System

Abstract

To provide a system.SOLUTION: The system includes: first generative AI means for determining the content of posting to the social media; second generative AI means for editing the posted content when the first generative AI means has determined that the posted content is dangerous; means for automatically generating a report concerning the posted content edited by the second generative AI means; means for allowing an employee to confirm the report; means for sharing the report with an organization; means for checking a keyword or a key phrase in the posted content; emotion engine means for recognizing the emotion of the poster from the posted content; and means for notifying of the result of determination by the emotion engine means.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] When employees or part-time workers post on social media, there is a risk that the content could spark outrage or reveal confidential company information. Furthermore, there is no efficient way to collect information about employees' complaints and positive points. [Means for solving the problem]

[0005] Generative AI is used to judge the content of posts on social media and edit any dangerous content. Furthermore, a report on the edited content is automatically generated and employees are asked to review it for training purposes. Furthermore, by sharing this report with the company, information on employee complaints and positive points is collected. [Brief explanation of the drawings]

[0006] [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. 2 is a sequence diagram showing a flow of processing in the data processing system according to the first embodiment of the first form example. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Embodiment 1. [Figure 13] FIG. 10 is a sequence diagram showing a processing flow of a data processing system in a second embodiment of the second form example. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Embodiment Example 2. [Figure 15] FIG. 10 is a sequence diagram showing the flow of processing in a data processing system according to a third embodiment of the third embodiment. [Figure 16] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Embodiment 3. [Figure 17] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the first embodiment of the first form example when an emotion engine is combined. [Figure 18] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Form Example 1 when an emotion engine is combined. [Figure 19] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the second embodiment of the second form example when an emotion engine is combined. [Figure 20] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Form Example 2 when an emotion engine is combined. [Figure 21] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the third embodiment of the third form example when an emotion engine is combined. [Figure 22] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Form Example 3 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0007] 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.

[0008] First, the terms used in the following description will be explained.

[0009] 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, the 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), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0010] 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.

[0011] 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.

[0012] 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.

[0013] 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."

[0014] [First embodiment]

[0015] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0016] 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.

[0017] 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).

[0018] 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.

[0019] 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.

[0020] 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.

[0021] 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.

[0022] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0023] 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.

[0024] 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.

[0025] 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.

[0026] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0027] "Example 1"

[0028] In one embodiment of the present invention, before employees or part-timers post on social media, the content of the post is input into a generative AI. The generative AI determines whether the content of the post is likely to spark an outrage or whether it contains confidential company information. Specifically, the AI ​​checks whether the content of the post contains keywords or phrases that could spark an outrage or keywords or phrases related to confidential company information.

[0029] "Example 2"

[0030] If the generative AI judges a post to be dangerous, it will edit it. Specifically, it will delete or rewrite any parts that could spark outrage or that touch on confidential company information. For example, if a post says, "Our new product is better than our competitors' products," the AI ​​will rewrite it to say, "Our new product has superior performance."

[0031] "Example 3"

[0032] After the generative AI edits the post, it automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. This report is then reviewed by employees, and used to educate them.

[0033] "Example 4"

[0034] The reports created by the generative AI are shared with the company, allowing the company to efficiently collect information on employees' complaints and positive points. For example, the report can reveal that employees have high opinions of a new product, or that some employees are dissatisfied with their work.

[0035] The processing flow of each embodiment will be described below.

[0036] "Example 1"

[0037] Step 1: Before employees or part-time workers post on social media, the content of the post is input into a generative AI.

[0038] Step 2: The generative AI determines whether the post is likely to spark an uproar or contain confidential company information. Specifically, it checks whether the post contains keywords or phrases that could spark an uproar or keywords or phrases related to confidential company information.

[0039] "Example 2"

[0040] Step 1: If the generative AI determines that the post is dangerous, it will edit it.

[0041] Step 2: Specifically, parts that could cause controversy or parts that touch on confidential company information are deleted or rewritten. For example, if a post says, "Our new product is better than our competitors' products," the AI ​​will rewrite it to say, "Our new product has superior performance."

[0042] "Example 3"

[0043] Step 1: After the generative AI edits the post, it automatically generates a report on the edits.

[0044] Step 2: The report will detail which parts were dangerous, why they were, and how they were edited. This report will be reviewed by employees and used to educate them.

[0045] "Example 4"

[0046] Step 1: The report created by the generative AI is shared with the company.

[0047] Step 2: This allows the company to efficiently collect information on employees' dissatisfaction and positive points. For example, the report can reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[0048] Example 1

[0049] Next, a description will be given of Example 1 of Form 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."

[0050] When employees or part-time workers post on social media, if there is a possibility that the content of the post may spark outrage or may contain confidential company information, it is necessary to assess the risk in advance and take appropriate action. However, manual checking is time-consuming and labor-intensive, and is inefficient. Furthermore, the accuracy of the checks can vary depending on the person. This poses a challenge, as it increases the risk of damaging a company's brand image and leaking confidential information.

[0051] 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.

[0052] In this invention, the server includes a generative AI means for determining the content of posts to the SNS, a generative AI means for editing the post if the post is determined to be risky by the generative AI, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for a user to input the post content before posting it to the SNS, a means for transmitting the input post content to the server, a means for the server to input the post content to a generative AI model, a means for the generative AI model to analyze the post content and return a determination result to the server, a means for the server to return the determination result to a terminal, and a means for the terminal to display the determination result to the user. This makes it possible to automatically determine the risk of the post content in advance and take appropriate measures quickly and efficiently.

[0053] "Generative AI methods" refers to artificial intelligence technology that analyzes the content of posts on social media and determines the possibility of a controversy or the risk of confidential information being leaked.

[0054] "Generative AI means for editing posted content" refers to artificial intelligence technology that automatically corrects and edits posted content that is deemed dangerous.

[0055] "Means for automatically generating reports" refers to technology that automatically creates reports that explain the risks of the posted content and the details of any corrections based on the posted content edited by the generative AI means.

[0056] "Means for employee review" refers to a technique for presenting the generated report to an employee for review or approval.

[0057] "Means for sharing with the organization" refers to the technology for sharing the generated report with relevant parties within the organization and communicating the information.

[0058] "Means for users to input content before posting on SNS" refers to technology that provides an interface for users to input content they intend to post on SNS in advance.

[0059] The term "means for transmitting the input posting content to the server" refers to a communication technology for transmitting the posting content input by the user to the server.

[0060] "Means by which the server inputs the posted content into the generative AI model" refers to the technology by which the server inputs the posted content received into the generative AI model and has it analyzed.

[0061] "Means for the generative AI model to analyze the content of posts and return the results to the server" refers to the technology that enables the generative AI model to analyze the content of posts and return the results to the server.

[0062] "Means by which the server returns the judgment results to the terminal" refers to the communication technology that enables the server to return the judgment results received from the generative AI model to the terminal.

[0063] "Means for the terminal to display the determination result to the user" refers to a technology that provides an interface for the terminal to display the determination result received from the server to the user.

[0064] MODE FOR CARRYING OUT THE INVENTION

[0065] This invention is a system that checks the content of posts made by employees and part-time workers on social media in advance to determine the possibility of them causing controversy or the risk of confidential information being leaked.The system uses a generative AI model to analyze the content of posts and automatically determine the risk.

[0066] Hardware and software used

[0067] Server: Runs the generative AI model and analyzes and judges the content of posts.

[0068] Terminal: Provides an interface for users to input their posting content and check the results of the evaluation.

[0069] Generative AI model: For example, use GPT-4 (registered trademark) from OpenAI (registered trademark).

[0070] Data processing and calculation

[0071] 1. Users enter their post content before posting to social media:

[0072] Users can use a dedicated application or web interface on their device to input the content they want to post to SNS. For example, a user might input, "I'm sharing details about our new product on SNS! This is still confidential, but I want everyone to know about it soon!"

[0073] 2. The device sends the entered post content to the server:

[0074] The terminal sends the content of the post entered by the user to the server. At this time, the terminal converts the content of the post into an appropriate format (for example, JSON format) and sends it to the server using a secure communication protocol (for example, HTTPS).

[0075] 3. The server inputs the post into the generative AI model:

[0076] The server inputs the received post content into the generative AI model. Specifically, the server generates a prompt like this:

[0077] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[0078] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[0079] This prompt is sent to the generative AI model.

[0080] 4. The generative AI model analyzes the post and returns the results to the server:

[0081] The generative AI model analyzes the prompt text and checks whether the post contains keywords or phrases that could spark controversy or keywords or phrases related to confidential company information. For example, it detects the keyword "confidential" and determines that the post contains confidential company information. The generative AI model then returns this determination result to the server.

[0082] 5. The server returns the result to the device:

[0083] The server receives the results of the generative AI model and sends them back to the device, including information on whether the content of the post is likely to spark an uproar or whether it contains confidential company information.

[0084] 6. The device displays the result to the user:

[0085] The device displays the result of the judgment received from the server to the user. For example, it may display a message saying, "This post contains confidential company information. Please correct the post." Based on this information, the user can either correct the post or cancel it.

[0086] Specific examples

[0087] Suppose a user enters the following post content:

[0088] "We'll be sharing details about our new products on social media! This is still confidential, but we want you all to know soon!"

[0089] Here's an example prompt to feed this post to a generative AI model:

[0090] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[0091] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[0092] The generative AI model analyzes this prompt, detects that the post contains keywords such as "confidential," and determines that it may contain confidential company information.

[0093] The server returns the result of this determination to the terminal, and the terminal displays to the user a message saying, "This post contains confidential company information. Please correct your post."

[0094] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0095] Step 1:

[0096] The user inputs the content to post before posting to the SNS.

[0097] Users use a dedicated application or web interface on their device to input the content they want to post to SNS. For example, a user might input, "I'm sharing details about our new product on SNS! This is still confidential, but I want everyone to know about it soon!" The input content is temporarily stored in the device's memory.

[0098] Step 2:

[0099] The terminal transmits the input post content to the server.

[0100] The terminal converts the content of the post entered by the user into an appropriate format (e.g., JSON format) and sends it to the server using a secure communication protocol (e.g., HTTPS). The input data is the text information of the post, and the output data is the post sent to the server.

[0101] Step 3:

[0102] The server inputs the post content into a generative AI model.

[0103] The server inputs the received post content into the generative AI model. Specifically, the server generates a prompt like this:

[0104] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[0105] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[0106] The input data is the text information of the post, and the output data is the generated prompt.

[0107] Step 4:

[0108] The generative AI model analyzes the content of the post and returns the results to the server.

[0109] The generative AI model analyzes the prompt text and checks whether the post contains keywords or phrases that could spark an uproar or keywords or phrases related to confidential company information. For example, it detects the keyword "confidential" and determines that the post contains confidential company information. The input data is the prompt text, and the output data is the judgment result.

[0110] Step 5:

[0111] The server returns the result of the determination to the terminal.

[0112] The server receives the judgment results from the generative AI model and sends them back to the device. The judgment results include information on whether the content of the post has the potential to spark a controversy or whether it contains confidential company information. The input data is the judgment result, and the output data is the judgment result sent to the device.

[0113] Step 6:

[0114] The terminal displays the judgment result to the user.

[0115] The device displays the judgment result received from the server to the user. For example, it may display a message such as "This post contains confidential company information. Please correct the post." The input data is the judgment result, and the output data is the judgment result displayed to the user. Based on this information, the user can correct the post or cancel it.

[0116] (Application example 1)

[0117] Next, a description will be given of Application Example 1 of Embodiment 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."

[0118] When employees or part-time workers post on social media, there is a risk that they may unintentionally post content that will spark controversy or include confidential information. Such posts not only damage the company's credibility, but can also lead to legal issues. Furthermore, manually checking and correcting posted content is time-consuming and labor-intensive, and inefficient. To solve these issues, a system is needed that can automatically assess the content of posts and prompt corrections as necessary.

[0119] 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.

[0120] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the post if the post is determined to be dangerous by the generative AI, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for converting the post into a prompt sentence and generating a determination result using a generative AI model, a means for displaying the determination result to the user, and a means for prompting the user to modify the post as necessary. This makes it possible to automatically check the content posted by employees and part-time workers on SNS, reducing the risk of online outrage and the risk of confidential information being leaked.

[0121] "Generative AI methods" are methods that use artificial intelligence to analyze the content of social media posts and determine the risk of online outrage or the risk of confidential information being leaked.

[0122] "Generative AI means for editing posted content when the determined posted content is dangerous" refers to a means for automatically correcting posted content that the generative AI has determined to be dangerous.

[0123] "Means for automatically generating a report on the content of posts edited by generative AI" means means for automatically creating a detailed report on the content of posts edited by generative AI.

[0124] The "means for having the employee check the report" refers to a means for presenting the generated report to the employee and having the employee check it.

[0125] The "means for sharing the report with the organization" is a means for sharing the generated report within the organization.

[0126] "Means for converting the content of a post into a prompt sentence and generating a judgment result using a generative AI model" refers to means for converting the content of a post entered by a user into a prompt sentence, analyzing the content using a generative AI model, and generating a judgment result.

[0127] "Means for displaying the judgment results to the user" refers to means for displaying the judgment results generated by the generative AI model to the user.

[0128] The "means of encouraging the user to correct the posted content as necessary" refers to a means of encouraging the user to correct the posted content based on the judgment result.

[0129] As an embodiment of the present invention, a system is constructed that automatically judges the content of posts on SNS and prompts users to make corrections as necessary. A specific embodiment of this system is described below.

