System

Abstract

A system is provided.SOLUTION: A system comprising: means for receiving user-input post content; means for analyzing the post content using natural language processing to extract keyword, context, and tone information; means for evaluating the analysis results based on a pre-defined rule-based checklist; means for predicting a fire risk of the post content using a machine learning model that has learned past fire cases and post patterns; and means for providing feedback to a user based on the evaluation and prediction results.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] Individuals and companies using social media face the risk of offensive language or inappropriate comments becoming a social problem because their posts are instantly spread far and wide. Such flame wars can have a significant negative impact on the reputation of individuals and brands. However, existing countermeasures are primarily reactive, and there are few effective systems for preventing flame wars. This can lead to users posting content that poses a risk of flame wars without realizing it. The present invention aims to solve these issues and provide a system that enables real-time post analysis and flame war prevention. [Means for solving the problem]

[0005] The system of the present invention includes a means for receiving a post entered by a user, a means for analyzing the post using natural language processing and extracting information on keywords, context, and tone, a means for evaluating the analysis results based on a predefined rule-based checklist, and a means for predicting the risk of the post causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, and a means for providing feedback to the user based on the evaluation and prediction results. This allows the system to identify the risk of a user's post causing an uproar in advance and provide appropriate suggestions for correction, thereby maintaining healthy communication.

[0006] "User" refers to an individual or company that creates and publishes posts using the SNS platform.

[0007] "Posted content" refers to the text information entered by a user to be published on an SNS.

[0008] "Means for receiving" refers to an input interface or communication means for acquiring the content posted by the user and transmitting it to the system.

[0009] "Natural language processing" refers to computer technology that analyzes human language and understands and processes its meaning and structure.

[0010] "Analysis" refers to the process of breaking down a post and extracting and understanding its keywords, context, and tone.

[0011] "Keywords" refer to words or phrases that are particularly important in the content of a post.

[0012] "Context" refers to the relationship between words and sentences that help understand the overall meaning of the post.

[0013] "Tone" refers to the choice of words and expressions that convey the emotion or attitude of the post.

[0014] A "rules-based checklist" refers to a predefined criteria or list that determines whether a particular keyword or phrase is included.

[0015] "Means of evaluation" refers to the process of checking the analysis results against a rule-based checklist to determine whether there are any problems.

[0016] A "machine learning model" refers to an algorithm that learns patterns in posted content based on past data and makes predictions and classifications for new data.

[0017] "Risk of a firestorm" refers to the likelihood that the content of a post will cause criticism or trouble after it is published.

[0018] "Means for providing feedback" refers to a process or system that notifies users of analysis results and suggested modifications in real time. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8]FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0020] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

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

[0022] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0023] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0024] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0025] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0026] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0027] [First embodiment]

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

[0029] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0030] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0031] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0032] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0033] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0034] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

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

[0036] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0037] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0038] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0039] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0040] The present invention is a system that analyzes user's SNS postings in real time and identifies and prevents the risk of flame wars in advance. The system is implemented using the following specific means.

[0041] System Configuration

[0042] The system consists of the following main components:

[0043] 1. Terminal

[0044] 2. Server

[0045] 3. Natural Language Processing (NLP) Engine

[0046] 4. Machine Learning Models

[0047] 5. Database

[0048] Processing Flow

[0049] 1. Enter and submit your post

[0050] The device inputs user posts and receives them through a social networking app or web interface. For example, a user might post something like, "These services are really no good these days!" The device then generates and sends an HTTP request to send this post to the server.

[0051] 2. Analysis of posted content

[0052] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[0053] Keyword analysis: Extracting important words and phrases within posts. Examples: "no" and "really."

[0054] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[0055] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0056] 3. Rule-based checks

[0057] The server evaluates the output of the NLP engine based on a rule-based checklist, for example determining whether "no" or "really" is offensive.

[0058] 4. Application of the fire prediction model

[0059] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. This model learns from past cases of uproars and posting patterns, and provides a risk score for the post. For example, "The risk of uproar is 70%."

[0060] 5. Providing real-time feedback

[0061] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0062] 6. Corrections and resubmissions

[0063] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post (e.g., "I feel there is room for improvement in the recent service"), and the device resends the final post to the server.

[0064] 7. Save and publish your final post

[0065] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0066] The above is a specific embodiment of the present invention. This system allows users to understand in advance the risk of their posts causing a controversy and make appropriate corrections, thereby enabling them to maintain healthy communication.

[0067] The processing flow will be explained below.

[0068] Step 1:

[0069] The terminal receives user posts via a text area or form. For example, the user might write, "These services are really no good these days!"

[0070] Step 2:

[0071] The device constructs the input post content as an HTTP request and sends this request to the server, including the user ID and timestamp.

[0072] Step 3:

[0073] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[0074] Step 4:

[0075] The server's NLP engine analyzes the post, specifically:

[0076] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0077] Contextual analysis: Understand the overall context of a post. Analyze the relationships between words.

[0078] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0079] Step 5:

[0080] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[0081] Step 6:

[0082] The server uses a machine learning model to predict the risk of a post causing an uproar. The model learns from past cases of uproars and posting patterns, and assigns a risk score to the post (e.g., a 70% risk of causing an uproar).

[0083] Step 7:

[0084] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0085] Step 8:

[0086] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[0087] Step 9:

[0088] The user reviews the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[0089] Step 10:

[0090] The device resends the modified or original post in its final form to the server, which receives the final post.

[0091] Step 11:

[0092] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0093] Example 1

[0094] 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."

[0095] In recent years, the impact of social media posts has become increasingly large, and inappropriate posts by users often have widespread repercussions in the form of flame wars. Therefore, there is a strong demand for methods that allow users to understand in advance whether their posts pose a risk of causing a flame war and to appropriately avoid that risk. However, current systems have the problem of insufficient support for users to quickly and accurately evaluate the flame war risk of their posts and make appropriate corrections.

[0096] 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.

[0097] In this invention, the server includes means for receiving posts entered by users, means for analyzing the posts using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the posts causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation and prediction results, means for the user to review the proposed revisions and choose whether to accept or ignore them, and means for receiving the revised posts again, saving them, and publishing them on the SNS platform. This allows users to understand the risk of their posts causing an uproar in advance and make appropriate revisions, thereby significantly reducing the risk of an uproar.

[0098] "User" refers to an individual or corporation that uses this system to post on SNS.

[0099] A "terminal" is a device used by a user, and includes a computer, a smartphone, a tablet, and the like.

[0100] "Server" refers to the computer system that receives, analyzes, provides feedback on, stores, and publishes Submissions.

[0101] "Posted content" refers to the text information posted by a user on a social media platform.

[0102] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[0103] "Keywords" refer to important words or phrases extracted from the content of a post.

[0104] "Context" refers to the relevance of the meaning of a word or phrase within the overall content of a post.

[0105] "Tone" refers to facial expressions that convey the emotion and intent of the post.

[0106] A "rule-based checklist" is a list for evaluating posts based on predefined criteria.

[0107] A "machine learning model" refers to an algorithm that learns from past data and makes predictions and classifications for new data.

[0108] "Flame risk" refers to the possibility that the content of a post will provoke a negative reaction.

[0109] "Feedback" refers to advice and suggestions provided to users based on analysis and prediction results.

[0110] "Suggested Revisions" means suggestions or alternatives for a User to revise a Submission.

[0111] "SNS Platform" refers to an online service that allows users to publish and share their posts with other users.

[0112] "Storage" refers to recording the posted content in a database or storage device.

[0113] "Public" means displaying the posted content on the social media platform so that other users can view it.

[0114] This invention is a system that analyzes users' social media posts in real time to identify and prevent the risk of flame wars in advance. This system receives the content posted by users and analyzes, evaluates, and provides feedback using a natural language processing (NLP) engine and machine learning model, making it possible to maintain healthy communication.

[0115] The main hardware and software components of this system are:

[0116] The device used by the user (computer, smartphone, tablet, etc.)

[0117] Server (responsible for receiving, analyzing, providing feedback on, storing and publishing posted content)

[0118] Natural Language Processing (NLP) Engine

[0119] Machine learning models

[0120] Database

[0121] First, a user uses a device to input content for a post via a social networking app or web interface. For example, the user might input "These services are really no good these days!" The device then generates this content as an HTTP request and sends it to the server.

[0122] The server then passes the content received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts information about keywords, context, and tone. Specifically, it identifies keywords like "no" and "really," analyzes the meaning of the entire sentence, and calculates a negative tone.

[0123] The server then evaluates the output of the NLP engine based on a rule-based checklist. The parsed keywords and tones are compared against the rule-based list to determine whether they are offensive. For example, words like "no" and "really" are considered offensive.

[0124] Next, the server uses a machine learning model to predict whether the post has a risk of causing a controversy. This model learns from past cases of controversy and posting patterns. It calculates a controversy risk score for the post and outputs, for example, "Controversy risk is 70%."

[0125] The server provides feedback to the user based on the evaluation and prediction results. This feedback includes specific suggestions for rephrasing or correction. For example, it may provide a suggestion such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0126] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post. For example, they might revise the post to say, "I feel there is room for improvement in the recent service." The device then sends this revised post back to the server.

[0127] Finally, the server saves the modified post to a database and publishes it on the social media platform, so that it appears on the timeline and can be viewed by other users.

[0128] For example, the following prompt could be fed to a generative AI model: "Please use an NLP engine to analyze social media posts entered by users and detect offensive language. Also, please suggest modifications to reduce this risk."

[0129] This system allows users to understand in advance the risk of their posts causing a backlash and make appropriate corrections, significantly reducing the risk of a backlash.

[0130] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0131] Step 1:

[0132] A user uses a device to enter content to post via a social networking app or web interface and presses the "Send" button. For example, a user enters a post such as "These services these days are really no good!" The device generates the entered content as an HTTP request and sends this request to the server. The input is the text data posted by the user, and the output is an HTTP request to the server.

[0133] Step 2:

[0134] The server passes the content of posts received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts the following information: the input is the received post data, and the output is extracted keywords, context, and tone information.

[0135] Keyword analysis: Identify important words and phrases within your posts. Examples include "no" and "really."

[0136] Contextual analysis: Analyzing the relationships between words to understand the meaning of the whole sentence. For example, the context of "Dameda."

[0137] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0138] Step 3:

[0139] The server evaluates the output of the NLP engine based on a rule-based checklist. The input is the analyzed data, and the output is the evaluation result. Specifically, it checks the analyzed keywords and tones against the rule-based list to see if they are offensive. For example, "no" and "really" are considered offensive.

[0140] Step 4:

[0141] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. The input is the rule-based evaluation result, and the output is a score for the risk of causing an uproar. Specifically, it uses a model that has learned from past cases of uproars and posting patterns to calculate the risk score for the post. Example: "The risk of causing an uproar is 70%."

[0142] Step 5:

[0143] The server provides feedback to the user based on the analysis and prediction results. The input is the flame risk score, and the output is the content of the feedback. For example, the following suggestion is generated as a specific action: "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0144] Step 6:

[0145] The user reviews the proposed revisions and chooses whether to accept or ignore them. The input is the feedback content, and the output is the revised post content or the original post content. Specifically, if the user accepts the revisions, the post content is updated (e.g., "I feel there is room for improvement in the recent service"). The device then sends the revised post content back to the server.

[0146] Step 7:

[0147] The server saves the final post in a database and publishes it on the social media platform. The input is the final post, and the output is saving it to the database and publishing it. Specifically, the revised post is saved in the database and displayed on the social media platform's timeline for other users to view.

[0148] (Application example 1)

[0149] 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."

[0150] In modern advertising campaigns, posting text to digital platforms such as social media is essential. However, posts containing incorrect expressions or a negative tone can damage a company's brand image and risk sparking outrage. In particular, in an age where past posts are easily shared, negative reactions can spread instantly. Therefore, there is a need for a system that can proactively identify advertising copy that may spark outrage and suggest revisions.

[0151] 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.

[0152] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and prediction results, and means for analyzing the risk of advertising copy in the post content causing an uproar and proposing revisions. This allows users to understand the risk of uproar in the post content of their advertising campaigns in advance and make appropriate revisions, thereby enabling safe and effective advertising.

[0153] "User" refers to a person or organization that uses the system to input and post advertising copy.

[0154] "Post content" refers to the text entered by a user and intended to be published on social media or digital platforms.

[0155] "Natural language processing" refers to the technology that enables computers to understand and analyze human language.

[0156] "Keywords" refers to important words and phrases extracted from the content of a post.

[0157] "Context" refers to information that allows us to understand the meaning of the entire sentence through the meaning and relationships of each word and phrase.

[0158] "Tone" refers to information that indicates the emotional nuance or attitude of a piece of writing.

[0159] A "rule-based checklist" refers to a list for evaluating posts based on predefined evaluation criteria.

[0160] A "machine learning model" refers to an algorithm that learns specific tasks based on past data and makes predictions and evaluations for new data.

[0161] "Flame risk" refers to an indicator of the likelihood that a post will provoke a negative reaction.

[0162] "Feedback" refers to information provided to a user, including analysis results and suggested modifications.

[0163] "Proposed amendments" refer to proposed changes to the content of a post to reduce the risk of a backlash.

[0164] This invention is a system that analyzes the content of advertising text in real time before users post it on social media or digital platforms, and identifies and prevents the risk of online outrage in advance. The system includes the following main components and processing flow:

[0165] System Configuration

[0166] The system consists of the following main components:

[0167] 1. Terminal: The device on which the user enters advertising text.

[0168] 2. Server: The central computer that performs analysis and predictions.

[0169] 3. Natural Language Processing (NLP) engine: Software that analyzes the content of posts.

[0170] 4. Machine learning model: An algorithm for predicting the risk of a firestorm.

[0171] 5. Database: An information repository that stores past posts and analysis results.

[0172] Explanation of program processing

[0173] The server first passes the ad copy received from the user device to a natural language processing (NLP) engine, which performs the following analysis:

[0174] Keyword analysis: Extracting important words and phrases within ad copy.

[0175] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[0176] Tone analysis: Calculating a sentiment score for text to infer sentiment and intent.

[0177] The server then evaluates the output of the NLP engine based on a rule-based checklist, determining, for example, that the phrase "better than any other product" is offensive.

[0178] The server then uses a machine learning model to predict the risk of the ad copy causing an uproar. The model learns from past cases of uproar and posting patterns, and provides an uproar risk score for the ad copy.

[0179] Based on the results, the server will provide feedback to the user, such as "This expression may provoke a negative reaction. We recommend changing it to something like 'Many customers are satisfied.'"

[0180] The user can then review the proposed revisions and choose to accept or ignore them. If the user accepts the revisions, the ad copy is updated and the device sends the final copy back to the server. The server then stores the final copy in a database and publishes it on the social media platform.

[0181] Specific examples

[0182] For example, if a user enters the ad copy, "This product is better than any other product," the server will analyze that expression and determine that there is a high risk of a negative reaction. Therefore, the server will provide feedback such as, "This expression may cause a negative reaction. We recommend changing it to something like, 'Many customers are satisfied with this product.'"

[0183] Example prompt sentence:

[0184] Please analyze whether this ad copy poses a risk of causing controversy. Also, please explain why and suggest modifications.

[0185] Hardware and software used

[0186] Hardware: General PCs and smartphones

[0187] Software: nltk (natural language processing library), sklearn (machine learning model library)

[0188] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0189] Step 1:

[0190] To receive the ad copy entered by the user, the device inputs the ad copy from the user and receives it through a social networking app or web interface. For example, the user may input ad copy such as "This product is better than any other product." The user inputs text data through a browser or mobile app, and the data is saved on the device.

[0191] Step 2:

[0192] The terminal generates and sends an HTTP request to send the entered ad copy to the server. The sent data is the ad copy in text format. After the server receives this data, it saves it as text data.

[0193] Step 3:

[0194] The server passes the received ad copy to a natural language processing (NLP) engine. The input is the text data received in step 2. The NLP engine analyzes the data to extract keywords, context, and tone information from the ad copy. Specifically, it extracts keywords, analyzes the context of the text, and calculates a sentiment score. The output is the analysis results (keyword list, context information, and tone score).

[0195] Step 4:

[0196] Based on the analysis results, the server performs an evaluation according to a predefined rule-based checklist. The analysis results obtained in step 3 are used as input. The rule-based evaluation determines whether certain keywords or tones pose a risk. For example, the keyword "better than any other product" is evaluated as offensive. The evaluation results (a list of problematic keywords and tone analysis results) are generated as output.

[0197] Step 5:

[0198] The server uses a machine learning model to predict the risk of the ad copy causing an uproar. The evaluation results obtained in step 4 are used as input. The machine learning model has learned from past cases of uproars and posting patterns, and calculates a risk score for new data. Specifically, the text data is vectorized, and the vector is input into the model for scoring. The output is an uproar risk score.

[0199] Step 6:

[0200] The server provides feedback to the user based on the evaluation results and risk score. The evaluation results and the flame risk score are used as inputs. The server generates feedback for the user, including problematic expressions and recommended corrections. For example, the server generates feedback such as, "This expression may provoke negative reactions. We recommend changing it to something like, 'Many customers are satisfied.'" The server generates a feedback message as output.

[0201] Step 7:

[0202] The server sends the generated feedback message to the terminal and displays it to the user. The feedback message generated in step 6 is used as input. The user checks the feedback message and chooses whether to accept or ignore the proposed revision. For example, if the user accepts the proposed revision, the revised advertisement copy is displayed.

[0203] Step 8:

[0204] If the user accepts the proposed revision, the device sends the revised copy to the server again. The revised copy is used as input. The server stores the received final copy in a database and publishes it on the social networking platform. The published copy is generated as output.

