server preprocesses the data it receives. For text data, it removes HTML tags and special characters and normalizes it. For image data, it uses OCR technology (pytesseract library) to extract text. This step results in clean data that can be analyzed.

[0368] Input: Raw data sent to the server

[0369] Output: Preprocessed, clean text data

[0370] Step 3:

[0371] The server performs sentiment analysis on the preprocessed text data. It uses the TextBlob library to evaluate the sentiment polarity of the text and determine whether it is "positive" or "negative." This step quantifies the user's sentiment.

[0372] Input: Preprocessed clean text data

[0373] Output: Sentiment analysis result (positive or negative)

[0374] Step 4:

[0375] The server uses a natural language processing model to assess risk. Using Keras and a text tokenizer, the text data is passed to the model, which then compares it with a database of past scandals to calculate a risk score. This step quantifies the risk of a particular piece of content becoming a scandal.

[0376] Input: Preprocessed clean text data

[0377] Output: Risk score

[0378] Step 5:

[0379] The server generates specific correction suggestions based on the risk assessment results. If the risk score is high, it will suggest corrections to the problematic expression. For example, instead of "50% off all products!", it will suggest "50% off some products!"

[0380] Input: Risk Score

[0381] Output: Revision proposal

[0382] Step 6:

[0383] The server then presents the generated remediation proposals and risk assessment results to the user, who can then view this information and consider appropriate responses through a visual dashboard.

[0384] Input: Remediation proposals and risk assessment results

[0385] Output: Dashboard display

[0386] Step 7:

[0387] Depending on the risk assessment result, the server generates a notification recommending that the user consult with the legal department. If the risk score is particularly high, a notification is generated and the user is prompted to check with the legal department.

[0388] Input: Risk Score

[0389] Output: Notice recommending consultation with legal department

[0390] Step 8:

[0391] The user can provide additional input (audio and video data) via a smart device. This is also received by the server and similarly preprocessed and analyzed, enabling more accurate evaluation.

[0392] Input: Additional audio and video data

[0393] Output: Analysis results

[0394] 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.

[0395] 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.

[0396] 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.

[0397] [Second embodiment]

[0398] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0399] 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.

[0400] 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).

[0401] 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.

[0402] 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.

[0403] 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).

[0404] 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.

[0405] 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.

[0406] 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.

[0407] 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.

[0408] 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.

[0409] 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."

[0410] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and suggests appropriate corrections and actions. Below, we will generate a program for this system and explain its processing in natural language.

[0411] What the program does

[0412] Receiving and Preprocessing Input Data

[0413] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[0414] Examples:

[0415] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a banner image for the campaign. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into a format that can be analyzed.

[0416] Data analysis and risk assessment

[0417] The server receives the preprocessed data and begins analyzing it. It uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the data with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[0418] Examples:

[0419] When the user types "50% off all products!", the server determines that the expression "all products" poses a risk. A database of past cases records similar campaigns that have caused misunderstandings and sparked outrage, so the server issues a warning based on this.

[0420] Generating results and presenting suggestions

[0421] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[0422] Examples:

[0423] The server generates a correction suggestion for the risky phrase "50% off all products!" to "50% off some products!" and also creates a notification recommending confirmation by the legal department. The user can check this information on the dashboard and make the necessary corrections.

[0424] Specific program operation scenarios

[0425] Users input new campaign ideas and upload campaign banners. The server receives and pre-processes them. It then uses natural language processing and deep learning models to perform a risk assessment and compares it with a database of past cases. If a risk is detected, a notification is generated and presented to the user, suggesting specific remediation measures and consultation with the legal department. Users can review the information through a dashboard and take appropriate remediation or action.

[0426] The above is an embodiment of the present invention and its specific example. This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0427] The processing flow will be explained below.

[0428] Step 1:

[0429] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[0430] Step 2:

[0431] The server receives text data and image data sent from the user via the Internet.

[0432] Step 3:

[0433] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[0434] Step 4:

[0435] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[0436] Step 5:

[0437] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[0438] Step 6:

[0439] The server detects negative elements and the risk of a backlash based on data analyzed by an NLP model.

[0440] Step 7:

[0441] The server refers to a database of past flame war cases, performs pattern matching with the preprocessed data, and checks whether similar cases exist.

[0442] Step 8:

[0443] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[0444] Step 9:

[0445] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction proposal (for example, changing "50% off all products!" to "50% off some products!").

[0446] Step 10:

[0447] The server generates a notification recommending legal consultation if necessary.

[0448] Step 11:

[0449] The server visualizes the generated correction proposals and risk assessment results and sends the results to the terminal in a dashboard format that is easy for the user to understand.

[0450] Step 12:

[0451] The user checks the proposed modifications and risk assessment results presented on the dashboard and makes any necessary modifications.

[0452] Example 1

[0453] 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."

[0454] Content published online by companies and individuals often carries the risk of causing a firestorm. Therefore, there is a need for a system that can perform risk assessments before content is published and propose appropriate revisions and actions. Furthermore, because content is diverse, it is necessary to support multiple data formats, such as text and images. Furthermore, to improve the accuracy of risk assessments, it is important to compare the content with a database of past cases, and a method of presenting information that takes user ease of use into consideration is also required.

[0455] 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.

[0456] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and preprocessing it by removing unnecessary information, and means for assessing risk using a natural language processing model based on the preprocessed text data. This makes it possible to assess risk before content is made public and to propose appropriate modifications and actions.

[0457] The "Internet" is an infrastructure for sending and receiving information through networks around the world.

[0458] "User" means any person or entity that uses the System to input or upload Content.

[0459] "Data" means electronic files in any format that contain information, including text, images, audio, and video.

[0460] "Receiving" is the process by which the server takes in data that a user enters or uploads over the Internet.

[0461] "Text format" is a data format expressed as character information.

[0462] "Unnecessary information" refers to metadata, special characters, HTML tags, etc. that are not necessary for analysis or evaluation, and refers to removing this information.

[0463] "Preprocessing" refers to the preliminary data cleansing operations performed to convert received data into an analyzable format.

[0464] A "natural language processing model" is an artificial intelligence algorithm used to understand and analyze the meaning of text data.

[0465] "Risk assessment" is an analytical process used to determine the likelihood that words or phrases contained in data will spark outrage, misunderstanding, or negative reactions.

[0466] A "fix proposal" is a specific change proposal based on the results of a risk assessment to make the content published by users safer.

[0467] The "dashboard format" is an interface that allows users to visually check risk assessment results and proposed modifications.

[0468] A "legal department" is a department within a company or organization that handles legal matters.

[0469] "Image data" is a digital file containing visual information.

[0470] "OCR technology" is a technology that extracts character information from image data and converts it into text format.

[0471] "Inappropriate content" refers to information or expressions that are considered socially, legally, or morally problematic.

[0472]

[0473] The present invention is an AI-based system that proactively diagnoses the risk of online content being published by companies and individuals causing outrage and proposes appropriate corrections and actions. Detailed embodiments of this system are described below.

[0474] System Overview

[0475] The server receives data (text and images) entered or uploaded by users via the Internet. It then converts the received data into text format and performs preprocessing by removing unnecessary information. Specifically, this includes removing HTML tags and special characters and normalizing the text. For image data, OCR technology is used to extract character information and convert it into an analyzable text format. This completes the preprocessing of the text and image data.

[0476] The server then analyzes the preprocessed data and performs a risk assessment using a natural language processing (NLP) model. This involves extracting keywords and phrases from the text and determining whether they pose a negative or potential risk. Specifically, it compares them with a database of past scandals to see if there are any similar patterns. A deep learning model is also applied to image data to assess whether it contains inappropriate content.

[0477] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. These results and suggestions are presented to the user in the form of a visual dashboard.

[0478] Hardware and software used

[0479] Server: The main processing computer system, responsible for receiving, preprocessing, analyzing, and generating recommendations on data.

[0480] Internet: The communications infrastructure used to send and receive data.

[0481] Natural language processing model (NLP model): An AI model used to analyze text data.

[0482] OCR technology: A technology that extracts text information from image data.

[0483] Deep learning model: An AI model used to assess the content of image data.

[0484] Database: A database that stores past incidents of online outrage and is used to check against them when assessing risk.

[0485] Dashboard: An interface for presenting results and recommendations to the user.

[0486] Specific examples

[0487] For example, suppose a user uploads the text "New campaign begins! 50% off all products!" along with a banner image for the campaign. The server receives this data, first removes HTML tags and special characters from the text data, and then uses OCR technology to extract text information from the banner image. Next, it uses a natural language processing model to analyze the phrase "50% off all products!" and determines that the "all products" part poses a risk. It then compares this with a database of past cases and issues a warning based on this.

[0488] The server then generates a suggested change to change the phrase "50% off everything!" to "50% off some items!" and creates a notification recommending that the legal department review the change. This information is presented to the user via a dashboard, allowing them to make the necessary changes.

[0489] Prompt Sentence Examples

[0490] The prompt sentence to be input by the user may have the following format, for example:

[0491] Examples include "New campaign announcement! 50% off all products!" or "This year's new product sale, all items are half price!"

[0492] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0493] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0494] Step 1:

[0495] The server receives data entered or uploaded by users via the Internet. This data is often in text or image format. A specific example of input is the text "New campaign starts! All products 50% off!" and a campaign banner image. The output is the received data itself.

[0496] Step 2:

[0497] The server performs preprocessing on the received text data. Specifically, it removes HTML tags and special characters and normalizes the text. For example, 50% off all products! ", the HTML tags will be removed and the text "50% off all items!" will be output.

[0498] Step 3:

[0499] The server applies OCR technology to the received image data and extracts the text from the image. Specifically, it extracts the text information "50% off all products!" from the campaign banner image and converts it into an analyzable text format. The extracted text is output.

[0500] Step 4:

[0501] The server sends the preprocessed text data to a natural language processing (NLP) model to extract keywords and phrases. Specifically, it extracts the phrase "50% off all items!" and analyzes it. The extracted keywords and phrases are the output.

[0502] Step 5:

[0503] The server uses the extracted keywords and phrases to perform a risk assessment. Specifically, it compares them with a database of past online scandals to see if similar patterns exist. For example, it determines that the phrase "all products" poses a risk. The assessed risk is the output.

[0504] Step 6:

[0505] The server generates a specific correction proposal based on the risk assessment. For example, it generates a correction proposal to change the expression "50% off all products!" to "50% off some products!" The generated correction proposal is the output.

[0506] Step 7:

[0507] The server presents the risk assessment results and suggested modifications to the user in a dashboard format. Specifically, the results are displayed in a format that is easy for the user to understand visually. For example, the dashboard displays suggested modifications such as "50% off all products!" and "50% off some products!" The presented information is the output.

[0508] Step 8:

[0509] The server generates a notification recommending consulting the legal department based on the risk assessment results. Specifically, if high-risk content is detected, it creates a notification such as "We recommend checking with the legal department." The generated notification is the output.

[0510] The above is the processing flow and specific operation of the program for this system.

[0511] (Application example 1)

[0512] 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."

[0513] In recent years, the amount of information published on the Internet has rapidly increased, and the risk of content posted by companies and individuals becoming a hot topic has also increased. However, there is currently a lack of means to identify the risk of a hot topic in advance and respond appropriately. For this reason, there is a need for a system that can efficiently diagnose the risk of a hot topic and propose specific corrections and actions. This system must be realized especially when using mobile devices such as smartphones.

[0514] 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.

[0515] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for referencing a database of past cases and generating a revision plan based on the risk assessment result, means for presenting the generated revision plan and the risk assessment result to the user, and means for suggesting specific revision plans and actions when a risk is detected. This enables users to take measures to quickly and accurately avoid the risk of a flame war using a mobile device such as a smartphone.

[0516] "User" refers to any individual or company that uses the System.

[0517] "Data" refers to text, images, or other information entered or uploaded by Users.

[0518] "Means for receiving" refers to the function of taking data entered or uploaded by the user into the server via the Internet.

[0519] "Means for converting to text format and preprocessing by removing unnecessary information" refers to the process of converting received data into a format that can be analyzed and removing irrelevant information.

[0520] A "natural language processing model" refers to algorithms and techniques for analyzing text data, understanding its content, and extracting its intent.

[0521] "Means for assessing risk" refers to a function that assesses the risk of content causing a backlash based on preprocessed data.

[0522] The "past case database" refers to a database that accumulates past cases of online outrage.

[0523] "Means of generation" refers to the function of proposing appropriate corrections or actions based on the results of risk assessment.

[0524] "Means of presentation" refers to the function of notifying the user of generated correction proposals and risk assessment results.

[0525] "Means to suggest specific corrections and actions" refers to a function that shows the user specific countermeasures and corrections for detected risks.

[0526] "Dashboard function" refers to an interface that displays risk assessment results and correction suggestions in a format that makes it easy for users to visually confirm them.

[0527] "OCR technology" refers to the technology that optically recognizes text in an image and extracts it as electronic data.

[0528] The system of the present invention receives data (text and images) entered or uploaded by users via the Internet, analyzes the received data, and evaluates the risk of a social media outcry. If a risk is detected, the system proposes specific corrections and actions. The system has the following processing flow:

[0529] Hardware and Software

[0530] This system mainly uses the following hardware and software.