[0130] First, the server is equipped with a generative AI means for judging the content of posts on social media. This generative AI means analyzes the content of posts entered by users and judges the risk of flame wars and the risk of confidential information leaks. Specifically, it converts the content of posts into prompt statements and generates judgment results using a generative AI model (for example, OpenAI's API).

[0131] Next, the server includes a generative AI means for editing the posted content when the generative AI determines that the posted content is dangerous. This means automatically corrects the posted content that the generative AI determines to be dangerous and provides the user with safe posted content.

[0132] Additionally, the server has a means for automatically generating a report on the posts edited by the generative AI, which will detail why the posts were dangerous and how they were corrected.

[0133] The server also provides a means for employees to review the generated reports and for them to be shared with the organization, allowing employees to understand the risks of what they post and for information to be shared across the organization.

[0134] As a concrete example, consider the case where a user posts something like, "Details of the new product haven't been released yet, but it's going to be great!" This post is converted into a prompt sentence and input into the generative AI model.

[0135] Example prompt sentence:

[0136] Please determine whether the following posts are inflammatory or contain confidential information.

[0137] Posted: Details of the new product have not been released yet, but it's going to be awesome!

[0138] Judgment result:

[0139] The server sends this prompt to the generative AI model and obtains the judgment result. If the post is deemed dangerous based on the judgment result, the server automatically modifies the post and provides safe content to the user. In addition, the generated report can be reviewed by employees and shared with the organization for thorough risk management.

[0140] In this way, this invention automatically checks the content posted on social media by employees and part-time workers, making it possible to reduce the risk of online outrage and the risk of confidential information being leaked.

[0141] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0142] Step 1:

[0143] The user inputs the content to post on social media. The user uses a smartphone or computer to enter the content to post into the application. This input becomes the starting point for the system's processing.

[0144] Step 2:

[0145] The server receives the post content and converts it into a prompt sentence. The server receives the post content entered by the user and converts it into a prompt sentence suitable for the generative AI model. Specifically, the post content is converted into the following format.

[0146] Please determine whether the following posts are inflammatory or contain confidential information.

[0147] Post content: [User post content]

[0148] Judgment result:

[0149] Input: User post

[0150] Output: A prompt suitable for a generative AI model

[0151] Step 3:

[0152] The server generates a judgment result using a generative AI model. The server sends the prompt text to the generative AI model (e.g., OpenAI's API) and receives the judgment result. The generative AI model analyzes the prompt text and determines whether the post content poses a risk of flaming or leaking confidential information.

[0153] Input: prompt statement

[0154] Output: Judgment result

[0155] Step 4:

[0156] The server displays the judgment result to the user. The server displays the judgment result obtained from the generative AI model to the user. The judgment result includes whether the posted content is dangerous and the specific risk content.

[0157] Input: Judgment result

[0158] Output: The result displayed to the user

[0159] Step 5:

[0160] The server prompts the user to modify the posted content as necessary. Based on the judgment result, the server prompts the user to modify the posted content. For example, if the posted content is judged to be dangerous, the server will present the user with specific suggestions for modification.

[0161] Input: Judgment result

[0162] Output: suggested fixes presented to the user

[0163] Step 6:

[0164] The server automatically generates a report on the post content edited by the generative AI. The server automatically creates a detailed report on the post content edited by the generative AI. This report details why the post content was dangerous and how it was corrected.

[0165] Input: Edited post

[0166] Output: Detailed report

[0167] Step 7:

[0168] The server presents the generated report to employees for review and shares it with the organization. The server also shares the report within the organization to ensure thorough risk management.

[0169] Input: Detailed report

[0170] Output: Reports presented to employees, reports shared with the organization

[0171] Example 2

[0172] Next, a description will be given of Example 2 of Form 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."

[0173] On social media and other online platforms, when user posts pose a risk of causing controversy or leaking confidential information, there is a need to properly manage the content and edit it safely. However, manual monitoring and editing takes time and effort, and is inefficient. Therefore, a system is needed that automatically analyzes posted content, identifies risky parts, and edits them.

[0174] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0175] In this invention, the server includes means for receiving posted content entered by a user, means for transmitting the received posted content to a generative AI model, means for the generative AI model to analyze the posted content and identify dangerous parts, means for the generative AI model to edit the identified dangerous parts, and means for returning the edited posted content to the user. This makes it possible to automatically analyze the posted content of users and reduce the risk of flame wars and the leakage of confidential information.

[0176] "User" means an individual or group that uses the system to input posting content.

[0177] "Posted content" refers to information such as text, images, and videos that users post on social media or other online platforms.

[0178] The "server" is a computer system that receives user submissions and sends them to the generative AI model.

[0179] A "generative AI model" is a model that uses artificial intelligence technology to analyze posted content, identify risky parts, and edit them.

[0180] A "prompt sentence" is an input sentence that instructs the generative AI model to analyze and edit the post content.

[0181] "Analysis" is the process by which the generative AI model understands the context and meaning of the post and identifies risky parts.

[0182] "Dangerous parts" are parts of the post that pose a risk of causing an uproar or that contain confidential information.

[0183] "Editing" is the process by which a generative AI model removes or rewrites identified dangerous parts.

[0184] "Return" is the process by which the server sends back to the user the post content edited by the generative AI model.

[0185] This invention relates to a system that automatically analyzes content posted by users on social networking sites and other online platforms, identifies risky parts, and edits them. Specific embodiments of this system are described below.

[0186] The server receives the posting content entered by the user. The user uses their own device (PC, smartphone, etc.) to enter the posting content. For example, they might enter, "Our new product is better than our competitor's product."

[0187] The terminal sends the content of the post entered by the user to the server, using a communication protocol such as an HTTP request.

[0188] The server sends a prompt to the generative AI model to analyze the content of the post received from the device. The prompt includes the content of the post and instructions for analysis. For example, the server sends a prompt such as, "Analyze the content posted by the user, identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'"

[0189] A generative AI model (such as OpenAI's GPT-4) analyzes the content of posts based on prompts received from the server. The model uses natural language processing techniques to understand the context and meaning of the posts and identify parts that may be controversial or contain sensitive information.

[0190] The generative AI model then removes or rewrites the identified risky parts, for example, rewriting "Our new product is better than our competitors' products" to "Our new product has superior performance."

[0191] The server then sends the edited post content received from the generative AI model back to the device, again using a communication protocol such as an HTTP response.

[0192] The terminal displays the edited post content received from the server to the user, who can then check the edited content and revise it again if necessary.

[0193] As a concrete example, if a user posts, "Our new product is better than our competitor's product," the process will proceed as follows:

[0194] 1. The user enters the content to post on the device.

[0195] 2. The device sends the post content to the server.

[0196] 3. The server receives the post and sends the following prompt to the generative AI model: "Analyze the content posted by the user and identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'"

[0197] 4. The generative AI model analyzes the content of the post and identifies any part that claims to be "better than competitors' products" as dangerous.

[0198] 5. The generative AI model rewrites this part to "has excellent performance."

[0199] 6. The server sends the edited post back to the device.

[0200] 7. The device displays the edited post "Our new product has excellent performance" to the user.

[0201] In this way, the system can safely edit user posts.

[0202] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0203] Step 1:

[0204] The user enters the content to post.

[0205] Users use their own devices (PCs, smartphones, etc.) to enter their post content. For example, they might enter, "Our new product is better than our competitors' products." The entered post content is saved on the device.

[0206] Step 2:

[0207] The device sends the post content to the server.

[0208] The terminal sends the content posted by the user to the server. This transmission uses a communication protocol such as an HTTP request. The input is the content posted by the user, and the output is the content sent to the server.

[0209] Step 3:

[0210] The server receives the post and sends a prompt to the generative AI model.

[0211] The server sends a prompt to the generative AI model to analyze the content of the post received from the device. The prompt includes the content of the post and instructions for analysis. For example, it generates a prompt such as, "Analyze the content posted by the user, identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'" The input is the content of the post received from the device, and the output is the prompt sent to the generative AI model.

[0212] Step 4:

[0213] A generative AI model analyzes the content of posts and identifies risky parts.

[0214] A generative AI model (e.g., OpenAI's GPT-4) analyzes the content of posts based on prompts received from the server. The model uses natural language processing techniques to understand the context and meaning of the posts and identify parts that pose a risk of flaming or contain confidential information. The input is the prompt, and the output is the identified risky parts.

[0215] Step 5:

[0216] The generative AI model edits out dangerous parts.

[0217] The generative AI model removes or rewrites the identified risky parts. For example, it rewrites "Our new product is better than our competitors' products" to "Our new product has superior performance." The input is the identified risky parts, and the output is the edited post.

[0218] Step 6:

[0219] The server sends the edited post back to the device.

[0220] The server returns the edited post content received from the generative AI model to the device. This return also uses a communication protocol such as an HTTP response. The input is the edited post content received from the generative AI model, and the output is transmission to the device.

[0221] Step 7:

[0222] The device displays the edited post content to the user.

[0223] The terminal displays the edited post content received from the server to the user. The user can check the edited content and revise the post content again if necessary. The input is the edited post content received from the server, and the output is the display to the user.

[0224] (Application example 2)

[0225] Next, a description will be given of Application Example 2 of Form 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."

[0226] On online platforms such as social media and blogs, if a user posts content that could spark controversy or leak confidential information, the post could cause serious damage to a company or individual. To prevent such risks, a system is needed to check posts in advance and edit them as necessary. However, manual checking and editing takes time and effort, and there are challenges in that it is difficult to respond in real time.

[0227] 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.

[0228] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for analyzing the posted content in real time, a means for detecting expressions that pose a risk of causing an uproar, a means for detecting confidential information, and a means for automatically rewriting the posted content to safe expressions. This makes it possible to detect the risk of posted content in real time and automatically rewrite it to safe expressions.

[0229] "Generative AI means" refers to means that use artificial intelligence technology to analyze input data and generate output according to a specific purpose.

[0230] "Posted content" refers to information such as text, images, and videos that users make public on online platforms such as social media and blogs.

[0231] "If judged to be dangerous" means that the generative AI means analyzes the content of the post and determines that there is a risk of it causing a firestorm or leaking confidential information.

[0232] "Generative editing AI means" refers to a means that uses artificial intelligence technology to rewrite posted content that is deemed dangerous into safer expression.

[0233] "Means for automatically generating reports" refers to means for automatically creating a report summarizing the edits and reasons for edits made to posted content by generative AI means.

[0234] The "means for having employees check" refers to a means for presenting the automatically generated report to employees and having them check the contents.

[0235] The "means for sharing with the organization" is a means for sharing the generated report with the people involved in the organization.

[0236] "Means of real-time analysis" refers to means for instantly analyzing the content of posts and determining whether they are dangerous.

[0237] "Means for detecting expressions that pose a risk of causing an outrage" refers to means for identifying expressions contained in posted content that could cause an outrage.

[0238] "Means for detecting confidential information" refers to means for identifying confidential information of companies or individuals contained in posted content.

[0239] "Means for automatically rewriting content into safe expressions" refers to a means for automatically rewriting content that is deemed dangerous into safe expressions that do not pose a risk of online outrage or the leakage of confidential information.

[0240] As an embodiment of the present invention, a system is provided that analyzes content posted on online platforms such as social networking sites and blogs in real time and automatically edits any dangerous parts. Specific embodiments of this system are described below.

[0241] System Configuration

[0242] The system consists of the following main components:

[0243] 1. Generative AI means: Evaluates the content of posts and edits them if deemed dangerous.

[0244] 2. Automatic report generation method: Automatically generate reports on edited post content.

[0245] 3. Employee verification method: Have employees verify the generated reports.

[0246] 4. Organizational sharing means: Share reports within the organization.

[0247] 5. Real-time analysis: Analyze posted content instantly.

[0248] 6. Flaming risk detection method: Detect expressions that pose a risk of flaming.

[0249] 7. Confidential information detection methods: Detect confidential information.

[0250] 8. Automatic rewriting method: Automatically rewrite dangerous posts into safe expressions.

[0251] Program processing

[0252] The server does the following:

[0253] 1. Receiving posted content: Receives content that a user intends to post on social media or a blog.

[0254] 2. Analysis by a generative AI model: The received post content is sent to a generative AI model for analysis. The generative AI model used is, for example, OpenAI's "text-davinci-003."

[0255] 3. Risk assessment: The generative AI model analyzes the content of the post and determines whether there is a risk of a controversy or the leakage of confidential information.

[0256] 4. Automatic editing: If a post is deemed dangerous, the generative AI model automatically rewrites it to make it safer.

[0257] 5. Automatic report generation: Automatically generate a report for edited posts, summarizing the edits made and the reasons for them.

[0258] 6. Notify employees: Notify employees of the generated reports and have them review them.

[0259] 7. Share within your organization: Share the reviewed report within your organization.

[0260] Hardware and software used

[0261] Hardware: Servers, user devices (smartphones, PCs, etc.)

[0262] Software: Generative AI model (OpenAI API), real-time analysis software, report generation software

[0263] Specific examples

[0264] For example, if a user enters the following post:

[0265] Original post: "Our new product is better than our competitors' products."

[0266] In this case, the prompt sent to the generative AI model would be:

[0267] Prompt: "Please reword the following post to make it safer:\n\nOur new product is better than our competitors' products.\n\nPlease remove or rewrite any content that is inflammatory or contains confidential information."

[0268] The generative AI model analyzes the post based on this prompt and rewrites it to a safer expression. For example, the rewritten content might read, "Our new product has excellent performance."