[0205] 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.

[0206] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[0207] System Configuration

[0208] The system consists of the following main components:

[0209] 1. Terminal

[0210] 2. Server

[0211] 3. Natural Language Processing (NLP) Engine

[0212] 4. Emotion Engine

[0213] 5. Machine Learning Models

[0214] 6. Database

[0215] Processing Flow

[0216] 1. Enter and submit your post

[0217] The device receives user submissions via a text area or form. For example, the user might enter a submission such as "These services are really no good these days!" The device then generates and sends an HTTP request to the server to send the submission.

[0218] 2. Analysis of posted content

[0219] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[0220] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0221] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[0222] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0223] 3. Running the Emotion Engine

[0224] The server uses an emotion engine to analyze the type and intensity of emotions (e.g., joy, anger, sadness) contained in the user's post. For example, it may determine that "this post has a strong emotion of anger."

[0225] 4. Rule-based checks

[0226] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[0227] 5. Application of the Fire Prediction Model

[0228] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[0229] 6. Providing real-time feedback

[0230] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[0231] 7. Corrections and resubmissions

[0232] The user checks the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent back to the server via the device. For example, change it to "I feel there is room for improvement in the recent service."

[0233] 8. Save and publish your final post

[0234] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0235] The above is a specific embodiment of the present invention. By incorporating an emotion engine, it is possible to grasp user emotions in detail and improve the accuracy of predicting the risk of a controversy. This system allows users to understand the risk of a post causing a controversy in advance and make appropriate corrections, thereby maintaining healthy communication.

[0236] The processing flow will be explained below.

[0237] Step 1:

[0238] The terminal receives user posts via a text area or form. For example, the user may post, "These services are really no good these days!"

[0239] Step 2:

[0240] The device constructs an HTTP request containing the entered post content and sends this request to the server, including the user ID and timestamp.

[0241] Step 3:

[0242] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[0243] Step 4:

[0244] The server's NLP engine analyzes the post, specifically:

[0245] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0246] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[0247] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0248] Step 5:

[0249] The server uses an emotion engine to analyze the type of emotion (e.g., joy, anger, sadness) and intensity of the emotion contained in the post. For example, it may determine that "this post contains strong anger."

[0250] Step 6:

[0251] The server evaluates the results of the analysis using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate. For example, checking if words like "bad" or "terrible" are present in the list.

[0252] Step 7:

[0253] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[0254] Step 8:

[0255] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0256] Step 9:

[0257] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[0258] Step 10:

[0259] The user checks the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[0260] Step 11:

[0261] The device resends the modified or original post in its final form to the server, which receives the final post.

[0262] Step 12:

[0263] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0264] Example 2

[0265] 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."

[0266] Conventional SNS systems lacked sufficient filtering and moderation functions to determine whether user posts were appropriate, resulting in a high risk of unexpected outrage. Furthermore, when a user unintentionally used offensive language, the feedback function was insufficient, making it difficult to prompt the user to correct the post. To solve these problems, a system that analyzes user posts in more detail and provides real-time feedback is needed.

[0267] 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 post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for using an emotion engine to analyze the type and intensity of emotion contained in the post content, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned past cases of uproars and posting patterns, and means for providing feedback to the user based on the evaluation results and prediction results. This makes it possible to grasp the risk of a user's post content causing an uproar in advance and provide appropriate feedback in real time.

[0268] "User" means any individual or entity that uses a social media platform to enter and submit posts.

[0269] "Terminal" refers to an information processing device used by a user, such as a computer, smartphone, or tablet.

[0270] "Server" refers to a computer system that has the function of analyzing and evaluating the content posted by users and providing feedback.

[0271] "Natural language processing" refers to the technology of using computers to analyze human language, extract information, and understand meaning.

[0272] "Keyword analysis" refers to the technology of extracting important words and phrases from the content of posts.

[0273] "Contextual analysis" refers to a technology that understands the relationships between words in a sentence and grasps the overall meaning of the post.

[0274] "Tone analysis" refers to the technique of calculating an emotional score to infer the sentiment or intent of a piece of text.

[0275] "Emotion engine" refers to software or hardware for analyzing the type and intensity of emotions contained in posted content.

[0276] A "rules-based checklist" is a list for evaluating posts based on predefined keywords and phrases.

[0277] A "machine learning model" refers to a computational model that uses algorithms to learn patterns and rules based on data.

[0278] "Flame risk prediction" refers to a technology that predicts the risk of a post causing a flame war based on past flame war cases and posting patterns.

[0279] "Means for providing feedback" refers to a function that provides appropriate advice and suggestions for correction to the user in real time based on the analysis and prediction results.

[0280] "Suggestions to modify post content" refers to suggestions to change a user's post to an appropriate expression when the post is deemed offensive or inappropriate.

[0281] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[0282] System Configuration

[0283] The system consists of the following main components:

[0284] 1. Devices (computers, smartphones, tablets, etc.)

[0285] 2. Server (AWS or on-premise)

[0286] 3. Natural Language Processing (NLP) engines (e.g., SpaCy, nltk)

[0287] 4. Emotion engines (e.g., IBM Watson, Microsoft Text Analytics)

[0288] 5. Machine learning models (e.g., custom models based on TensorFlow or Scikit-Learn)

[0289] 6. Database (e.g. MySQL, PostgreSQL)

[0290] Function details

[0291] 1. Enter and submit your post

[0292] A user uses a device to enter text into a posting field on a social networking site. For example, the user enters the content of a post, such as "The services these days are really no good!" The device generates an HTTP request to send this content to the server, and sends it to the server.

[0293] 2. Analysis of posted content

[0294] The server passes the content received from the device to a natural language processing (NLP) engine, which then performs the following tasks:

[0295] Keyword analysis: Extracting important words and phrases, such as "no" and "really."

[0296] Contextual analysis: Understand the overall context of the post and analyze the relationships between words.

[0297] Tone analysis: Infers the sentiment and intent of a post and calculates a sentiment score. For example, it may be determined to have a "very negative tone."

[0298] 3. Running the Emotion Engine

[0299] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post. For example, it may determine that "this post contains a strong emotion of anger."

[0300] 4. Rule-based checks

[0301] The server uses a rule-based checklist to evaluate the results of the analysis, checking whether certain keywords or phrases are offensive or inappropriate, for example, whether words like "bad" or "terrible" exist in the checklist.

[0302] 5. Application of the Fire Prediction Model

[0303] The server uses a machine learning model to predict the risk of a post causing an uproar. This model learns from past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine. For example, it may assess the risk of a post causing an uproar as 70%.

[0304] 6. Providing real-time feedback

[0305] The server generates the analysis results and proposed modifications and sends them to the device. The device then displays feedback to the user. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[0306] 7. Corrections and resubmissions

[0307] The user reviews the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent again to the server via the device. For example, the user could change the post to "I feel there is room for improvement in the recent service."

[0308] 8. Save and publish your final post

[0309] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0310] Specific examples in the description

[0311] A user posts on social media, "These services are really no good!" This post is analyzed by the system and the following feedback is provided:

[0312] Keyword analysis determines "no" as offensive

[0313] The sentiment engine analyzes the tone of the post as "very negative."

[0314] The flame war prediction model predicts a "70% risk of flame war"

[0315] The server sends feedback to the device saying, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there are areas for improvement in your recent service.'"

[0316] Prompt Sentence Examples

[0317] "Please provide a detailed explanation of the specific processing steps involved in the system for analyzing the sentiment of social media posts and predicting the risk of a social media outcry. Furthermore, please provide specific examples of how this system provides feedback to users."

[0318] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0319] Step 1:

[0320] The user enters the content to post and submits it.

[0321] Specific actions: The user uses a device to enter the text "These services are really no good!" into the post field of a social networking site and clicks the send button.

[0322] Input: User-entered post content (e.g., "These services are really no good these days!")

[0323] Output: HTTP request generated by the device

[0324] Step 2:

[0325] The server receives the post and passes it to a natural language processing (NLP) engine.

[0326] Specific operation: The server receives the HTTP request generated by the device and passes the text data in the request to the NLP engine.

[0327] Input: HTTP request sent from the terminal

[0328] Output: Passing text data to the NLP engine

[0329] Step 3:

[0330] A natural language processing (NLP) engine analyzes posts to extract keywords, context, and tone of voice.

[0331] Specific operation: The NLP engine extracts important keywords such as "bad" and "really" from the text "These services these days are really no good!", performs contextual analysis, and then calculates an emotional score to determine the sentence as "negative."

[0332] Input: Text data of the post content

[0333] Output: Keyword, context, and tone analysis results (e.g., "Keyword: No, really," "Context: Negative," "Tone: Very negative")

[0334] Step 4:

[0335] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post.

[0336] Specific operation: The analysis results from the NLP engine are passed to the emotion engine, which analyzes the type and intensity of emotions, such as "anger" or "sadness."

[0337] Input: Analysis results of the NLP engine

[0338] Output: Analysis results of the emotion engine (e.g., "Emotion type: anger" and "Intensity: strong")

[0339] Step 5:

[0340] The server evaluates the submission using a rule-based checklist.

[0341] What it does: Parsed keywords are checked against a predefined rule-based checklist to see if they contain offensive words or phrases.

[0342] Input: Analysis results of NLP engine, analysis results of emotion engine

[0343] Output: Rule-based check results (e.g., "Aggressive: No (exists)", "Non-aggressive: Really (does not exist)", "Overall rating: Aggressive")

[0344] Step 6:

[0345] The server uses a machine learning model to predict the risk of a post causing an uproar.

[0346] Specific operation: The analysis results and the results of the emotion engine are input into a machine learning model, which then calculates the risk of a controversy based on past cases of controversy and posting patterns.

[0347] Input: Analysis results, emotion engine results

[0348] Output: Flame risk prediction (e.g., "Flame risk: 70%")

[0349] Step 7:

[0350] The server generates analysis results and suggested corrections and sends them to the device.

[0351] Specific operation: The server generates a proposal to modify the post content based on the results of the flame war risk prediction and sends it to the terminal as an HTTP response.

[0352] Input: Flame risk prediction results

[0353] Output: Generates a suggested fix and an HTTP response (e.g., "Suggested fix: This expression may be considered offensive. We recommend changing it to something like, 'We feel your recent service has room for improvement.'")

[0354] Step 8:

[0355] The user checks the feedback and selects whether to make corrections.

[0356] Specific operation: The device displays the feedback received from the server to the user, and the user decides whether to accept the suggested revisions or continue posting as is.

[0357] Input: Feedback from the server

[0358] Output: User's choice (e.g. "Accept proposed revision" or "Keep original post")

[0359] Step 9:

[0360] The server stores the final post in a database and publishes it on the social media platform.

[0361] Specific operation: The user resubmits the revised or original post to the server, which then stores the post in its database and publishes it on the social media platform.

[0362] Input: User's last post

[0363] Output: Saving to a database and publishing on a social media platform (e.g. "Saved to database" and "Published on social media")

[0364] (Application example 2)

[0365] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0366] Conventional systems for analyzing the content of posts on social media platforms are able to provide appropriate feedback by gaining a detailed understanding of user emotions and predicting the risk of a controversy in advance. However, advertising content created by advertising agencies in particular has a higher risk of causing a controversy than general user posts, and could potentially damage a company's brand image. Current systems lack specialized analysis and feedback for advertising content, and there is a need for technology that automatically provides appropriate revision suggestions to allow advertising agencies to post with confidence.

[0367] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0368] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and the prediction results, means for evaluating the emotions of advertising content using an emotion engine and scoring the risk of uproar, and means for providing suggestions for modifying the advertising content based on the scoring results. This enables advertising agencies to thoroughly evaluate the risk of uproaring advertising content before posting it and automatically receive appropriate suggestions for modification.

[0369] A "user" is a user who enters content to post on a social networking site or advertising platform.

[0370] "Posted content" refers to content such as text, images, and videos that users enter and submit to social media or advertising platforms.

[0371] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[0372] "Analysis" is the process of examining input data (in this case, posts) in detail and extracting information such as keywords, context, and tone.

[0373] "Keywords" refer to the main words or phrases in the post content and are an important source of information for analysis.

[0374] "Context" refers to the relationship or background information in which a keyword or phrase is used.

[0375] "Tone" refers to the overall sentiment or intent of a post (e.g., positive, negative, neutral).

[0376] A "rule-based checklist" is a list for evaluating analysis results based on predefined conditions or rules.

[0377] "Evaluation" is the process of checking the analysis results against a rule-based checklist to determine whether the content is appropriate.

[0378] A "machine learning model" is an algorithm that learns patterns and rules based on past data and makes predictions and analyses on new data.

[0379] "Flame risk" refers to the degree to which a post's content, if made public, is likely to provoke a negative reaction among users.

[0380] The "emotion engine" is a system that analyzes the type and intensity of emotions contained in posts and provides feedback based on that analysis.

[0381] "Scoring" is the process of quantifying evaluation and prediction results to quantitatively indicate the degree of risk.

[0382] "Feedback" refers to advice and suggestions for corrections provided to users based on analysis and evaluation results.

[0383] "Suggested fixes" are specific suggestions for improvements provided to users to make their posts safe and appropriate.

[0384] This invention is a system that analyzes user posts on social media and advertising platforms in real time to identify and prevent the risk of online outrage. This system uses an emotion engine, especially for advertising content, to score the risk of online outrage and provide feedback.

[0385] System Configuration

[0386] The system consists of the following main components:

[0387] 1. Terminal

[0388] 2. Server

[0389] 3. Natural Language Processing (NLP) Engine

[0390] 4. Emotion Engine

[0391] 5. Machine Learning Models

[0392] 6. Database

[0393] Hardware and software used

[0394] Device: The device (e.g., smartphone, tablet, or computer) through which a user enters their post.

[0395] Server: A central processing unit for receiving and analyzing submissions and generating feedback.

[0396] Natural Language Processing (NLP) engine: Analyzes text for keywords, context, and tone using libraries such as TextBlob.

[0397] Emotion engine: Software for analyzing the type and intensity of emotions contained in posts.

[0398] Machine learning model: Predicts the risk of a backlash based on past data.

[0399] Database: Stores submitted content and analysis results.

[0400] Details of data processing and calculation

[0401] Receiving posted content

[0402] Users use their devices to input content and send it to the server. Content can be in a variety of formats, including text, images, and videos.

[0403] Natural Language Processing (NLP)

[0404] The server passes the content received from the device to a natural language processing engine, which extracts information about keywords, context, and tone.

[0405] Applying the Emotion Engine

[0406] The server uses an emotion engine to evaluate the emotions contained in the advertising content and analyze their type and intensity.

[0407] Using machine learning models

[0408] The server uses a machine learning model to score the risk of a post causing an uproar based on past examples of uproars and posting patterns.

[0409] Providing Feedback

[0410] Based on the analysis results and the flame risk score, the server generates and provides feedback to the user, which may include suggested modifications.

[0411] Specific examples

[0412] Below are examples of advertising content and corresponding prompt sentences.

[0413] Examples:

[0414] Let's say an advertising agency creates the following advertising content:

[0415] "This product is really no good. Don't buy it!"

[0416] Example prompt sentence:

[0417] plaintext

[0418] Ad content: This product is really bad. Don't buy it!

[0419] Is this content potentially controversial? What's your sentiment rating and suggested changes?

[0420] Example output:

[0421] plaintext

[0422] This content has a high risk of causing controversy (controversy score: 85%, sentiment: negative). We recommend changing the wording to something like, "This product has some improvements. Please check out our other products."

[0423] In this way, the present invention realizes a system that detects the risk of a controversy in advance and provides appropriate correction suggestions, allowing advertising agencies to post advertising content with peace of mind.

[0424] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0425] Step 1:

[0426] The user inputs advertising content using the device. The input content can be in various formats such as text, images, videos, etc. After inputting, the user clicks the post button, and the device sends the posted content to the server.

[0427] Step 2:

[0428] The server passes the content of the post received from the device to a natural language processing (NLP) engine. The NLP engine analyzes the text data and extracts information about keywords, context, and tone. For example, if a post such as "This product is really no good" is entered, the keywords "product" and "no good" are extracted, and the negative tone is identified through contextual analysis.

[0429] Step 3:

[0430] The server passes the extracted keywords, context, and tone information to an emotion engine, which analyzes the type (e.g., anger, sadness) and intensity of the emotion contained in the post. The analysis results indicate that the emotion is "very negative."

[0431] Step 4:

[0432] The server evaluates the results of the analysis against a predefined rule-based checklist, for example, checking for the presence of offensive words like "bad" or "terrible," and then makes an initial assessment of the appropriateness of the post.

[0433] Step 5:

[0434] The server uses a machine learning model to predict the risk of a post causing an uproar. The machine learning model has learned from past cases of uproars and posting patterns, and also uses the analysis results as additional information. At this step, the server predicts an "85% risk of uproar."

[0435] Step 6:

[0436] The server generates feedback based on the analysis and prediction results and provides it to the user. The feedback includes problematic expressions and suggestions for rephrasing. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'This product has some improvements. Please check out our other products.'"

[0437] Step 7:

[0438] The user checks the feedback and chooses whether to accept the suggested revisions or keep the original post. If the user accepts the revisions, the post is updated and sent back to the server via the device.

[0439] Step 8:

[0440] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0441] 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.

[0442] 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.

[0443] 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.

[0444] [Second embodiment]

[0445] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0446] 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.

[0447] 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).

[0448] 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.

[0449] 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.

[0450] 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).

[0451] 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.