[0531] 1. Server: The central device that receives, preprocesses, analyzes data, and generates and presents results.

[0532] 2. Smartphone or computer: This is the device where users enter and upload data and check the results.

[0533] 3. Natural language processing model (spacy): A library for analyzing text data.

[0534] 4. OCR technology (pytesseract and PIL): Technology for extracting text from images and converting it into an analyzable format.

[0535] Data reception and preprocessing

[0536] The server receives text data entered by users or image data uploaded via the Internet. The received text data is preprocessed by removing HTML tags and special characters and normalizing it. For image data, OCR technology is used to extract the text within the image and normalize it.

[0537] Data analysis and risk assessment

[0538] The preprocessed text and image data are analyzed using a natural language processing model (SPACY). This analysis extracts keywords and phrases from the text and evaluates negative elements and potential risks. The results are then compared with a database of past online outrage cases to assess risk.

[0539] Suggested fixes and actions

[0540] The server generates specific correction suggestions based on the risk assessment results. For example, it determines that the expression "50% off all products!" is risky and suggests changing it to "50% off some products!" It also generates a notification recommending consulting the legal department if necessary. These results are presented to the user in the form of a visual dashboard.

[0541] Specific operation scenario

[0542] 1. User: Enter a new campaign idea and upload a campaign banner. Example: "New campaign starts! 50% off all products!"

[0543] 2. Server: Receives this and performs preprocessing.

[0544] 3. Server: Performs risk assessment using natural language processing and deep learning models and compares the results with a database of past cases.

[0545] 4. Server: If a risk is detected, a notification is generated and presented to the user with specific remediation suggestions and recommendations for consulting the legal department.

[0546] 5. User: Check the information through the dashboard and take appropriate corrections or actions.

[0547] Example prompt sentence:

[0548] content = "New campaign starts! 50% off all products!"

[0549] image_path = "path / to / banner_image.jpg"

[0550] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0551] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0552] Step 1:

[0553] The server receives text data entered by the user via the Internet and image data uploaded by the user.

[0554] Input: Text data entered by the user and image data uploaded by the user.

[0555] Output: Received data.

[0556] Specific operation: The user uploads the text data "New campaign starts! All products 50% off!" and the campaign banner image. The server receives this data via the Internet.

[0557] Step 2:

[0558] The server pre-processes the received text data, i.e., removes HTML tags and special characters, and normalizes it.

[0559] Input: Received data (text data).

[0560] Output: Preprocessed text data.

[0561] Specific operation: Removes HTML tags and special characters from the text data "New campaign starts! All products 50% off!" and performs normalization such as lowercasing.

[0562] Step 3:

[0563] The server uses OCR technology on the received image data to extract and normalize the text within the image.

[0564] Input: Received data (image data).

[0565] Output: Preprocessed text data.

[0566] Specific operation: Extract text from campaign banner images using pytesseract and PIL, and normalize the text by lowercasing it, etc.

[0567] Step 4:

[0568] The server analyzes the preprocessed text data using a natural language processing model (spacy), which extracts keywords and phrases and evaluates negative elements and risks.

[0569] Input: Preprocessed text data.

[0570] Output: Analysis results (keywords, phrases, risk assessment).

[0571] Specific operation: The spacy model analyzes text data containing the phrase "50% off all products!", extracts keywords and phrases, and compares them with a past database to assess risk.

[0572] Step 5:

[0573] Based on the risk assessment results, the server refers to a database of past controversy cases and generates appropriate correction proposals.

[0574] Input: Analysis results, database of past flame war cases.

[0575] Output: Revision proposal.

[0576] Specific behavior: Generate a suggestion to correct the phrase "50% off all items!" to "50% off some items!"

[0577] Step 6:

[0578] The server presents the generated remediation suggestions and risk assessment results to the user, and if necessary, generates a notification recommending consultation with the legal department.

[0579] Input: Correction proposal, risk assessment results.

[0580] Output: Presentation information, notifications.

[0581] Specific behavior: The user is presented with a visual dashboard showing suggested fixes such as "50% off selected items!" and the results of the risk assessment, and a notification is generated recommending that the user contact the legal department.

[0582] Step 7:

[0583] Users can view the information through the dashboard and take appropriate corrective action.

[0584] Input: Presented information, notifications.

[0585] Output: Modified content, behavior.

[0586] Specific actions: The user checks the proposed revisions and risk assessment results on the dashboard, changes "50% off all items!" to "50% off some items!", and consults with the legal department if necessary.

[0587] 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.

[0588] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and proposes appropriate corrections and actions. Furthermore, by combining it with an emotion engine that recognizes user emotions, the accuracy of risk assessment is improved.

[0589] What the program does

[0590] Receiving and Preprocessing Input Data

[0591] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[0592] Examples:

[0593] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a campaign banner image. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into an analyzable format.

[0594] Data analysis and risk assessment

[0595] The server receives the preprocessed data and first analyzes user sentiment using an emotion engine. It then uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the results with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[0596] Examples:

[0597] When the user inputs "50% off all products!", the server uses an emotion engine to recognize the emotion of "excitement, anticipation." The NLP model then determines that the expression "all products" poses a risk, and issues a warning because a database of past cases records similar campaigns that have caused misunderstandings and sparked outrage.

[0598] Generating results and presenting suggestions

[0599] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[0600] Examples:

[0601] The server generates a suggested correction for the risky phrase "50% off all items!" to "50% off some items!" and also creates a notification recommending confirmation by the legal department. Taking into account the results of sentiment analysis, the server reevaluates whether this suggested correction is appropriate for the user. The user can check this information on the dashboard and make any necessary corrections.

[0602] Specific program operation scenarios

[0603] Users input new campaign ideas and upload campaign banners. The server receives and preprocesses them. It then uses an emotion engine, natural language processing models, and deep learning models to perform a risk assessment. It compares the results with a database of past cases, and if a risk is detected, it generates a notification to the user suggesting specific corrections and consultation with the legal department. Users can view the information through a dashboard and take appropriate corrections or actions.

[0604] The above is an embodiment of the present invention and its specific example. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and maintain credibility.

[0605] The processing flow will be explained below.

[0606] What the program does

[0607] Processing steps of a system that combines emotion engines

[0608] Step 1:

[0609] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[0610] Step 2:

[0611] The server receives text data and image data sent from the user via the Internet.

[0612] Step 3:

[0613] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[0614] Step 4:

[0615] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[0616] Step 5:

[0617] The server applies an emotion engine to the preprocessed text data to recognize the user's emotions, which are then recorded in a database for subsequent processing.

[0618] Step 6:

[0619] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[0620] Step 7:

[0621] The server detects negative elements and potential risks based on the data analyzed by the NLP model, taking into account the results of the sentiment analysis recognized in step 5.

[0622] Step 8:

[0623] The server compares the data with a database of past flame war cases, performs pattern matching with the given data, and checks whether similar cases exist.

[0624] Step 9:

[0625] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[0626] Step 10:

[0627] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction suggestion. For example, it suggests correcting the expression "50% off all products!" to "50% off some products!"

[0628] Step 11:

[0629] The server generates notifications recommending legal consultation as needed, with specific justification based on past experience and current risk assessment.

[0630] Step 12:

[0631] The server visualizes the generated correction proposals and risk assessment results and sends the results to the user's terminal in a dashboard format that is easy for the user to understand.

[0632] Step 13:

[0633] Users can check the proposed modifications and risk assessment results presented on the dashboard and make any necessary modifications. Once the modifications are complete, they can also perform a reassessment.

[0634] Example 2

[0635] 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."

[0636] There is a lack of methods to proactively diagnose the risk of online content becoming controversial and propose appropriate revisions and actions. In particular, there is a need for a system that can assess the risk of not only text data but also image data, and that also takes user emotions into account.

[0637] 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 means for receiving data entered by a user via the Internet, means for performing preprocessing of the received data by removing HTML tags and special characters and normalizing the text, means for extracting text from image data using OCR technology, means for performing sentiment analysis using an emotion engine based on the preprocessed data, means for assessing risk using a natural language processing model, means for confirming the risk assessment result by comparing it with a database of similar past cases, means for generating a revision proposal based on the risk assessment result, and means for presenting the generated revision proposal and the risk assessment result to the user. This makes it possible to prevent the risk of a controversy in advance and increase the reliability of content.

[0638] "User" means any person or entity that uses the System to input or upload data.

[0639] "Server" refers to the hardware or software that receives, processes, and analyzes data entered or uploaded by users.

[0640] "Data" refers to information entered or uploaded by a user into the system, in the form of text, images, etc.

[0641] "Internet" refers to the global information network that enables the communication of data between users and servers.

[0642] "Preprocessing" refers to the processing performed on received data, specifically tasks such as removing HTML tags and special characters, and normalizing text.

[0643] "OCR technology" refers to optical character recognition technology for extracting character information from image data and converting it into text format.

[0644] An "emotion engine" refers to software or algorithms that analyze and determine user emotions from text data.

[0645] A "natural language processing model" refers to an algorithm or software that analyzes text data, extracts keywords and phrases, and assesses risk.

[0646] "Deep learning model" refers to a machine learning algorithm that analyzes image data and automatically evaluates inappropriate content or problematic elements.

[0647] A "database" refers to an information resource that stores past cases and information and is referenced when conducting risk assessments.

[0648] "Proposed fixes" refer to proposals that are generated based on the results of risk assessment and indicate specific fixes or countermeasures that the user should take.

[0649] A "dashboard" refers to an interface that visually displays risk assessment results and suggested corrections to users.

[0650] This invention is an AI-based system that proactively diagnoses the risk of online content published by companies and individuals becoming controversial and proposes appropriate corrections and actions. The system receives data (text and images) entered or uploaded by users, and performs a comprehensive process from data preprocessing, sentiment analysis, risk assessment, generation of correction suggestions, and presentation of the results to the user.

[0651] Receiving and Preprocessing Input Data

[0652] The server receives data entered or uploaded by users via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. This converts the data into analyzable data. For image data, OCR (Optical Character Recognition) technology is used to extract text from the image and convert it into an analyzable text format.

[0653] sentiment analysis

[0654] The server performs sentiment analysis on the preprocessed text data using an emotion engine, which is software that uses natural language processing (NLP) technology to determine the emotions contained in the text, such as "excitement" or "anticipation."

[0655] Risk Assessment

[0656] The server not only evaluates content based on sentiment analysis, but also uses natural language processing models to extract keywords and phrases from the text to detect negative elements and potential risks. It also uses deep learning models to analyze preprocessed image data to assess whether it contains inappropriate content or problematic elements. These processes also involve cross-checking the data with a database of past scandals.

[0657] Generating and Presenting Results

[0658] The server generates specific correction suggestions based on the risk assessment results. For example, it suggests changing the phrase "50% off all products!" to "50% off some products!" It also generates a notification recommending consultation with the legal department if necessary. These results and correction suggestions are presented to the user in the form of a visual dashboard.

[0659] An example of a specific prompt sentence is the action of entering the text "New campaign starts! All products 50% off!" and uploading a campaign banner image. In this case, the server receives the data corresponding to the prompt, preprocesses it, analyzes sentiment, and evaluates risks, and presents optimal correction suggestions to the user. The user can check this information on the dashboard and make any necessary corrections.

[0660] This completes the description of the embodiment of the present invention. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and increase the reliability of their content.

[0661] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0662] Step 1: Receive the database

[0663] The server supports users. (PC, 스마트폰) HTTP

[0664] Specific behavior:

[0665] User 가 "New campaign starts! 50% off all products!" Campaign Yes, the server is private.

[0666] Input: User input

[0667] Pre-processing: Pre-processing

[0668] Step 2: Data collection

[0669] Server performs preprocessing on the received data. For text data, it removes HTML tags and special characters and normalizes the text. For image data, it uses OCR technology to extract the text within the image and converts it into an analyzable text format.

[0670] Artistic work:

[0671] HTML tags and special characters are removed from the text "New campaign opening! All products 50% off!" entered by Yuza, and the text "All products 50% off!" is extracted from the deployed campaign banner image using OCR technology.

[0672] Input: Received data (text and images)

[0673] Output: Preprocessed data, normalized and converted to text.

[0674] Step 3: Emotion Analysis

[0675] サーバ inputs preprocessed text data into the emotion engine to analyze Yuza's emotions. This emotion engine is software that uses natural language processing (NLP) technology to determine the emotions of the text.

[0676] Artistic work:

[0677] The emotion engine analyzes the text "New campaign opening! 50% off all products!" and detects emotions such as "excitement" and "anticipation."

[0678] Input: Normalized text data

[0679] Output: Detected emotion results

[0680] ステップ4:リスク Evaluation

[0681] The server analyzes the preprocessed data by combining the NLP model and the deep learning model. In the text data, keywords and phrases are extracted to detect negative elements and potential risks. Similarly, for image data, the deep learning model is applied to evaluate whether there is inappropriate content or problem elements. In this evaluation, the risk assessment results are reviewed by querying the database of similar cases that occurred previously.