[0269] In this way, users can check the content before posting and, if necessary, rewrite it to safer language, thereby preventing the risk of flame wars and the leakage of confidential information.

[0270] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0271] Step 1:

[0272] The user enters the content they want to post on social media or a blog.

[0273] Input: User-typed post content (e.g., "Our new product is better than our competitors' products.")

[0274] Output: The post is sent to the server.

[0275] Specific operation: A user uses a smartphone or PC to enter text into the posting screen of a social networking site or blog and presses the "Post" button.

[0276] Step 2:

[0277] The server receives the posted content and sends it to the generative AI model.

[0278] Input: User submitted post

[0279] Output: The prompt sent to the generative AI model

[0280] What happens: The server receives the post and generates a prompt, such as "Please rewrite the following post to make it safer:\n\nOur new product is superior to our competitors' products.\n\nPlease remove or rewrite any parts that may be controversial or contain confidential information."

[0281] Step 3:

[0282] A generative AI model analyzes the content of posts and determines whether they are dangerous.

[0283] Input: A prompt sent to the generative AI model

[0284] Output: Analysis results (presence or absence of danger, identification of dangerous parts)

[0285] How it works: A generative AI model (e.g., OpenAI's "text-davinci-003") analyzes the prompt text and determines whether the post content poses a risk of inciting controversy or leaking confidential information.

[0286] Step 4:

[0287] The server receives the analysis results of the generative AI model and edits the post content as necessary.

[0288] Input: Analysis results of generative AI model

[0289] Output: Edited post

[0290] Specific operation: The server receives the analysis results of the generative AI model and rewrites any parts judged to be dangerous into safe expressions. For example, it could rewrite it as "Our new product has excellent performance."

[0291] Step 5:

[0292] The server automatically generates a report of edited posts.

[0293] Input: Edited post

[0294] Output: Auto-generated report

[0295] Specific operation: The server automatically generates a report summarizing the edits and the reasons for the edits based on the edited post content. The report includes the original post content, the edited content, and the reason for the edit.

[0296] Step 6:

[0297] The server notifies the employee of the generated report for review.

[0298] Input: Auto-generated report

[0299] Output: Report notified to employees

[0300] What happens: The server sends an automatically generated report to the employee's email address or internal system for review.

[0301] Step 7:

[0302] Employees review the reports and provide feedback as needed.

[0303] Input: Report notified to employee

[0304] Output: Employee feedback

[0305] Specific actions: The employee checks the report and checks if there are any problems with the content. If necessary, feedback is sent to the server.

[0306] Step 8:

[0307] Share the server verification report with your organization.

[0308] Input: Reviewed report with employee feedback

[0309] Output: Reports shared within the organization

[0310] Specific actions: The server shares the confirmed report with relevant parties within the organization and prompts them to take necessary action, for example by posting it on the internal portal site.

[0311] Example 3

[0312] Next, a description will be given of a third embodiment of the third embodiment. 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."

[0313] Conventional SNS post management systems lack the functionality to automatically determine the risk of posts and edit them appropriately, which makes it difficult to efficiently educate employees and collect company feedback.In addition, they do not generate reports that explain in detail the reasons for editing posts and specific areas for improvement, which makes it difficult for employees to understand and the effectiveness of training to be achieved.

[0314] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0315] In this invention, the server includes a means for a user to input posted content, a means for the server to receive the posted content, a means for the server to send a prompt to the generative AI model, a means for the generative AI model to edit the posted content, a means for the server to receive the edited content, a means for the server to generate a report on the edited content, a means for the server to send the report to a terminal, a means for the terminal to display the report to employees, and a means for employees to review the report and use it for training. This enables automatic determination of the riskiness of posted content and appropriate editing, enabling efficient employee training and company feedback collection. Furthermore, a report is generated that details the reasons for edits and specific improvements, improving employee understanding and the effectiveness of training.

[0316] A "user" is an individual or group that inputs posting content into the system.

[0317] A "server" is a device or system that receives user submissions, sends prompts to the generative AI model, receives edited content, generates a report, and sends it to the terminal.

[0318] A "generative AI model" is an artificial intelligence algorithm or program that edits posts based on prompts received from a server.

[0319] A "prompt" is text that contains instructions for the generative AI model to edit the post content.

[0320] A "terminal" is a device or system that displays reports sent from the server to employees.

[0321] A "report" is a document that details which parts of the post were edited by the generative AI model, why they were dangerous, and how they were edited.

[0322] An "employee" is an individual or entity that uses the device to review reports and assist in education.

[0323] This invention is a system in which a user inputs post content, a server receives the post content and sends a prompt message to a generative AI model, the generative AI model edits the post content, the server receives the edited content and generates a report, and a terminal displays the report to employees.

[0324] First, a user uses their own device (e.g., a PC or smartphone) to input their post into the system interface. For example, they might type, "I feel the quality of the new product is poor. It needs further improvement."

[0325] Next, the server receives the post content entered by the user. This post content is temporarily stored in a database on the server. The server sends the received post content to the generative AI model as a prompt text. The prompt text is structured as follows: "Please improve the following post content and generate a report: Post content: 'I feel the quality of the new product is poor. Further improvement is needed.'"

[0326] A generative AI model (e.g., OpenAI's GPT-4) analyzes the prompt received from the server and edits the post, which might look something like, "I feel there are some areas for improvement in the quality of the new product. Further improvements are expected."

[0327] The server receives the edited content from the generative AI model. This edited content is also stored in a database. The server compares the original and edited content and automatically generates a report on the edits. The report contains the following information:

[0328] Which parts were dangerous and why?

[0329] How did you edit it?

[0330] For example, the report states, "Because saying 'poor quality' could damage the company's image, we softened the expression to 'there are some areas for improvement in quality.'"

[0331] The server sends the generated report to the employee's device. The report is displayed on a dedicated interface that the employee can access. The device displays the report received from the server to the employee. The employee can check the report on their own device and learn which parts of the posted content were problematic and how they could be improved. This can be used to educate employees.

[0332] This system automatically determines the risk of posted content and enables appropriate editing, thereby enabling efficient employee training and company feedback collection. Furthermore, a report detailing the reasons for editing and specific areas for improvement is generated, improving employee understanding and the effectiveness of training. The flow of the identification process in the third embodiment will be described with reference to FIG. 15.

[0333] Step 1:

[0334] The user enters the content to post.

[0335] Users use their own devices (PCs or smartphones) to input their posts into the system interface. For example, they might type, "I feel the quality of the new product is poor. It needs further improvement." The entered post is then sent from the device to the server.

[0336] Step 2:

[0337] The server receives the post.

[0338] The server receives the posts sent by the user. The received posts are temporarily stored in a database on the server. The input is the user's post, and the output is the post stored in the database.

[0339] Step 3:

[0340] The server sends a prompt to the generative AI model.

[0341] The server sends the received post content to the generative AI model as a prompt. The prompt is structured as follows: "Please improve the following post content and generate a report: Post content: 'I feel the quality of the new product is poor. It needs further improvement.'" The input is the post content stored in the database, and the output is the prompt sent to the generative AI model.

[0342] Step 4:

[0343] A generative AI model edits posts.

[0344] A generative AI model (e.g., OpenAI's GPT-4) analyzes the prompt received from the server and edits the post content. The edited content might look something like, "I feel there are some areas for improvement in the quality of the new product. Further improvements are expected." The input is the prompt, and the output is the edited post content.

[0345] Step 5:

[0346] The server receives the edited content.

[0347] The server receives the edited content from the generative AI model. The received edited content is also stored in the database. The input is the edited post from the generative AI model, and the output is the edited post stored in the database.

[0348] Step 6:

[0349] The server generates a report of the edits.

[0350] The server compares the content before and after editing and automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. For example, it might say, "Because saying 'poor quality' could damage the company's image, we softened the expression to 'there are some areas for improvement in quality.'" The input is the post content before and after editing, and the output is the generated report.

[0351] Step 7:

[0352] The server sends the report to the device.

[0353] The server sends the generated report to the employee's terminal. The report is displayed in a dedicated interface that the employee can access. The input is the generated report, and the output is the report sent to the terminal.

[0354] Step 8:

[0355] The terminal displays the report to the employee.

[0356] The terminal displays the report received from the server to the employee. The employee can check the report on his / her own terminal. The input is the report sent from the server, and the output is the report displayed on the terminal.

[0357] Step 9:

[0358] Employees review the reports and use them for training purposes.

[0359] Employees can check the displayed report and learn which parts of the post were problematic and how they were improved. This can be used to educate employees. The input is the report displayed on the terminal, and the output is the employee's understanding and the effectiveness of the education.

[0360] (Application example 3)

[0361] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[0362] On online platforms such as social media and blogs, if users post content that is inappropriate or contains dangerous information, it can become public and pose a significant risk to companies and individuals. Manually monitoring and editing these posts is extremely time-consuming and inefficient. Furthermore, the lack of detailed feedback on edited content makes it difficult for users and security personnel to take appropriate action.

[0363] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0364] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having an employee review the report, a means for sharing the report with the company, a means for sharing the report with a security officer, a means for describing in detail which parts were dangerous and why, and how they were edited, when generating the report, and a means for sharing the report with the user. This makes it possible to automatically detect and edit the riskiness of posted content and provide detailed feedback.

[0365] "Generative AI methods" are artificial intelligence technologies that analyze content posted on online platforms such as social media and blogs, detect dangerous information or inappropriate expressions, and edit them as necessary.

[0366] "Posted content" refers to information such as text, images, and videos that a user intends to publish on online platforms such as social media and blogs.

[0367] A "report" is a document that details which parts of a post were edited by generative AI tools, why they were dangerous, and how they were edited.

[0368] An "employee" is an individual who belongs to a company or organization and performs work for that company or organization.

[0369] A "company" is a legal entity established to carry out a specific business.

[0370] A "security officer" is an individual responsible for managing and monitoring information security within a company or organization.

[0371] "User" refers to an individual who creates and publishes content using online platforms such as social media and blogs.

[0372] "Automatic generation means" refers to technology for automatically creating a report on the content of posts edited by the generative AI means.

[0373] "Means of sharing" refers to the technology used to send and provide generated reports to specific individuals or organizations.

[0374] A system for implementing this invention monitors content posted to online platforms such as social networking sites and blogs in real time, automatically edits dangerous information and inappropriate language, and generates and shares detailed reports on the edited content. Specific embodiments of this system are described below.

[0375] System Configuration

[0376] 1. Hardware Configuration

[0377] Server: Use a server equipped with a high-performance processor and large memory capacity. For example, use an EC2 instance from Amazon Web Services (AWS (registered trademark)).

[0378] Device: The smartphone or head-mounted display used by the user. For example, we will use an iPhone (registered trademark) or Oculus Quest.

[0379] 2. Software Configuration

[0380] Generative AI models: Use natural language processing models such as GPT-4.

[0381] Social Media API: Use APIs to retrieve posts from social media platforms, such as Twitter API or Facebook Graph API.

[0382] Report generation libraries: Use libraries such as ReportLab to generate reports.

[0383] Emailing API: Share reports using email sending services like SendGrid.

[0384] Cloud Storage API: Save and share your reports using cloud storage services such as AWS S3.

[0385] Data processing and calculation

[0386] 1. Obtaining input data

[0387] The server uses the SNS API to obtain the content that the user is about to post in real time.

[0388] 2. Data analysis and editing

[0389] The server uses a generative AI model (e.g., GPT-4) to analyze the acquired posts. Specifically, it uses natural language processing technology to detect risk and inappropriate expressions in the posts.

[0390] The generative AI model automatically edits the detected parts.

[0391] 3. Generate a report

[0392] The server generates a detailed report of the edited post, detailing which parts were dangerous, why they were dangerous, and how they were edited.

[0393] Use a report generation library (e.g. ReportLab) to generate reports in PDF format.

[0394] 4. Share the report

[0395] The server sends the generated report to the user and security officer using an email sending API (e.g., SendGrid).

[0396] It also uses cloud storage APIs (e.g. AWS S3) to store reports in cloud storage and make them accessible when needed.

[0397] Specific examples

[0398] For example, if a user posts on social media, "This product is completely unusable, it's the worst!", the server retrieves this post via the social media API. A generative AI model (e.g., GPT-4) analyzes the post and edits it to say, "This product didn't work for me, but it might work for other people." A detailed report of the edits is then generated and shared with the user and security personnel.

[0399] Prompt Sentence Examples

[0400] Analyze the following content that users are attempting to post on social media, detect and edit any potentially dangerous or inappropriate content, and generate a detailed report of the edits.

[0401] Posting: "This product is completely unusable and awful!"

[0402] In this way, the system can automatically detect and edit potentially dangerous content and provide detailed feedback.

[0403] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0404] Step 1:

[0405] The server uses the SNS API to obtain the content that the user is about to post in real time. The input is the user's post content, and the output is the obtained post content. Specifically, the server sends a request to the SNS API to obtain the user's post content.

[0406] Step 2:

[0407] The server uses a generative AI model (e.g., GPT-4) to analyze the acquired post content. The input is the acquired post content, and the output is the analysis result. Specifically, the server inputs prompt sentences into the generative AI model to detect risk and inappropriate expressions in the post content.

[0408] Step 3:

[0409] The server uses a generative AI model to automatically edit the detected dangerous parts and inappropriate expressions. The input is the analysis result, and the output is the edited post content. Specifically, the server gives editing instructions to the generative AI model and appropriately corrects the post content.