[0452] 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.

[0453] 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.

[0454] 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.

[0455] 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.

[0456] 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."

[0457] The present invention is a system that analyzes user's SNS postings in real time and identifies and prevents the risk of flame wars in advance. The system is implemented using the following specific means.

[0458] System Configuration

[0459] The system consists of the following main components:

[0460] 1. Terminal

[0461] 2. Server

[0462] 3. Natural Language Processing (NLP) Engine

[0463] 4. Machine Learning Models

[0464] 5. Database

[0465] Processing Flow

[0466] 1. Enter and submit your post

[0467] The device inputs user posts and receives them through a social networking app or web interface. For example, a user might post something like, "These services are really no good these days!" The device then generates and sends an HTTP request to send this post to the server.

[0468] 2. Analysis of posted content

[0469] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[0470] Keyword analysis: Extracting important words and phrases within posts. Examples: "no" and "really."

[0471] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[0472] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0473] 3. Rule-based checks

[0474] The server evaluates the output of the NLP engine based on a rule-based checklist, for example determining whether "no" or "really" is offensive.

[0475] 4. Application of the fire prediction model

[0476] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. This model learns from past cases of uproars and posting patterns, and provides a risk score for the post. For example, "The risk of uproar is 70%."

[0477] 5. Providing real-time feedback

[0478] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0479] 6. Corrections and resubmissions

[0480] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post (e.g., "I feel there is room for improvement in the recent service"), and the device resends the final post to the server.

[0481] 7. Save and publish your final post

[0482] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0483] The above is a specific embodiment of the present invention. This system allows users to understand in advance the risk of their posts causing a controversy and make appropriate corrections, thereby enabling them to maintain healthy communication.

[0484] The processing flow will be explained below.

[0485] Step 1:

[0486] The terminal receives user posts via a text area or form. For example, the user might write, "These services are really no good these days!"

[0487] Step 2:

[0488] The device constructs the input post content as an HTTP request and sends this request to the server, including the user ID and timestamp.

[0489] Step 3:

[0490] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[0491] Step 4:

[0492] The server's NLP engine analyzes the post, specifically:

[0493] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0494] Contextual analysis: Understand the overall context of a post. Analyze the relationships between words.

[0495] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0496] Step 5:

[0497] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[0498] Step 6:

[0499] The server uses a machine learning model to predict the risk of a post causing an uproar. The model learns from past cases of uproars and posting patterns, and assigns a risk score to the post (e.g., a 70% risk of causing an uproar).

[0500] Step 7:

[0501] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0502] Step 8:

[0503] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[0504] Step 9:

[0505] The user reviews the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[0506] Step 10:

[0507] The device resends the modified or original post in its final form to the server, which receives the final post.

[0508] Step 11:

[0509] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0510] Example 1

[0511] 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."

[0512] In recent years, the impact of social media posts has become increasingly large, and inappropriate posts by users often have widespread repercussions in the form of flame wars. Therefore, there is a strong demand for methods that allow users to understand in advance whether their posts pose a risk of causing a flame war and to appropriately avoid that risk. However, current systems have the problem of insufficient support for users to quickly and accurately evaluate the flame war risk of their posts and make appropriate corrections.

[0513] 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.

[0514] In this invention, the server includes means for receiving posts entered by users, means for analyzing the posts using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the posts causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation and prediction results, means for the user to review the proposed revisions and choose whether to accept or ignore them, and means for receiving the revised posts again, saving them, and publishing them on the SNS platform. This allows users to understand the risk of their posts causing an uproar in advance and make appropriate revisions, thereby significantly reducing the risk of an uproar.

[0515] "User" refers to an individual or corporation that uses this system to post on SNS.

[0516] A "terminal" is a device used by a user, and includes a computer, a smartphone, a tablet, and the like.

[0517] "Server" refers to the computer system that receives, analyzes, provides feedback on, stores, and publishes Submissions.

[0518] "Posted content" refers to the text information posted by a user on a social media platform.

[0519] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[0520] "Keywords" refer to important words or phrases extracted from the content of a post.

[0521] "Context" refers to the relevance of the meaning of a word or phrase within the overall content of a post.

[0522] "Tone" refers to facial expressions that convey the emotion and intent of the post.

[0523] A "rule-based checklist" is a list for evaluating posts based on predefined criteria.

[0524] A "machine learning model" refers to an algorithm that learns from past data and makes predictions and classifications for new data.

[0525] "Flame risk" refers to the possibility that the content of a post will provoke a negative reaction.

[0526] "Feedback" refers to advice and suggestions provided to users based on analysis and prediction results.

[0527] "Suggested Revisions" means suggestions or alternatives for a User to revise a Submission.

[0528] "SNS Platform" refers to an online service that allows users to publish and share their posts with other users.

[0529] "Storage" refers to recording the posted content in a database or storage device.

[0530] "Public" means displaying the posted content on the social media platform so that other users can view it.

[0531] This invention is a system that analyzes users' social media posts in real time to identify and prevent the risk of flame wars in advance. This system receives the content posted by users and analyzes, evaluates, and provides feedback using a natural language processing (NLP) engine and machine learning model, making it possible to maintain healthy communication.

[0532] The main hardware and software components of this system are:

[0533] The device used by the user (computer, smartphone, tablet, etc.)

[0534] Server (responsible for receiving, analyzing, providing feedback on, storing and publishing posted content)

[0535] Natural Language Processing (NLP) Engine

[0536] Machine learning models

[0537] Database

[0538] First, a user uses a device to input content for a post via a social networking app or web interface. For example, the user might input "These services are really no good these days!" The device then generates this content as an HTTP request and sends it to the server.

[0539] The server then passes the content received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts information about keywords, context, and tone. Specifically, it identifies keywords like "no" and "really," analyzes the meaning of the entire sentence, and calculates a negative tone.

[0540] The server then evaluates the output of the NLP engine based on a rule-based checklist. The parsed keywords and tones are compared against the rule-based list to determine whether they are offensive. For example, words like "no" and "really" are considered offensive.

[0541] Next, the server uses a machine learning model to predict whether the post has a risk of causing a controversy. This model learns from past cases of controversy and posting patterns. It calculates a controversy risk score for the post and outputs, for example, "Controversy risk is 70%."

[0542] The server provides feedback to the user based on the evaluation and prediction results. This feedback includes specific suggestions for rephrasing or correction. For example, it may provide a suggestion such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0543] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post. For example, they might revise the post to say, "I feel there is room for improvement in the recent service." The device then sends this revised post back to the server.

[0544] Finally, the server saves the modified post to a database and publishes it on the social media platform, so that it appears on the timeline and can be viewed by other users.

[0545] For example, the following prompt could be fed to a generative AI model: "Please use an NLP engine to analyze social media posts entered by users and detect offensive language. Also, please suggest modifications to reduce this risk."

[0546] This system allows users to understand in advance the risk of their posts causing a backlash and make appropriate corrections, significantly reducing the risk of a backlash.

[0547] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0548] Step 1:

[0549] A user uses a device to enter content to post via a social networking app or web interface and presses the "Send" button. For example, a user enters a post such as "These services these days are really no good!" The device generates the entered content as an HTTP request and sends this request to the server. The input is the text data posted by the user, and the output is an HTTP request to the server.

[0550] Step 2:

[0551] The server passes the content of posts received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts the following information: the input is the received post data, and the output is extracted keywords, context, and tone information.

[0552] Keyword analysis: Identify important words and phrases within your posts. Examples include "no" and "really."

[0553] Contextual analysis: Analyzing the relationships between words to understand the meaning of the whole sentence. For example, the context of "Dameda."

[0554] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0555] Step 3:

[0556] The server evaluates the output of the NLP engine based on a rule-based checklist. The input is the analyzed data, and the output is the evaluation result. Specifically, it checks the analyzed keywords and tones against the rule-based list to see if they are offensive. For example, "no" and "really" are considered offensive.

[0557] Step 4:

[0558] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. The input is the rule-based evaluation result, and the output is a score for the risk of causing an uproar. Specifically, it uses a model that has learned from past cases of uproars and posting patterns to calculate the risk score for the post. Example: "The risk of causing an uproar is 70%."

[0559] Step 5:

[0560] The server provides feedback to the user based on the analysis and prediction results. The input is the flame risk score, and the output is the content of the feedback. For example, the following suggestion is generated as a specific action: "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0561] Step 6:

[0562] The user reviews the proposed revisions and chooses whether to accept or ignore them. The input is the feedback content, and the output is the revised post content or the original post content. Specifically, if the user accepts the revisions, the post content is updated (e.g., "I feel there is room for improvement in the recent service"). The device then sends the revised post content back to the server.

[0563] Step 7:

[0564] The server saves the final post in a database and publishes it on the social media platform. The input is the final post, and the output is saving it to the database and publishing it. Specifically, the revised post is saved in the database and displayed on the social media platform's timeline for other users to view.

[0565] (Application example 1)

[0566] 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."

[0567] In modern advertising campaigns, posting text to digital platforms such as social media is essential. However, posts containing incorrect expressions or a negative tone can damage a company's brand image and risk sparking outrage. In particular, in an age where past posts are easily shared, negative reactions can spread instantly. Therefore, there is a need for a system that can proactively identify advertising copy that may spark outrage and suggest revisions.

[0568] 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.

[0569] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and prediction results, and means for analyzing the risk of advertising copy in the post content causing an uproar and proposing revisions. This allows users to understand the risk of uproar in the post content of their advertising campaigns in advance and make appropriate revisions, thereby enabling safe and effective advertising.

[0570] "User" refers to a person or organization that uses the system to input and post advertising copy.

[0571] "Post content" refers to the text entered by a user and intended to be published on social media or digital platforms.

[0572] "Natural language processing" refers to the technology that enables computers to understand and analyze human language.

[0573] "Keywords" refers to important words and phrases extracted from the content of a post.

[0574] "Context" refers to information that allows us to understand the meaning of the entire sentence through the meaning and relationships of each word and phrase.

[0575] "Tone" refers to information that indicates the emotional nuance or attitude of a piece of writing.

[0576] A "rule-based checklist" refers to a list for evaluating posts based on predefined evaluation criteria.

[0577] A "machine learning model" refers to an algorithm that learns specific tasks based on past data and makes predictions and evaluations for new data.

[0578] "Flame risk" refers to an indicator of the likelihood that a post will provoke a negative reaction.

[0579] "Feedback" refers to information provided to a user, including analysis results and suggested modifications.

[0580] "Proposed amendments" refer to proposed changes to the content of a post to reduce the risk of a backlash.

[0581] This invention is a system that analyzes the content of advertising text in real time before users post it on social media or digital platforms, and identifies and prevents the risk of online outrage in advance. The system includes the following main components and processing flow:

[0582] System Configuration

[0583] The system consists of the following main components:

[0584] 1. Terminal: The device on which the user enters advertising text.

[0585] 2. Server: The central computer that performs analysis and predictions.

[0586] 3. Natural Language Processing (NLP) engine: Software that analyzes the content of posts.

[0587] 4. Machine learning model: An algorithm for predicting the risk of a firestorm.

[0588] 5. Database: An information repository that stores past posts and analysis results.

[0589] Explanation of program processing

[0590] The server first passes the ad copy received from the user device to a natural language processing (NLP) engine, which performs the following analysis:

[0591] Keyword analysis: Extracting important words and phrases within ad copy.

[0592] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[0593] Tone analysis: Calculating a sentiment score for text to infer sentiment and intent.

[0594] The server then evaluates the output of the NLP engine based on a rule-based checklist, determining, for example, that the phrase "better than any other product" is offensive.

[0595] The server then uses a machine learning model to predict the risk of the ad copy causing an uproar. The model learns from past cases of uproar and posting patterns, and provides an uproar risk score for the ad copy.

[0596] Based on the results, the server will provide feedback to the user, such as "This expression may provoke a negative reaction. We recommend changing it to something like 'Many customers are satisfied.'"

[0597] The user can then review the proposed revisions and choose to accept or ignore them. If the user accepts the revisions, the ad copy is updated and the device sends the final copy back to the server. The server then stores the final copy in a database and publishes it on the social media platform.

[0598] Specific examples

[0599] For example, if a user enters the ad copy, "This product is better than any other product," the server will analyze that expression and determine that there is a high risk of a negative reaction. Therefore, the server will provide feedback such as, "This expression may cause a negative reaction. We recommend changing it to something like, 'Many customers are satisfied with this product.'"

[0600] Example prompt sentence:

[0601] Please analyze whether this ad copy poses a risk of causing controversy. Also, please explain why and suggest modifications.

[0602] Hardware and software used

[0603] Hardware: General PCs and smartphones

[0604] Software: nltk (natural language processing library), sklearn (machine learning model library)

[0605] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0606] Step 1:

[0607] To receive the ad copy entered by the user, the device inputs the ad copy from the user and receives it through a social networking app or web interface. For example, the user may input ad copy such as "This product is better than any other product." The user inputs text data through a browser or mobile app, and the data is saved on the device.

[0608] Step 2:

[0609] The terminal generates and sends an HTTP request to send the entered ad copy to the server. The sent data is the ad copy in text format. After the server receives this data, it saves it as text data.

[0610] Step 3:

[0611] The server passes the received ad copy to a natural language processing (NLP) engine. The input is the text data received in step 2. The NLP engine analyzes the data to extract keywords, context, and tone information from the ad copy. Specifically, it extracts keywords, analyzes the context of the text, and calculates a sentiment score. The output is the analysis results (keyword list, context information, and tone score).

[0612] Step 4:

[0613] Based on the analysis results, the server performs an evaluation according to a predefined rule-based checklist. The analysis results obtained in step 3 are used as input. The rule-based evaluation determines whether certain keywords or tones pose a risk. For example, the keyword "better than any other product" is evaluated as offensive. The evaluation results (a list of problematic keywords and tone analysis results) are generated as output.

[0614] Step 5:

[0615] The server uses a machine learning model to predict the risk of the ad copy causing an uproar. The evaluation results obtained in step 4 are used as input. The machine learning model has learned from past cases of uproars and posting patterns, and calculates a risk score for new data. Specifically, the text data is vectorized, and the vector is input into the model for scoring. The output is an uproar risk score.

[0616] Step 6:

[0617] The server provides feedback to the user based on the evaluation results and risk score. The evaluation results and the flame risk score are used as inputs. The server generates feedback for the user, including problematic expressions and recommended corrections. For example, the server generates feedback such as, "This expression may provoke negative reactions. We recommend changing it to something like, 'Many customers are satisfied.'" The server generates a feedback message as output.

[0618] Step 7:

[0619] The server sends the generated feedback message to the terminal and displays it to the user. The feedback message generated in step 6 is used as input. The user checks the feedback message and chooses whether to accept or ignore the proposed revision. For example, if the user accepts the proposed revision, the revised advertisement copy is displayed.

[0620] Step 8:

[0621] If the user accepts the proposed revision, the device sends the revised copy to the server again. The revised copy is used as input. The server stores the received final copy in a database and publishes it on the social networking platform. The published copy is generated as output.

[0622] 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.

[0623] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[0624] System Configuration

[0625] The system consists of the following main components:

[0626] 1. Terminal

[0627] 2. Server

[0628] 3. Natural Language Processing (NLP) Engine

[0629] 4. Emotion Engine

[0630] 5. Machine Learning Models

[0631] 6. Database

[0632] Processing Flow

[0633] 1. Enter and submit your post

[0634] The device receives user submissions via a text area or form. For example, the user might enter a submission such as "These services are really no good these days!" The device then generates and sends an HTTP request to the server to send the submission.

[0635] 2. Analysis of posted content

[0636] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[0637] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0638] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[0639] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0640] 3. Running the Emotion Engine

[0641] The server uses an emotion engine to analyze the type and intensity of emotions (e.g., joy, anger, sadness) contained in the user's post. For example, it may determine that "this post has a strong emotion of anger."

[0642] 4. Rule-based checks

[0643] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[0644] 5. Application of the Fire Prediction Model

[0645] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[0646] 6. Providing real-time feedback

[0647] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[0648] 7. Corrections and resubmissions

[0649] The user checks the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent back to the server via the device. For example, change it to "I feel there is room for improvement in the recent service."

[0650] 8. Save and publish your final post

[0651] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0652] The above is a specific embodiment of the present invention. By incorporating an emotion engine, it is possible to grasp user emotions in detail and improve the accuracy of predicting the risk of a controversy. This system allows users to understand the risk of a post causing a controversy in advance and make appropriate corrections, thereby maintaining healthy communication.

[0653] The processing flow will be explained below.

[0654] Step 1:

[0655] The terminal receives user posts via a text area or form. For example, the user may post, "These services are really no good these days!"

[0656] Step 2:

[0657] The device constructs an HTTP request containing the entered post content and sends this request to the server, including the user ID and timestamp.

[0658] Step 3:

[0659] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[0660] Step 4:

[0661] The server's NLP engine analyzes the post, specifically:

[0662] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0663] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[0664] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0665] Step 5:

[0666] The server uses an emotion engine to analyze the type of emotion (e.g., joy, anger, sadness) and intensity of the emotion contained in the post. For example, it may determine that "this post contains strong anger."

[0667] Step 6:

[0668] The server evaluates the results of the analysis using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate. For example, checking if words like "bad" or "terrible" are present in the list.

[0669] Step 7:

[0670] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[0671] Step 8:

[0672] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0673] Step 9:

[0674] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[0675] Step 10:

[0676] The user checks the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[0677] Step 11:

[0678] The device resends the modified or original post in its final form to the server, which receives the final post.

[0679] Step 12:

[0680] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0681] Example 2

[0682] 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."