[0682] Specific operations:

[0683] The NLP model determines that the keyword "all products" has a risk. There is a record in the previous case database where a campaign was conducted with a similar expression, so a warning is issued.

[0684] Input: Sentiment analysis results, preprocessed text, and image data

[0685] Output: Risk assessment results and detected risks

[0686] Step 5: Result generation

[0687] When the server detects a risk, it generates specific modification suggestions. For example, for the expression "50% off all products!", it suggests changing it to "50% off some products!". Additionally, it generates a notice recommending consultation with the legal department if necessary.

[0688] Specific operations:

[0689] The server creates a modification suggestion of "50% off some products!" for the expression "50% off all products!" and generates a notice recommending confirmation to the legal department.

[0690] Input: Risk assessment results

[0691] Output: Specific modification suggestions and recommendation notice

[0692] Step 6: Result presentation

[0693] Create a server, create a user, and create a user. If the user wants to know what they need, what they need.

[0694] Specific behavior:

[0695] User terminal 에서 대시보드를 열어"50% off all products!" 50% off on all products!" 50% off on all products!" Home. I want to know what I want.

[0696] Input: Sorry for the inconvenience.

[0697] user:user name

[0698] The above are the specific processing steps performed by the system of the present invention.

[0699] (Application example 2)

[0700] 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."

[0701] There is a need for a system that can proactively diagnose the risk of online content becoming a hot topic and propose appropriate revisions and actions. It is also necessary to provide a more accurate risk assessment that can handle a variety of data formats, including images and audio, and that takes user emotions into account.

[0702] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for generating a revision plan based on the risk assessment result by referring to a database of past cases, means for presenting the generated revision plan and the risk assessment result to the user, means for recognizing the user's emotions using an emotion engine to improve the accuracy of the risk assessment, and means for collecting and analyzing audio and video data via a smart device. This enables advance diagnosis of the risk of a controversy in content that a user intends to distribute and appropriate measures to be taken.

[0703] The "Internet" is a global information and communications network that interconnects computer networks around the world.

[0704] "User" means any person or entity that uses a system or device.

[0705] "Data" is a collection of information and includes various formats such as text, images, and audio.

[0706] A "server" is a computer system used to receive, process, and transmit data.

[0707] "Preprocessing" is the initial processing of data to convert it into an analyzable format.

[0708] "Natural language processing model" is a general term for algorithms and technologies for understanding and manipulating human language.

[0709] "Risk assessment" is the process of evaluating the potential dangers or problems of a particular activity or situation.

[0710] A "case database" is a database that systematically collects past cases and data.

[0711] "Fixation Suggestions" refers to specific improvements or proposals for addressing detected issues or risks.

[0712] "Emotion engine" is an artificial intelligence technology for analyzing and recognizing user emotions.

[0713] A "smart device" is an advanced device that has internet connectivity and is capable of collecting and analyzing data.

[0714] "Audio data" refers to audio information recorded in digital format.

[0715] "Video data" refers to visual information such as images and videos recorded in digital format.

[0716] "Analysis" is the process of analyzing data in detail and extracting its meaning and value.

[0717] This invention is a system that proactively diagnoses the risk of online content becoming controversial and suggests appropriate modifications and actions. The system receives data entered by users and performs risk assessment using an emotion engine and natural language processing model. Furthermore, the accuracy of the assessment is improved by collecting and analyzing audio and video data via smart devices.

[0718] The server first receives data entered by the user via the Internet. The data may be in a variety of formats, including text and image formats, but after receiving it, it undergoes preprocessing to make it analyzable. For example, text data is normalized by removing HTML tags and special characters. For image data, OCR technology (pytesseract library) is used to extract text from the image, and further preprocessing is performed.

[0719] Next, we use the preprocessed data to run the emotion engine, which performs sentiment analysis on the content users are about to post. For example, we can recognize emotions such as "excitement," "expectation," and "dissatisfaction." We then use the TextBlob library to evaluate the polarity of the emotions.

[0720] Once the sentiment analysis is complete, a natural language processing model (Keras and a text tokenizer) is used to assess the risk of the received text data. During this process, the data is compared with a database of past scandals to determine whether there is any risk. Specific revisions are then proposed based on the generated risk assessment results. For example, if "50% off all products!" is judged to be a risky expression, a suggestion is made to change it to "50% off some products!"

[0721] Finally, the server presents the risk assessment results and suggested remediation to the user in a visual dashboard, allowing the user to take concrete corrective action. Depending on the level of risk, a notification may also be generated recommending that the user consult with the legal department.

[0722] These processes are carried out via smart devices, making it easy to collect audio and video data. For example, content can be evaluated on the spot using a smartphone or tablet, and suggestions for revisions can be received immediately.

[0723] Specific examples

[0724] The user inputs the text "New campaign starts! All products 50% off!" as a new campaign proposal and uploads a campaign banner image. The server receives this and preprocesses the text and image. It then uses an emotion engine and natural language processing model to perform sentiment analysis and risk assessment, generating a revised proposal: "50% off some products!"

[0725] Example prompts to input to a generative AI model:

[0726] test_text = "New campaign starts! 50% off all items!"

[0727] test_image_path = "path_to_campaign_banner.jpg"

[0728] main(test_text, test_image_path)

[0729] By entering this prompt, users can check the risk assessment results and suggested corrections on the dashboard, enabling safer content delivery.

[0730] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0731] Step 1:

[0732] The user sends text data or image data from their terminal to the server via the Internet. The user's input data is content information such as campaign proposals. In this step, it is confirmed whether the user's input reaches the server.

[0733] Input: Text data and image data entered by the user

[0734] Output: Raw data sent to the server

[0735] Step 2:

[0736] The server preprocesses the data it receives. For text data, it removes HTML tags and special characters and normalizes it. For image data, it uses OCR technology (pytesseract library) to extract text. This step results in clean data that can be analyzed.

[0737] Input: Raw data sent to the server

[0738] Output: Preprocessed, clean text data

[0739] Step 3:

[0740] The server performs sentiment analysis on the preprocessed text data. It uses the TextBlob library to evaluate the sentiment polarity of the text and determine whether it is "positive" or "negative." This step quantifies the user's sentiment.

[0741] Input: Preprocessed clean text data

[0742] Output: Sentiment analysis result (positive or negative)

[0743] Step 4:

[0744] The server uses a natural language processing model to assess risk. Using Keras and a text tokenizer, the text data is passed to the model, which then compares it with a database of past scandals to calculate a risk score. This step quantifies the risk of a particular piece of content becoming a scandal.

[0745] Input: Preprocessed clean text data

[0746] Output: Risk score

[0747] Step 5:

[0748] The server generates specific correction suggestions based on the risk assessment results. If the risk score is high, it will suggest corrections to the problematic expression. For example, instead of "50% off all products!", it will suggest "50% off some products!"

[0749] Input: Risk Score

[0750] Output: Revision proposal

[0751] Step 6:

[0752] The server then presents the generated remediation proposals and risk assessment results to the user, who can then view this information and consider appropriate responses through a visual dashboard.

[0753] Input: Remediation proposals and risk assessment results

[0754] Output: Dashboard display

[0755] Step 7:

[0756] Depending on the risk assessment result, the server generates a notification recommending that the user consult with the legal department. If the risk score is particularly high, a notification is generated and the user is prompted to check with the legal department.

[0757] Input: Risk Score

[0758] Output: Notice recommending consultation with legal department

[0759] Step 8:

[0760] The user can provide additional input (audio and video data) via a smart device. This is also received by the server and similarly preprocessed and analyzed, enabling more accurate evaluation.

[0761] Input: Additional audio and video data

[0762] Output: Analysis results

[0763] 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.

[0764] 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.

[0765] 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.

[0766] [Third embodiment]

[0767] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0768] 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.

[0769] 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).

[0770] 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.

[0771] 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.

[0772] 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).

[0773] 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.

[0774] 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.

[0775] 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.

[0776] 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.

[0777] 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.

[0778] 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."

[0779] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and suggests appropriate corrections and actions. Below, we will generate a program for this system and explain its processing in natural language.

[0780] What the program does

[0781] Receiving and Preprocessing Input Data

[0782] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[0783] Examples:

[0784] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a banner image for the campaign. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into a format that can be analyzed.

[0785] Data analysis and risk assessment

[0786] The server receives the preprocessed data and begins analyzing it. It uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the data with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[0787] Examples:

[0788] When the user types "50% off all products!", the server determines that the expression "all products" poses a risk. A database of past cases records similar campaigns that have caused misunderstandings and sparked outrage, so the server issues a warning based on this.

[0789] Generating results and presenting suggestions

[0790] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[0791] Examples:

[0792] The server generates a correction suggestion for the risky phrase "50% off all products!" to "50% off some products!" and also creates a notification recommending confirmation by the legal department. The user can check this information on the dashboard and make the necessary corrections.

[0793] Specific program operation scenarios

[0794] Users input new campaign ideas and upload campaign banners. The server receives and pre-processes them. It then uses natural language processing and deep learning models to perform a risk assessment and compares it with a database of past cases. If a risk is detected, a notification is generated and presented to the user, suggesting specific remediation measures and consultation with the legal department. Users can review the information through a dashboard and take appropriate remediation or action.

[0795] The above is an embodiment of the present invention and its specific example. This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0796] The processing flow will be explained below.

[0797] Step 1:

[0798] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[0799] Step 2:

[0800] The server receives text data and image data sent from the user via the Internet.

[0801] Step 3:

[0802] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[0803] Step 4:

[0804] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[0805] Step 5:

[0806] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[0807] Step 6:

[0808] The server detects negative elements and the risk of a backlash based on data analyzed by an NLP model.

[0809] Step 7:

[0810] The server refers to a database of past flame war cases, performs pattern matching with the preprocessed data, and checks whether similar cases exist.

[0811] Step 8:

[0812] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[0813] Step 9:

[0814] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction proposal (for example, changing "50% off all products!" to "50% off some products!").

[0815] Step 10:

[0816] The server generates a notification recommending legal consultation if necessary.

[0817] Step 11:

[0818] The server visualizes the generated correction proposals and risk assessment results and sends the results to the terminal in a dashboard format that is easy for the user to understand.

[0819] Step 12:

[0820] The user checks the proposed modifications and risk assessment results presented on the dashboard and makes any necessary modifications.

[0821] Example 1

[0822] 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."

[0823] Content published online by companies and individuals often carries the risk of causing a firestorm. Therefore, there is a need for a system that can perform risk assessments before content is published and propose appropriate revisions and actions. Furthermore, because content is diverse, it is necessary to support multiple data formats, such as text and images. Furthermore, to improve the accuracy of risk assessments, it is important to compare the content with a database of past cases, and a method of presenting information that takes user ease of use into consideration is also required.

[0824] 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.

[0825] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and preprocessing it by removing unnecessary information, and means for assessing risk using a natural language processing model based on the preprocessed text data. This makes it possible to assess risk before content is made public and to propose appropriate modifications and actions.

[0826] The "Internet" is an infrastructure for sending and receiving information through networks around the world.

[0827] "User" means any person or entity that uses the System to input or upload Content.

[0828] "Data" means electronic files in any format that contain information, including text, images, audio, and video.

[0829] "Receiving" is the process by which the server takes in data that a user enters or uploads over the Internet.

[0830] "Text format" is a data format expressed as character information.

[0831] "Unnecessary information" refers to metadata, special characters, HTML tags, etc. that are not necessary for analysis or evaluation, and refers to removing this information.

[0832] "Preprocessing" refers to the preliminary data cleansing operations performed to convert received data into an analyzable format.

[0833] A "natural language processing model" is an artificial intelligence algorithm used to understand and analyze the meaning of text data.

[0834] "Risk assessment" is an analytical process used to determine the likelihood that words or phrases contained in data will spark outrage, misunderstanding, or negative reactions.

[0835] A "fix proposal" is a specific change proposal based on the results of a risk assessment to make the content published by users safer.

[0836] The "dashboard format" is an interface that allows users to visually check risk assessment results and proposed modifications.

[0837] A "legal department" is a department within a company or organization that handles legal matters.

[0838] "Image data" is a digital file containing visual information.

[0839] "OCR technology" is a technology that extracts character information from image data and converts it into text format.

[0840] "Inappropriate content" refers to information or expressions that are considered socially, legally, or morally problematic.

[0841]

[0842] The present invention is an AI-based system that proactively diagnoses the risk of online content being published by companies and individuals causing outrage and proposes appropriate corrections and actions. Detailed embodiments of this system are described below.

[0843] System Overview

[0844] The server receives data (text and images) entered or uploaded by users via the Internet. It then converts the received data into text format and performs preprocessing by removing unnecessary information. Specifically, this includes removing HTML tags and special characters and normalizing the text. For image data, OCR technology is used to extract character information and convert it into an analyzable text format. This completes the preprocessing of the text and image data.

[0845] The server then analyzes the preprocessed data and performs a risk assessment using a natural language processing (NLP) model. This involves extracting keywords and phrases from the text and determining whether they pose a negative or potential risk. Specifically, it compares them with a database of past scandals to see if there are any similar patterns. A deep learning model is also applied to image data to assess whether it contains inappropriate content.