[0410] Step 4:

[0411] The server generates a detailed report of the edited post. The input is the edited post and the output is the generated report. Specifically, the server uses a report generation library (e.g. ReportLab) to create a report detailing which parts were dangerous, why, and how they were edited.

[0412] Step 5:

[0413] The server sends the generated report to the user and security officer using an email sending API (e.g., SendGrid). The input is the generated report, and the output is the sent email. Specifically, the server attaches the report to the email sending API and sends it to the specified email address.

[0414] Step 6:

[0415] The server stores the generated report using a cloud storage API (e.g. AWS S3) and makes it accessible when needed. The input is the generated report and the output is the report stored in cloud storage. Specifically, the server uploads the report to the cloud storage API and obtains the URL of the storage location.

[0416] 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.

[0417] "Example 1"

[0418] In one embodiment of the present invention, generative AI is used in combination with an emotion engine that recognizes user emotions. Specifically, before a user posts to a social networking site, the content of the post is input into the generative AI. The generative AI then determines whether the content of the post is likely to spark a controversy or contain confidential company information.

[0419] "Example 2"

[0420] The emotion engine extracts emotions from the content of users' social media posts. For example, it extracts the emotion of "joy" from a post such as "Today was fun!", and the emotion of "distress" from a post such as "Today was tough...".

[0421] "Example 3"

[0422] If the generative AI judges the content of a post to be dangerous, or if the emotion engine judges the user's emotions to be negative, the AI ​​will edit the post, specifically deleting or rewriting any parts that could spark outrage or that contain confidential company information.

[0423] "Example 4"

[0424] After the generative AI edits the post, it automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. This report is then reviewed by employees, and used to educate them.

[0425] The processing flow of each embodiment will be described below.

[0426] "Example 1"

[0427] Step 1: Before a user posts on social media, the content of the post is input into the generative AI.

[0428] Step 2: The generative AI determines whether the content of the post is likely to spark an uproar or contain confidential company information.

[0429] "Example 2"

[0430] Step 1: The emotion engine extracts emotions from the user's social media posts.

[0431] Step 2: For example, extract the emotion of "joy" from a post such as "Today was fun!", and the emotion of "distress" from a post such as "Today was tough...".

[0432] "Example 3"

[0433] Step 1: If the generative AI determines that the post is dangerous, or if the emotion engine determines that the user's emotion is negative, the AI ​​will edit the post.

[0434] Step 2: Specifically, delete or rewrite any parts that could cause controversy or that contain confidential company information.

[0435] "Example 4"

[0436] Step 1: After the generative AI edits the post, it automatically generates a report on the edits.

[0437] Step 2: The report will detail which parts were unsafe, why, and how they were edited.

[0438] Step 3: Have employees review this report and use it to educate them.

[0439] Example 1

[0440] Next, a description will be given of Example 1 of Form 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."

[0441] There is a need to reduce the risk of posts on social media causing outrage and leaking confidential company information. However, in conventional systems, post content is often checked manually, which is inefficient. In addition, there is a lack of means to recognize the sentiment of posts and provide appropriate feedback, making it difficult for users to make appropriate decisions.

[0442] 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.

[0443] In this invention, the server includes generative AI means for determining the content of posts to SNS, generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, means for automatically generating a report on the posted content edited by the generative AI, means for having employees review the report, means for sharing the report with the organization, means for checking keywords and phrases included in the posted content, emotion engine means for recognizing emotions from the posted content, and means for notifying the emotion engine of the determination result. This reduces the risk of content posted to SNS causing a flame war or the risk of confidential information being leaked, and enables users to make appropriate decisions.

[0444] "Generative AI methods" refers to artificial intelligence technology that analyzes the content of posts on social media and determines the risk of online outrage or the risk of confidential information being leaked.

[0445] "Generative AI editing means" refers to artificial intelligence technology that automatically corrects and edits posted content that is deemed dangerous by generative AI.

[0446] "Means for automatically generating reports" refers to technology that automatically creates detailed reports about the content of posts edited by generative AI.

[0447] "Means for employees to check" refers to a technique for presenting the generated report to employees and having them check its contents.

[0448] "Means for sharing within the organization" refers to the technology used to share the generated reports within the organization and make them accessible to relevant parties.

[0449] "Keyword and phrase checking measures" refers to technology that detects specific keywords and phrases in posts and assesses risk.

[0450] "Emotion engine means" refers to technology for recognizing a user's emotions from the content of their posts and determining the type and intensity of the emotions.

[0451] "Means for notifying the judgment results" refers to technology for notifying the user of the analysis results of the generative AI and emotion engine and providing appropriate feedback.

[0452] This invention is a system that checks the content of posts to social media in advance to reduce the risk of flame wars and leaks of confidential information. The system includes a generative AI means, an emotion engine means, a report generation means, etc. Specific embodiments of this system are described below.

[0453] Hardware and software used

[0454] Hardware:

[0455] Server: A central processing unit for analyzing submissions and generating reports.

[0456] Device: The device (PC, smartphone, etc.) that a user uses to enter content to post on a social networking site.

[0457] software:

[0458] Generative AI model: An artificial intelligence technology that analyzes the content of posts and determines the risk of flame wars and the risk of confidential information leaks.

[0459] Emotion engine: Technology for recognizing user emotions from posted content and determining the type and intensity of emotions.

[0460] Report generation software: Software for automatically generating reports on posts edited by generative AI.

[0461] Data processing and calculation

[0462] Enter your post content:

[0463] The user uses the device to input the content to post on the social networking site. For example, the user might type, "The new project isn't progressing at all. My boss's instructions are too vague!"

[0464] Generative AI analysis:

[0465] The device sends the input post content to the generative AI, which checks the keywords and phrases contained in the post to determine the risk of a controversy or the leakage of confidential information. For example, it detects keywords such as "project," "instructions," and "ambiguous."

[0466] Emotion Recognition with Emotion Engine:

[0467] The server uses an emotion engine to recognize the user's emotion from the post, for example, determining whether the post expresses anger or frustration. The emotion engine analyzes the context and wording of the post to identify the user's emotion.

[0468] Report generation and notification:

[0469] The server automatically generates a report based on the analysis results of the generative AI and emotion engine. This report explains why the post was deemed dangerous and what emotions it contains. The generated report is sent to employees' devices for review and is then shared within the organization.

[0470] Examples and prompts

[0471] Examples:

[0472] If a user types, "I'm not making any progress on my new project. My boss's instructions are too vague!", the following process will occur:

[0473] Example prompt sentence:

[0474] Check out the post below:

[0475] "The new project isn't progressing at all. My boss's instructions are too vague!"

[0476] Determine whether this post is likely to cause controversy or contain confidential company information.

[0477] The server inputs this prompt into the generative AI, which then analyzes the content of the post and notifies the user that there is a high risk of a controversy or that the post may contain confidential information about the project.

[0478] In this way, users can reduce the risk of a social media firestorm or the leakage of confidential information before posting on social media.

[0479] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0480] Step 1:

[0481] The user enters the post content

[0482] The user uses the device to input the content to be posted on the social networking site. For example, the user might input, "The new project isn't progressing at all. My boss's instructions are too vague!" This input content is then processed in the next step.

[0483] Step 2:

[0484] The device sends the post content to the generative AI

[0485] The device sends the input post content to the generative AI. Specifically, it generates a prompt sentence: "Please check the following post content: 'The new project isn't progressing at all. My boss's instructions are too vague!'" and sends it to the generative AI. This prompt sentence becomes the input data for the generative AI.

[0486] Step 3:

[0487] The server analyzes the content of posts using generative AI

[0488] The server uses generative AI to analyze the content of the post. The generative AI analyzes the prompt and checks for keywords and phrases contained in the post. For example, it detects keywords such as "project," "instructions," and "ambiguous" to determine the risk of a controversy or the risk of confidential information being leaked. The results of this analysis become the input data for the next step.

[0489] Step 4:

[0490] The server recognizes emotions using an emotion engine

[0491] The server uses an emotion engine to recognize the user's emotions from the posted content. The emotion engine analyzes the context and wording of the posted content to identify the user's emotions. For example, it determines whether the posted content expresses anger or frustration. The emotion recognition results become input data for the next step.

[0492] Step 5:

[0493] The server generates a report based on the analysis results.

[0494] The server automatically generates a report based on the analysis results of the generative AI and emotion engine. This report explains why the post was deemed dangerous and what emotions it contains. The generated report serves as input data for the next step.

[0495] Step 6:

[0496] The server notifies the user of the report

[0497] The server notifies the user of the generated report, for example, by displaying a message such as "This post may cause controversy" or "This post contains confidential information." The user can then decide whether to modify the post or refrain from posting.

[0498] Step 7:

[0499] The server shares the report with the organization

[0500] The server shares the generated reports within the organization and makes them accessible to relevant parties, allowing the entire organization to manage the risks of posted content and take appropriate action.

[0501] (Application example 1)

[0502] Next, a description will be given of Application Example 1 of Embodiment 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."

[0503] In today's companies, there is a growing risk that content posted by employees or part-time workers on social media will spark controversy or leak confidential company information. Effective measures to prevent such risks are needed. Furthermore, it is necessary to protect the company's brand image by analyzing the sentiment of the posts and displaying appropriate warnings if negative sentiment is included.

[0504] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means. In this invention, the server includes: a generative AI means for determining the content of posts to SNS; a generative AI means for editing the posted content if the generative AI determines that the posted content is risky; a means for automatically generating a report on the posted content edited by the generative AI; a means for having an employee review the report; a means for sharing the report with the company; an emotion engine means for analyzing the emotion of the posted content; a means for displaying a warning if the emotion engine detects a negative emotion; and a means for determining, by the generative AI, whether the posted content has the potential to cause a controversy or contains confidential information. This makes it possible to evaluate the risk of the posted content to SNS in advance and take appropriate measures.

[0505] "Generative AI methods" are artificial intelligence technologies that analyze the content of posts on social media and determine whether the content has the potential to cause a firestorm or whether it contains confidential information.

[0506] The "emotion engine means" is a technology for analyzing the emotions in the posted content and determining whether or not it contains negative emotions.

[0507] The "means for automatically generating reports" refers to a technology that automatically creates detailed reports about the content of posts edited by generative AI.

[0508] The "means for having employees check" is a technique for presenting the generated report to employees and having them check the contents.

[0509] The "means for sharing with the company" is a technique for sharing the generated report with relevant parties in the company.

[0510] The "means for displaying a warning" is a technique for displaying a warning to the user when a negative emotion is detected by the emotion engine.

[0511] In order to implement this invention, it is necessary to build a system that includes a generative AI means for determining the content of posts on SNS, an emotion engine means, a means for automatically generating reports, a means for employees to check the reports, a means for sharing the reports with the company, and a means for displaying warnings.

[0512] System Program

[0513] This system is realized using the following hardware and software.

[0514] Hardware: Smartphones, servers

[0515] Software: Python, OpenAI API, TextBlob

[0516] Processing Description

[0517] Generative AI method for determining the content of posts on social media

[0518] The server receives the content of posts on social media and analyzes it using a generative AI model to determine whether the content has the potential to spark controversy or whether it contains confidential information.

[0519] Emotion Engine Means

[0520] The server analyzes the sentiment of the post using TextBlob and displays a warning to the user if a negative sentiment is detected by the sentiment engine.

[0521] A means of automatically generating reports

[0522] The server automatically generates a detailed report of the post edited by the generative AI, including an explanation of why the post was dangerous.

[0523] How to get employees to check

[0524] The server presents the generated report to the employee for confirmation, and the employee checks the report and corrects the posted content as necessary.

[0525] How to share with the company

[0526] The server then shares the generated report with relevant parties within the company, allowing the entire company to understand the risks of posted content and take appropriate action.

[0527] How to display a warning

[0528] If the emotion engine detects negative emotions, the server displays a warning to the user, giving them an opportunity to reconsider their post.

[0529] Specific examples

[0530] For example, if an employee tries to post, "Today we'll be announcing details about our new product!", the system works as follows:

[0531] 1. Generative AI method: The server inputs the posted content into the generative AI model and analyzes it.

[0532] 2. Sentiment Engine Method: The server uses TextBlob to analyze the sentiment of the post. In this case, the sentiment is positive, so it proceeds to the next step.

[0533] 3. Means for automatically generating reports: The server automatically generates detailed reports about the posts edited by the generative AI.

[0534] 4. Means for employee to check: The server presents the generated report to the employee and asks them to check the contents.

[0535] 5. Means of sharing with the company: The server shares the generated report with the company's stakeholders.

[0536] 6. Means for displaying a warning: If a negative emotion is detected by the emotion engine, the server displays a warning to the user.

[0537] Prompt Sentence Examples

[0538] Please assess whether the following posts have the potential to cause controversy or contain confidential company information:

[0539] Today we're announcing new product details!

[0540] In this way, the system can assess the risks of posts on social media in advance and take appropriate action.

[0541] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0542] Step 1:

[0543] The server receives posts from users to the SNS. The input is the post entered by the user, and the output is the text data of the post.

[0544] Step 2:

[0545] The server inputs the received post content into a generative AI model for analysis. The input is the text data of the post content, and the output is the analysis results. This analysis determines whether the post content has the potential to cause a controversy or whether it contains confidential information.

[0546] Step 3:

[0547] The server analyzes the sentiment of the post using TextBlob. The input is the text data of the post, and the output is the result of the sentiment analysis. If the sentiment engine detects a negative sentiment, the next step is to display a warning.