[0683] Conventional SNS systems lacked sufficient filtering and moderation functions to determine whether user posts were appropriate, resulting in a high risk of unexpected outrage. Furthermore, when a user unintentionally used offensive language, the feedback function was insufficient, making it difficult to prompt the user to correct the post. To solve these problems, a system that analyzes user posts in more detail and provides real-time feedback is needed.

[0684] 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 post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for using an emotion engine to analyze the type and intensity of emotion contained in the post content, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned past cases of uproars and posting patterns, and means for providing feedback to the user based on the evaluation results and prediction results. This makes it possible to grasp the risk of a user's post content causing an uproar in advance and provide appropriate feedback in real time.

[0685] "User" means any individual or entity that uses a social media platform to enter and submit posts.

[0686] "Terminal" refers to an information processing device used by a user, such as a computer, smartphone, or tablet.

[0687] "Server" refers to a computer system that has the function of analyzing and evaluating the content posted by users and providing feedback.

[0688] "Natural language processing" refers to the technology of using computers to analyze human language, extract information, and understand meaning.

[0689] "Keyword analysis" refers to the technology of extracting important words and phrases from the content of posts.

[0690] "Contextual analysis" refers to a technology that understands the relationships between words in a sentence and grasps the overall meaning of the post.

[0691] "Tone analysis" refers to the technique of calculating an emotional score to infer the sentiment or intent of a piece of text.

[0692] "Emotion engine" refers to software or hardware for analyzing the type and intensity of emotions contained in posted content.

[0693] A "rules-based checklist" is a list for evaluating posts based on predefined keywords and phrases.

[0694] A "machine learning model" refers to a computational model that uses algorithms to learn patterns and rules based on data.

[0695] "Flame risk prediction" refers to a technology that predicts the risk of a post causing a flame war based on past flame war cases and posting patterns.

[0696] "Means for providing feedback" refers to a function that provides appropriate advice and suggestions for correction to the user in real time based on the analysis and prediction results.

[0697] "Suggestions to modify post content" refers to suggestions to change a user's post to an appropriate expression when the post is deemed offensive or inappropriate.

[0698] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[0699] System Configuration

[0700] The system consists of the following main components:

[0701] 1. Devices (computers, smartphones, tablets, etc.)

[0702] 2. Server (AWS or on-premise)

[0703] 3. Natural Language Processing (NLP) engines (e.g., SpaCy, nltk)

[0704] 4. Emotion engines (e.g., IBM Watson, Microsoft Text Analytics)

[0705] 5. Machine learning models (e.g., custom models based on TensorFlow or Scikit-Learn)

[0706] 6. Database (e.g. MySQL, PostgreSQL)

[0707] Function details

[0708] 1. Enter and submit your post

[0709] A user uses a device to enter text into a posting field on a social networking site. For example, the user enters the content of a post, such as "The services these days are really no good!" The device generates an HTTP request to send this content to the server, and sends it to the server.

[0710] 2. Analysis of posted content

[0711] The server passes the content received from the device to a natural language processing (NLP) engine, which then performs the following tasks:

[0712] Keyword analysis: Extracting important words and phrases, such as "no" and "really."

[0713] Contextual analysis: Understand the overall context of the post and analyze the relationships between words.

[0714] Tone analysis: Infers the sentiment and intent of a post and calculates a sentiment score. For example, it may be determined to have a "very negative tone."

[0715] 3. Running the Emotion Engine

[0716] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post. For example, it may determine that "this post contains a strong emotion of anger."

[0717] 4. Rule-based checks

[0718] The server uses a rule-based checklist to evaluate the results of the analysis, checking whether certain keywords or phrases are offensive or inappropriate, for example, whether words like "bad" or "terrible" exist in the checklist.

[0719] 5. Application of the Fire Prediction Model

[0720] The server uses a machine learning model to predict the risk of a post causing an uproar. This model learns from past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine. For example, it may assess the risk of a post causing an uproar as 70%.

[0721] 6. Providing real-time feedback

[0722] The server generates the analysis results and proposed modifications and sends them to the device. The device then displays feedback to the user. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[0723] 7. Corrections and resubmissions

[0724] The user reviews the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent again to the server via the device. For example, the user could change the post to "I feel there is room for improvement in the recent service."

[0725] 8. Save and publish your final post

[0726] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0727] Specific examples in the description

[0728] A user posts on social media, "These services are really no good!" This post is analyzed by the system and the following feedback is provided:

[0729] Keyword analysis determines "no" as offensive

[0730] The sentiment engine analyzes the tone of the post as "very negative."

[0731] The flame war prediction model predicts a "70% risk of flame war"

[0732] The server sends feedback to the device saying, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there are areas for improvement in your recent service.'"

[0733] Prompt Sentence Examples

[0734] "Please provide a detailed explanation of the specific processing steps involved in the system for analyzing the sentiment of social media posts and predicting the risk of a social media outcry. Furthermore, please provide specific examples of how this system provides feedback to users."

[0735] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0736] Step 1:

[0737] The user enters the content to post and submits it.

[0738] Specific actions: The user uses a device to enter the text "These services are really no good!" into the post field of a social networking site and clicks the send button.

[0739] Input: User-entered post content (e.g., "These services are really no good these days!")

[0740] Output: HTTP request generated by the device

[0741] Step 2:

[0742] The server receives the post and passes it to a natural language processing (NLP) engine.

[0743] Specific operation: The server receives the HTTP request generated by the device and passes the text data in the request to the NLP engine.

[0744] Input: HTTP request sent from the terminal

[0745] Output: Passing text data to the NLP engine

[0746] Step 3:

[0747] A natural language processing (NLP) engine analyzes posts to extract keywords, context, and tone of voice.

[0748] Specific operation: The NLP engine extracts important keywords such as "bad" and "really" from the text "These services these days are really no good!", performs contextual analysis, and then calculates an emotional score to determine the sentence as "negative."

[0749] Input: Text data of the post content

[0750] Output: Keyword, context, and tone analysis results (e.g., "Keyword: No, really," "Context: Negative," "Tone: Very negative")

[0751] Step 4:

[0752] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post.

[0753] Specific operation: The analysis results from the NLP engine are passed to the emotion engine, which analyzes the type and intensity of emotions, such as "anger" or "sadness."

[0754] Input: Analysis results of the NLP engine

[0755] Output: Analysis results of the emotion engine (e.g., "Emotion type: anger" and "Intensity: strong")

[0756] Step 5:

[0757] The server evaluates the submission using a rule-based checklist.

[0758] What it does: Parsed keywords are checked against a predefined rule-based checklist to see if they contain offensive words or phrases.

[0759] Input: Analysis results of NLP engine, analysis results of emotion engine

[0760] Output: Rule-based check results (e.g., "Aggressive: No (exists)", "Non-aggressive: Really (does not exist)", "Overall rating: Aggressive")

[0761] Step 6:

[0762] The server uses a machine learning model to predict the risk of a post causing an uproar.

[0763] Specific operation: The analysis results and the results of the emotion engine are input into a machine learning model, which then calculates the risk of a controversy based on past cases of controversy and posting patterns.

[0764] Input: Analysis results, emotion engine results

[0765] Output: Flame risk prediction (e.g., "Flame risk: 70%")

[0766] Step 7:

[0767] The server generates analysis results and suggested corrections and sends them to the device.

[0768] Specific operation: The server generates a proposal to modify the post content based on the results of the flame war risk prediction and sends it to the terminal as an HTTP response.

[0769] Input: Flame risk prediction results

[0770] Output: Generates a suggested fix and an HTTP response (e.g., "Suggested fix: This expression may be considered offensive. We recommend changing it to something like, 'We feel your recent service has room for improvement.'")

[0771] Step 8:

[0772] The user checks the feedback and selects whether to make corrections.

[0773] Specific operation: The device displays the feedback received from the server to the user, and the user decides whether to accept the suggested revisions or continue posting as is.

[0774] Input: Feedback from the server

[0775] Output: User's choice (e.g. "Accept proposed revision" or "Keep original post")

[0776] Step 9:

[0777] The server stores the final post in a database and publishes it on the social media platform.

[0778] Specific operation: The user resubmits the revised or original post to the server, which then stores the post in its database and publishes it on the social media platform.

[0779] Input: User's last post

[0780] Output: Saving to a database and publishing on a social media platform (e.g. "Saved to database" and "Published on social media")

[0781] (Application example 2)

[0782] 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."

[0783] Conventional systems for analyzing the content of posts on social media platforms are able to provide appropriate feedback by gaining a detailed understanding of user emotions and predicting the risk of a controversy in advance. However, advertising content created by advertising agencies in particular has a higher risk of causing a controversy than general user posts, and could potentially damage a company's brand image. Current systems lack specialized analysis and feedback for advertising content, and there is a need for technology that automatically provides appropriate revision suggestions to allow advertising agencies to post with confidence.

[0784] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0785] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and the prediction results, means for evaluating the emotions of advertising content using an emotion engine and scoring the risk of uproar, and means for providing suggestions for modifying the advertising content based on the scoring results. This enables advertising agencies to thoroughly evaluate the risk of uproaring advertising content before posting it and automatically receive appropriate suggestions for modification.

[0786] A "user" is a user who enters content to post on a social networking site or advertising platform.

[0787] "Posted content" refers to content such as text, images, and videos that users enter and submit to social media or advertising platforms.

[0788] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[0789] "Analysis" is the process of examining input data (in this case, posts) in detail and extracting information such as keywords, context, and tone.

[0790] "Keywords" refer to the main words or phrases in the post content and are an important source of information for analysis.

[0791] "Context" refers to the relationship or background information in which a keyword or phrase is used.

[0792] "Tone" refers to the overall sentiment or intent of a post (e.g., positive, negative, neutral).

[0793] A "rule-based checklist" is a list for evaluating analysis results based on predefined conditions or rules.

[0794] "Evaluation" is the process of checking the analysis results against a rule-based checklist to determine whether the content is appropriate.

[0795] A "machine learning model" is an algorithm that learns patterns and rules based on past data and makes predictions and analyses on new data.

[0796] "Flame risk" refers to the degree to which a post's content, if made public, is likely to provoke a negative reaction among users.

[0797] The "emotion engine" is a system that analyzes the type and intensity of emotions contained in posts and provides feedback based on that analysis.

[0798] "Scoring" is the process of quantifying evaluation and prediction results to quantitatively indicate the degree of risk.

[0799] "Feedback" refers to advice and suggestions for corrections provided to users based on analysis and evaluation results.

[0800] "Suggested fixes" are specific suggestions for improvements provided to users to make their posts safe and appropriate.

[0801] This invention is a system that analyzes user posts on social media and advertising platforms in real time to identify and prevent the risk of online outrage. This system uses an emotion engine, especially for advertising content, to score the risk of online outrage and provide feedback.

[0802] System Configuration

[0803] The system consists of the following main components:

[0804] 1. Terminal

[0805] 2. Server

[0806] 3. Natural Language Processing (NLP) Engine

[0807] 4. Emotion Engine

[0808] 5. Machine Learning Models

[0809] 6. Database

[0810] Hardware and software used

[0811] Device: The device (e.g., smartphone, tablet, or computer) through which a user enters their post.

[0812] Server: A central processing unit for receiving and analyzing submissions and generating feedback.

[0813] Natural Language Processing (NLP) engine: Analyzes text for keywords, context, and tone using libraries such as TextBlob.

[0814] Emotion engine: Software for analyzing the type and intensity of emotions contained in posts.

[0815] Machine learning model: Predicts the risk of a backlash based on past data.

[0816] Database: Stores submitted content and analysis results.

[0817] Details of data processing and calculation

[0818] Receiving posted content

[0819] Users use their devices to input content and send it to the server. Content can be in a variety of formats, including text, images, and videos.

[0820] Natural Language Processing (NLP)

[0821] The server passes the content received from the device to a natural language processing engine, which extracts information about keywords, context, and tone.

[0822] Applying the Emotion Engine

[0823] The server uses an emotion engine to evaluate the emotions contained in the advertising content and analyze their type and intensity.

[0824] Using machine learning models

[0825] The server uses a machine learning model to score the risk of a post causing an uproar based on past examples of uproars and posting patterns.

[0826] Providing Feedback

[0827] Based on the analysis results and the flame risk score, the server generates and provides feedback to the user, which may include suggested modifications.

[0828] Specific examples

[0829] Below are examples of advertising content and corresponding prompt sentences.

[0830] Examples:

[0831] Let's say an advertising agency creates the following advertising content:

[0832] "This product is really no good. Don't buy it!"

[0833] Example prompt sentence:

[0834] plaintext

[0835] Ad content: This product is really bad. Don't buy it!

[0836] Is this content potentially controversial? What's your sentiment rating and suggested changes?

[0837] Example output:

[0838] plaintext

[0839] This content has a high risk of causing controversy (controversy score: 85%, sentiment: negative). We recommend changing the wording to something like, "This product has some improvements. Please check out our other products."

[0840] In this way, the present invention realizes a system that detects the risk of a controversy in advance and provides appropriate correction suggestions, allowing advertising agencies to post advertising content with peace of mind.

[0841] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0842] Step 1:

[0843] The user inputs advertising content using the device. The input content can be in various formats such as text, images, videos, etc. After inputting, the user clicks the post button, and the device sends the posted content to the server.

[0844] Step 2:

[0845] The server passes the content of the post received from the device to a natural language processing (NLP) engine. The NLP engine analyzes the text data and extracts information about keywords, context, and tone. For example, if a post such as "This product is really no good" is entered, the keywords "product" and "no good" are extracted, and the negative tone is identified through contextual analysis.

[0846] Step 3:

[0847] The server passes the extracted keywords, context, and tone information to an emotion engine, which analyzes the type (e.g., anger, sadness) and intensity of the emotion contained in the post. The analysis results indicate that the emotion is "very negative."

[0848] Step 4:

[0849] The server evaluates the results of the analysis against a predefined rule-based checklist, for example, checking for the presence of offensive words like "bad" or "terrible," and then makes an initial assessment of the appropriateness of the post.

[0850] Step 5:

[0851] The server uses a machine learning model to predict the risk of a post causing an uproar. The machine learning model has learned from past cases of uproars and posting patterns, and also uses the analysis results as additional information. At this step, the server predicts an "85% risk of uproar."

[0852] Step 6:

[0853] The server generates feedback based on the analysis and prediction results and provides it to the user. The feedback includes problematic expressions and suggestions for rephrasing. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'This product has some improvements. Please check out our other products.'"

[0854] Step 7:

[0855] The user checks the feedback and chooses whether to accept the suggested revisions or keep the original post. If the user accepts the revisions, the post is updated and sent back to the server via the device.

[0856] Step 8:

[0857] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0858] 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.

[0859] 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.

[0860] 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.

[0861] [Third embodiment]

[0862] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0863] 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.

[0864] 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).

[0865] 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.

[0866] 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.

[0867] 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).

[0868] 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.

[0869] 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.

[0870] 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.

[0871] 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.

[0872] 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.

[0873] 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."

[0874] The present invention is a system that analyzes user's SNS postings in real time and identifies and prevents the risk of flame wars in advance. The system is implemented using the following specific means.

[0875] System Configuration

[0876] The system consists of the following main components:

[0877] 1. Terminal

[0878] 2. Server

[0879] 3. Natural Language Processing (NLP) Engine

[0880] 4. Machine Learning Models

[0881] 5. Database

[0882] Processing Flow

[0883] 1. Enter and submit your post

[0884] The device inputs user posts and receives them through a social networking app or web interface. For example, a user might post something like, "These services are really no good these days!" The device then generates and sends an HTTP request to send this post to the server.

[0885] 2. Analysis of posted content

[0886] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[0887] Keyword analysis: Extracting important words and phrases within posts. Examples: "no" and "really."

[0888] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[0889] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0890] 3. Rule-based checks

[0891] The server evaluates the output of the NLP engine based on a rule-based checklist, for example determining whether "no" or "really" is offensive.

[0892] 4. Application of the fire prediction model

[0893] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. This model learns from past cases of uproars and posting patterns, and provides a risk score for the post. For example, "The risk of uproar is 70%."

[0894] 5. Providing real-time feedback

[0895] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0896] 6. Corrections and resubmissions

[0897] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post (e.g., "I feel there is room for improvement in the recent service"), and the device resends the final post to the server.

[0898] 7. Save and publish your final post

[0899] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0900] The above is a specific embodiment of the present invention. This system allows users to understand in advance the risk of their posts causing a controversy and make appropriate corrections, thereby enabling them to maintain healthy communication.

[0901] The processing flow will be explained below.

[0902] Step 1:

[0903] The terminal receives user posts via a text area or form. For example, the user might write, "These services are really no good these days!"

[0904] Step 2:

[0905] The device constructs the input post content as an HTTP request and sends this request to the server, including the user ID and timestamp.

[0906] Step 3:

[0907] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[0908] Step 4:

[0909] The server's NLP engine analyzes the post, specifically:

[0910] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[0911] Contextual analysis: Understand the overall context of a post. Analyze the relationships between words.

[0912] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0913] Step 5:

[0914] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[0915] Step 6:

[0916] The server uses a machine learning model to predict the risk of a post causing an uproar. The model learns from past cases of uproars and posting patterns, and assigns a risk score to the post (e.g., a 70% risk of causing an uproar).

[0917] Step 7:

[0918] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0919] Step 8:

[0920] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[0921] Step 9:

[0922] The user reviews the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[0923] Step 10:

[0924] The device resends the modified or original post in its final form to the server, which receives the final post.