[0846] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. These results and suggestions are presented to the user in the form of a visual dashboard.

[0847] Hardware and software used

[0848] Server: The main processing computer system, responsible for receiving, preprocessing, analyzing, and generating recommendations on data.

[0849] Internet: The communications infrastructure used to send and receive data.

[0850] Natural language processing model (NLP model): An AI model used to analyze text data.

[0851] OCR technology: A technology that extracts text information from image data.

[0852] Deep learning model: An AI model used to assess the content of image data.

[0853] Database: A database that stores past incidents of online outrage and is used to check against them when assessing risk.

[0854] Dashboard: An interface for presenting results and recommendations to the user.

[0855] Specific examples

[0856] For example, suppose a user uploads the text "New campaign begins! 50% off all products!" along with a banner image for the campaign. The server receives this data, first removes HTML tags and special characters from the text data, and then uses OCR technology to extract text information from the banner image. Next, it uses a natural language processing model to analyze the phrase "50% off all products!" and determines that the "all products" part poses a risk. It then compares this with a database of past cases and issues a warning based on this.

[0857] The server then generates a suggested change to change the phrase "50% off everything!" to "50% off some items!" and creates a notification recommending that the legal department review the change. This information is presented to the user via a dashboard, allowing them to make the necessary changes.

[0858] Prompt Sentence Examples

[0859] The prompt sentence to be input by the user may have the following format, for example:

[0860] Examples include "New campaign announcement! 50% off all products!" or "This year's new product sale, all items are half price!"

[0861] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0862] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0863] Step 1:

[0864] The server receives data entered or uploaded by users via the Internet. This data is often in text or image format. A specific example of input is the text "New campaign starts! All products 50% off!" and a campaign banner image. The output is the received data itself.

[0865] Step 2:

[0866] The server performs preprocessing on the received text data. Specifically, it removes HTML tags and special characters and normalizes the text. For example, 50% off all products! ", the HTML tags will be removed and the text "50% off all items!" will be output.

[0867] Step 3:

[0868] The server applies OCR technology to the received image data and extracts the text from the image. Specifically, it extracts the text information "50% off all products!" from the campaign banner image and converts it into an analyzable text format. The extracted text is output.

[0869] Step 4:

[0870] The server sends the preprocessed text data to a natural language processing (NLP) model to extract keywords and phrases. Specifically, it extracts the phrase "50% off all items!" and analyzes it. The extracted keywords and phrases are the output.

[0871] Step 5:

[0872] The server uses the extracted keywords and phrases to perform a risk assessment. Specifically, it compares them with a database of past online scandals to see if similar patterns exist. For example, it determines that the phrase "all products" poses a risk. The assessed risk is the output.

[0873] Step 6:

[0874] The server generates a specific correction proposal based on the risk assessment. For example, it generates a correction proposal to change the expression "50% off all products!" to "50% off some products!" The generated correction proposal is the output.

[0875] Step 7:

[0876] The server presents the risk assessment results and suggested modifications to the user in a dashboard format. Specifically, the results are displayed in a format that is easy for the user to understand visually. For example, the dashboard displays suggested modifications such as "50% off all products!" and "50% off some products!" The presented information is the output.

[0877] Step 8:

[0878] The server generates a notification recommending consulting the legal department based on the risk assessment results. Specifically, if high-risk content is detected, it creates a notification such as "We recommend checking with the legal department." The generated notification is the output.

[0879] The above is the processing flow and specific operation of the program for this system.

[0880] (Application example 1)

[0881] 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."

[0882] In recent years, the amount of information published on the Internet has rapidly increased, and the risk of content posted by companies and individuals becoming a hot topic has also increased. However, there is currently a lack of means to identify the risk of a hot topic in advance and respond appropriately. For this reason, there is a need for a system that can efficiently diagnose the risk of a hot topic and propose specific corrections and actions. This system must be realized especially when using mobile devices such as smartphones.

[0883] 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.

[0884] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for referencing a database of past cases and generating a revision plan based on the risk assessment result, means for presenting the generated revision plan and the risk assessment result to the user, and means for suggesting specific revision plans and actions when a risk is detected. This enables users to take measures to quickly and accurately avoid the risk of a flame war using a mobile device such as a smartphone.

[0885] "User" refers to any individual or company that uses the System.

[0886] "Data" refers to text, images, or other information entered or uploaded by Users.

[0887] "Means for receiving" refers to the function of taking data entered or uploaded by the user into the server via the Internet.

[0888] "Means for converting to text format and preprocessing by removing unnecessary information" refers to the process of converting received data into a format that can be analyzed and removing irrelevant information.

[0889] A "natural language processing model" refers to algorithms and techniques for analyzing text data, understanding its content, and extracting its intent.

[0890] "Means for assessing risk" refers to a function that assesses the risk of content causing a backlash based on preprocessed data.

[0891] The "past case database" refers to a database that accumulates past cases of online outrage.

[0892] "Means of generation" refers to the function of proposing appropriate corrections or actions based on the results of risk assessment.

[0893] "Means of presentation" refers to the function of notifying the user of generated correction proposals and risk assessment results.

[0894] "Means to suggest specific corrections and actions" refers to a function that shows the user specific countermeasures and corrections for detected risks.

[0895] "Dashboard function" refers to an interface that displays risk assessment results and correction suggestions in a format that makes it easy for users to visually confirm them.

[0896] "OCR technology" refers to the technology that optically recognizes text in an image and extracts it as electronic data.

[0897] The system of the present invention receives data (text and images) entered or uploaded by users via the Internet, analyzes the received data, and evaluates the risk of a social media outcry. If a risk is detected, the system proposes specific corrections and actions. The system has the following processing flow:

[0898] Hardware and Software

[0899] This system mainly uses the following hardware and software.

[0900] 1. Server: The central device that receives, preprocesses, analyzes data, and generates and presents results.

[0901] 2. Smartphone or computer: This is the device where users enter and upload data and check the results.

[0902] 3. Natural language processing model (spacy): A library for analyzing text data.

[0903] 4. OCR technology (pytesseract and PIL): Technology for extracting text from images and converting it into an analyzable format.

[0904] Data reception and preprocessing

[0905] The server receives text data entered by users or image data uploaded via the Internet. The received text data is preprocessed by removing HTML tags and special characters and normalizing it. For image data, OCR technology is used to extract the text within the image and normalize it.

[0906] Data analysis and risk assessment

[0907] The preprocessed text and image data are analyzed using a natural language processing model (SPACY). This analysis extracts keywords and phrases from the text and evaluates negative elements and potential risks. The results are then compared with a database of past online outrage cases to assess risk.

[0908] Suggested fixes and actions

[0909] The server generates specific correction suggestions based on the risk assessment results. For example, it determines that the expression "50% off all products!" is risky and suggests changing it to "50% off some products!" It also generates a notification recommending consulting the legal department if necessary. These results are presented to the user in the form of a visual dashboard.

[0910] Specific operation scenario

[0911] 1. User: Enter a new campaign idea and upload a campaign banner. Example: "New campaign starts! 50% off all products!"

[0912] 2. Server: Receives this and performs preprocessing.

[0913] 3. Server: Performs risk assessment using natural language processing and deep learning models and compares the results with a database of past cases.

[0914] 4. Server: If a risk is detected, a notification is generated and presented to the user with specific remediation suggestions and recommendations for consulting the legal department.

[0915] 5. User: Check the information through the dashboard and take appropriate corrections or actions.

[0916] Example prompt sentence:

[0917] content = "New campaign starts! 50% off all products!"

[0918] image_path = "path / to / banner_image.jpg"

[0919] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[0920] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0921] Step 1:

[0922] The server receives text data entered by the user via the Internet and image data uploaded by the user.

[0923] Input: Text data entered by the user and image data uploaded by the user.

[0924] Output: Received data.

[0925] Specific operation: The user uploads the text data "New campaign starts! All products 50% off!" and the campaign banner image. The server receives this data via the Internet.

[0926] Step 2:

[0927] The server pre-processes the received text data, i.e., removes HTML tags and special characters, and normalizes it.

[0928] Input: Received data (text data).

[0929] Output: Preprocessed text data.

[0930] Specific operation: Removes HTML tags and special characters from the text data "New campaign starts! All products 50% off!" and performs normalization such as lowercasing.

[0931] Step 3:

[0932] The server uses OCR technology on the received image data to extract and normalize the text within the image.

[0933] Input: Received data (image data).

[0934] Output: Preprocessed text data.

[0935] Specific operation: Extract text from campaign banner images using pytesseract and PIL, and normalize the text by lowercasing it, etc.

[0936] Step 4:

[0937] The server analyzes the preprocessed text data using a natural language processing model (spacy), which extracts keywords and phrases and evaluates negative elements and risks.

[0938] Input: Preprocessed text data.

[0939] Output: Analysis results (keywords, phrases, risk assessment).

[0940] Specific operation: The spacy model analyzes text data containing the phrase "50% off all products!", extracts keywords and phrases, and compares them with a past database to assess risk.

[0941] Step 5:

[0942] Based on the risk assessment results, the server refers to a database of past controversy cases and generates appropriate correction proposals.

[0943] Input: Analysis results, database of past flame war cases.

[0944] Output: Revision proposal.

[0945] Specific behavior: Generate a suggestion to correct the phrase "50% off all items!" to "50% off some items!"

[0946] Step 6:

[0947] The server presents the generated remediation suggestions and risk assessment results to the user, and if necessary, generates a notification recommending consultation with the legal department.

[0948] Input: Correction proposal, risk assessment results.

[0949] Output: Presentation information, notifications.

[0950] Specific behavior: The user is presented with a visual dashboard showing suggested fixes such as "50% off selected items!" and the results of the risk assessment, and a notification is generated recommending that the user contact the legal department.

[0951] Step 7:

[0952] Users can view the information through the dashboard and take appropriate corrective action.

[0953] Input: Presented information, notifications.

[0954] Output: Modified content, behavior.

[0955] Specific actions: The user checks the proposed revisions and risk assessment results on the dashboard, changes "50% off all items!" to "50% off some items!", and consults with the legal department if necessary.

[0956] 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.

[0957] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and proposes appropriate corrections and actions. Furthermore, by combining it with an emotion engine that recognizes user emotions, the accuracy of risk assessment is improved.

[0958] What the program does

[0959] Receiving and Preprocessing Input Data

[0960] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[0961] Examples:

[0962] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a campaign banner image. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into an analyzable format.

[0963] Data analysis and risk assessment

[0964] The server receives the preprocessed data and first analyzes user sentiment using an emotion engine. It then uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the results with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[0965] Examples:

[0966] When the user inputs "50% off all products!", the server uses an emotion engine to recognize the emotion of "excitement, anticipation." The NLP model then determines that the expression "all products" poses a risk, and issues a warning because a database of past cases records similar campaigns that have caused misunderstandings and sparked outrage.

[0967] Generating results and presenting suggestions

[0968] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[0969] Examples:

[0970] The server generates a suggested correction for the risky phrase "50% off all items!" to "50% off some items!" and also creates a notification recommending confirmation by the legal department. Taking into account the results of sentiment analysis, the server reevaluates whether this suggested correction is appropriate for the user. The user can check this information on the dashboard and make any necessary corrections.

[0971] Specific program operation scenarios

[0972] Users input new campaign ideas and upload campaign banners. The server receives and preprocesses them. It then uses an emotion engine, natural language processing models, and deep learning models to perform a risk assessment. It compares the results with a database of past cases, and if a risk is detected, it generates a notification to the user suggesting specific corrections and consultation with the legal department. Users can view the information through a dashboard and take appropriate corrections or actions.

[0973] The above is an embodiment of the present invention and its specific example. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and maintain credibility.

[0974] The processing flow will be explained below.

[0975] What the program does

[0976] Processing steps of a system that combines emotion engines

[0977] Step 1:

[0978] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[0979] Step 2:

[0980] The server receives text data and image data sent from the user via the Internet.

[0981] Step 3:

[0982] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[0983] Step 4:

[0984] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[0985] Step 5:

[0986] The server applies an emotion engine to the preprocessed text data to recognize the user's emotions, which are then recorded in a database for subsequent processing.

[0987] Step 6:

[0988] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[0989] Step 7:

[0990] The server detects negative elements and potential risks based on the data analyzed by the NLP model, taking into account the results of the sentiment analysis recognized in step 5.

[0991] Step 8:

[0992] The server compares the data with a database of past flame war cases, performs pattern matching with the given data, and checks whether similar cases exist.

[0993] Step 9:

[0994] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[0995] Step 10:

[0996] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction suggestion. For example, it suggests correcting the expression "50% off all products!" to "50% off some products!"

[0997] Step 11:

[0998] The server generates notifications recommending legal consultation as needed, with specific justification based on past experience and current risk assessment.

[0999] Step 12:

[1000] The server visualizes the generated correction proposals and risk assessment results and sends the results to the user's terminal in a dashboard format that is easy for the user to understand.

[1001] Step 13:

[1002] Users can check the proposed modifications and risk assessment results presented on the dashboard and make any necessary modifications. Once the modifications are complete, they can also perform a reassessment.