[0548] Step 4:

[0549] If the emotion engine detects a negative emotion, the server displays a warning to the user. The input is the result of the emotion analysis, and the output is a warning message. The user receives this warning and has the opportunity to reconsider what they posted.

[0550] Step 5:

[0551] The server automatically generates a detailed report of the post edited by the generative AI. The input is the analysis result of the generative AI model, and the output is a detailed report, including an explanation of why the post was dangerous.

[0552] Step 6:

[0553] The server presents the generated report to the employee for review. The input is the detailed report, and the output is the employee's review. The employee reviews the report and corrects the submission as necessary.

[0554] Step 7:

[0555] The server shares the generated report with company stakeholders. The input is a detailed report, and the output is a report shared with company stakeholders. This allows the entire company to understand the risks of posted content and take appropriate measures.

[0556] Example 2

[0557] Next, a description will be given of Example 2 of Form 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."

[0558] Posts on social media often have the potential to spark outrage and pose a risk of leaking confidential company information. This can threaten a company's reputation and security. Furthermore, it is difficult to accurately grasp a user's psychological state because it is difficult to properly extract and analyze emotions from user posts. A system that can solve these issues is needed.

[0559] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0560] In this invention, the server includes generative AI means for determining the content of posts to the SNS, generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, means for returning the edited posted content by the generative AI to the user, emotion engine means for extracting emotions from the posted content, means for analyzing the emotions extracted by the emotion engine, and means for providing the analysis results to the user. This makes it possible to accurately grasp the emotions of users while ensuring the safety of the content posted to the SNS.

[0561] "Generative AI methods" are artificial intelligence technologies that analyze the content of posts on social media, detect parts that could cause outrage or confidential information, and edit them as necessary.

[0562] The "emotion engine means" is a natural language processing technology for extracting user emotions from the content of social media posts.

[0563] The "means of returning to the user" refers to the communication means for sending the posted content edited by the generative AI to the user's device.

[0564] The "means for analyzing emotions" is a technique for analyzing the emotion data extracted by the emotion engine and evaluating the user's psychological state.

[0565] "Means for providing analysis results to users" refers to means for notifying users of the analysis results from the emotion engine and generative AI.

[0566] This invention is a system for safely and appropriately managing content posted to a social networking service. The system includes a generative AI means, an emotion engine means, a means for sending feedback to a user, a means for analyzing emotions, and a means for providing the analysis results to a user.

[0567] The server provides the hardware and software required to run the generative AI model and emotion engine, specifically using servers equipped with NVIDIA GPUs and deep learning frameworks such as TENSORFLOW® and PyTorch.

[0568] When a user posts on an SNS, the device sends the content of the post to the server. The server first uses an emotion engine to extract emotions from the content of the post. For example, from a post such as "Today was fun!", the emotion of "joy" is extracted.

[0569] Next, generative AI analyzes the content of posts to detect parts that could spark controversy or touch on confidential company information. For example, if a post says, "Our new product is better than our competitors' products," the generative AI rewrites this part to read, "Our new product has superior performance."

[0570] The server then sends the edited post back to the user, who then makes a final confirmation.

[0571] As a concrete example, suppose a user posts the following:

[0572] "Today was tough... but our new product is better than our competitors' products!"

[0573] When this post is sent to the server, the emotion engine extracts the emotions of "distress" and "happiness." The generative AI then rewrites the part that says, "Our new product is better than our competitors' products," to "Our new product has superior performance."

[0574] Example prompt sentence:

[0575] Please analyze the following social media posts and remove or rewrite any potentially inflammatory or confidential information:

[0576] "Today was tough... but our new product is better than our competitors' products!"

[0577] By inputting this prompt into a generative AI model, appropriate editing is performed.

[0578] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0579] Step 1:

[0580] A user posts to a social networking site.

[0581] A user uses their own device (such as a smartphone or PC) to input content to post on a social networking site and presses the post button. For example, a user might post, "Today was tough... but our new product is better than our competitors' products!" The input is the user's post, and the output is the post displayed on the device.

[0582] Step 2:

[0583] The device sends the post content to the server.

[0584] The terminal sends the content entered by the user to the server. This data is sent securely using the HTTPS protocol. The input is the user's post, and the output is the post sent to the server.

[0585] Step 3:

[0586] The server uses an emotion engine to extract emotions from the posted content.

[0587] The server inputs the received post content into the emotion engine. The emotion engine uses natural language processing technology to extract emotions from the post content. For example, it extracts the emotion of "distress" from the part "Today was tough..." and the emotion of "joy" from the part "But our new product is better than our competitors' products!" The input is the post content sent to the server, and the output is the extracted emotion data.

[0588] Step 4:

[0589] The server uses generative AI to analyze the content of the post and edit it as necessary.

[0590] The server inputs the content of the post into a generative AI model to detect potentially inflammatory parts and confidential information. For example, it detects the part that says, "Our new product is better than our competitors' products," and rewrites it to, "Our new product has superior performance." The generative AI model uses a pre-trained dataset. The input is the extracted emotion data and the post content, and the output is the edited post content.

[0591] Step 5:

[0592] The server sends the edited post back to the user.

[0593] The server sends the edited post back to the user's device, again securely using the HTTPS protocol. The input is the edited post, and the output is the post sent back to the user's device.

[0594] Step 6:

[0595] The user provides final confirmation.

[0596] The user checks the edited post displayed on the device. If necessary, they can make further corrections. Finally, the user confirms the post and publishes it to the SNS. The input is the post sent back to the user's device, and the output is the final confirmed post.

[0597] (Application example 2)

[0598] Next, a description will be given of Application Example 2 of Form 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."

[0599] In today's social networking sites, users often unintentionally post risky content, leading to problems such as online outrage and information leaks. Another issue is the difficulty of properly extracting emotions from user posts and providing feedback. To solve these problems, a system is needed that automatically assesses the riskiness of posts and edits them as necessary. It is also necessary to support users' psychological health by extracting their emotions and providing appropriate feedback.

[0600] 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.

[0601] In this invention, the server includes a generative AI means for determining the content of posts to the social networking site, a generative AI means for editing the post if the generative AI determines that the post is dangerous, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, an emotion engine means for extracting emotions from the post, and a means for providing feedback on the extracted emotions to the user. This prevents users from unintentionally posting dangerous content and reduces the risk of online outrage and information leaks. Furthermore, by appropriately extracting and providing feedback on the user's emotions, the user's psychological health can be supported.

[0602] "Generative AI methods" are methods that use artificial intelligence technology to analyze the content of posts on social media and determine whether they are dangerous.

[0603] "Generative AI editing means" refers to means that use artificial intelligence technology to automatically modify or delete posted content that is deemed dangerous by generative AI.

[0604] "Means for automatically generating reports" refers to a means for automatically creating a report that explains the reasons for the danger and the edited content based on the edited content of a post using generative AI.

[0605] The "means for employee confirmation" refers to a means for presenting the automatically generated report to an employee for confirmation and approval.

[0606] "Means for sharing within the organization" refers to the means for sharing confirmed reports within the organization and disseminating information widely.

[0607] The "emotion engine means" is a means that uses technology to extract user emotions from the content posted on SNS.

[0608] The "feedback means" is a means for appropriately conveying the extracted emotion to the user.

[0609] A system for carrying out this invention includes a generative AI means for determining the content of posts to SNS and editing it as necessary, an emotion engine means for extracting emotions from the posted content, and a means for providing feedback of the extracted emotions to the user. A specific embodiment of this system is described below.

[0610] System Configuration

[0611] The system mainly uses the following hardware and software:

[0612] Hardware: Smartphones, servers

[0613] Software: Python, transformers library, sentiment analysis model

[0614] Program processing

[0615] The server receives the content posted to the SNS and analyzes it using the generative AI means. The generative AI means determines whether the content is dangerous, and if it is determined to be dangerous, the editing generative AI means automatically edits or deletes the content. For edited content, the automatic report generation means creates a report explaining the reason for the edit and the edited content.

[0616] Next, the emotion engine means extracts the user's emotion from the posted content. The extracted emotion is conveyed to the user through the feedback means. This allows the user to understand how their posted content is received and supports their psychological well-being.

[0617] Specific examples

[0618] For example, if a user posts the following:

[0619] Our new product is better than our competitors' products. Today was tough...

[0620] This post content is sent to a server and analyzed by the generative AI means. The generative AI means judges the part "better than competitors' products" to be dangerous, and the editing generative AI means rewrites it to "has excellent performance." After that, the emotion engine means extracts the emotion of "distress" from the part "Today was tough..."

[0621] Prompt Sentence Examples

[0622] An example of a prompt to input to a generative AI model is as follows:

[0623] Check your posts, edit out the risky parts, and extract the emotion.

[0624] Post: "Our new product is better than our competitors' products. Today was tough..."

[0625] Using this prompt, the generative AI model can check and edit posts and extract sentiment, helping users to make safe and appropriate posts and understand their own emotions.

[0626] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0627] Step 1:

[0628] The server receives posts from users to the SNS. The input is the user's post, and the output is the text data of the post. Specifically, the user enters the post content using a smartphone and sends it to the server.

[0629] Step 2:

[0630] The server analyzes the received post using a generative AI method. The input is the text data of the post received in step 1, and the output is the analysis result. Specifically, the generative AI model scans the post and detects dangerous keywords and phrases.

[0631] Step 3:

[0632] If the server determines that the posted content is dangerous, it uses the generative AI means to edit or delete the posted content. The input is the analysis result obtained in step 2, and the output is the corrected posted content. Specifically, the dangerous parts are rewritten to safe expressions.

[0633] Step 4:

[0634] The server creates a report using a means for automatically generating a report on the edited post content using the generative AI means. The input is the post content corrected in step 3, and the output is a report. Specifically, the server generates a report that includes content explaining why the part was dangerous and how it was corrected.

[0635] Step 5:

[0636] The server presents the generated report to the employee using a means for the employee to review it. The input is the report generated in step 4, and the output is the employee's review result. Specifically, the employee reviews the report and provides feedback on approval or correction.

[0637] Step 6:

[0638] The server shares the confirmed report within the organization using a means for sharing the report with the organization. The input is the confirmation result obtained in step 5, and the output is the report shared within the organization. Specifically, the server saves the report in the organization's database and notifies the relevant parties.

[0639] Step 7:

[0640] The server uses an emotion engine to extract the user's emotions from the posted content. The input is the text data of the posted content received in step 1, and the output is the extracted emotion data. Specifically, the emotion analysis model analyzes the posted content and identifies emotions such as joy, sadness, and anger.

[0641] Step 8:

[0642] The server communicates the extracted emotion to the user using a means of providing feedback to the user. The input is the emotion data extracted in step 7, and the output is a feedback message to the user. Specifically, the server sends a notification to the user's smartphone, explaining the extracted emotion and the reason for it.

[0643] Example 3

[0644] Next, a description will be given of a third embodiment of the third embodiment. 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."

[0645] When content posted on social media poses risks such as flame wars or the leaking of confidential information, it is necessary to appropriately detect and edit it. It is also necessary to generate detailed reports on the edited content to help educate employees and gather information for the company. However, performing these processes manually is time-consuming, labor-intensive, and inefficient. Therefore, there is a need for a system that can automatically detect risks in posted content, edit it, and generate and share reports on the content.

[0646] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0647] In this invention, the server includes means for receiving posted content, means for passing the received posted content to a generative AI, means for the generative AI to analyze the posted content and detect risky parts or negative emotions, means for the generative AI to edit the posted content based on the analysis results, means for the generative AI to automatically generate a report on the edited content, means for sending the generated report to employees to check, and means for sharing the generated report with the company and collecting information on employees' complaints and good points. This makes it possible to automatically detect risks in posted content, edit it, and generate and share reports on the content.

[0648] "Posted content" refers to text or media information that a user sends to a social networking site or other platform.

[0649] "Generative AI" is a system that uses artificial intelligence technology to analyze data and automatically perform specific tasks.

[0650] "Analysis" is the process by which generative AI evaluates the content of posts and detects risky parts or negative emotions.

[0651] "Editing" is the process by which generative AI changes the content of posts based on the analysis results to prevent outrage or the leaking of confidential information.

[0652] A "report" is a document automatically generated by generative AI about the edits, detailing which parts were dangerous, why, and how they were edited.

[0653] "Employees" are individuals employed by a company who review the generated reports and use them for training purposes.

[0654] The "company" refers to an organization or business entity that shares the generated reports and collects information on employee complaints and positive points.

[0655] "Dangerous parts" are parts of the post that could cause an uproar or parts that touch on confidential company information.

[0656] "Negative emotions" are negative emotions of users detected by the emotion engine, which affect the tone and expression of the content of posts.

[0657] The "emotion engine" is software that evaluates the emotional tone of posts and functions as part of a generative AI.

[0658] This invention is a system that automatically analyzes content posted to social networking sites, detects and edits dangerous parts and negative emotions, and generates and shares reports on the edited content. Specific embodiments of this system are described below.

[0659] Hardware and software used

[0660] Server: A central computer system that receives posts and passes them on to the generative AI.

[0661] Device: The device (e.g., computer, smartphone) on which a user enters and submits their post.

[0662] Generative AI models: Artificial intelligence systems that analyze and edit posts (e.g., OpenAI GPT-4).

[0663] Emotion engine: Software to assess the emotional tone of posts (e.g., IBM Watson® Tone Analyzer).

[0664] System Operation

[0665] 1. The user enters the content to post on their device and presses the send button.

[0666] 2. The server receives this post and temporarily stores it in a database.