[0925] Step 11:

[0926] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[0927] Example 1

[0928] 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."

[0929] In recent years, the impact of social media posts has become increasingly large, and inappropriate posts by users often have widespread repercussions in the form of flame wars. Therefore, there is a strong demand for methods that allow users to understand in advance whether their posts pose a risk of causing a flame war and to appropriately avoid that risk. However, current systems have the problem of insufficient support for users to quickly and accurately evaluate the flame war risk of their posts and make appropriate corrections.

[0930] 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.

[0931] In this invention, the server includes means for receiving posts entered by users, means for analyzing the posts using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the posts causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation and prediction results, means for the user to review the proposed revisions and choose whether to accept or ignore them, and means for receiving the revised posts again, saving them, and publishing them on the SNS platform. This allows users to understand the risk of their posts causing an uproar in advance and make appropriate revisions, thereby significantly reducing the risk of an uproar.

[0932] "User" refers to an individual or corporation that uses this system to post on SNS.

[0933] A "terminal" is a device used by a user, and includes a computer, a smartphone, a tablet, and the like.

[0934] "Server" refers to the computer system that receives, analyzes, provides feedback on, stores, and publishes Submissions.

[0935] "Posted content" refers to the text information posted by a user on a social media platform.

[0936] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[0937] "Keywords" refer to important words or phrases extracted from the content of a post.

[0938] "Context" refers to the relevance of the meaning of a word or phrase within the overall content of a post.

[0939] "Tone" refers to facial expressions that convey the emotion and intent of the post.

[0940] A "rule-based checklist" is a list for evaluating posts based on predefined criteria.

[0941] A "machine learning model" refers to an algorithm that learns from past data and makes predictions and classifications for new data.

[0942] "Flame risk" refers to the possibility that the content of a post will provoke a negative reaction.

[0943] "Feedback" refers to advice and suggestions provided to users based on analysis and prediction results.

[0944] "Suggested Revisions" means suggestions or alternatives for a User to revise a Submission.

[0945] "SNS Platform" refers to an online service that allows users to publish and share their posts with other users.

[0946] "Storage" refers to recording the posted content in a database or storage device.

[0947] "Public" means displaying the posted content on the social media platform so that other users can view it.

[0948] This invention is a system that analyzes users' social media posts in real time to identify and prevent the risk of flame wars in advance. This system receives the content posted by users and analyzes, evaluates, and provides feedback using a natural language processing (NLP) engine and machine learning model, making it possible to maintain healthy communication.

[0949] The main hardware and software components of this system are:

[0950] The device used by the user (computer, smartphone, tablet, etc.)

[0951] Server (responsible for receiving, analyzing, providing feedback on, storing and publishing posted content)

[0952] Natural Language Processing (NLP) Engine

[0953] Machine learning models

[0954] Database

[0955] First, a user uses a device to input content for a post via a social networking app or web interface. For example, the user might input "These services are really no good these days!" The device then generates this content as an HTTP request and sends it to the server.

[0956] The server then passes the content received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts information about keywords, context, and tone. Specifically, it identifies keywords like "no" and "really," analyzes the meaning of the entire sentence, and calculates a negative tone.

[0957] The server then evaluates the output of the NLP engine based on a rule-based checklist. The parsed keywords and tones are compared against the rule-based list to determine whether they are offensive. For example, words like "no" and "really" are considered offensive.

[0958] Next, the server uses a machine learning model to predict whether the post has a risk of causing a controversy. This model learns from past cases of controversy and posting patterns. It calculates a controversy risk score for the post and outputs, for example, "Controversy risk is 70%."

[0959] The server provides feedback to the user based on the evaluation and prediction results. This feedback includes specific suggestions for rephrasing or correction. For example, it may provide a suggestion such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0960] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post. For example, they might revise the post to say, "I feel there is room for improvement in the recent service." The device then sends this revised post back to the server.

[0961] Finally, the server saves the modified post to a database and publishes it on the social media platform, so that it appears on the timeline and can be viewed by other users.

[0962] For example, the following prompt could be fed to a generative AI model: "Please use an NLP engine to analyze social media posts entered by users and detect offensive language. Also, please suggest modifications to reduce this risk."

[0963] This system allows users to understand in advance the risk of their posts causing a backlash and make appropriate corrections, significantly reducing the risk of a backlash.

[0964] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0965] Step 1:

[0966] A user uses a device to enter content to post via a social networking app or web interface and presses the "Send" button. For example, a user enters a post such as "These services these days are really no good!" The device generates the entered content as an HTTP request and sends this request to the server. The input is the text data posted by the user, and the output is an HTTP request to the server.

[0967] Step 2:

[0968] The server passes the content of posts received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts the following information: the input is the received post data, and the output is extracted keywords, context, and tone information.

[0969] Keyword analysis: Identify important words and phrases within your posts. Examples include "no" and "really."

[0970] Contextual analysis: Analyzing the relationships between words to understand the meaning of the whole sentence. For example, the context of "Dameda."

[0971] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[0972] Step 3:

[0973] The server evaluates the output of the NLP engine based on a rule-based checklist. The input is the analyzed data, and the output is the evaluation result. Specifically, it checks the analyzed keywords and tones against the rule-based list to see if they are offensive. For example, "no" and "really" are considered offensive.

[0974] Step 4:

[0975] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. The input is the rule-based evaluation result, and the output is a score for the risk of causing an uproar. Specifically, it uses a model that has learned from past cases of uproars and posting patterns to calculate the risk score for the post. Example: "The risk of causing an uproar is 70%."

[0976] Step 5:

[0977] The server provides feedback to the user based on the analysis and prediction results. The input is the flame risk score, and the output is the content of the feedback. For example, the following suggestion is generated as a specific action: "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[0978] Step 6:

[0979] The user reviews the proposed revisions and chooses whether to accept or ignore them. The input is the feedback content, and the output is the revised post content or the original post content. Specifically, if the user accepts the revisions, the post content is updated (e.g., "I feel there is room for improvement in the recent service"). The device then sends the revised post content back to the server.

[0980] Step 7:

[0981] The server saves the final post in a database and publishes it on the social media platform. The input is the final post, and the output is saving it to the database and publishing it. Specifically, the revised post is saved in the database and displayed on the social media platform's timeline for other users to view.

[0982] (Application example 1)

[0983] 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."

[0984] In modern advertising campaigns, posting text to digital platforms such as social media is essential. However, posts containing incorrect expressions or a negative tone can damage a company's brand image and risk sparking outrage. In particular, in an age where past posts are easily shared, negative reactions can spread instantly. Therefore, there is a need for a system that can proactively identify advertising copy that may spark outrage and suggest revisions.

[0985] 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.

[0986] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and prediction results, and means for analyzing the risk of advertising copy in the post content causing an uproar and proposing revisions. This allows users to understand the risk of uproar in the post content of their advertising campaigns in advance and make appropriate revisions, thereby enabling safe and effective advertising.

[0987] "User" refers to a person or organization that uses the system to input and post advertising copy.

[0988] "Post content" refers to the text entered by a user and intended to be published on social media or digital platforms.

[0989] "Natural language processing" refers to the technology that enables computers to understand and analyze human language.

[0990] "Keywords" refers to important words and phrases extracted from the content of a post.

[0991] "Context" refers to information that allows us to understand the meaning of the entire sentence through the meaning and relationships of each word and phrase.

[0992] "Tone" refers to information that indicates the emotional nuance or attitude of a piece of writing.

[0993] A "rule-based checklist" refers to a list for evaluating posts based on predefined evaluation criteria.

[0994] A "machine learning model" refers to an algorithm that learns specific tasks based on past data and makes predictions and evaluations for new data.

[0995] "Flame risk" refers to an indicator of the likelihood that a post will provoke a negative reaction.

[0996] "Feedback" refers to information provided to a user, including analysis results and suggested modifications.

[0997] "Proposed amendments" refer to proposed changes to the content of a post to reduce the risk of a backlash.

[0998] This invention is a system that analyzes the content of advertising text in real time before users post it on social media or digital platforms, and identifies and prevents the risk of online outrage in advance. The system includes the following main components and processing flow:

[0999] System Configuration

[1000] The system consists of the following main components:

[1001] 1. Terminal: The device on which the user enters advertising text.

[1002] 2. Server: The central computer that performs analysis and predictions.

[1003] 3. Natural Language Processing (NLP) engine: Software that analyzes the content of posts.

[1004] 4. Machine learning model: An algorithm for predicting the risk of a firestorm.

[1005] 5. Database: An information repository that stores past posts and analysis results.

[1006] Explanation of program processing

[1007] The server first passes the ad copy received from the user device to a natural language processing (NLP) engine, which performs the following analysis:

[1008] Keyword analysis: Extracting important words and phrases within ad copy.

[1009] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[1010] Tone analysis: Calculating a sentiment score for text to infer sentiment and intent.

[1011] The server then evaluates the output of the NLP engine based on a rule-based checklist, determining, for example, that the phrase "better than any other product" is offensive.

[1012] The server then uses a machine learning model to predict the risk of the ad copy causing an uproar. The model learns from past cases of uproar and posting patterns, and provides an uproar risk score for the ad copy.

[1013] Based on the results, the server will provide feedback to the user, such as "This expression may provoke a negative reaction. We recommend changing it to something like 'Many customers are satisfied.'"

[1014] The user can then review the proposed revisions and choose to accept or ignore them. If the user accepts the revisions, the ad copy is updated and the device sends the final copy back to the server. The server then stores the final copy in a database and publishes it on the social media platform.

[1015] Specific examples

[1016] For example, if a user enters the ad copy, "This product is better than any other product," the server will analyze that expression and determine that there is a high risk of a negative reaction. Therefore, the server will provide feedback such as, "This expression may cause a negative reaction. We recommend changing it to something like, 'Many customers are satisfied with this product.'"

[1017] Example prompt sentence:

[1018] Please analyze whether this ad copy poses a risk of causing controversy. Also, please explain why and suggest modifications.

[1019] Hardware and software used

[1020] Hardware: General PCs and smartphones

[1021] Software: nltk (natural language processing library), sklearn (machine learning model library)

[1022] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1023] Step 1:

[1024] To receive the ad copy entered by the user, the device inputs the ad copy from the user and receives it through a social networking app or web interface. For example, the user may input ad copy such as "This product is better than any other product." The user inputs text data through a browser or mobile app, and the data is saved on the device.

[1025] Step 2:

[1026] The terminal generates and sends an HTTP request to send the entered ad copy to the server. The sent data is the ad copy in text format. After the server receives this data, it saves it as text data.

[1027] Step 3:

[1028] The server passes the received ad copy to a natural language processing (NLP) engine. The input is the text data received in step 2. The NLP engine analyzes the data to extract keywords, context, and tone information from the ad copy. Specifically, it extracts keywords, analyzes the context of the text, and calculates a sentiment score. The output is the analysis results (keyword list, context information, and tone score).

[1029] Step 4:

[1030] Based on the analysis results, the server performs an evaluation according to a predefined rule-based checklist. The analysis results obtained in step 3 are used as input. The rule-based evaluation determines whether certain keywords or tones pose a risk. For example, the keyword "better than any other product" is evaluated as offensive. The evaluation results (a list of problematic keywords and tone analysis results) are generated as output.

[1031] Step 5:

[1032] The server uses a machine learning model to predict the risk of the ad copy causing an uproar. The evaluation results obtained in step 4 are used as input. The machine learning model has learned from past cases of uproars and posting patterns, and calculates a risk score for new data. Specifically, the text data is vectorized, and the vector is input into the model for scoring. The output is an uproar risk score.

[1033] Step 6:

[1034] The server provides feedback to the user based on the evaluation results and risk score. The evaluation results and the flame risk score are used as inputs. The server generates feedback for the user, including problematic expressions and recommended corrections. For example, the server generates feedback such as, "This expression may provoke negative reactions. We recommend changing it to something like, 'Many customers are satisfied.'" The server generates a feedback message as output.

[1035] Step 7:

[1036] The server sends the generated feedback message to the terminal and displays it to the user. The feedback message generated in step 6 is used as input. The user checks the feedback message and chooses whether to accept or ignore the proposed revision. For example, if the user accepts the proposed revision, the revised advertisement copy is displayed.

[1037] Step 8:

[1038] If the user accepts the proposed revision, the device sends the revised copy to the server again. The revised copy is used as input. The server stores the received final copy in a database and publishes it on the social networking platform. The published copy is generated as output.

[1039] 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.

[1040] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[1041] System Configuration

[1042] The system consists of the following main components:

[1043] 1. Terminal

[1044] 2. Server

[1045] 3. Natural Language Processing (NLP) Engine

[1046] 4. Emotion Engine

[1047] 5. Machine Learning Models

[1048] 6. Database

[1049] Processing Flow

[1050] 1. Enter and submit your post

[1051] The device receives user submissions via a text area or form. For example, the user might enter a submission such as "These services are really no good these days!" The device then generates and sends an HTTP request to the server to send the submission.

[1052] 2. Analysis of posted content

[1053] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[1054] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[1055] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[1056] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1057] 3. Running the Emotion Engine

[1058] The server uses an emotion engine to analyze the type and intensity of emotions (e.g., joy, anger, sadness) contained in the user's post. For example, it may determine that "this post has a strong emotion of anger."

[1059] 4. Rule-based checks

[1060] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[1061] 5. Application of the Fire Prediction Model

[1062] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[1063] 6. Providing real-time feedback

[1064] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[1065] 7. Corrections and resubmissions

[1066] The user checks the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent back to the server via the device. For example, change it to "I feel there is room for improvement in the recent service."

[1067] 8. Save and publish your final post

[1068] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1069] The above is a specific embodiment of the present invention. By incorporating an emotion engine, it is possible to grasp user emotions in detail and improve the accuracy of predicting the risk of a controversy. This system allows users to understand the risk of a post causing a controversy in advance and make appropriate corrections, thereby maintaining healthy communication.

[1070] The processing flow will be explained below.

[1071] Step 1:

[1072] The terminal receives user posts via a text area or form. For example, the user may post, "These services are really no good these days!"

[1073] Step 2:

[1074] The device constructs an HTTP request containing the entered post content and sends this request to the server, including the user ID and timestamp.

[1075] Step 3:

[1076] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[1077] Step 4:

[1078] The server's NLP engine analyzes the post, specifically:

[1079] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[1080] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[1081] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1082] Step 5:

[1083] The server uses an emotion engine to analyze the type of emotion (e.g., joy, anger, sadness) and intensity of the emotion contained in the post. For example, it may determine that "this post contains strong anger."

[1084] Step 6:

[1085] The server evaluates the results of the analysis using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate. For example, checking if words like "bad" or "terrible" are present in the list.

[1086] Step 7:

[1087] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[1088] Step 8:

[1089] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[1090] Step 9:

[1091] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[1092] Step 10:

[1093] The user checks the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[1094] Step 11:

[1095] The device resends the modified or original post in its final form to the server, which receives the final post.

[1096] Step 12:

[1097] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1098] Example 2

[1099] 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."

[1100] Conventional SNS systems lacked sufficient filtering and moderation functions to determine whether user posts were appropriate, resulting in a high risk of unexpected outrage. Furthermore, when a user unintentionally used offensive language, the feedback function was insufficient, making it difficult to prompt the user to correct the post. To solve these problems, a system that analyzes user posts in more detail and provides real-time feedback is needed.

[1101] 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 post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for using an emotion engine to analyze the type and intensity of emotion contained in the post content, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned past cases of uproars and posting patterns, and means for providing feedback to the user based on the evaluation results and prediction results. This makes it possible to grasp the risk of a user's post content causing an uproar in advance and provide appropriate feedback in real time.

[1102] "User" means any individual or entity that uses a social media platform to enter and submit posts.

[1103] "Terminal" refers to an information processing device used by a user, such as a computer, smartphone, or tablet.

[1104] "Server" refers to a computer system that has the function of analyzing and evaluating the content posted by users and providing feedback.

[1105] "Natural language processing" refers to the technology of using computers to analyze human language, extract information, and understand meaning.

[1106] "Keyword analysis" refers to the technology of extracting important words and phrases from the content of posts.

[1107] "Contextual analysis" refers to a technology that understands the relationships between words in a sentence and grasps the overall meaning of the post.

[1108] "Tone analysis" refers to the technique of calculating an emotional score to infer the sentiment or intent of a piece of text.

[1109] "Emotion engine" refers to software or hardware for analyzing the type and intensity of emotions contained in posted content.

[1110] A "rules-based checklist" is a list for evaluating posts based on predefined keywords and phrases.

[1111] A "machine learning model" refers to a computational model that uses algorithms to learn patterns and rules based on data.

[1112] "Flame risk prediction" refers to a technology that predicts the risk of a post causing a flame war based on past flame war cases and posting patterns.

[1113] "Means for providing feedback" refers to a function that provides appropriate advice and suggestions for correction to the user in real time based on the analysis and prediction results.

[1114] "Suggestions to modify post content" refers to suggestions to change a user's post to an appropriate expression when the post is deemed offensive or inappropriate.

[1115] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[1116] System Configuration

[1117] The system consists of the following main components:

[1118] 1. Devices (computers, smartphones, tablets, etc.)

[1119] 2. Server (AWS or on-premise)

[1120] 3. Natural Language Processing (NLP) engines (e.g., SpaCy, nltk)

[1121] 4. Emotion engines (e.g., IBM Watson, Microsoft Text Analytics)

[1122] 5. Machine learning models (e.g., custom models based on TensorFlow or Scikit-Learn)

[1123] 6. Database (e.g. MySQL, PostgreSQL)

[1124] Function details

[1125] 1. Enter and submit your post

[1126] A user uses a device to enter text into a posting field on a social networking site. For example, the user enters the content of a post, such as "The services these days are really no good!" The device generates an HTTP request to send this content to the server, and sends it to the server.