[1003] Example 2

[1004] 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."

[1005] There is a lack of methods to proactively diagnose the risk of online content becoming controversial and propose appropriate revisions and actions. In particular, there is a need for a system that can assess the risk of not only text data but also image data, and that also takes user emotions into account.

[1006] 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 means for receiving data entered by a user via the Internet, means for performing preprocessing of the received data by removing HTML tags and special characters and normalizing the text, means for extracting text from image data using OCR technology, means for performing sentiment analysis using an emotion engine based on the preprocessed data, means for assessing risk using a natural language processing model, means for confirming the risk assessment result by comparing it with a database of similar past cases, means for generating a revision proposal based on the risk assessment result, and means for presenting the generated revision proposal and the risk assessment result to the user. This makes it possible to prevent the risk of a controversy in advance and increase the reliability of content.

[1007] "User" means any person or entity that uses the System to input or upload data.

[1008] "Server" refers to the hardware or software that receives, processes, and analyzes data entered or uploaded by users.

[1009] "Data" refers to information entered or uploaded by a user into the system, in the form of text, images, etc.

[1010] "Internet" refers to the global information network that enables the communication of data between users and servers.

[1011] "Preprocessing" refers to the processing performed on received data, specifically tasks such as removing HTML tags and special characters, and normalizing text.

[1012] "OCR technology" refers to optical character recognition technology for extracting character information from image data and converting it into text format.

[1013] An "emotion engine" refers to software or algorithms that analyze and determine user emotions from text data.

[1014] A "natural language processing model" refers to an algorithm or software that analyzes text data, extracts keywords and phrases, and assesses risk.

[1015] "Deep learning model" refers to a machine learning algorithm that analyzes image data and automatically evaluates inappropriate content or problematic elements.

[1016] A "database" refers to an information resource that stores past cases and information and is referenced when conducting risk assessments.

[1017] "Proposed fixes" refer to proposals that are generated based on the results of risk assessment and indicate specific fixes or countermeasures that the user should take.

[1018] A "dashboard" refers to an interface that visually displays risk assessment results and suggested corrections to users.

[1019] This invention is an AI-based system that proactively diagnoses the risk of online content published by companies and individuals becoming controversial and proposes appropriate corrections and actions. The system receives data (text and images) entered or uploaded by users, and performs a comprehensive process from data preprocessing, sentiment analysis, risk assessment, generation of correction suggestions, and presentation of the results to the user.

[1020] Receiving and Preprocessing Input Data

[1021] The server receives data entered or uploaded by users via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. This converts the data into analyzable data. For image data, OCR (Optical Character Recognition) technology is used to extract text from the image and convert it into an analyzable text format.

[1022] sentiment analysis

[1023] The server performs sentiment analysis on the preprocessed text data using an emotion engine, which is software that uses natural language processing (NLP) technology to determine the emotions contained in the text, such as "excitement" or "anticipation."

[1024] Risk Assessment

[1025] The server not only evaluates content based on sentiment analysis, but also uses natural language processing models to extract keywords and phrases from the text to detect negative elements and potential risks. It also uses deep learning models to analyze preprocessed image data to assess whether it contains inappropriate content or problematic elements. These processes also involve cross-checking the data with a database of past scandals.

[1026] Generating and Presenting Results

[1027] The server generates specific correction suggestions based on the risk assessment results. For example, it suggests changing the phrase "50% off all products!" to "50% off some products!" It also generates a notification recommending consultation with the legal department if necessary. These results and correction suggestions are presented to the user in the form of a visual dashboard.

[1028] An example of a specific prompt sentence is the action of entering the text "New campaign starts! All products 50% off!" and uploading a campaign banner image. In this case, the server receives the data corresponding to the prompt, preprocesses it, analyzes sentiment, and evaluates risks, and presents optimal correction suggestions to the user. The user can check this information on the dashboard and make any necessary corrections.

[1029] This completes the description of the embodiment of the present invention. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and increase the reliability of their content.

[1030] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1031] Step 1: Receive the database

[1032] The server supports users. (PC, 스마트폰) HTTP

[1033] Specific behavior:

[1034] User 가 "New campaign starts! 50% off all products!" Campaign Yes, the server is private.

[1035] Input: Data entered or uploaded by Yuzaga (text and images)

[1036] Output: Received data requiring preprocessing

[1037] ステップ2:Preprocessing of データ

[1038] Server performs preprocessing on the received data. For text data, it removes HTML tags and special characters and normalizes the text. For image data, it uses OCR technology to extract the text within the image and converts it into an analyzable text format.

[1039] Artistic work:

[1040] HTML tags and special characters are removed from the text "New campaign opening! All products 50% off!" entered by Yuza, and the text "All products 50% off!" is extracted from the deployed campaign banner image using OCR technology.

[1041] Input: Received data (text and images)

[1042] Output: Preprocessed data, normalized and converted to text.

[1043] Step 3: Emotion Analysis

[1044] サーバ inputs preprocessed text data into the emotion engine to analyze Yuza's emotions. This emotion engine is software that uses natural language processing (NLP) technology to determine the emotions of the text.

[1045] Artistic work:

[1046] The emotion engine analyzes the text "New campaign opening! 50% off all products!" and detects emotions such as "excitement" and "anticipation."

[1047] Input: Normalized text data

[1048] Output: Detected emotion result

[1049] Step 4: Risk assessment

[1050] The server combines the NLP model and the deep learning model to analyze the preprocessed data. In the text data, keywords and phrases are extracted to detect negative elements and potential risks. Similarly, for image data, the deep learning model is applied to evaluate whether there is inappropriate content or problem elements. In this evaluation, the risk assessment result is reviewed by querying the database of similar cases that occurred previously.

[1051] Specific operation:

[1052] The NLP model determines that the keyword "all products" has a risk. There is a record in the previous case database where a campaign was flagged with a similar expression, so a warning is issued.

[1053] Input: Emotion analysis result, preprocessed text and image data

[1054] Output: Risk assessment result and detected risks

[1055] Step 5: Result generation

[1056] When risks are detected, the server generates specific modification suggestions. For example, for the expression "All products 50% off!", it is proposed to change it to "Some products 50% off!". Also, a notice recommending consultation with the legal department is generated as needed.

[1057] Specific operation:

[1058] The server creates a notice recommending verification by the legal department, and writes a revision to "50% off for some products!" for the expression "50% off for all products!".

[1059] Input: Risk assessment result

[1060] Output: Specific revision and notice of recommendation

[1061] Step 6: Present the result

[1062] The server presents the generated revision or notice to the user in a visualized dashboard format. The user can check the information through the dashboard and make necessary revisions.

[1063] Specific operations:

[1064] The user opens the dashboard on the terminal to check the risk assessment result of "50% off for all products!" and the revision of "50% off for some products!". A notice of verification to the legal department is also displayed.

[1065] Input: Specific revision and notice of recommendation

[1066] Output: Visualized assessment result and revision to the user

[1067] The above are the specific processing steps performed by the system of the present invention.

[1068] (Application Example 2)

[1069] Next, Application Example 2 will be described. In the following description, the data processing device 12 is referred to as the "server", and the headset type terminal 314 is referred to as the "terminal".

[1070] There is a need for a system that can proactively diagnose the risk of online content becoming a hot topic and propose appropriate revisions and actions. It is also necessary to provide a more accurate risk assessment that can handle a variety of data formats, including images and audio, and that takes user emotions into account.

[1071] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for generating a revision plan based on the risk assessment result by referring to a database of past cases, means for presenting the generated revision plan and the risk assessment result to the user, means for recognizing the user's emotions using an emotion engine to improve the accuracy of the risk assessment, and means for collecting and analyzing audio and video data via a smart device. This enables advance diagnosis of the risk of a controversy in content that a user intends to distribute and appropriate measures to be taken.

[1072] The "Internet" is a global information and communications network that interconnects computer networks around the world.

[1073] "User" means any person or entity that uses a system or device.

[1074] "Data" is a collection of information and includes various formats such as text, images, and audio.

[1075] A "server" is a computer system used to receive, process, and transmit data.

[1076] "Preprocessing" is the initial processing of data to convert it into an analyzable format.

[1077] "Natural language processing model" is a general term for algorithms and technologies for understanding and manipulating human language.

[1078] "Risk assessment" is the process of evaluating the potential dangers or problems of a particular activity or situation.

[1079] A "case database" is a database that systematically collects past cases and data.

[1080] "Fixation Suggestions" refers to specific improvements or proposals for addressing detected issues or risks.

[1081] "Emotion engine" is an artificial intelligence technology for analyzing and recognizing user emotions.

[1082] A "smart device" is an advanced device that has internet connectivity and is capable of collecting and analyzing data.

[1083] "Audio data" refers to audio information recorded in digital format.

[1084] "Video data" refers to visual information such as images and videos recorded in digital format.

[1085] "Analysis" is the process of analyzing data in detail and extracting its meaning and value.

[1086] This invention is a system that proactively diagnoses the risk of online content becoming controversial and suggests appropriate modifications and actions. The system receives data entered by users and performs risk assessment using an emotion engine and natural language processing model. Furthermore, the accuracy of the assessment is improved by collecting and analyzing audio and video data via smart devices.

[1087] The server first receives data entered by the user via the Internet. The data may be in a variety of formats, including text and image formats, but after receiving it, it undergoes preprocessing to make it analyzable. For example, text data is normalized by removing HTML tags and special characters. For image data, OCR technology (pytesseract library) is used to extract text from the image, and further preprocessing is performed.

[1088] Next, we use the preprocessed data to run the emotion engine, which performs sentiment analysis on the content users are about to post. For example, we can recognize emotions such as "excitement," "expectation," and "dissatisfaction." We then use the TextBlob library to evaluate the polarity of the emotions.

[1089] Once the sentiment analysis is complete, a natural language processing model (Keras and a text tokenizer) is used to assess the risk of the received text data. During this process, the data is compared with a database of past scandals to determine whether there is any risk. Specific revisions are then proposed based on the generated risk assessment results. For example, if "50% off all products!" is judged to be a risky expression, a suggestion is made to change it to "50% off some products!"

[1090] Finally, the server presents the risk assessment results and suggested remediation to the user in a visual dashboard, allowing the user to take concrete corrective action. Depending on the level of risk, a notification may also be generated recommending that the user consult with the legal department.

[1091] These processes are carried out via smart devices, making it easy to collect audio and video data. For example, content can be evaluated on the spot using a smartphone or tablet, and suggestions for revisions can be received immediately.

[1092] Specific examples

[1093] The user inputs the text "New campaign starts! All products 50% off!" as a new campaign proposal and uploads a campaign banner image. The server receives this and preprocesses the text and image. It then uses an emotion engine and natural language processing model to perform sentiment analysis and risk assessment, generating a revised proposal: "50% off some products!"

[1094] Example prompts to input to a generative AI model:

[1095] test_text = "New campaign starts! 50% off all items!"

[1096] test_image_path = "path_to_campaign_banner.jpg"

[1097] main(test_text, test_image_path)

[1098] By entering this prompt, users can check the risk assessment results and suggested corrections on the dashboard, enabling safer content delivery.

[1099] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1100] Step 1:

[1101] The user sends text data or image data from their terminal to the server via the Internet. The user's input data is content information such as campaign proposals. In this step, it is confirmed whether the user's input reaches the server.

[1102] Input: Text data and image data entered by the user

[1103] Output: Raw data sent to the server

[1104] Step 2:

[1105] The server preprocesses the data it receives. For text data, it removes HTML tags and special characters and normalizes it. For image data, it uses OCR technology (pytesseract library) to extract text. This step results in clean data that can be analyzed.

[1106] Input: Raw data sent to the server

[1107] Output: Preprocessed, clean text data

[1108] Step 3:

[1109] The server performs sentiment analysis on the preprocessed text data. It uses the TextBlob library to evaluate the sentiment polarity of the text and determine whether it is "positive" or "negative." This step quantifies the user's sentiment.

[1110] Input: Preprocessed clean text data

[1111] Output: Sentiment analysis result (positive or negative)

[1112] Step 4:

[1113] The server uses a natural language processing model to assess risk. Using Keras and a text tokenizer, the text data is passed to the model, which then compares it with a database of past scandals to calculate a risk score. This step quantifies the risk of a particular piece of content becoming a scandal.

[1114] Input: Preprocessed clean text data

[1115] Output: Risk score

[1116] Step 5:

[1117] The server generates specific correction suggestions based on the risk assessment results. If the risk score is high, it will suggest corrections to the problematic expression. For example, instead of "50% off all products!", it will suggest "50% off some products!"

[1118] Input: Risk Score

[1119] Output: Revision proposal

[1120] Step 6:

[1121] The server then presents the generated remediation proposals and risk assessment results to the user, who can then view this information and consider appropriate responses through a visual dashboard.

[1122] Input: Remediation proposals and risk assessment results

[1123] Output: Dashboard display

[1124] Step 7:

[1125] Depending on the risk assessment result, the server generates a notification recommending that the user consult with the legal department. If the risk score is particularly high, a notification is generated and the user is prompted to check with the legal department.