[0667] 3. The server creates an API request to pass the received post content to the generative AI.

[0668] 4. Generative AI analyzes posts to detect risk and negative sentiment. Specifically, it uses an emotion engine to evaluate the emotional tone of posts.

[0669] 5. The generative AI edits the post based on the analysis results. For example, a post that says, "The new product's functionality is insufficient, and we've received many complaints from customers" can be edited to say, "We are considering improving the functionality of our new product based on feedback from customers."

[0670] 6. The generative AI automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. For example, it might say, "The expression 'many complaints' was negative, so to emphasize customer dissatisfaction, we edited it to 'we are considering improvements based on feedback.'"

[0671] 7. The server sends the generated report to the employee for review. Specifically, the report is sent to the employee's email address as an attachment.

[0672] 8. The server also saves the report to the company database and collects information on employees' dissatisfaction and positive points. For example, the report may reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[0673] Examples of concrete examples and prompts

[0674] Examples:

[0675] A user posts, "The new product's functionality is insufficient, and we're receiving a lot of complaints from customers."

[0676] The generative AI edits the article to say, "We are currently considering improving the functionality of our new product based on feedback from customers."

[0677] The report states, "The expression 'many complaints' was negative, and to emphasize customer dissatisfaction, we edited it to 'we are considering improvements in response to feedback.'"

[0678] Example prompt sentence:

[0679] "Analyze user posts, detect risky or negative sentiment, and edit them. Generate detailed reports on edits."

[0680] This system makes it possible to automatically detect risks in posted content, edit it, and generate and share reports on the content. This makes it possible to efficiently educate employees and collect information about the company. The flow of the identification process in Example 3 will be explained using FIG. 21.

[0681] Step 1:

[0682] The user inputs the content to be posted on the terminal and presses the send button.

[0683] Input: User-typed post content (e.g., "Our new product's functionality is insufficient, and customers are complaining about it.").

[0684] Output: The post is sent to the server.

[0685] Specific behavior: The user enters the content to post into the input field on the device and clicks the send button.

[0686] Step 2:

[0687] The server receives the posted content and temporarily stores it in a database.

[0688] Input: The post submitted by the user.

[0689] Output: Post content temporarily saved in the database.

[0690] Specific operation: The server receives the HTTP request and saves the post content to the database.

[0691] Step 3:

[0692] The server creates an API request to pass the received post content to the generative AI.

[0693] Input: Post content saved in database.

[0694] Output: API request to be passed to the generative AI.

[0695] Specific operation: The server retrieves the post content and sends a request to the generative AI's API endpoint.

[0696] Step 4:

[0697] Generative AI analyzes the content of posts and detects dangerous parts and negative emotions.

[0698] Input: The post content sent from the server.

[0699] Output: Analysis results (detection of dangerous parts and negative emotions).

[0700] How it works: Generative AI uses an emotion engine to assess the emotional tone of posts and identify risky areas.

[0701] Step 5:

[0702] The generative AI edits the post content based on the analysis results.

[0703] Input: Analysis results and original post.

[0704] Output: The edited post.

[0705] How it works: Based on the analysis results, the generative AI edits dangerous parts and negative expressions in the post. For example, it edits "The new product's functionality is insufficient, resulting in many customer complaints" to "We are currently considering improving the functionality of the new product based on customer feedback."

[0706] Step 6:

[0707] Generative AI automatically generates a report on the edits.

[0708] Input: Edited post and analysis results.

[0709] Output: A report of your edits.

[0710] What it does: The generative AI generates a report detailing the reasons for the edits and the changes made. For example, it might say, "The expression 'many complaints' is negative, and to emphasize customer dissatisfaction, we edited it to 'we are considering improvements based on feedback.'"

[0711] Step 7:

[0712] The server sends the generated report to the employee for review.

[0713] Input: The generated report.

[0714] Output: Report sent to employee.

[0715] Specific operation: The server attaches the report and sends it to the employee's email address.

[0716] Step 8:

[0717] The server also stores the reports in the company's database, collecting information on employee complaints and positive points.

[0718] Input: The generated report.

[0719] Output: Reports saved in the company database.

[0720] What it does: The server saves the report in the company database and uses it for analysis. For example, the report may reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[0721] (Application example 3)

[0722] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[0723] The content posted by employees and users on social media and other online platforms can have a negative impact on a company's reputation and security. In particular, posts that have the potential to spark controversy or contain confidential information pose a significant risk to companies. Furthermore, there is a lack of efficient ways to collect information about employees' complaints and positive points and use it to educate them. To address these issues, a system is needed that monitors posts in real time, automatically edits dangerous content, and generates detailed reports on the edits.

[0724] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0725] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having employees review the report, a means for sharing the report with the company, a means for monitoring the posted content in real time, a means for displaying the edited content of the posted content, and a means for using the report to help educate employees. This makes it possible to monitor the dangers of posted content in real time and edit it appropriately to reduce corporate reputation and security risks, and also to help educate employees and improve the company.

[0726] "Generative AI methods" refers to artificial intelligence technology that judges the content of posts on social media and edits them as necessary.

[0727] "Posted content" refers to information such as text, images, and videos that users make public on social media and other online platforms.

[0728] "If deemed dangerous" refers to a situation where the generative AI determines that the content of a post could have a negative impact on a company's reputation or security.

[0729] "Edit" means to delete, change, or add any part of a post.

[0730] "Means for automatically generating reports" refers to technology that automatically creates a report on the content of posts edited by generative AI, including the reasons for the edits and details.

[0731] "Means for employees to check" refers to a method or system for presenting the generated report to employees and having them check the contents.

[0732] "Means for sharing within the company" refers to the methods and systems for sharing generated reports throughout the company and efficiently collecting and utilizing information.

[0733] "Real-time monitoring measures" refers to technology that allows for immediate monitoring of posted content and rapid detection of dangerous content.

[0734] "Means for displaying edited content" refers to a method or system for displaying the post content after it has been edited by generative AI to users and employees.

[0735] "Means for employee education" refers to a method or system for educating or training employees using the generated reports.

[0736] A system for implementing this invention monitors content posted to social networking sites in real time, automatically edits dangerous content using generative AI, and generates a detailed report on the edited content. A specific embodiment of this system is described below.

[0737] System Configuration

[0738] The system consists of the following main components:

[0739] 1. Generative AI: Evaluates the content of posts on social media and edits them as necessary.

[0740] 2. Real-time monitoring measures: Immediately monitor posted content and quickly detect dangerous content.

[0741] 3. Automatic report generation: Automatically generate a report for edited posts, detailing the reasons for the edits and other details.

[0742] 4. Employee confirmation method: Present the generated report to employees and have them confirm the contents.

[0743] 5. Company sharing tool: Generated reports can be shared across the company to efficiently collect and utilize information.

[0744] 6. Display of edited content: The post content after being edited by the generative AI is displayed to users and employees.

[0745] 7. Employee education tool: Use the generated reports to educate and train employees.

[0746] Program processing

[0747] The server monitors the content posted on social media in real time and judges the content using a generative AI. If the judged content is dangerous, the generative AI edits it. A detailed report of the edited content is created by an automatic report generation means. This report is presented to employees through an employee confirmation means, and is further shared throughout the company using a company sharing means. The edited content is displayed to users and employees using a post-edit content display means. Finally, the generated report is used for employee education and training through an employee education means.

[0748] Hardware and software used

[0749] Hardware: Servers, user devices (PCs, smartphones, etc.)

[0750] Software: OpenAI API, real-time monitoring system, report generation system

[0751] Specific examples

[0752] For example, consider the following post on a social networking site:

[0753] Example of post content:

[0754] This new product is completely unusable and the company's response is terrible.

[0755] The server monitors the content of this post in real time and uses generative AI to generate the following prompt:

[0756] Example prompt sentence:

[0757] Please analyze the post below and edit out any dangerous parts.

[0758] Post content: This new product is completely unusable. The company's response is terrible.

[0759] Edited content:

[0760] The generative AI analyzes the posts based on these prompts and edits out any dangerous parts. The edited content and a detailed report are then shared with the entire company for review. Ultimately, this report can be used to educate and train employees.

[0761] In this way, the system can reduce reputational and security risks to the company and help educate employees and improve the company.

[0762] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0763] Step 1:

[0764] The server monitors posts on social media in real time. It receives posts obtained from the social media platform as input and passes the monitored posts to the generative AI as output. Specifically, the server connects with the social media platform via an API and obtains the content of each new post.

[0765] Step 2:

[0766] The server uses generative AI to judge the content of the post. As input, it receives the post content obtained in step 1, and as output, it obtains a judgment result as to whether the post content is dangerous. Specifically, the server generates the following prompt sentence for the generative AI and makes a judgment.

[0767] Please analyze the post below and edit out any dangerous parts.

[0768] Post content: [Post content]

[0769] Edited content:

[0770] Step 3:

[0771] If the generative AI determines that the posted content is dangerous, the server edits the posted content. The server receives the judgment result from step 2 and the posted content as input, and obtains the edited posted content as output. Specifically, the generative AI deletes or changes the dangerous parts and edits the content to be safe.

[0772] Step 4:

[0773] The server automatically generates a report on the post content edited by the generative AI. It receives the edited post content as input and generates a detailed report of the edited content as output. Specifically, the server generates the following prompt for the generative AI and creates the report:

[0774] I edited the post below.

[0775] Original post: [Original post]

[0776] Edited content: [Edited content]

[0777] Please provide details about which parts were dangerous, why, and how you edited them.

[0778] Step 5:

[0779] The server allows the employee to review the generated report. It receives the report as input and presents it to the employee as output. Specifically, the server sends the report to the employee's terminal and notifies them to review it.

[0780] Step 6:

[0781] The server shares the generated report with the company. It receives the report as input and generates information to be shared across the company as output. Specifically, the server uploads the report to the company's internal system and notifies relevant parties.

[0782] Step 7:

[0783] The server displays the edited content to users and employees. It receives the edited content as input and generates the display data as output. Specifically, the server updates the edited content on the social media platform or internal systems.

[0784] Step 8:

[0785] The server uses the generated reports to educate and train employees. It receives the reports as input and generates training materials as output. Specifically, the server creates an educational program based on the reports and provides it to employees.

[0786] 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.

[0787] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ) 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.

[0788] Another example of generative AI is Gemini (registered trademark) (Internet search engine). <url: https: gemini.google.com ?hl="ja">) are listed.

[0789] 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.

[0790] [Second embodiment]

[0791] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0792] 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.

[0793] 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).

[0794] 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.

[0795] 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.

[0796] 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).

[0797] 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.

[0798] 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.

[0799] 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.

[0800] 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.

[0801] 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.

[0802] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0803] "Example 1"

[0804] In one embodiment of the present invention, before employees or part-timers post on social media, the content of the post is input into a generative AI. The generative AI determines whether the content of the post is likely to spark an outrage or whether it contains confidential company information. Specifically, the AI ​​checks whether the content of the post contains keywords or phrases that could spark an outrage or keywords or phrases related to confidential company information.

[0805] "Example 2"

[0806] If the generative AI judges a post to be dangerous, it will edit it. Specifically, it will delete or rewrite any parts that could spark outrage or that touch on confidential company information. For example, if a post says, "Our new product is better than our competitors' products," the AI ​​will rewrite it to say, "Our new product has superior performance."

[0807] "Example 3"

[0808] After the generative AI edits the post, it automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. This report is then reviewed by employees, and used to educate them.

[0809] "Example 4"

[0810] The reports created by the generative AI are shared with the company, allowing the company to efficiently collect information on employees' complaints and positive points. For example, the report can reveal that employees have high opinions of a new product, or that some employees are dissatisfied with their work.

[0811] The processing flow of each embodiment will be described below.

[0812] "Example 1"

[0813] Step 1: Before employees or part-time workers post on social media, the content of the post is input into a generative AI.

[0814] Step 2: The generative AI determines whether the post is likely to spark an uproar or contain confidential company information. Specifically, it checks whether the post contains keywords or phrases that could spark an uproar or keywords or phrases related to confidential company information.

[0815] "Example 2"

[0816] Step 1: If the generative AI determines that the post is dangerous, it will edit it.

[0817] Step 2: Specifically, parts that could cause controversy or parts that touch on confidential company information are deleted or rewritten. For example, if a post says, "Our new product is better than our competitors' products," the AI ​​will rewrite it to say, "Our new product has superior performance."

[0818] "Example 3"

[0819] Step 1: After the generative AI edits the post, it automatically generates a report on the edits.

[0820] Step 2: The report will detail which parts were dangerous, why they were, and how they were edited. This report will be reviewed by employees and used to educate them.

[0821] "Example 4"

[0822] Step 1: The report created by the generative AI is shared with the company.

[0823] Step 2: This allows the company to efficiently collect information on employees' dissatisfaction and positive points. For example, the report can reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[0824] Example 1

[0825] Next, a description will be given of Example 1 of Form 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."

[0826] When employees or part-time workers post on social media, if there is a possibility that the content of the post may spark outrage or may contain confidential company information, it is necessary to assess the risk in advance and take appropriate action. However, manual checking is time-consuming and labor-intensive, and is inefficient. Furthermore, the accuracy of the checks can vary depending on the person. This poses a challenge, as it increases the risk of damaging a company's brand image and leaking confidential information.

[0827] 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.