[1127] 2. Analysis of posted content

[1128] The server passes the content received from the device to a natural language processing (NLP) engine, which then performs the following tasks:

[1129] Keyword analysis: Extracting important words and phrases, such as "no" and "really."

[1130] Contextual analysis: Understand the overall context of the post and analyze the relationships between words.

[1131] Tone analysis: Infers the sentiment and intent of a post and calculates a sentiment score. For example, it may be determined to have a "very negative tone."

[1132] 3. Running the Emotion Engine

[1133] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post. For example, it may determine that "this post contains a strong emotion of anger."

[1134] 4. Rule-based checks

[1135] The server uses a rule-based checklist to evaluate the results of the analysis, checking whether certain keywords or phrases are offensive or inappropriate, for example, whether words like "bad" or "terrible" exist in the checklist.

[1136] 5. Application of the Fire Prediction Model

[1137] The server uses a machine learning model to predict the risk of a post causing an uproar. This model learns from past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine. For example, it may assess the risk of a post causing an uproar as 70%.

[1138] 6. Providing real-time feedback

[1139] The server generates the analysis results and proposed modifications and sends them to the device. The device then displays feedback to the user. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[1140] 7. Corrections and resubmissions

[1141] The user reviews the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent again to the server via the device. For example, the user could change the post to "I feel there is room for improvement in the recent service."

[1142] 8. Save and publish your final post

[1143] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1144] Specific examples in the description

[1145] A user posts on social media, "These services are really no good!" This post is analyzed by the system and the following feedback is provided:

[1146] Keyword analysis determines "no" as offensive

[1147] The sentiment engine analyzes the tone of the post as "very negative."

[1148] The flame war prediction model predicts a "70% risk of flame war"

[1149] The server sends feedback to the device saying, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there are areas for improvement in your recent service.'"

[1150] Prompt Sentence Examples

[1151] "Please provide a detailed explanation of the specific processing steps involved in the system for analyzing the sentiment of social media posts and predicting the risk of a social media outcry. Furthermore, please provide specific examples of how this system provides feedback to users."

[1152] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1153] Step 1:

[1154] The user enters the content to post and submits it.

[1155] Specific actions: The user uses a device to enter the text "These services are really no good!" into the post field of a social networking site and clicks the send button.

[1156] Input: User-entered post content (e.g., "These services are really no good these days!")

[1157] Output: HTTP request generated by the device

[1158] Step 2:

[1159] The server receives the post and passes it to a natural language processing (NLP) engine.

[1160] Specific operation: The server receives the HTTP request generated by the device and passes the text data in the request to the NLP engine.

[1161] Input: HTTP request sent from the terminal

[1162] Output: Passing text data to the NLP engine

[1163] Step 3:

[1164] A natural language processing (NLP) engine analyzes posts to extract keywords, context, and tone of voice.

[1165] Specific operation: The NLP engine extracts important keywords such as "bad" and "really" from the text "These services these days are really no good!", performs contextual analysis, and then calculates an emotional score to determine the sentence as "negative."

[1166] Input: Text data of the post content

[1167] Output: Keyword, context, and tone analysis results (e.g., "Keyword: No, really," "Context: Negative," "Tone: Very negative")

[1168] Step 4:

[1169] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post.

[1170] Specific operation: The analysis results from the NLP engine are passed to the emotion engine, which analyzes the type and intensity of emotions, such as "anger" or "sadness."

[1171] Input: Analysis results of the NLP engine

[1172] Output: Analysis results of the emotion engine (e.g., "Emotion type: anger" and "Intensity: strong")

[1173] Step 5:

[1174] The server evaluates the submission using a rule-based checklist.

[1175] What it does: Parsed keywords are checked against a predefined rule-based checklist to see if they contain offensive words or phrases.

[1176] Input: Analysis results of NLP engine, analysis results of emotion engine

[1177] Output: Rule-based check results (e.g., "Aggressive: No (exists)", "Non-aggressive: Really (does not exist)", "Overall rating: Aggressive")

[1178] Step 6:

[1179] The server uses a machine learning model to predict the risk of a post causing an uproar.

[1180] Specific operation: The analysis results and the results of the emotion engine are input into a machine learning model, which then calculates the risk of a controversy based on past cases of controversy and posting patterns.

[1181] Input: Analysis results, emotion engine results

[1182] Output: Flame risk prediction (e.g., "Flame risk: 70%")

[1183] Step 7:

[1184] The server generates analysis results and suggested corrections and sends them to the device.

[1185] Specific operation: The server generates a proposal to modify the post content based on the results of the flame war risk prediction and sends it to the terminal as an HTTP response.

[1186] Input: Flame risk prediction results

[1187] Output: Generates a suggested fix and an HTTP response (e.g., "Suggested fix: This expression may be considered offensive. We recommend changing it to something like, 'We feel your recent service has room for improvement.'")

[1188] Step 8:

[1189] The user checks the feedback and selects whether to make corrections.

[1190] Specific operation: The device displays the feedback received from the server to the user, and the user decides whether to accept the suggested revisions or continue posting as is.

[1191] Input: Feedback from the server

[1192] Output: User's choice (e.g. "Accept proposed revision" or "Keep original post")

[1193] Step 9:

[1194] The server stores the final post in a database and publishes it on the social media platform.

[1195] Specific operation: The user resubmits the revised or original post to the server, which then stores the post in its database and publishes it on the social media platform.

[1196] Input: User's last post

[1197] Output: Saving to a database and publishing on a social media platform (e.g. "Saved to database" and "Published on social media")

[1198] (Application example 2)

[1199] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1200] Conventional systems for analyzing the content of posts on social media platforms are able to provide appropriate feedback by gaining a detailed understanding of user emotions and predicting the risk of a controversy in advance. However, advertising content created by advertising agencies in particular has a higher risk of causing a controversy than general user posts, and could potentially damage a company's brand image. Current systems lack specialized analysis and feedback for advertising content, and there is a need for technology that automatically provides appropriate revision suggestions to allow advertising agencies to post with confidence.

[1201] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1202] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and the prediction results, means for evaluating the emotions of advertising content using an emotion engine and scoring the risk of uproar, and means for providing suggestions for modifying the advertising content based on the scoring results. This enables advertising agencies to thoroughly evaluate the risk of uproaring advertising content before posting it and automatically receive appropriate suggestions for modification.

[1203] A "user" is a user who enters content to post on a social networking site or advertising platform.

[1204] "Posted content" refers to content such as text, images, and videos that users enter and submit to social media or advertising platforms.

[1205] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[1206] "Analysis" is the process of examining input data (in this case, posts) in detail and extracting information such as keywords, context, and tone.

[1207] "Keywords" refer to the main words or phrases in the post content and are an important source of information for analysis.

[1208] "Context" refers to the relationship or background information in which a keyword or phrase is used.

[1209] "Tone" refers to the overall sentiment or intent of a post (e.g., positive, negative, neutral).

[1210] A "rule-based checklist" is a list for evaluating analysis results based on predefined conditions or rules.

[1211] "Evaluation" is the process of checking the analysis results against a rule-based checklist to determine whether the content is appropriate.

[1212] A "machine learning model" is an algorithm that learns patterns and rules based on past data and makes predictions and analyses on new data.

[1213] "Flame risk" refers to the degree to which a post's content, if made public, is likely to provoke a negative reaction among users.

[1214] The "emotion engine" is a system that analyzes the type and intensity of emotions contained in posts and provides feedback based on that analysis.

[1215] "Scoring" is the process of quantifying evaluation and prediction results to quantitatively indicate the degree of risk.

[1216] "Feedback" refers to advice and suggestions for corrections provided to users based on analysis and evaluation results.

[1217] "Suggested fixes" are specific suggestions for improvements provided to users to make their posts safe and appropriate.

[1218] This invention is a system that analyzes user posts on social media and advertising platforms in real time to identify and prevent the risk of online outrage. This system uses an emotion engine, especially for advertising content, to score the risk of online outrage and provide feedback.

[1219] System Configuration

[1220] The system consists of the following main components:

[1221] 1. Terminal

[1222] 2. Server

[1223] 3. Natural Language Processing (NLP) Engine

[1224] 4. Emotion Engine

[1225] 5. Machine Learning Models

[1226] 6. Database

[1227] Hardware and software used

[1228] Device: The device (e.g., smartphone, tablet, or computer) through which a user enters their post.

[1229] Server: A central processing unit for receiving and analyzing submissions and generating feedback.

[1230] Natural Language Processing (NLP) engine: Analyzes text for keywords, context, and tone using libraries such as TextBlob.

[1231] Emotion engine: Software for analyzing the type and intensity of emotions contained in posts.

[1232] Machine learning model: Predicts the risk of a backlash based on past data.

[1233] Database: Stores submitted content and analysis results.

[1234] Details of data processing and calculation

[1235] Receiving posted content

[1236] Users use their devices to input content and send it to the server. Content can be in a variety of formats, including text, images, and videos.

[1237] Natural Language Processing (NLP)

[1238] The server passes the content received from the device to a natural language processing engine, which extracts information about keywords, context, and tone.

[1239] Applying the Emotion Engine

[1240] The server uses an emotion engine to evaluate the emotions contained in the advertising content and analyze their type and intensity.

[1241] Using machine learning models

[1242] The server uses a machine learning model to score the risk of a post causing an uproar based on past examples of uproars and posting patterns.

[1243] Providing Feedback

[1244] Based on the analysis results and the flame risk score, the server generates and provides feedback to the user, which may include suggested modifications.

[1245] Specific examples

[1246] Below are examples of advertising content and corresponding prompt sentences.

[1247] Examples:

[1248] Let's say an advertising agency creates the following advertising content:

[1249] "This product is really no good. Don't buy it!"

[1250] Example prompt sentence:

[1251] plaintext

[1252] Ad content: This product is really bad. Don't buy it!

[1253] Is this content potentially controversial? What's your sentiment rating and suggested changes?

[1254] Example output:

[1255] plaintext

[1256] This content has a high risk of causing controversy (controversy score: 85%, sentiment: negative). We recommend changing the wording to something like, "This product has some improvements. Please check out our other products."

[1257] In this way, the present invention realizes a system that detects the risk of a controversy in advance and provides appropriate correction suggestions, allowing advertising agencies to post advertising content with peace of mind.

[1258] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1259] Step 1:

[1260] The user inputs advertising content using the device. The input content can be in various formats such as text, images, videos, etc. After inputting, the user clicks the post button, and the device sends the posted content to the server.

[1261] Step 2:

[1262] The server passes the content of the post received from the device to a natural language processing (NLP) engine. The NLP engine analyzes the text data and extracts information about keywords, context, and tone. For example, if a post such as "This product is really no good" is entered, the keywords "product" and "no good" are extracted, and the negative tone is identified through contextual analysis.

[1263] Step 3:

[1264] The server passes the extracted keywords, context, and tone information to an emotion engine, which analyzes the type (e.g., anger, sadness) and intensity of the emotion contained in the post. The analysis results indicate that the emotion is "very negative."

[1265] Step 4:

[1266] The server evaluates the results of the analysis against a predefined rule-based checklist, for example, checking for the presence of offensive words like "bad" or "terrible," and then makes an initial assessment of the appropriateness of the post.

[1267] Step 5:

[1268] The server uses a machine learning model to predict the risk of a post causing an uproar. The machine learning model has learned from past cases of uproars and posting patterns, and also uses the analysis results as additional information. At this step, the server predicts an "85% risk of uproar."

[1269] Step 6:

[1270] The server generates feedback based on the analysis and prediction results and provides it to the user. The feedback includes problematic expressions and suggestions for rephrasing. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'This product has some improvements. Please check out our other products.'"

[1271] Step 7:

[1272] The user checks the feedback and chooses whether to accept the suggested revisions or keep the original post. If the user accepts the revisions, the post is updated and sent back to the server via the device.

[1273] Step 8:

[1274] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1275] 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.

[1276] 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.

[1277] 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.

[1278] [Fourth embodiment]

[1279] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1280] 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.

[1281] 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).

[1282] 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.

[1283] 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.

[1284] 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).

[1285] 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.

[1286] 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.

[1287] 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.

[1288] 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.

[1289] 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.

[1290] 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.

[1291] 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."

[1292] The present invention is a system that analyzes user's SNS postings in real time and identifies and prevents the risk of flame wars in advance. The system is implemented using the following specific means.

[1293] System Configuration

[1294] The system consists of the following main components:

[1295] 1. Terminal

[1296] 2. Server

[1297] 3. Natural Language Processing (NLP) Engine

[1298] 4. Machine Learning Models

[1299] 5. Database

[1300] Processing Flow

[1301] 1. Enter and submit your post

[1302] The device inputs user posts and receives them through a social networking app or web interface. For example, a user might post something like, "These services are really no good these days!" The device then generates and sends an HTTP request to send this post to the server.

[1303] 2. Analysis of posted content

[1304] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[1305] Keyword analysis: Extracting important words and phrases within posts. Examples: "no" and "really."

[1306] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[1307] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1308] 3. Rule-based checks

[1309] The server evaluates the output of the NLP engine based on a rule-based checklist, for example determining whether "no" or "really" is offensive.

[1310] 4. Application of the fire prediction model

[1311] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. This model learns from past cases of uproars and posting patterns, and provides a risk score for the post. For example, "The risk of uproar is 70%."

[1312] 5. Providing real-time feedback

[1313] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[1314] 6. Corrections and resubmissions

[1315] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post (e.g., "I feel there is room for improvement in the recent service"), and the device resends the final post to the server.

[1316] 7. Save and publish your final post

[1317] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1318] The above is a specific embodiment of the present invention. This system allows users to understand in advance the risk of their posts causing a controversy and make appropriate corrections, thereby enabling them to maintain healthy communication.

[1319] The processing flow will be explained below.

[1320] Step 1:

[1321] The terminal receives user posts via a text area or form. For example, the user might write, "These services are really no good these days!"

[1322] Step 2:

[1323] The device constructs the input post content as an HTTP request and sends this request to the server, including the user ID and timestamp.

[1324] Step 3:

[1325] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[1326] Step 4:

[1327] The server's NLP engine analyzes the post, specifically:

[1328] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[1329] Contextual analysis: Understand the overall context of a post. Analyze the relationships between words.

[1330] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1331] Step 5:

[1332] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[1333] Step 6:

[1334] The server uses a machine learning model to predict the risk of a post causing an uproar. The model learns from past cases of uproars and posting patterns, and assigns a risk score to the post (e.g., a 70% risk of causing an uproar).

[1335] Step 7:

[1336] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[1337] Step 8:

[1338] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[1339] Step 9:

[1340] The user reviews the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[1341] Step 10:

[1342] The device resends the modified or original post in its final form to the server, which receives the final post.

[1343] Step 11:

[1344] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1345] Example 1

[1346] 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."

[1347] In recent years, the impact of social media posts has become increasingly large, and inappropriate posts by users often have widespread repercussions in the form of flame wars. Therefore, there is a strong demand for methods that allow users to understand in advance whether their posts pose a risk of causing a flame war and to appropriately avoid that risk. However, current systems have the problem of insufficient support for users to quickly and accurately evaluate the flame war risk of their posts and make appropriate corrections.

[1348] 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.

[1349] In this invention, the server includes means for receiving posts entered by users, means for analyzing the posts using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the posts causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation and prediction results, means for the user to review the proposed revisions and choose whether to accept or ignore them, and means for receiving the revised posts again, saving them, and publishing them on the SNS platform. This allows users to understand the risk of their posts causing an uproar in advance and make appropriate revisions, thereby significantly reducing the risk of an uproar.

[1350] "User" refers to an individual or corporation that uses this system to post on SNS.

[1351] A "terminal" is a device used by a user, and includes a computer, a smartphone, a tablet, and the like.

[1352] "Server" refers to the computer system that receives, analyzes, provides feedback on, stores, and publishes Submissions.

[1353] "Posted content" refers to the text information posted by a user on a social media platform.

[1354] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[1355] "Keywords" refer to important words or phrases extracted from the content of a post.

[1356] "Context" refers to the relevance of the meaning of a word or phrase within the overall content of a post.

[1357] "Tone" refers to facial expressions that convey the emotion and intent of the post.

[1358] A "rule-based checklist" is a list for evaluating posts based on predefined criteria.

[1359] A "machine learning model" refers to an algorithm that learns from past data and makes predictions and classifications for new data.

[1360] "Flame risk" refers to the possibility that the content of a post will provoke a negative reaction.

[1361] "Feedback" refers to advice and suggestions provided to users based on analysis and prediction results.

[1362] "Suggested Revisions" means suggestions or alternatives for a User to revise a Submission.

[1363] "SNS Platform" refers to an online service that allows users to publish and share their posts with other users.

[1364] "Storage" refers to recording the posted content in a database or storage device.

[1365] "Public" means displaying the posted content on the social media platform so that other users can view it.

[1366] This invention is a system that analyzes users' social media posts in real time to identify and prevent the risk of flame wars in advance. This system receives the content posted by users and analyzes, evaluates, and provides feedback using a natural language processing (NLP) engine and machine learning model, making it possible to maintain healthy communication.

[1367] The main hardware and software components of this system are:

[1368] The device used by the user (computer, smartphone, tablet, etc.)