[1126] Input: Risk Score

[1127] Output: Notice recommending consultation with legal department

[1128] Step 8:

[1129] The user can provide additional input (audio and video data) via a smart device. This is also received by the server and similarly preprocessed and analyzed, enabling more accurate evaluation.

[1130] Input: Additional audio and video data

[1131] Output: Analysis results

[1132] 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.

[1133] 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.

[1134] 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.

[1135] [Fourth embodiment]

[1136] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1137] 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.

[1138] 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).

[1139] 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.

[1140] 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.

[1141] 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).

[1142] 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.

[1143] 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.

[1144] 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.

[1145] 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.

[1146] 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.

[1147] 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.

[1148] 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."

[1149] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and suggests appropriate corrections and actions. Below, we will generate a program for this system and explain its processing in natural language.

[1150] What the program does

[1151] Receiving and Preprocessing Input Data

[1152] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[1153] Examples:

[1154] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a banner image for the campaign. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into a format that can be analyzed.

[1155] Data analysis and risk assessment

[1156] The server receives the preprocessed data and begins analyzing it. It uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the data with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[1157] Examples:

[1158] When the user types "50% off all products!", the server determines that the expression "all products" poses a risk. A database of past cases records similar campaigns that have caused misunderstandings and sparked outrage, so the server issues a warning based on this.

[1159] Generating results and presenting suggestions

[1160] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[1161] Examples:

[1162] The server generates a correction suggestion for the risky phrase "50% off all products!" to "50% off some products!" and also creates a notification recommending confirmation by the legal department. The user can check this information on the dashboard and make the necessary corrections.

[1163] Specific program operation scenarios

[1164] Users input new campaign ideas and upload campaign banners. The server receives and pre-processes them. It then uses natural language processing and deep learning models to perform a risk assessment and compares it with a database of past cases. If a risk is detected, a notification is generated and presented to the user, suggesting specific remediation measures and consultation with the legal department. Users can review the information through a dashboard and take appropriate remediation or action.

[1165] The above is an embodiment of the present invention and its specific example. This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[1166] The processing flow will be explained below.

[1167] Step 1:

[1168] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[1169] Step 2:

[1170] The server receives text data and image data sent from the user via the Internet.

[1171] Step 3:

[1172] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[1173] Step 4:

[1174] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[1175] Step 5:

[1176] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[1177] Step 6:

[1178] The server detects negative elements and the risk of a backlash based on data analyzed by an NLP model.

[1179] Step 7:

[1180] The server refers to a database of past flame war cases, performs pattern matching with the preprocessed data, and checks whether similar cases exist.

[1181] Step 8:

[1182] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[1183] Step 9:

[1184] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction proposal (for example, changing "50% off all products!" to "50% off some products!").

[1185] Step 10:

[1186] The server generates a notification recommending legal consultation if necessary.

[1187] Step 11:

[1188] The server visualizes the generated correction proposals and risk assessment results and sends the results to the terminal in a dashboard format that is easy for the user to understand.

[1189] Step 12:

[1190] The user checks the proposed modifications and risk assessment results presented on the dashboard and makes any necessary modifications.

[1191] Example 1

[1192] 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."

[1193] Content published online by companies and individuals often carries the risk of causing a firestorm. Therefore, there is a need for a system that can perform risk assessments before content is published and propose appropriate revisions and actions. Furthermore, because content is diverse, it is necessary to support multiple data formats, such as text and images. Furthermore, to improve the accuracy of risk assessments, it is important to compare the content with a database of past cases, and a method of presenting information that takes user ease of use into consideration is also required.

[1194] 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.

[1195] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and preprocessing it by removing unnecessary information, and means for assessing risk using a natural language processing model based on the preprocessed text data. This makes it possible to assess risk before content is made public and to propose appropriate modifications and actions.

[1196] The "Internet" is an infrastructure for sending and receiving information through networks around the world.

[1197] "User" means any person or entity that uses the System to input or upload Content.

[1198] "Data" means electronic files in any format that contain information, including text, images, audio, and video.

[1199] "Receiving" is the process by which the server takes in data that a user enters or uploads over the Internet.

[1200] "Text format" is a data format expressed as character information.

[1201] "Unnecessary information" refers to metadata, special characters, HTML tags, etc. that are not necessary for analysis or evaluation, and refers to removing this information.

[1202] "Preprocessing" refers to the preliminary data cleansing operations performed to convert received data into an analyzable format.

[1203] A "natural language processing model" is an artificial intelligence algorithm used to understand and analyze the meaning of text data.

[1204] "Risk assessment" is an analytical process used to determine the likelihood that words or phrases contained in data will spark outrage, misunderstanding, or negative reactions.

[1205] A "fix proposal" is a specific change proposal based on the results of a risk assessment to make the content published by users safer.

[1206] The "dashboard format" is an interface that allows users to visually check risk assessment results and proposed modifications.

[1207] A "legal department" is a department within a company or organization that handles legal matters.

[1208] "Image data" is a digital file containing visual information.

[1209] "OCR technology" is a technology that extracts character information from image data and converts it into text format.

[1210] "Inappropriate content" refers to information or expressions that are considered socially, legally, or morally problematic.

[1211]

[1212] The present invention is an AI-based system that proactively diagnoses the risk of online content being published by companies and individuals causing outrage and proposes appropriate corrections and actions. Detailed embodiments of this system are described below.

[1213] System Overview

[1214] The server receives data (text and images) entered or uploaded by users via the Internet. It then converts the received data into text format and performs preprocessing by removing unnecessary information. Specifically, this includes removing HTML tags and special characters and normalizing the text. For image data, OCR technology is used to extract character information and convert it into an analyzable text format. This completes the preprocessing of the text and image data.

[1215] The server then analyzes the preprocessed data and performs a risk assessment using a natural language processing (NLP) model. This involves extracting keywords and phrases from the text and determining whether they pose a negative or potential risk. Specifically, it compares them with a database of past scandals to see if there are any similar patterns. A deep learning model is also applied to image data to assess whether it contains inappropriate content.

[1216] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. These results and suggestions are presented to the user in the form of a visual dashboard.

[1217] Hardware and software used

[1218] Server: The main processing computer system, responsible for receiving, preprocessing, analyzing, and generating recommendations on data.

[1219] Internet: The communications infrastructure used to send and receive data.

[1220] Natural language processing model (NLP model): An AI model used to analyze text data.

[1221] OCR technology: A technology that extracts text information from image data.

[1222] Deep learning model: An AI model used to assess the content of image data.

[1223] Database: A database that stores past incidents of online outrage and is used to check against them when assessing risk.

[1224] Dashboard: An interface for presenting results and recommendations to the user.

[1225] Specific examples

[1226] For example, suppose a user uploads the text "New campaign begins! 50% off all products!" along with a banner image for the campaign. The server receives this data, first removes HTML tags and special characters from the text data, and then uses OCR technology to extract text information from the banner image. Next, it uses a natural language processing model to analyze the phrase "50% off all products!" and determines that the "all products" part poses a risk. It then compares this with a database of past cases and issues a warning based on this.

[1227] The server then generates a suggested change to change the phrase "50% off everything!" to "50% off some items!" and creates a notification recommending that the legal department review the change. This information is presented to the user via a dashboard, allowing them to make the necessary changes.

[1228] Prompt Sentence Examples

[1229] The prompt sentence to be input by the user may have the following format, for example:

[1230] Examples include "New campaign announcement! 50% off all products!" or "This year's new product sale, all items are half price!"

[1231] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[1232] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1233] Step 1:

[1234] The server receives data entered or uploaded by users via the Internet. This data is often in text or image format. A specific example of input is the text "New campaign starts! All products 50% off!" and a campaign banner image. The output is the received data itself.

[1235] Step 2:

[1236] The server performs preprocessing on the received text data. Specifically, it removes HTML tags and special characters and normalizes the text. For example, 50% off all products! ", the HTML tags will be removed and the text "50% off all items!" will be output.

[1237] Step 3:

[1238] The server applies OCR technology to the received image data and extracts the text from the image. Specifically, it extracts the text information "50% off all products!" from the campaign banner image and converts it into an analyzable text format. The extracted text is output.

[1239] Step 4:

[1240] The server sends the preprocessed text data to a natural language processing (NLP) model to extract keywords and phrases. Specifically, it extracts the phrase "50% off all items!" and analyzes it. The extracted keywords and phrases are the output.

[1241] Step 5:

[1242] The server uses the extracted keywords and phrases to perform a risk assessment. Specifically, it compares them with a database of past online scandals to see if similar patterns exist. For example, it determines that the phrase "all products" poses a risk. The assessed risk is the output.

[1243] Step 6:

[1244] The server generates a specific correction proposal based on the risk assessment. For example, it generates a correction proposal to change the expression "50% off all products!" to "50% off some products!" The generated correction proposal is the output.

[1245] Step 7:

[1246] The server presents the risk assessment results and suggested modifications to the user in a dashboard format. Specifically, the results are displayed in a format that is easy for the user to understand visually. For example, the dashboard displays suggested modifications such as "50% off all products!" and "50% off some products!" The presented information is the output.

[1247] Step 8:

[1248] The server generates a notification recommending consulting the legal department based on the risk assessment results. Specifically, if high-risk content is detected, it creates a notification such as "We recommend checking with the legal department." The generated notification is the output.

[1249] The above is the processing flow and specific operation of the program for this system.

[1250] (Application example 1)

[1251] 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."

[1252] In recent years, the amount of information published on the Internet has rapidly increased, and the risk of content posted by companies and individuals becoming a hot topic has also increased. However, there is currently a lack of means to identify the risk of a hot topic in advance and respond appropriately. For this reason, there is a need for a system that can efficiently diagnose the risk of a hot topic and propose specific corrections and actions. This system must be realized especially when using mobile devices such as smartphones.

[1253] 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.

[1254] In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for referencing a database of past cases and generating a revision plan based on the risk assessment result, means for presenting the generated revision plan and the risk assessment result to the user, and means for suggesting specific revision plans and actions when a risk is detected. This enables users to take measures to quickly and accurately avoid the risk of a flame war using a mobile device such as a smartphone.

[1255] "User" refers to any individual or company that uses the System.

[1256] "Data" refers to text, images, or other information entered or uploaded by Users.

[1257] "Means for receiving" refers to the function of taking data entered or uploaded by the user into the server via the Internet.

[1258] "Means for converting to text format and preprocessing by removing unnecessary information" refers to the process of converting received data into a format that can be analyzed and removing irrelevant information.

[1259] A "natural language processing model" refers to algorithms and techniques for analyzing text data, understanding its content, and extracting its intent.

[1260] "Means for assessing risk" refers to a function that assesses the risk of content causing a backlash based on preprocessed data.

[1261] The "past case database" refers to a database that accumulates past cases of online outrage.

[1262] "Means of generation" refers to the function of proposing appropriate corrections or actions based on the results of risk assessment.

[1263] "Means of presentation" refers to the function of notifying the user of generated correction proposals and risk assessment results.

[1264] "Means to suggest specific corrections and actions" refers to a function that shows the user specific countermeasures and corrections for detected risks.

[1265] "Dashboard function" refers to an interface that displays risk assessment results and correction suggestions in a format that makes it easy for users to visually confirm them.

[1266] "OCR technology" refers to the technology that optically recognizes text in an image and extracts it as electronic data.

[1267] The system of the present invention receives data (text and images) entered or uploaded by users via the Internet, analyzes the received data, and evaluates the risk of a social media outcry. If a risk is detected, the system proposes specific corrections and actions. The system has the following processing flow:

[1268] Hardware and Software

[1269] This system mainly uses the following hardware and software.

[1270] 1. Server: The central device that receives, preprocesses, analyzes data, and generates and presents results.

[1271] 2. Smartphone or computer: This is the device where users enter and upload data and check the results.

[1272] 3. Natural language processing model (spacy): A library for analyzing text data.

[1273] 4. OCR technology (pytesseract and PIL): Technology for extracting text from images and converting it into an analyzable format.

[1274] Data reception and preprocessing

[1275] The server receives text data entered by users or image data uploaded via the Internet. The received text data is preprocessed by removing HTML tags and special characters and normalizing it. For image data, OCR technology is used to extract the text within the image and normalize it.

[1276] Data analysis and risk assessment

[1277] The preprocessed text and image data are analyzed using a natural language processing model (SPACY). This analysis extracts keywords and phrases from the text and evaluates negative elements and potential risks. The results are then compared with a database of past online outrage cases to assess risk.

[1278] Suggested fixes and actions

[1279] The server generates specific correction suggestions based on the risk assessment results. For example, it determines that the expression "50% off all products!" is risky and suggests changing it to "50% off some products!" It also generates a notification recommending consulting the legal department if necessary. These results are presented to the user in the form of a visual dashboard.

[1280] Specific operation scenario

[1281] 1. User: Enter a new campaign idea and upload a campaign banner. Example: "New campaign starts! 50% off all products!"

[1282] 2. Server: Receives this and performs preprocessing.

[1283] 3. Server: Performs risk assessment using natural language processing and deep learning models and compares the results with a database of past cases.

[1284] 4. Server: If a risk is detected, a notification is generated and presented to the user with specific remediation suggestions and recommendations for consulting the legal department.