[0828] In this invention, the server includes a generative AI means for determining the content of posts to the SNS, a generative AI means for editing the post if the post is determined to be risky by the generative AI, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for a user to input the post content before posting it to the SNS, a means for transmitting the input post content to the server, a means for the server to input the post content to a generative AI model, a means for the generative AI model to analyze the post content and return a determination result to the server, a means for the server to return the determination result to a terminal, and a means for the terminal to display the determination result to the user. This makes it possible to automatically determine the risk of the post content in advance and take appropriate measures quickly and efficiently.

[0829] "Generative AI methods" refers to artificial intelligence technology that analyzes the content of posts on social media and determines the possibility of a controversy or the risk of confidential information being leaked.

[0830] "Generative AI means for editing posted content" refers to artificial intelligence technology that automatically corrects and edits posted content that is deemed dangerous.

[0831] "Means for automatically generating reports" refers to technology that automatically creates reports that explain the risks of the posted content and the details of any corrections based on the posted content edited by the generative AI means.

[0832] "Means for employee review" refers to a technique for presenting the generated report to an employee for review or approval.

[0833] "Means for sharing with the organization" refers to the technology for sharing the generated report with relevant parties within the organization and communicating the information.

[0834] "Means for users to input content before posting on SNS" refers to technology that provides an interface for users to input content they intend to post on SNS in advance.

[0835] The term "means for transmitting the input posting content to the server" refers to a communication technology for transmitting the posting content input by the user to the server.

[0836] "Means by which the server inputs the posted content into the generative AI model" refers to the technology by which the server inputs the posted content received into the generative AI model and has it analyzed.

[0837] "Means for the generative AI model to analyze the content of posts and return the results to the server" refers to the technology that enables the generative AI model to analyze the content of posts and return the results to the server.

[0838] "Means by which the server returns the judgment results to the terminal" refers to the communication technology that enables the server to return the judgment results received from the generative AI model to the terminal.

[0839] "Means for the terminal to display the determination result to the user" refers to a technology that provides an interface for the terminal to display the determination result received from the server to the user.

[0840] MODE FOR CARRYING OUT THE INVENTION

[0841] This invention is a system that checks the content of posts made by employees and part-time workers on social media in advance to determine the possibility of them causing controversy or the risk of confidential information being leaked.The system uses a generative AI model to analyze the content of posts and automatically determine the risk.

[0842] Hardware and software used

[0843] Server: Runs the generative AI model and analyzes and judges the content of posts.

[0844] Terminal: Provides an interface for users to input their posting content and check the results of the evaluation.

[0845] Generative AI models: For example, use OpenAI's GPT-4.

[0846] Data processing and calculation

[0847] 1. Users enter their post content before posting to social media:

[0848] Users can use a dedicated application or web interface on their device to input the content they want to post to SNS. For example, a user might input, "I'm sharing details about our new product on SNS! This is still confidential, but I want everyone to know about it soon!"

[0849] 2. The device sends the entered post content to the server:

[0850] The terminal sends the content of the post entered by the user to the server. At this time, the terminal converts the content of the post into an appropriate format (for example, JSON format) and sends it to the server using a secure communication protocol (for example, HTTPS).

[0851] 3. The server inputs the post into the generative AI model:

[0852] The server inputs the received post content into the generative AI model. Specifically, the server generates a prompt like this:

[0853] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[0854] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[0855] This prompt is sent to the generative AI model.

[0856] 4. The generative AI model analyzes the post and returns the results to the server:

[0857] The generative AI model analyzes the prompt text and checks whether the post contains keywords or phrases that could spark controversy or keywords or phrases related to confidential company information. For example, it detects the keyword "confidential" and determines that the post contains confidential company information. The generative AI model then returns this determination result to the server.

[0858] 5. The server returns the result to the device:

[0859] The server receives the results of the generative AI model and sends them back to the device, including information on whether the content of the post is likely to spark an uproar or whether it contains confidential company information.

[0860] 6. The device displays the result to the user:

[0861] The device displays the result of the judgment received from the server to the user. For example, it may display a message saying, "This post contains confidential company information. Please correct the post." Based on this information, the user can either correct the post or cancel it.

[0862] Specific examples

[0863] Suppose a user enters the following post content:

[0864] "We'll be sharing details about our new products on social media! This is still confidential, but we want you all to know soon!"

[0865] Here's an example prompt to feed this post to a generative AI model:

[0866] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[0867] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[0868] The generative AI model analyzes this prompt, detects that the post contains keywords such as "confidential," and determines that it may contain confidential company information.

[0869] The server returns the result of this determination to the terminal, and the terminal displays to the user a message saying, "This post contains confidential company information. Please correct your post."

[0870] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0871] Step 1:

[0872] The user inputs the content to post before posting to the SNS.

[0873] Users use a dedicated application or web interface on their device to input the content they want to post to SNS. For example, a user might input, "I'm sharing details about our new product on SNS! This is still confidential, but I want everyone to know about it soon!" The input content is temporarily stored in the device's memory.

[0874] Step 2:

[0875] The terminal transmits the input post content to the server.

[0876] The terminal converts the content of the post entered by the user into an appropriate format (e.g., JSON format) and sends it to the server using a secure communication protocol (e.g., HTTPS). The input data is the text information of the post, and the output data is the post sent to the server.

[0877] Step 3:

[0878] The server inputs the post content into a generative AI model.

[0879] The server inputs the received post content into the generative AI model. Specifically, the server generates a prompt like this:

[0880] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[0881] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[0882] The input data is the text information of the post, and the output data is the generated prompt.

[0883] Step 4:

[0884] The generative AI model analyzes the content of the post and returns the results to the server.

[0885] The generative AI model analyzes the prompt text and checks whether the post contains keywords or phrases that could spark an uproar or keywords or phrases related to confidential company information. For example, it detects the keyword "confidential" and determines that the post contains confidential company information. The input data is the prompt text, and the output data is the judgment result.

[0886] Step 5:

[0887] The server returns the result of the determination to the terminal.

[0888] The server receives the judgment results from the generative AI model and sends them back to the device. The judgment results include information on whether the content of the post has the potential to spark a controversy or whether it contains confidential company information. The input data is the judgment result, and the output data is the judgment result sent to the device.

[0889] Step 6:

[0890] The terminal displays the judgment result to the user.

[0891] The device displays the judgment result received from the server to the user. For example, it may display a message such as "This post contains confidential company information. Please correct the post." The input data is the judgment result, and the output data is the judgment result displayed to the user. Based on this information, the user can correct the post or cancel it.

[0892] (Application example 1)

[0893] Next, a description will be given of Application Example 1 of Form 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."

[0894] When employees or part-time workers post on social media, there is a risk that they may unintentionally post content that will spark controversy or include confidential information. Such posts not only damage the company's credibility, but can also lead to legal issues. Furthermore, manually checking and correcting posted content is time-consuming and labor-intensive, and inefficient. To solve these issues, a system is needed that can automatically assess the content of posts and prompt corrections as necessary.

[0895] 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.

[0896] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the post if the post is determined to be dangerous by the generative AI, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for converting the post into a prompt sentence and generating a determination result using a generative AI model, a means for displaying the determination result to the user, and a means for prompting the user to modify the post as necessary. This makes it possible to automatically check the content posted by employees and part-time workers on SNS, reducing the risk of online outrage and the risk of confidential information being leaked.

[0897] "Generative AI methods" are methods that use artificial intelligence to analyze the content of social media posts and determine the risk of online outrage or the risk of confidential information being leaked.

[0898] "Generative AI means for editing posted content when the determined posted content is dangerous" refers to a means for automatically correcting posted content that the generative AI has determined to be dangerous.

[0899] "Means for automatically generating a report on the content of posts edited by generative AI" means means for automatically creating a detailed report on the content of posts edited by generative AI.

[0900] The "means for having the employee check the report" refers to a means for presenting the generated report to the employee and having the employee check it.

[0901] The "means for sharing the report with the organization" is a means for sharing the generated report within the organization.

[0902] "Means for converting the content of a post into a prompt sentence and generating a judgment result using a generative AI model" refers to means for converting the content of a post entered by a user into a prompt sentence, analyzing the content using a generative AI model, and generating a judgment result.

[0903] "Means for displaying the judgment results to the user" refers to means for displaying the judgment results generated by the generative AI model to the user.

[0904] The "means of encouraging the user to correct the posted content as necessary" refers to a means of encouraging the user to correct the posted content based on the judgment result.

[0905] As an embodiment of the present invention, a system is constructed that automatically judges the content of posts on SNS and prompts users to make corrections as necessary. A specific embodiment of this system is described below.

[0906] First, the server is equipped with a generative AI means for judging the content of posts on social media. This generative AI means analyzes the content of posts entered by users and judges the risk of flame wars and the risk of confidential information leaks. Specifically, it converts the content of posts into prompt statements and generates judgment results using a generative AI model (for example, OpenAI's API).

[0907] Next, the server includes a generative AI means for editing the posted content when the generative AI determines that the posted content is dangerous. This means automatically corrects the posted content that the generative AI determines to be dangerous and provides the user with safe posted content.

[0908] Additionally, the server has a means for automatically generating a report on the posts edited by the generative AI, which will detail why the posts were dangerous and how they were corrected.

[0909] The server also provides a means for employees to review the generated reports and for them to be shared with the organization, allowing employees to understand the risks of what they post and for information to be shared across the organization.

[0910] As a concrete example, consider the case where a user posts something like, "Details of the new product haven't been released yet, but it's going to be great!" This post is converted into a prompt sentence and input into the generative AI model.

[0911] Example prompt sentence:

[0912] Please determine whether the following posts are inflammatory or contain confidential information.

[0913] Posted: Details of the new product have not been released yet, but it's going to be awesome!

[0914] Judgment result:

[0915] The server sends this prompt to the generative AI model and obtains the judgment result. If the post is deemed dangerous based on the judgment result, the server automatically modifies the post and provides safe content to the user. In addition, the generated report can be reviewed by employees and shared with the organization for thorough risk management.

[0916] In this way, this invention automatically checks the content posted on social media by employees and part-time workers, making it possible to reduce the risk of online outrage and the risk of confidential information being leaked.

[0917] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0918] Step 1:

[0919] The user inputs the content to post on social media. The user uses a smartphone or computer to enter the content to post into the application. This input becomes the starting point for the system's processing.

[0920] Step 2:

[0921] The server receives the post content and converts it into a prompt sentence. The server receives the post content entered by the user and converts it into a prompt sentence suitable for the generative AI model. Specifically, the post content is converted into the following format.

[0922] Please determine whether the following posts are inflammatory or contain confidential information.

[0923] Post content: [User post content]

[0924] Judgment result:

[0925] Input: User post

[0926] Output: A prompt suitable for a generative AI model

[0927] Step 3:

[0928] The server generates a judgment result using a generative AI model. The server sends the prompt text to the generative AI model (e.g., OpenAI's API) and receives the judgment result. The generative AI model analyzes the prompt text and determines whether the post content poses a risk of flaming or leaking confidential information.

[0929] Input: prompt statement

[0930] Output: Judgment result

[0931] Step 4:

[0932] The server displays the judgment result to the user. The server displays the judgment result obtained from the generative AI model to the user. The judgment result includes whether the posted content is dangerous and the specific risk content.

[0933] Input: Judgment result

[0934] Output: The result displayed to the user

[0935] Step 5:

[0936] The server prompts the user to modify the posted content as necessary. Based on the judgment result, the server prompts the user to modify the posted content. For example, if the posted content is judged to be dangerous, the server will present the user with specific suggestions for modification.

[0937] Input: Judgment result

[0938] Output: suggested fixes presented to the user

[0939] Step 6:

[0940] The server automatically generates a report on the post content edited by the generative AI. The server automatically creates a detailed report on the post content edited by the generative AI. This report details why the post content was dangerous and how it was corrected.

[0941] Input: Edited post

[0942] Output: Detailed report

[0943] Step 7:

[0944] The server presents the generated report to employees for review and shares it with the organization. The server also shares the report within the organization to ensure thorough risk management.

[0945] Input: Detailed report

[0946] Output: Reports presented to employees, reports shared with the organization

[0947] Example 2

[0948] Next, a description will be given of Example 2 of Form 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."

[0949] On social media and other online platforms, when user posts pose a risk of causing controversy or leaking confidential information, there is a need to properly manage the content and edit it safely. However, manual monitoring and editing takes time and effort, and is inefficient. Therefore, a system is needed that automatically analyzes posted content, identifies risky parts, and edits them.

[0950] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0951] In this invention, the server includes means for receiving posted content entered by a user, means for transmitting the received posted content to a generative AI model, means for the generative AI model to analyze the posted content and identify dangerous parts, means for the generative AI model to edit the identified dangerous parts, and means for returning the edited posted content to the user. This makes it possible to automatically analyze the posted content of users and reduce the risk of flame wars and the leakage of confidential information.

[0952] "User" means an individual or group that uses the system to input posting content.

[0953] "Posted content" refers to information such as text, images, and videos that users post on social media or other online platforms.

[0954] The "server" is a computer system that receives user submissions and sends them to the generative AI model.

[0955] A "generative AI model" is a model that uses artificial intelligence technology to analyze posted content, identify risky parts, and edit them.

[0956] A "prompt sentence" is an input sentence that instructs the generative AI model to analyze and edit the post content.

[0957] "Analysis" is the process by which the generative AI model understands the context and meaning of the post and identifies risky parts.

[0958] "Dangerous parts" are parts of the post that pose a risk of causing an uproar or that contain confidential information.

[0959] "Editing" is the process by which a generative AI model removes or rewrites identified dangerous parts.