[1369] Server (responsible for receiving, analyzing, providing feedback on, storing and publishing posted content)

[1370] Natural Language Processing (NLP) Engine

[1371] Machine learning models

[1372] Database

[1373] First, a user uses a device to input content for a post via a social networking app or web interface. For example, the user might input "These services are really no good these days!" The device then generates this content as an HTTP request and sends it to the server.

[1374] The server then passes the content received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts information about keywords, context, and tone. Specifically, it identifies keywords like "no" and "really," analyzes the meaning of the entire sentence, and calculates a negative tone.

[1375] The server then evaluates the output of the NLP engine based on a rule-based checklist. The parsed keywords and tones are compared against the rule-based list to determine whether they are offensive. For example, words like "no" and "really" are considered offensive.

[1376] Next, the server uses a machine learning model to predict whether the post has a risk of causing a controversy. This model learns from past cases of controversy and posting patterns. It calculates a controversy risk score for the post and outputs, for example, "Controversy risk is 70%."

[1377] The server provides feedback to the user based on the evaluation and prediction results. This feedback includes specific suggestions for rephrasing or correction. For example, it may provide a suggestion such as, "This expression may be deemed offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[1378] The user reviews the proposed revisions and chooses whether to accept or ignore them. If the user accepts the revisions, they update the post. For example, they might revise the post to say, "I feel there is room for improvement in the recent service." The device then sends this revised post back to the server.

[1379] Finally, the server saves the modified post to a database and publishes it on the social media platform, so that it appears on the timeline and can be viewed by other users.

[1380] For example, the following prompt could be fed to a generative AI model: "Please use an NLP engine to analyze social media posts entered by users and detect offensive language. Also, please suggest modifications to reduce this risk."

[1381] This system allows users to understand in advance the risk of their posts causing a backlash and make appropriate corrections, significantly reducing the risk of a backlash.

[1382] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1383] Step 1:

[1384] A user uses a device to enter content to post via a social networking app or web interface and presses the "Send" button. For example, a user enters a post such as "These services these days are really no good!" The device generates the entered content as an HTTP request and sends this request to the server. The input is the text data posted by the user, and the output is an HTTP request to the server.

[1385] Step 2:

[1386] The server passes the content of posts received from the device to a natural language processing (NLP) engine, which analyzes the content and extracts the following information: the input is the received post data, and the output is extracted keywords, context, and tone information.

[1387] Keyword analysis: Identify important words and phrases within your posts. Examples include "no" and "really."

[1388] Contextual analysis: Analyzing the relationships between words to understand the meaning of the whole sentence. For example, the context of "Dameda."

[1389] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1390] Step 3:

[1391] The server evaluates the output of the NLP engine based on a rule-based checklist. The input is the analyzed data, and the output is the evaluation result. Specifically, it checks the analyzed keywords and tones against the rule-based list to see if they are offensive. For example, "no" and "really" are considered offensive.

[1392] Step 4:

[1393] The server uses a machine learning model to predict whether a post has a risk of causing an uproar. The input is the rule-based evaluation result, and the output is a score for the risk of causing an uproar. Specifically, it uses a model that has learned from past cases of uproars and posting patterns to calculate the risk score for the post. Example: "The risk of causing an uproar is 70%."

[1394] Step 5:

[1395] The server provides feedback to the user based on the analysis and prediction results. The input is the flame risk score, and the output is the content of the feedback. For example, the following suggestion is generated as a specific action: "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[1396] Step 6:

[1397] The user reviews the proposed revisions and chooses whether to accept or ignore them. The input is the feedback content, and the output is the revised post content or the original post content. Specifically, if the user accepts the revisions, the post content is updated (e.g., "I feel there is room for improvement in the recent service"). The device then sends the revised post content back to the server.

[1398] Step 7:

[1399] The server saves the final post in a database and publishes it on the social media platform. The input is the final post, and the output is saving it to the database and publishing it. Specifically, the revised post is saved in the database and displayed on the social media platform's timeline for other users to view.

[1400] (Application example 1)

[1401] 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."

[1402] In modern advertising campaigns, posting text to digital platforms such as social media is essential. However, posts containing incorrect expressions or a negative tone can damage a company's brand image and risk sparking outrage. In particular, in an age where past posts are easily shared, negative reactions can spread instantly. Therefore, there is a need for a system that can proactively identify advertising copy that may spark outrage and suggest revisions.

[1403] 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.

[1404] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and prediction results, and means for analyzing the risk of advertising copy in the post content causing an uproar and proposing revisions. This allows users to understand the risk of uproar in the post content of their advertising campaigns in advance and make appropriate revisions, thereby enabling safe and effective advertising.

[1405] "User" refers to a person or organization that uses the system to input and post advertising copy.

[1406] "Post content" refers to the text entered by a user and intended to be published on social media or digital platforms.

[1407] "Natural language processing" refers to the technology that enables computers to understand and analyze human language.

[1408] "Keywords" refers to important words and phrases extracted from the content of a post.

[1409] "Context" refers to information that allows us to understand the meaning of the entire sentence through the meaning and relationships of each word and phrase.

[1410] "Tone" refers to information that indicates the emotional nuance or attitude of a piece of writing.

[1411] A "rule-based checklist" refers to a list for evaluating posts based on predefined evaluation criteria.

[1412] A "machine learning model" refers to an algorithm that learns specific tasks based on past data and makes predictions and evaluations for new data.

[1413] "Flame risk" refers to an indicator of the likelihood that a post will provoke a negative reaction.

[1414] "Feedback" refers to information provided to a user, including analysis results and suggested modifications.

[1415] "Proposed amendments" refer to proposed changes to the content of a post to reduce the risk of a backlash.

[1416] This invention is a system that analyzes the content of advertising text in real time before users post it on social media or digital platforms, and identifies and prevents the risk of online outrage in advance. The system includes the following main components and processing flow:

[1417] System Configuration

[1418] The system consists of the following main components:

[1419] 1. Terminal: The device on which the user enters advertising text.

[1420] 2. Server: The central computer that performs analysis and predictions.

[1421] 3. Natural Language Processing (NLP) engine: Software that analyzes the content of posts.

[1422] 4. Machine learning model: An algorithm for predicting the risk of a firestorm.

[1423] 5. Database: An information repository that stores past posts and analysis results.

[1424] Explanation of program processing

[1425] The server first passes the ad copy received from the user device to a natural language processing (NLP) engine, which performs the following analysis:

[1426] Keyword analysis: Extracting important words and phrases within ad copy.

[1427] Context analysis: Analyzing the relationships between words to understand the meaning of the entire sentence.

[1428] Tone analysis: Calculating a sentiment score for text to infer sentiment and intent.

[1429] The server then evaluates the output of the NLP engine based on a rule-based checklist, determining, for example, that the phrase "better than any other product" is offensive.

[1430] The server then uses a machine learning model to predict the risk of the ad copy causing an uproar. The model learns from past cases of uproar and posting patterns, and provides an uproar risk score for the ad copy.

[1431] Based on the results, the server will provide feedback to the user, such as "This expression may provoke a negative reaction. We recommend changing it to something like 'Many customers are satisfied.'"

[1432] The user can then review the proposed revisions and choose to accept or ignore them. If the user accepts the revisions, the ad copy is updated and the device sends the final copy back to the server. The server then stores the final copy in a database and publishes it on the social media platform.

[1433] Specific examples

[1434] For example, if a user enters the ad copy, "This product is better than any other product," the server will analyze that expression and determine that there is a high risk of a negative reaction. Therefore, the server will provide feedback such as, "This expression may cause a negative reaction. We recommend changing it to something like, 'Many customers are satisfied with this product.'"

[1435] Example prompt sentence:

[1436] Please analyze whether this ad copy poses a risk of causing controversy. Also, please explain why and suggest modifications.

[1437] Hardware and software used

[1438] Hardware: General PCs and smartphones

[1439] Software: nltk (natural language processing library), sklearn (machine learning model library)

[1440] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1441] Step 1:

[1442] To receive the ad copy entered by the user, the device inputs the ad copy from the user and receives it through a social networking app or web interface. For example, the user may input ad copy such as "This product is better than any other product." The user inputs text data through a browser or mobile app, and the data is saved on the device.

[1443] Step 2:

[1444] The terminal generates and sends an HTTP request to send the entered ad copy to the server. The sent data is the ad copy in text format. After the server receives this data, it saves it as text data.

[1445] Step 3:

[1446] The server passes the received ad copy to a natural language processing (NLP) engine. The input is the text data received in step 2. The NLP engine analyzes the data to extract keywords, context, and tone information from the ad copy. Specifically, it extracts keywords, analyzes the context of the text, and calculates a sentiment score. The output is the analysis results (keyword list, context information, and tone score).

[1447] Step 4:

[1448] Based on the analysis results, the server performs an evaluation according to a predefined rule-based checklist. The analysis results obtained in step 3 are used as input. The rule-based evaluation determines whether certain keywords or tones pose a risk. For example, the keyword "better than any other product" is evaluated as offensive. The evaluation results (a list of problematic keywords and tone analysis results) are generated as output.

[1449] Step 5:

[1450] The server uses a machine learning model to predict the risk of the ad copy causing an uproar. The evaluation results obtained in step 4 are used as input. The machine learning model has learned from past cases of uproars and posting patterns, and calculates a risk score for new data. Specifically, the text data is vectorized, and the vector is input into the model for scoring. The output is an uproar risk score.

[1451] Step 6:

[1452] The server provides feedback to the user based on the evaluation results and risk score. The evaluation results and the flame risk score are used as inputs. The server generates feedback for the user, including problematic expressions and recommended corrections. For example, the server generates feedback such as, "This expression may provoke negative reactions. We recommend changing it to something like, 'Many customers are satisfied.'" The server generates a feedback message as output.

[1453] Step 7:

[1454] The server sends the generated feedback message to the terminal and displays it to the user. The feedback message generated in step 6 is used as input. The user checks the feedback message and chooses whether to accept or ignore the proposed revision. For example, if the user accepts the proposed revision, the revised advertisement copy is displayed.

[1455] Step 8:

[1456] If the user accepts the proposed revision, the device sends the revised copy to the server again. The revised copy is used as input. The server stores the received final copy in a database and publishes it on the social networking platform. The published copy is generated as output.

[1457] 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.

[1458] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[1459] System Configuration

[1460] The system consists of the following main components:

[1461] 1. Terminal

[1462] 2. Server

[1463] 3. Natural Language Processing (NLP) Engine

[1464] 4. Emotion Engine

[1465] 5. Machine Learning Models

[1466] 6. Database

[1467] Processing Flow

[1468] 1. Enter and submit your post

[1469] The device receives user submissions via a text area or form. For example, the user might enter a submission such as "These services are really no good these days!" The device then generates and sends an HTTP request to the server to send the submission.

[1470] 2. Analysis of posted content

[1471] The server passes the content of the post received from the device to a natural language processing (NLP) engine, which performs the following analysis:

[1472] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[1473] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[1474] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1475] 3. Running the Emotion Engine

[1476] The server uses an emotion engine to analyze the type and intensity of emotions (e.g., joy, anger, sadness) contained in the user's post. For example, it may determine that "this post has a strong emotion of anger."

[1477] 4. Rule-based checks

[1478] The server evaluates the results using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate, for example, checking whether words like "bad" or "terrible" are present in the list.

[1479] 5. Application of the Fire Prediction Model

[1480] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[1481] 6. Providing real-time feedback

[1482] The server generates analysis results and suggested corrections and sends them to the device. The device then displays the problematic expressions and suggested rephrasing to the user. For example, it might say, "This expression may be deemed offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[1483] 7. Corrections and resubmissions

[1484] The user checks the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent back to the server via the device. For example, change it to "I feel there is room for improvement in the recent service."

[1485] 8. Save and publish your final post

[1486] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1487] The above is a specific embodiment of the present invention. By incorporating an emotion engine, it is possible to grasp user emotions in detail and improve the accuracy of predicting the risk of a controversy. This system allows users to understand the risk of a post causing a controversy in advance and make appropriate corrections, thereby maintaining healthy communication.

[1488] The processing flow will be explained below.

[1489] Step 1:

[1490] The terminal receives user posts via a text area or form. For example, the user may post, "These services are really no good these days!"

[1491] Step 2:

[1492] The device constructs an HTTP request containing the entered post content and sends this request to the server, including the user ID and timestamp.

[1493] Step 3:

[1494] The server parses the received HTTP request to extract the post content, which is then sent to a natural language processing (NLP) engine.

[1495] Step 4:

[1496] The server's NLP engine analyzes the post, specifically:

[1497] Keyword analysis: Extracting important words and phrases, e.g. "no" or "really."

[1498] Contextual analysis: Analyzes the relationships between words to understand the overall context of the post.

[1499] Tone analysis: Calculating a sentiment score for text to infer its sentiment and intent. Example: "Very negative tone."

[1500] Step 5:

[1501] The server uses an emotion engine to analyze the type of emotion (e.g., joy, anger, sadness) and intensity of the emotion contained in the post. For example, it may determine that "this post contains strong anger."

[1502] Step 6:

[1503] The server evaluates the results of the analysis using a rule-based checklist to see if certain keywords or phrases are offensive or inappropriate. For example, checking if words like "bad" or "terrible" are present in the list.

[1504] Step 7:

[1505] The server uses a machine learning model to predict the risk of a post causing an uproar. This model is trained on past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine as additional information. Example: "The risk of an uproar is 70%."

[1506] Step 8:

[1507] The server generates feedback to provide to the user based on the analysis results and the predictions of the machine learning model. For example, it generates feedback such as, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there is room for improvement in your recent services.'"

[1508] Step 9:

[1509] The server generates feedback and sends it to the terminal, which receives it and displays it to the user.

[1510] Step 10:

[1511] The user checks the feedback and chooses whether to accept or ignore the proposed correction. If the feedback is accepted, the message is changed to, for example, "I feel there is room for improvement in your recent service."

[1512] Step 11:

[1513] The device resends the modified or original post in its final form to the server, which receives the final post.

[1514] Step 12:

[1515] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1516] Example 2

[1517] 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."

[1518] Conventional SNS systems lacked sufficient filtering and moderation functions to determine whether user posts were appropriate, resulting in a high risk of unexpected outrage. Furthermore, when a user unintentionally used offensive language, the feedback function was insufficient, making it difficult to prompt the user to correct the post. To solve these problems, a system that analyzes user posts in more detail and provides real-time feedback is needed.

[1519] 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 post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for using an emotion engine to analyze the type and intensity of emotion contained in the post content, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned past cases of uproars and posting patterns, and means for providing feedback to the user based on the evaluation results and prediction results. This makes it possible to grasp the risk of a user's post content causing an uproar in advance and provide appropriate feedback in real time.

[1520] "User" means any individual or entity that uses a social media platform to enter and submit posts.

[1521] "Terminal" refers to an information processing device used by a user, such as a computer, smartphone, or tablet.

[1522] "Server" refers to a computer system that has the function of analyzing and evaluating the content posted by users and providing feedback.

[1523] "Natural language processing" refers to the technology of using computers to analyze human language, extract information, and understand meaning.

[1524] "Keyword analysis" refers to the technology of extracting important words and phrases from the content of posts.

[1525] "Contextual analysis" refers to a technology that understands the relationships between words in a sentence and grasps the overall meaning of the post.

[1526] "Tone analysis" refers to the technique of calculating an emotional score to infer the sentiment or intent of a piece of text.

[1527] "Emotion engine" refers to software or hardware for analyzing the type and intensity of emotions contained in posted content.

[1528] A "rules-based checklist" is a list for evaluating posts based on predefined keywords and phrases.

[1529] A "machine learning model" refers to a computational model that uses algorithms to learn patterns and rules based on data.

[1530] "Flame risk prediction" refers to a technology that predicts the risk of a post causing a flame war based on past flame war cases and posting patterns.

[1531] "Means for providing feedback" refers to a function that provides appropriate advice and suggestions for correction to the user in real time based on the analysis and prediction results.

[1532] "Suggestions to modify post content" refers to suggestions to change a user's post to an appropriate expression when the post is deemed offensive or inappropriate.

[1533] This invention is a system that analyzes user posts on social media in real time to identify and prevent the risk of flame wars in advance. This system incorporates a new emotion engine, which recognizes and analyzes user emotions in detail, enabling it to provide more accurate flame war predictions and feedback.

[1534] System Configuration

[1535] The system consists of the following main components:

[1536] 1. Devices (computers, smartphones, tablets, etc.)

[1537] 2. Server (AWS or on-premise)

[1538] 3. Natural Language Processing (NLP) engines (e.g., SpaCy, nltk)

[1539] 4. Emotion engines (e.g., IBM Watson, Microsoft Text Analytics)

[1540] 5. Machine learning models (e.g., custom models based on TensorFlow or Scikit-Learn)

[1541] 6. Database (e.g. MySQL, PostgreSQL)

[1542] Function details

[1543] 1. Enter and submit your post

[1544] A user uses a device to enter text into a posting field on a social networking site. For example, the user enters the content of a post, such as "The services these days are really no good!" The device generates an HTTP request to send this content to the server, and sends it to the server.

[1545] 2. Analysis of posted content

[1546] The server passes the content received from the device to a natural language processing (NLP) engine, which then performs the following tasks:

[1547] Keyword analysis: Extracting important words and phrases, such as "no" and "really."

[1548] Contextual analysis: Understand the overall context of the post and analyze the relationships between words.

[1549] Tone analysis: Infers the sentiment and intent of a post and calculates a sentiment score. For example, it may be determined to have a "very negative tone."

[1550] 3. Running the Emotion Engine

[1551] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post. For example, it may determine that "this post contains a strong emotion of anger."