[1285] 5. User: Check the information through the dashboard and take appropriate corrections or actions.

[1286] Example prompt sentence:

[1287] content = "New campaign starts! 50% off all products!"

[1288] image_path = "path / to / banner_image.jpg"

[1289] This system allows companies and individuals to avoid the risk of online outrage in advance and maintain their credibility.

[1290] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1291] Step 1:

[1292] The server receives text data entered by the user via the Internet and image data uploaded by the user.

[1293] Input: Text data entered by the user and image data uploaded by the user.

[1294] Output: Received data.

[1295] Specific operation: The user uploads the text data "New campaign starts! All products 50% off!" and the campaign banner image. The server receives this data via the Internet.

[1296] Step 2:

[1297] The server pre-processes the received text data, i.e., removes HTML tags and special characters, and normalizes it.

[1298] Input: Received data (text data).

[1299] Output: Preprocessed text data.

[1300] Specific operation: Removes HTML tags and special characters from the text data "New campaign starts! All products 50% off!" and performs normalization such as lowercasing.

[1301] Step 3:

[1302] The server uses OCR technology on the received image data to extract and normalize the text within the image.

[1303] Input: Received data (image data).

[1304] Output: Preprocessed text data.

[1305] Specific operation: Extract text from campaign banner images using pytesseract and PIL, and normalize the text by lowercasing it, etc.

[1306] Step 4:

[1307] The server analyzes the preprocessed text data using a natural language processing model (spacy), which extracts keywords and phrases and evaluates negative elements and risks.

[1308] Input: Preprocessed text data.

[1309] Output: Analysis results (keywords, phrases, risk assessment).

[1310] Specific operation: The spacy model analyzes text data containing the phrase "50% off all products!", extracts keywords and phrases, and compares them with a past database to assess risk.

[1311] Step 5:

[1312] Based on the risk assessment results, the server refers to a database of past controversy cases and generates appropriate correction proposals.

[1313] Input: Analysis results, database of past flame war cases.

[1314] Output: Revision proposal.

[1315] Specific behavior: Generate a suggestion to correct the phrase "50% off all items!" to "50% off some items!"

[1316] Step 6:

[1317] The server presents the generated remediation suggestions and risk assessment results to the user, and if necessary, generates a notification recommending consultation with the legal department.

[1318] Input: Correction proposal, risk assessment results.

[1319] Output: Presentation information, notifications.

[1320] Specific behavior: The user is presented with a visual dashboard showing suggested fixes such as "50% off selected items!" and the results of the risk assessment, and a notification is generated recommending that the user contact the legal department.

[1321] Step 7:

[1322] Users can view the information through the dashboard and take appropriate corrective action.

[1323] Input: Presented information, notifications.

[1324] Output: Modified content, behavior.

[1325] Specific actions: The user checks the proposed revisions and risk assessment results on the dashboard, changes "50% off all items!" to "50% off some items!", and consults with the legal department if necessary.

[1326] 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.

[1327] This invention is an AI-based system that proactively diagnoses the risk of online outrage for content published by companies and individuals, and proposes appropriate corrections and actions. Furthermore, by combining it with an emotion engine that recognizes user emotions, the accuracy of risk assessment is improved.

[1328] What the program does

[1329] Receiving and Preprocessing Input Data

[1330] The server receives data (text and images) entered or uploaded by the user via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. For image data, OCR (optical character recognition) technology is used to extract the text within the image and convert it into an analyzable text format.

[1331] Examples:

[1332] Suppose a user enters the text "New campaign starts! 50% off all products!" and uploads a campaign banner image. The server receives this data, extracts the text of interest, removes unnecessary information, and converts it into an analyzable format.

[1333] Data analysis and risk assessment

[1334] The server receives the preprocessed data and first analyzes user sentiment using an emotion engine. It then uses a natural language processing (NLP) model to extract keywords and phrases from the text and detect negative elements and potential risks. It then compares the results with a database of past scandals to check for similar patterns. It also applies a deep learning model to the image data to assess whether it contains inappropriate content or problematic elements.

[1335] Examples:

[1336] When the user inputs "50% off all products!", the server uses an emotion engine to recognize the emotion of "excitement, anticipation." The NLP model then determines that the expression "all products" poses a risk, and issues a warning because a database of past cases records similar campaigns that have caused misunderstandings and sparked outrage.

[1337] Generating results and presenting suggestions

[1338] If a risk is detected, the server generates specific correction suggestions. For example, it suggests changing the phrase "50% off all products!" to "50% off selected products!" It also generates a notification recommending consultation with the legal department if necessary. The results and suggestions are presented to the user in the form of a visual dashboard.

[1339] Examples:

[1340] The server generates a suggested correction for the risky phrase "50% off all items!" to "50% off some items!" and also creates a notification recommending confirmation by the legal department. Taking into account the results of sentiment analysis, the server reevaluates whether this suggested correction is appropriate for the user. The user can check this information on the dashboard and make any necessary corrections.

[1341] Specific program operation scenarios

[1342] Users input new campaign ideas and upload campaign banners. The server receives and preprocesses them. It then uses an emotion engine, natural language processing models, and deep learning models to perform a risk assessment. It compares the results with a database of past cases, and if a risk is detected, it generates a notification to the user suggesting specific corrections and consultation with the legal department. Users can view the information through a dashboard and take appropriate corrections or actions.

[1343] The above is an embodiment of the present invention and its specific example. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and maintain credibility.

[1344] The processing flow will be explained below.

[1345] What the program does

[1346] Processing steps of a system that combines emotion engines

[1347] Step 1:

[1348] The user inputs or uploads text data of the campaign proposal or SNS post content and related image data.

[1349] Step 2:

[1350] The server receives text data and image data sent from the user via the Internet.

[1351] Step 3:

[1352] The server performs pre-processing on the received text data, removing HTML tags and special characters and normalizing the data.

[1353] Step 4:

[1354] The server uses OCR technology on the received image data to extract text from the image and convert it into an analyzable text format.

[1355] Step 5:

[1356] The server applies an emotion engine to the preprocessed text data to recognize the user's emotions, which are then recorded in a database for subsequent processing.

[1357] Step 6:

[1358] The server applies natural language processing (NLP) models to the preprocessed text data to perform keyword extraction, sentiment analysis, and topic modeling.

[1359] Step 7:

[1360] The server detects negative elements and potential risks based on the data analyzed by the NLP model, taking into account the results of the sentiment analysis recognized in step 5.

[1361] Step 8:

[1362] The server compares the data with a database of past flame war cases, performs pattern matching with the given data, and checks whether similar cases exist.

[1363] Step 9:

[1364] The server also applies deep learning models to image data to assess whether it contains inappropriate content or problematic elements.

[1365] Step 10:

[1366] If a risk is detected based on the analysis and evaluation results, the server generates a specific correction suggestion. For example, it suggests correcting the expression "50% off all products!" to "50% off some products!"

[1367] Step 11:

[1368] The server generates notifications recommending legal consultation as needed, with specific justification based on past experience and current risk assessment.

[1369] Step 12:

[1370] The server visualizes the generated correction proposals and risk assessment results and sends the results to the user's terminal in a dashboard format that is easy for the user to understand.

[1371] Step 13:

[1372] Users can check the proposed modifications and risk assessment results presented on the dashboard and make any necessary modifications. Once the modifications are complete, they can also perform a reassessment.

[1373] Example 2

[1374] 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."

[1375] There is a lack of methods to proactively diagnose the risk of online content becoming controversial and propose appropriate revisions and actions. In particular, there is a need for a system that can assess the risk of not only text data but also image data, and that also takes user emotions into account.

[1376] 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 means for receiving data entered by a user via the Internet, means for performing preprocessing of the received data by removing HTML tags and special characters and normalizing the text, means for extracting text from image data using OCR technology, means for performing sentiment analysis using an emotion engine based on the preprocessed data, means for assessing risk using a natural language processing model, means for confirming the risk assessment result by comparing it with a database of similar past cases, means for generating a revision proposal based on the risk assessment result, and means for presenting the generated revision proposal and the risk assessment result to the user. This makes it possible to prevent the risk of a controversy in advance and increase the reliability of content.

[1377] "User" means any person or entity that uses the System to input or upload data.

[1378] "Server" refers to the hardware or software that receives, processes, and analyzes data entered or uploaded by users.

[1379] "Data" refers to information entered or uploaded by a user into the system, in the form of text, images, etc.

[1380] "Internet" refers to the global information network that enables the communication of data between users and servers.

[1381] "Preprocessing" refers to the processing performed on received data, specifically tasks such as removing HTML tags and special characters, and normalizing text.

[1382] "OCR technology" refers to optical character recognition technology for extracting character information from image data and converting it into text format.

[1383] An "emotion engine" refers to software or algorithms that analyze and determine user emotions from text data.

[1384] A "natural language processing model" refers to an algorithm or software that analyzes text data, extracts keywords and phrases, and assesses risk.

[1385] "Deep learning model" refers to a machine learning algorithm that analyzes image data and automatically evaluates inappropriate content or problematic elements.

[1386] A "database" refers to an information resource that stores past cases and information and is referenced when conducting risk assessments.

[1387] "Proposed fixes" refer to proposals that are generated based on the results of risk assessment and indicate specific fixes or countermeasures that the user should take.

[1388] A "dashboard" refers to an interface that visually displays risk assessment results and suggested corrections to users.

[1389] This invention is an AI-based system that proactively diagnoses the risk of online content published by companies and individuals becoming controversial and proposes appropriate corrections and actions. The system receives data (text and images) entered or uploaded by users, and performs a comprehensive process from data preprocessing, sentiment analysis, risk assessment, generation of correction suggestions, and presentation of the results to the user.

[1390] Receiving and Preprocessing Input Data

[1391] The server receives data entered or uploaded by users via the Internet. The received data is first preprocessed. For text data, HTML tags and special characters are removed and the text is normalized. This converts the data into analyzable data. For image data, OCR (Optical Character Recognition) technology is used to extract text from the image and convert it into an analyzable text format.

[1392] sentiment analysis

[1393] The server performs sentiment analysis on the preprocessed text data using an emotion engine, which is software that uses natural language processing (NLP) technology to determine the emotions contained in the text, such as "excitement" or "anticipation."

[1394] Risk Assessment

[1395] The server not only evaluates content based on sentiment analysis, but also uses natural language processing models to extract keywords and phrases from the text to detect negative elements and potential risks. It also uses deep learning models to analyze preprocessed image data to assess whether it contains inappropriate content or problematic elements. These processes also involve cross-checking the data with a database of past scandals.

[1396] Generating and Presenting Results

[1397] The server generates specific correction suggestions based on the risk assessment results. For example, it suggests changing the phrase "50% off all products!" to "50% off some products!" It also generates a notification recommending consultation with the legal department if necessary. These results and correction suggestions are presented to the user in the form of a visual dashboard.

[1398] An example of a specific prompt sentence is the action of entering the text "New campaign starts! All products 50% off!" and uploading a campaign banner image. In this case, the server receives the data corresponding to the prompt, preprocesses it, analyzes sentiment, and evaluates risks, and presents optimal correction suggestions to the user. The user can check this information on the dashboard and make any necessary corrections.

[1399] This completes the description of the embodiment of the present invention. This system allows users to take into consideration the results of sentiment analysis, and thereby avoid the risk of online outrage and increase the reliability of their content.

[1400] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1401] Step 1: Receive the database

[1402] The server supports users. (PC, 스마트폰) HTTP

[1403] Specific behavior:

[1404] User 가 "New campaign starts! 50% off all products!" Campaign Yes, the server is private.

[1405] Input: Data entered or uploaded by Yuzaga (text and images)

[1406] Output: Received data requiring preprocessing

[1407] ステップ2:Preprocessing of データ

[1408] Server performs preprocessing on the received data. For text data, it removes HTML tags and special characters and normalizes the text. For image data, it uses OCR technology to extract the text within the image and converts it into an analyzable text format.

[1409] Artistic work:

[1410] HTML tags and special characters are removed from the text "New campaign opening! All products 50% off!" entered by Yuza, and the text "All products 50% off!" is extracted from the deployed campaign banner image using OCR technology.

[1411] Input: Received data (text and images)

[1412] Output: Preprocessed data, normalized and converted to text.

[1413] Step 3: Emotion Analysis

[1414] サーバ inputs preprocessed text data into the emotion engine to analyze Yuza's emotions. This emotion engine is software that uses natural language processing (NLP) technology to determine the emotions of the text.

[1415] Artistic work:

[1416] The emotion engine analyzes the text "New campaign opening! 50% off all products!" and detects emotions such as "excitement" and "anticipation."

[1417] Input: Normalized text data

[1418] Output: Detected emotion result

[1419] Step 4: Risk assessment

[1420] The server combines the NLP model and the deep learning model to analyze the pre - processed data. In text data, keywords and phrases are extracted to detect negative elements and potential risks. Similarly, for image data, the deep learning model is applied to evaluate whether there is inappropriate content or problem elements. In this evaluation, the risk assessment results are reviewed by querying the database of similar cases that occurred before.

[1421] Specific operation:

[1422] The NLP model determines that the keyword "all products" has risks. There is a record in the previous similar case database that a campaign was flagged for a similar expression, so a warning is issued. [[ID=........]]

[1423] Input: Emotion analysis results, pre - processed text and image data

[1424] Output: Risk assessment results and detected risks

[1425] Step 5: Result generation

[1426] When risks are detected, the server generates specific modification suggestions. For example, for the expression "50% off all products!", it is proposed to change it to "50% off some products!". Also, a notice is generated to recommend consulting the legal department if necessary.

[1427] Specific operation:

[1428] The server creates a notice recommending verification by the legal department and drafts a revision to the expression "50% off all products!" to "50% off some products!".

[1429] Input: Risk assessment result

[1430] Output: Specific revisions and notice of recommendation

[1431] Step 6: Present results

[1432] The server presents the generated revisions or notices to the user in the form of a visualized dashboard. The user can check the information through the dashboard and make necessary revisions.

[1433] Specific operations:

[1434] The user opens the dashboard on the terminal and checks the risk assessment result of "50% off all products!" and the revision "50% off some products!". A notice of verification to the legal department is also displayed.

[1435] Input: Specific revisions and notice of recommendation

[1436] Output: Visualized assessment results and revisions for the user

[1437] The above are the specific processing steps performed by the system of the present invention.

[1438] [[ID=4)0](Application Example 2)

[1439] Next, Application Example 2 will be described. In the following description, the data processing device 12 is referred to as the "server", and the robot 414 is referred to as the "terminal".

[1440] There is a need for a system that can proactively diagnose the risk of online content becoming a hot topic and propose appropriate revisions and actions. It is also necessary to provide a more accurate risk assessment that can handle a variety of data formats, including images and audio, and that takes user emotions into account.

[1441] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving data entered by a user via the Internet, means for converting the received data into text format and performing preprocessing by removing unnecessary information, means for assessing risk using a natural language processing model based on the preprocessed data, means for generating a revision plan based on the risk assessment result by referring to a database of past cases, means for presenting the generated revision plan and the risk assessment result to the user, means for recognizing the user's emotions using an emotion engine to improve the accuracy of the risk assessment, and means for collecting and analyzing audio and video data via a smart device. This enables advance diagnosis of the risk of a controversy in content that a user intends to distribute and appropriate measures to be taken.

[1442] The "Internet" is a global information and communications network that interconnects computer networks around the world.

[1443] "User" means any person or entity that uses a system or device.

[1444] "Data" is a collection of information and includes various formats such as text, images, and audio.

[1445] A "server" is a computer system used to receive, process, and transmit data.

[1446] "Preprocessing" is the initial processing of data to convert it into an analyzable format.

[1447] "Natural language processing model" is a general term for algorithms and technologies for understanding and manipulating human language.

[1448] "Risk assessment" is the process of evaluating the potential dangers or problems of a particular activity or situation.

[1449] A "case database" is a database that systematically collects past cases and data.

[1450] "Fixation Suggestions" refers to specific improvements or proposals for addressing detected issues or risks.

[1451] "Emotion engine" is an artificial intelligence technology for analyzing and recognizing user emotions.

[1452] A "smart device" is an advanced device that has internet connectivity and is capable of collecting and analyzing data.

[1453] "Audio data" refers to audio information recorded in digital format.

[1454] "Video data" refers to visual information such as images and videos recorded in digital format.

[1455] "Analysis" is the process of analyzing data in detail and extracting its meaning and value.

[1456] This invention is a system that proactively diagnoses the risk of online content becoming controversial and suggests appropriate modifications and actions. The system receives data entered by users and performs risk assessment using an emotion engine and natural language processing model. Furthermore, the accuracy of the assessment is improved by collecting and analyzing audio and video data via smart devices.

[1457] The server first receives data entered by the user via the Internet. The data may be in a variety of formats, including text and image formats, but after receiving it, it undergoes preprocessing to make it analyzable. For example, text data is normalized by removing HTML tags and special characters. For image data, OCR technology (pytesseract library) is used to extract text from the image, and further preprocessing is performed.

[1458] Next, we use the preprocessed data to run the emotion engine, which performs sentiment analysis on the content users are about to post. For example, we can recognize emotions such as "excitement," "expectation," and "dissatisfaction." We then use the TextBlob library to evaluate the polarity of the emotions.

[1459] Once the sentiment analysis is complete, a natural language processing model (Keras and a text tokenizer) is used to assess the risk of the received text data. During this process, the data is compared with a database of past scandals to determine whether there is any risk. Specific revisions are then proposed based on the generated risk assessment results. For example, if "50% off all products!" is judged to be a risky expression, a suggestion is made to change it to "50% off some products!"

[1460] Finally, the server presents the risk assessment results and suggested remediation to the user in a visual dashboard, allowing the user to take concrete corrective action. Depending on the level of risk, a notification may also be generated recommending that the user consult with the legal department.

[1461] These processes are carried out via smart devices, making it easy to collect audio and video data. For example, content can be evaluated on the spot using a smartphone or tablet, and suggestions for revisions can be received immediately.

[1462] Specific examples

[1463] The user inputs the text "New campaign starts! All products 50% off!" as a new campaign proposal and uploads a campaign banner image. The server receives this and preprocesses the text and image. It then uses an emotion engine and natural language processing model to perform sentiment analysis and risk assessment, generating a revised proposal: "50% off some products!"

[1464] Example prompts to input to a generative AI model:

[1465] test_text = "New campaign starts! 50% off all items!"

[1466] test_image_path = "path_to_campaign_banner.jpg"

[1467] main(test_text, test_image_path)

[1468] By entering this prompt, users can check the risk assessment results and suggested corrections on the dashboard, enabling safer content delivery.

[1469] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1470] Step 1:

[1471] The user sends text data or image data from their terminal to the server via the Internet. The user's input data is content information such as campaign proposals. In this step, it is confirmed whether the user's input reaches the server.

[1472] Input: Text data and image data entered by the user

[1473] Output: Raw data sent to the server

[1474] Step 2:

[1475] The server preprocesses the data it receives. For text data, it removes HTML tags and special characters and normalizes it. For image data, it uses OCR technology (pytesseract library) to extract text. This step results in clean data that can be analyzed.

[1476] Input: Raw data sent to the server

[1477] Output: Preprocessed, clean text data

[1478] Step 3:

[1479] The server performs sentiment analysis on the preprocessed text data. It uses the TextBlob library to evaluate the sentiment polarity of the text and determine whether it is "positive" or "negative." This step quantifies the user's sentiment.

[1480] Input: Preprocessed clean text data

[1481] Output: Sentiment analysis result (positive or negative)

[1482] Step 4:

[1483] The server uses a natural language processing model to assess risk. Using Keras and a text tokenizer, the text data is passed to the model, which then compares it with a database of past scandals to calculate a risk score. This step quantifies the risk of a particular piece of content becoming a scandal.

[1484] Input: Preprocessed clean text data

[1485] Output: Risk score

[1486] Step 5:

[1487] The server generates specific correction suggestions based on the risk assessment results. If the risk score is high, it will suggest corrections to the problematic expression. For example, instead of "50% off all products!", it will suggest "50% off some products!"

[1488] Input: Risk Score

[1489] Output: Revision proposal

[1490] Step 6:

[1491] The server then presents the generated remediation proposals and risk assessment results to the user, who can then view this information and consider appropriate responses through a visual dashboard.

[1492] Input: Remediation proposals and risk assessment results

[1493] Output: Dashboard display

[1494] Step 7:

[1495] Depending on the risk assessment result, the server generates a notification recommending that the user consult with the legal department. If the risk score is particularly high, a notification is generated and the user is prompted to check with the legal department.

[1496] Input: Risk Score

[1497] Output: Notice recommending consultation with legal department

[1498] Step 8:

[1499] The user can provide additional input (audio and video data) via a smart device. This is also received by the server and similarly preprocessed and analyzed, enabling more accurate evaluation.

[1500] Input: Additional audio and video data

[1501] Output: Analysis results

[1502] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1503] 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.

[1504] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1505] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1506] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1507] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1508] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1509] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1510] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1511] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1512] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1513] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1514] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1515] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1516] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1517] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1518] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1519] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1520] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1521] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1522] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1523] The following is further disclosed regarding the above embodiment.

[1524] (Claim 1)

[1525] means for receiving user input data via the Internet;

[1526] means for preprocessing the received data by converting it into a text format and removing unnecessary information;

[1527] a means for assessing risk using a natural language processing model based on the preprocessed data;

[1528] A means for generating correction proposals based on risk assessment results by referring to a database of past cases;

[1529] means for presenting the generated revision proposals and risk assessment results to a user;

[1530] A system including:

[1531] (Claim 2)

[1532] 10. The system of claim 1, further comprising: means for generating a notification recommending legal consultation in response to a risk assessment result.

[1533] (Claim 3)

[1534] 10. The system of claim 1, further comprising means for converting the received image data to text using OCR technology and evaluating the text for inappropriate content.

[1535] "Example 1"

[1536] (Claim 1)

[1537] means for receiving user input data via the Internet;

[1538] means for preprocessing the received data by converting it into a text format and removing unnecessary information;

[1539] a means for assessing risk using a natural language processing model based on the preprocessed text data;

[1540] A means for generating correction proposals based on risk assessment results by referring to a database of past cases;

[1541] a means for presenting the generated remediation proposals and risk assessment results to a user in a dashboard format;

[1542] A system including:

[1543] (Claim 2)

[1544] 10. The system of claim 1, further comprising: means for generating a notification recommending legal consultation in response to a risk assessment result.

[1545] (Claim 3)

[1546] 10. The system of claim 1, further comprising means for converting the received image data to text using OCR technology and evaluating the text for inappropriate content.

[1547] "Application Example 1"

[1548] (Claim 1)

[1549] means for receiving user input data via the Internet;

[1550] means for preprocessing the received data by converting it into a text format and removing unnecessary information;

[1551] a means for assessing risk using a natural language processing model based on the preprocessed data;

[1552] A means for generating correction proposals based on risk assessment results by referring to a database of past cases;

[1553] means for presenting the generated revision proposals and risk assessment results to a user;

[1554] A means to suggest specific remediations or actions when risks are detected;

[1555] A system including:

[1556] (Claim 2)

[1557] 10. The system of claim 1, further comprising: means for generating a notification recommending legal consultation in response to a risk assessment result.

[1558] (Claim 3)

[1559] 10. The system of claim 1, further comprising means for converting the received image data to text using OCR technology and evaluating the text for inappropriate content.

[1560] (Claim 4)

[1561] 10. The system of claim 1, further comprising: means for providing a dashboard function that displays risk visualization and remediation suggestions to a user.

[1562] "Example 2: Combining Emotion Engines"

[1563] (Claim 1)

[1564] means for receiving user-entered data via the Internet;

[1565] A means for pre-processing the received data to remove HTML tags and special characters and normalize the text;

[1566] means for extracting text from image data using OCR technology;

[1567] means for performing sentiment analysis using a sentiment engine based on the preprocessed data;

[1568] a means for assessing risk using a natural language processing model;

[1569] A means to check the results of risk assessment against a database of similar cases from the past;

[1570] a means for generating a remediation plan based on the risk assessment results;

[1571] means for presenting the generated revision proposals and risk assessment results to a user;

[1572] A system including:

[1573] (Claim 2)

[1574] 10. The system of claim 1, further comprising: means for generating a notification recommending legal consultation in response to a risk assessment result.

[1575] (Claim 3)

[1576] 10. The system of claim 1, further comprising means for analyzing the received image data using a deep learning model to evaluate whether the image data contains inappropriate content.

[1577] "Application example 2 when combining emotion engines"

[1578] (Claim 1)

[1579] means for receiving user input data via the Internet;

[1580] means for preprocessing the received data by converting it into a text format and removing unnecessary information;

[1581] a means for assessing risk using a natural language processing model based on the preprocessed data;

[1582] A means for generating correction proposals based on risk assessment results by referring to a database of past cases;

[1583] means for presenting the generated revision proposals and risk assessment results to a user;

[1584] a means for recognizing user emotions using an emotion engine to improve the accuracy of risk assessment;

[1585] a means for collecting and analyzing audio and video data via a smart device;

[1586] A system including:

[1587] (Claim 2)

[1588] 10. The system of claim 1, further comprising: means for generating a notification recommending legal consultation in response to a risk assessment result.

[1589] (Claim 3)

[1590] 10. The system of claim 1, further comprising means for converting the received image data to text using OCR technology and evaluating the text for inappropriate content. [Explanation of symbols]

[1591] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving user input data via the Internet; means for preprocessing the received data by converting it into a text format and removing unnecessary information; a means for assessing risk using a natural language processing model based on the preprocessed data; A means for generating correction proposals based on risk assessment results by referring to a database of past cases; means for presenting the generated revision proposals and risk assessment results to a user; A system including:

2. The system of claim 1 , further comprising: means for generating a notification recommending legal consultation in response to a risk assessment result.

3. 10. The system of claim 1, further comprising means for converting the received image data to text using OCR technology and evaluating the text for inappropriate content.

Images