[1552] 4. Rule-based checks

[1553] The server uses a rule-based checklist to evaluate the results of the analysis, checking whether certain keywords or phrases are offensive or inappropriate, for example, whether words like "bad" or "terrible" exist in the checklist.

[1554] 5. Application of the Fire Prediction Model

[1555] The server uses a machine learning model to predict the risk of a post causing an uproar. This model learns from past cases of uproars and posting patterns, and also uses the analysis results of the emotion engine. For example, it may assess the risk of a post causing an uproar as 70%.

[1556] 6. Providing real-time feedback

[1557] The server generates the analysis results and proposed modifications and sends them to the device. The device then displays feedback to the user. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'We feel there is room for improvement in your recent services.'"

[1558] 7. Corrections and resubmissions

[1559] The user reviews the feedback and chooses whether to adopt the proposed corrections or keep the original post. If the user adopts the proposed corrections, the post is updated and sent again to the server via the device. For example, the user could change the post to "I feel there is room for improvement in the recent service."

[1560] 8. Save and publish your final post

[1561] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1562] Specific examples in the description

[1563] A user posts on social media, "These services are really no good!" This post is analyzed by the system and the following feedback is provided:

[1564] Keyword analysis determines "no" as offensive

[1565] The sentiment engine analyzes the tone of the post as "very negative."

[1566] The flame war prediction model predicts a "70% risk of flame war"

[1567] The server sends feedback to the device saying, "This expression may be considered offensive. We recommend changing it to something like, 'I feel there are areas for improvement in your recent service.'"

[1568] Prompt Sentence Examples

[1569] "Please provide a detailed explanation of the specific processing steps involved in the system for analyzing the sentiment of social media posts and predicting the risk of a social media outcry. Furthermore, please provide specific examples of how this system provides feedback to users."

[1570] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1571] Step 1:

[1572] The user enters the content to post and submits it.

[1573] Specific actions: The user uses a device to enter the text "These services are really no good!" into the post field of a social networking site and clicks the send button.

[1574] Input: User-entered post content (e.g., "These services are really no good these days!")

[1575] Output: HTTP request generated by the device

[1576] Step 2:

[1577] The server receives the post and passes it to a natural language processing (NLP) engine.

[1578] Specific operation: The server receives the HTTP request generated by the device and passes the text data in the request to the NLP engine.

[1579] Input: HTTP request sent from the terminal

[1580] Output: Passing text data to the NLP engine

[1581] Step 3:

[1582] A natural language processing (NLP) engine analyzes posts to extract keywords, context, and tone of voice.

[1583] Specific operation: The NLP engine extracts important keywords such as "bad" and "really" from the text "These services these days are really no good!", performs contextual analysis, and then calculates an emotional score to determine the sentence as "negative."

[1584] Input: Text data of the post content

[1585] Output: Keyword, context, and tone analysis results (e.g., "Keyword: No, really," "Context: Negative," "Tone: Very negative")

[1586] Step 4:

[1587] The server uses an emotion engine to analyze the type and intensity of emotions contained in the post.

[1588] Specific operation: The analysis results from the NLP engine are passed to the emotion engine, which analyzes the type and intensity of emotions, such as "anger" or "sadness."

[1589] Input: Analysis results of the NLP engine

[1590] Output: Analysis results of the emotion engine (e.g., "Emotion type: anger" and "Intensity: strong")

[1591] Step 5:

[1592] The server evaluates the submission using a rule-based checklist.

[1593] What it does: Parsed keywords are checked against a predefined rule-based checklist to see if they contain offensive words or phrases.

[1594] Input: Analysis results of NLP engine, analysis results of emotion engine

[1595] Output: Rule-based check results (e.g., "Aggressive: No (exists)", "Non-aggressive: Really (does not exist)", "Overall rating: Aggressive")

[1596] Step 6:

[1597] The server uses a machine learning model to predict the risk of a post causing an uproar.

[1598] Specific operation: The analysis results and the results of the emotion engine are input into a machine learning model, which then calculates the risk of a controversy based on past cases of controversy and posting patterns.

[1599] Input: Analysis results, emotion engine results

[1600] Output: Flame risk prediction (e.g., "Flame risk: 70%")

[1601] Step 7:

[1602] The server generates analysis results and suggested corrections and sends them to the device.

[1603] Specific operation: The server generates a proposal to modify the post content based on the results of the flame war risk prediction and sends it to the terminal as an HTTP response.

[1604] Input: Flame risk prediction results

[1605] Output: Generates a suggested fix and an HTTP response (e.g., "Suggested fix: This expression may be considered offensive. We recommend changing it to something like, 'We feel your recent service has room for improvement.'")

[1606] Step 8:

[1607] The user checks the feedback and selects whether to make corrections.

[1608] Specific operation: The device displays the feedback received from the server to the user, and the user decides whether to accept the suggested revisions or continue posting as is.

[1609] Input: Feedback from the server

[1610] Output: User's choice (e.g. "Accept proposed revision" or "Keep original post")

[1611] Step 9:

[1612] The server stores the final post in a database and publishes it on the social media platform.

[1613] Specific operation: The user resubmits the revised or original post to the server, which then stores the post in its database and publishes it on the social media platform.

[1614] Input: User's last post

[1615] Output: Saving to a database and publishing on a social media platform (e.g. "Saved to database" and "Published on social media")

[1616] (Application example 2)

[1617] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1618] Conventional systems for analyzing the content of posts on social media platforms are able to provide appropriate feedback by gaining a detailed understanding of user emotions and predicting the risk of a controversy in advance. However, advertising content created by advertising agencies in particular has a higher risk of causing a controversy than general user posts, and could potentially damage a company's brand image. Current systems lack specialized analysis and feedback for advertising content, and there is a need for technology that automatically provides appropriate revision suggestions to allow advertising agencies to post with confidence.

[1619] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1620] In this invention, the server includes means for receiving post content entered by a user, means for analyzing the post content using natural language processing and extracting information on keywords, context, and tone, means for evaluating the analysis results based on a predefined rule-based checklist, means for predicting the risk of the post content causing an uproar using a machine learning model that has learned from past cases of uproars and posting patterns, means for providing feedback to the user based on the evaluation results and the prediction results, means for evaluating the emotions of advertising content using an emotion engine and scoring the risk of uproar, and means for providing suggestions for modifying the advertising content based on the scoring results. This enables advertising agencies to thoroughly evaluate the risk of uproaring advertising content before posting it and automatically receive appropriate suggestions for modification.

[1621] A "user" is a user who enters content to post on a social networking site or advertising platform.

[1622] "Posted content" refers to content such as text, images, and videos that users enter and submit to social media or advertising platforms.

[1623] "Natural language processing" is a technology that allows computers to understand and analyze human language.

[1624] "Analysis" is the process of examining input data (in this case, posts) in detail and extracting information such as keywords, context, and tone.

[1625] "Keywords" refer to the main words or phrases in the post content and are an important source of information for analysis.

[1626] "Context" refers to the relationship or background information in which a keyword or phrase is used.

[1627] "Tone" refers to the overall sentiment or intent of a post (e.g., positive, negative, neutral).

[1628] A "rule-based checklist" is a list for evaluating analysis results based on predefined conditions or rules.

[1629] "Evaluation" is the process of checking the analysis results against a rule-based checklist to determine whether the content is appropriate.

[1630] A "machine learning model" is an algorithm that learns patterns and rules based on past data and makes predictions and analyses on new data.

[1631] "Flame risk" refers to the degree to which a post's content, if made public, is likely to provoke a negative reaction among users.

[1632] The "emotion engine" is a system that analyzes the type and intensity of emotions contained in posts and provides feedback based on that analysis.

[1633] "Scoring" is the process of quantifying evaluation and prediction results to quantitatively indicate the degree of risk.

[1634] "Feedback" refers to advice and suggestions for corrections provided to users based on analysis and evaluation results.

[1635] "Suggested fixes" are specific suggestions for improvements provided to users to make their posts safe and appropriate.

[1636] This invention is a system that analyzes user posts on social media and advertising platforms in real time to identify and prevent the risk of online outrage. This system uses an emotion engine, especially for advertising content, to score the risk of online outrage and provide feedback.

[1637] System Configuration

[1638] The system consists of the following main components:

[1639] 1. Terminal

[1640] 2. Server

[1641] 3. Natural Language Processing (NLP) Engine

[1642] 4. Emotion Engine

[1643] 5. Machine Learning Models

[1644] 6. Database

[1645] Hardware and software used

[1646] Device: The device (e.g., smartphone, tablet, or computer) through which a user enters their post.

[1647] Server: A central processing unit for receiving and analyzing submissions and generating feedback.

[1648] Natural Language Processing (NLP) engine: Analyzes text for keywords, context, and tone using libraries such as TextBlob.

[1649] Emotion engine: Software for analyzing the type and intensity of emotions contained in posts.

[1650] Machine learning model: Predicts the risk of a backlash based on past data.

[1651] Database: Stores submitted content and analysis results.

[1652] Details of data processing and calculation

[1653] Receiving posted content

[1654] Users use their devices to input content and send it to the server. Content can be in a variety of formats, including text, images, and videos.

[1655] Natural Language Processing (NLP)

[1656] The server passes the content received from the device to a natural language processing engine, which extracts information about keywords, context, and tone.

[1657] Applying the Emotion Engine

[1658] The server uses an emotion engine to evaluate the emotions contained in the advertising content and analyze their type and intensity.

[1659] Using machine learning models

[1660] The server uses a machine learning model to score the risk of a post causing an uproar based on past examples of uproars and posting patterns.

[1661] Providing Feedback

[1662] Based on the analysis results and the flame risk score, the server generates and provides feedback to the user, which may include suggested modifications.

[1663] Specific examples

[1664] Below are examples of advertising content and corresponding prompt sentences.

[1665] Examples:

[1666] Let's say an advertising agency creates the following advertising content:

[1667] "This product is really no good. Don't buy it!"

[1668] Example prompt sentence:

[1669] plaintext

[1670] Ad content: This product is really bad. Don't buy it!

[1671] Is this content potentially controversial? What's your sentiment rating and suggested changes?

[1672] Example output:

[1673] plaintext

[1674] This content has a high risk of causing controversy (controversy score: 85%, sentiment: negative). We recommend changing the wording to something like, "This product has some improvements. Please check out our other products."

[1675] In this way, the present invention realizes a system that detects the risk of a controversy in advance and provides appropriate correction suggestions, allowing advertising agencies to post advertising content with peace of mind.

[1676] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1677] Step 1:

[1678] The user inputs advertising content using the device. The input content can be in various formats such as text, images, videos, etc. After inputting, the user clicks the post button, and the device sends the posted content to the server.

[1679] Step 2:

[1680] The server passes the content of the post received from the device to a natural language processing (NLP) engine. The NLP engine analyzes the text data and extracts information about keywords, context, and tone. For example, if a post such as "This product is really no good" is entered, the keywords "product" and "no good" are extracted, and the negative tone is identified through contextual analysis.

[1681] Step 3:

[1682] The server passes the extracted keywords, context, and tone information to an emotion engine, which analyzes the type (e.g., anger, sadness) and intensity of the emotion contained in the post. The analysis results indicate that the emotion is "very negative."

[1683] Step 4:

[1684] The server evaluates the results of the analysis against a predefined rule-based checklist, for example, checking for the presence of offensive words like "bad" or "terrible," and then makes an initial assessment of the appropriateness of the post.

[1685] Step 5:

[1686] The server uses a machine learning model to predict the risk of a post causing an uproar. The machine learning model has learned from past cases of uproars and posting patterns, and also uses the analysis results as additional information. At this step, the server predicts an "85% risk of uproar."

[1687] Step 6:

[1688] The server generates feedback based on the analysis and prediction results and provides it to the user. The feedback includes problematic expressions and suggestions for rephrasing. For example, it might say, "This expression may be considered offensive. We recommend changing it to something like, 'This product has some improvements. Please check out our other products.'"

[1689] Step 7:

[1690] The user checks the feedback and chooses whether to accept the suggested revisions or keep the original post. If the user accepts the revisions, the post is updated and sent back to the server via the device.

[1691] Step 8:

[1692] The server saves the final post in a database and publishes it on the social media platform, where it appears on the timeline and can be viewed by other users.

[1693] 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.

[1694] 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.

[1695] 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.

[1696] 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.

[1697] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1698] 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.

[1699] 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).

[1700] 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.

[1701] 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."

[1702] 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.

[1703] 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).

[1704] 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.

[1705] 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.

[1706] 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.

[1707] 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.

[1708] 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.

[1709] 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.

[1710] 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.

[1711] 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.

[1712] 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.

[1713] 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.

[1714] The following is further disclosed regarding the above embodiment.

[1715] (Claim 1)

[1716] A means for receiving the posting content entered by the user;

[1717] means for analyzing the posted content using natural language processing to extract information on keywords, context, and tone;

[1718] means for evaluating the analysis results based on a predefined rule-based checklist;

[1719] A means for predicting the risk of the content of said post becoming an incendiary entity using a machine learning model that has learned from past examples of incendiary entities and posting patterns;

[1720] means for providing feedback to a user based on the evaluation results and prediction results;

[1721] A system including:

[1722] (Claim 2)

[1723] 10. The system of claim 1, wherein the feedback provided to the user includes suggested modifications to the posting.

[1724] (Claim 3)

[1725] The system of claim 1, wherein the machine learning model scores the risk of the posted content causing a controversy.

[1726] "Example 1"

[1727] (Claim 1)

[1728] A means for receiving the posting content entered by the user;

[1729] means for analyzing the posted content using natural language processing to extract information on keywords, context, and tone;

[1730] means for evaluating the analysis results based on a predefined rule-based checklist;

[1731] A means for predicting the risk of the content of said post becoming an incendiary entity using a machine learning model that has learned from past examples of incendiary entities and posting patterns;

[1732] means for providing feedback to a user based on the evaluation results and prediction results;

[1733] a means for the user to review the proposed revisions and choose whether to accept or ignore them;

[1734] A means to receive the revised post again, store it and publish it on the social media platform;

[1735] A system including:

[1736] (Claim 2)

[1737] 10. The system of claim 1, wherein the feedback provided to the user includes suggested modifications to the posting.

[1738] (Claim 3)

[1739] The system of claim 1, wherein the machine learning model scores the risk of the posted content causing a controversy.

[1740] "Application Example 1"

[1741] (Claim 1)

[1742] A means for receiving the posting content entered by the user;

[1743] means for analyzing the posted content using natural language processing to extract information on keywords, context, and tone;

[1744] means for evaluating the analysis results based on a predefined rule-based checklist;

[1745] A means for predicting the risk of the content of said post becoming an incendiary entity using a machine learning model that has learned from past examples of incendiary entities and posting patterns;

[1746] means for providing feedback to a user based on the evaluation results and prediction results;

[1747] A means for analyzing the risk of the advertisement copy in the posted content causing a firestorm and proposing a revision;

[1748] A system including:

[1749] (Claim 2)

[1750] 10. The system of claim 1, wherein the feedback provided to the user includes suggested modifications to the posting.

[1751] (Claim 3)

[1752] The system of claim 1, wherein the machine learning model scores the risk of the posted content causing a controversy.

[1753] "Example 2: Combining Emotion Engines"

[1754] (Claim 1)

[1755] A means for receiving the posting content entered by the user;

[1756] means for analyzing the posted content using natural language processing to extract information on keywords, context, and tone;

[1757] means for using an emotion engine to analyze the type and intensity of emotions contained in the posted content;

[1758] means for evaluating the analysis results based on a predefined rule-based checklist;

[1759] A means for predicting the risk of the content of said post becoming an incendiary entity using a machine learning model that has learned from past examples of incendiary entities and posting patterns;

[1760] means for providing feedback to a user based on the evaluation results and prediction results;

[1761] A system including:

[1762] (Claim 2)

[1763] 10. The system of claim 1, wherein the feedback provided to the user includes suggested modifications to the posting.

[1764] (Claim 3)

[1765] The system of claim 1, wherein the machine learning model scores the risk of the posted content causing a controversy.

[1766] "Application example 2 when combining emotion engines"

[1767] (Claim 1)

[1768] A means for receiving the posting content entered by the user;

[1769] means for analyzing the posted content using natural language processing to extract information on keywords, context, and tone;

[1770] means for evaluating the analysis results based on a predefined rule-based checklist;

[1771] A means for predicting the risk of the content of said post becoming an incendiary entity using a machine learning model that has learned from past examples of incendiary entities and posting patterns;

[1772] means for providing feedback to a user based on the evaluation results and prediction results;

[1773] A means for evaluating the emotions of advertisements using an emotion engine for advertisement content and scoring the risk of a flame war;

[1774] means for providing suggested modifications to advertising content based on the scoring results;

[1775] A system including:

[1776] (Claim 2)

[1777] 10. The system of claim 1, wherein the feedback provided to the user includes suggested modifications to the posting.

[1778] (Claim 3)

[1779] The system of claim 1, wherein the machine learning model scores the risk of the posted content causing a controversy. [Explanation of symbols]

[1780] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for receiving the posting content entered by the user; means for analyzing the posted content using natural language processing to extract information on keywords, context, and tone; means for evaluating the analysis results based on a predefined rule-based checklist; A means for predicting the risk of the content of said post becoming an incendiary entity using a machine learning model that has learned from past examples of incendiary entities and posting patterns; means for providing feedback to a user based on the evaluation results and prediction results; A system including:

2. The system of claim 1 , wherein the feedback provided to the user includes suggested modifications to the posting.

3. The system according to claim 1 , wherein the machine learning model scores the risk of the posted content causing a controversy.

Images