[0960] "Return" is the process by which the server sends back to the user the post content edited by the generative AI model.

[0961] This invention relates to a system that automatically analyzes content posted by users on social networking sites and other online platforms, identifies risky parts, and edits them. Specific embodiments of this system are described below.

[0962] The server receives the posting content entered by the user. The user uses their own device (PC, smartphone, etc.) to enter the posting content. For example, they might enter, "Our new product is better than our competitor's product."

[0963] The terminal sends the content of the post entered by the user to the server, using a communication protocol such as an HTTP request.

[0964] The server sends a prompt to the generative AI model to analyze the content of the post received from the device. The prompt includes the content of the post and instructions for analysis. For example, the server sends a prompt such as, "Analyze the content posted by the user, identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'"

[0965] A generative AI model (such as OpenAI's GPT-4) analyzes the content of posts based on prompts received from the server. The model uses natural language processing techniques to understand the context and meaning of the posts and identify parts that may be controversial or contain sensitive information.

[0966] The generative AI model then removes or rewrites the identified risky parts, for example, rewriting "Our new product is better than our competitors' products" to "Our new product has superior performance."

[0967] The server then sends the edited post content received from the generative AI model back to the device, again using a communication protocol such as an HTTP response.

[0968] The terminal displays the edited post content received from the server to the user, who can then check the edited content and revise it again if necessary.

[0969] As a concrete example, if a user posts, "Our new product is better than our competitor's product," the process will proceed as follows:

[0970] 1. The user enters the content to post on the device.

[0971] 2. The device sends the post content to the server.

[0972] 3. The server receives the post and sends the following prompt to the generative AI model: "Analyze the content posted by the user and identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'"

[0973] 4. The generative AI model analyzes the content of the post and identifies any part that claims to be "better than competitors' products" as dangerous.

[0974] 5. The generative AI model rewrites this part to "has excellent performance."

[0975] 6. The server sends the edited post back to the device.

[0976] 7. The device displays the edited post "Our new product has excellent performance" to the user.

[0977] In this way, the system can safely edit user posts.

[0978] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0979] Step 1:

[0980] The user enters the content to post.

[0981] Users use their own devices (PCs, smartphones, etc.) to enter their post content. For example, they might enter, "Our new product is better than our competitors' products." The entered post content is saved on the device.

[0982] Step 2:

[0983] The device sends the post content to the server.

[0984] The terminal sends the content posted by the user to the server. This transmission uses a communication protocol such as an HTTP request. The input is the content posted by the user, and the output is the content sent to the server.

[0985] Step 3:

[0986] The server receives the post and sends a prompt to the generative AI model.

[0987] The server sends a prompt to the generative AI model to analyze the content of the post received from the device. The prompt includes the content of the post and instructions for analysis. For example, it generates a prompt such as, "Analyze the content posted by the user, identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'" The input is the content of the post received from the device, and the output is the prompt sent to the generative AI model.

[0988] Step 4:

[0989] A generative AI model analyzes the content of posts and identifies risky parts.

[0990] A generative AI model (e.g., OpenAI's GPT-4) analyzes the content of posts based on prompts received from the server. The model uses natural language processing techniques to understand the context and meaning of the posts and identify parts that pose a risk of flaming or contain confidential information. The input is the prompt, and the output is the identified risky parts.

[0991] Step 5:

[0992] The generative AI model edits out dangerous parts.

[0993] The generative AI model removes or rewrites the identified risky parts. For example, it rewrites "Our new product is better than our competitors' products" to "Our new product has superior performance." The input is the identified risky parts, and the output is the edited post.

[0994] Step 6:

[0995] The server sends the edited post back to the device.

[0996] The server returns the edited post content received from the generative AI model to the device. This return also uses a communication protocol such as an HTTP response. The input is the edited post content received from the generative AI model, and the output is transmission to the device.

[0997] Step 7:

[0998] The device displays the edited post content to the user.

[0999] The terminal displays the edited post content received from the server to the user. The user can check the edited content and revise the post content again if necessary. The input is the edited post content received from the server, and the output is the display to the user.

[1000] (Application example 2)

[1001] Next, a description will be given of Application Example 2 of Form 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."

[1002] On online platforms such as social media and blogs, if a user posts content that could spark controversy or leak confidential information, the post could cause serious damage to a company or individual. To prevent such risks, a system is needed to check posts in advance and edit them as necessary. However, manual checking and editing takes time and effort, and there are challenges in that it is difficult to respond in real time.

[1003] 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.

[1004] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for analyzing the posted content in real time, a means for detecting expressions that pose a risk of causing an uproar, a means for detecting confidential information, and a means for automatically rewriting the posted content to safe expressions. This makes it possible to detect the risk of posted content in real time and automatically rewrite it to safe expressions.

[1005] "Generative AI means" refers to means that use artificial intelligence technology to analyze input data and generate output according to a specific purpose.

[1006] "Posted content" refers to information such as text, images, and videos that users make public on online platforms such as social media and blogs.

[1007] "If judged to be dangerous" means that the generative AI means analyzes the content of the post and determines that there is a risk of it causing a firestorm or leaking confidential information.

[1008] "Generative editing AI means" refers to a means that uses artificial intelligence technology to rewrite posted content that is deemed dangerous into safer expression.

[1009] "Means for automatically generating reports" refers to means for automatically creating a report summarizing the edits and reasons for edits made to posted content by generative AI means.

[1010] The "means for having employees check" refers to a means for presenting the automatically generated report to employees and having them check the contents.

[1011] The "means for sharing with the organization" is a means for sharing the generated report with the people involved in the organization.

[1012] "Means of real-time analysis" refers to means for instantly analyzing the content of posts and determining whether they are dangerous.

[1013] "Means for detecting expressions that pose a risk of causing an outrage" refers to means for identifying expressions contained in posted content that could cause an outrage.

[1014] "Means for detecting confidential information" refers to means for identifying confidential information of companies or individuals contained in posted content.

[1015] "Means for automatically rewriting content into safe expressions" refers to a means for automatically rewriting content that is deemed dangerous into safe expressions that do not pose a risk of online outrage or the leakage of confidential information.

[1016] As an embodiment of the present invention, a system is provided that analyzes content posted on online platforms such as social networking sites and blogs in real time and automatically edits any dangerous parts. Specific embodiments of this system are described below.

[1017] System Configuration

[1018] The system consists of the following main components:

[1019] 1. Generative AI means: Evaluates the content of posts and edits them if deemed dangerous.

[1020] 2. Automatic report generation method: Automatically generate reports on edited post content.

[1021] 3. Employee verification method: Have employees verify the generated reports.

[1022] 4. Organizational sharing means: Share reports within the organization.

[1023] 5. Real-time analysis: Analyze posted content instantly.

[1024] 6. Flaming risk detection method: Detect expressions that pose a risk of flaming.

[1025] 7. Confidential information detection methods: Detect confidential information.

[1026] 8. Automatic rewriting method: Automatically rewrite dangerous posts into safe expressions.

[1027] Program processing

[1028] The server does the following:

[1029] 1. Receiving posted content: Receives content that a user intends to post on social media or a blog.

[1030] 2. Analysis by a generative AI model: The received post content is sent to a generative AI model for analysis. The generative AI model used is, for example, OpenAI's "text-davinci-003."

[1031] 3. Risk assessment: The generative AI model analyzes the content of the post and determines whether there is a risk of a controversy or the leakage of confidential information.

[1032] 4. Automatic editing: If a post is deemed dangerous, the generative AI model automatically rewrites it to make it safer.

[1033] 5. Automatic report generation: Automatically generate a report for edited posts, summarizing the edits made and the reasons for them.

[1034] 6. Notify employees: Notify employees of the generated reports and have them review them.

[1035] 7. Share within your organization: Share the reviewed report within your organization.

[1036] Hardware and software used

[1037] Hardware: Servers, user devices (smartphones, PCs, etc.)

[1038] Software: Generative AI model (OpenAI API), real-time analysis software, report generation software

[1039] Specific examples

[1040] For example, if a user enters the following post:

[1041] Original post: "Our new product is better than our competitors' products."

[1042] In this case, the prompt sent to the generative AI model would be:

[1043] Prompt: "Please reword the following post to make it safer:\n\nOur new product is better than our competitors' products.\n\nPlease remove or rewrite any content that is inflammatory or contains confidential information."

[1044] The generative AI model analyzes the post based on this prompt and rewrites it to a safer expression. For example, the rewritten content might read, "Our new product has excellent performance."

[1045] In this way, users can check the content before posting and, if necessary, rewrite it to safer language, thereby preventing the risk of flame wars and the leakage of confidential information.

[1046] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1047] Step 1:

[1048] The user enters the content they want to post on social media or a blog.

[1049] Input: User-typed post content (e.g., "Our new product is better than our competitors' products.")

[1050] Output: The post is sent to the server.

[1051] Specific operation: A user uses a smartphone or PC to enter text into the posting screen of a social networking site or blog and presses the "Post" button.

[1052] Step 2:

[1053] The server receives the posted content and sends it to the generative AI model.

[1054] Input: User submitted post

[1055] Output: The prompt sent to the generative AI model

[1056] What happens: The server receives the post and generates a prompt, such as "Please rewrite the following post to make it safer:\n\nOur new product is superior to our competitors' products.\n\nPlease remove or rewrite any parts that may be controversial or contain confidential information."

[1057] Step 3:

[1058] A generative AI model analyzes the content of posts and determines whether they are dangerous.

[1059] Input: A prompt sent to the generative AI model

[1060] Output: Analysis results (presence or absence of danger, identification of dangerous parts)

[1061] How it works: A generative AI model (e.g., OpenAI's "text-davinci-003") analyzes the prompt text and determines whether the post content poses a risk of inciting controversy or leaking confidential information.

[1062] Step 4:

[1063] The server receives the analysis results of the generative AI model and edits the post content as necessary.

[1064] Input: Analysis results of generative AI model

[1065] Output: Edited post

[1066] Specific operation: The server receives the analysis results of the generative AI model and rewrites any parts judged to be dangerous into safe expressions. For example, it could rewrite it as "Our new product has excellent performance."

[1067] Step 5:

[1068] The server automatically generates a report of edited posts.

[1069] Input: Edited post

[1070] Output: Auto-generated report

[1071] Specific operation: The server automatically generates a report summarizing the edits and the reasons for the edits based on the edited post content. The report includes the original post content, the edited content, and the reason for the edit.

[1072] Step 6:

[1073] The server notifies the employee of the generated report for review.

[1074] Input: Auto-generated report

[1075] Output: Report notified to employees

[1076] What happens: The server sends an automatically generated report to the employee's email address or internal system for review.

[1077] Step 7:

[1078] Employees review the reports and provide feedback as needed.

[1079] Input: Report notified to employee

[1080] Output: Employee feedback

[1081] Specific actions: The employee checks the report and checks if there are any problems with the content. If necessary, feedback is sent to the server.

[1082] Step 8:

[1083] Share the server verification report with your organization.

[1084] Input: Reviewed report with employee feedback

[1085] Output: Reports shared within the organization

[1086] Specific actions: The server shares the confirmed report with relevant parties within the organization and prompts them to take necessary action, for example by posting it on the internal portal site.

[1087] Example 3

[1088] Next, a description will be given of Example 3 of Form Example 3. 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."

[1089] Conventional SNS post management systems lack the functionality to automatically determine the risk of posts and edit them appropriately, which makes it difficult to efficiently educate employees and collect company feedback.In addition, they do not generate reports that explain in detail the reasons for editing posts and specific areas for improvement, which makes it difficult for employees to understand and the effectiveness of training to be achieved.

[1090] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1091] In this invention, the server includes a means for a user to input posted content, a means for the server to receive the posted content, a means for the server to send a prompt to the generative AI model, a means for the generative AI model to edit the posted content, a means for the server to receive the edited content, a means for the server to generate a report on the edited content, a means for the server to send the report to a terminal, a means for the terminal to display the report to employees, and a means for employees to review the report and use it for training. This enables automatic determination of the riskiness of posted content and appropriate editing, enabling efficient employee training and company feedback collection. Furthermore, a report is generated that details the reasons for edits and specific improvements, improving employee understanding and the effectiveness of training.

[1092] A "user" is an individual or group that inputs posting content into the system.

[1093] A "server" is a device or system that receives user submissions, sends prompts to the generative AI model, receives edited content, generates a report, and sends it to the terminal.

[1094] A "generative AI model" is an artificial intelligence algorithm or program that edits posts based on prompts received from a server.

[1095] A "prompt" is text that contains instructions for the generative AI model to edit the post content.

[1096] A "terminal" is a device or system that displays reports sent from the server to employees.

[1097] A "report" is a document that details which parts of the post were edited by the generative AI model, why they were dangerous, and how they were edited.

[1098] An "employee" is an individual or entity that uses the device to review reports and assist in education.

[1099] This invention is a system in which a user inputs post content, a server receives the post content and sends a prompt message to a generative AI model, the generative AI model edits the post content, the server receives the edited content and generates a report, and a terminal displays the report to employees.

[1100] First, a user uses their own device (e.g., a PC or smartphone) to input their post into the system interface. For example, they might type, "I feel the quality of the new product is poor. It needs further improvement."

[1101] Next, the server receives the post content entered by the user. This post content is temporarily stored in a database on the server. The server sends the received post content to the generative AI model as a prompt text. The prompt text is structured as follows: "Please improve the following post content and generate a report: Post content: 'I feel the quality of the new product is poor. Further improvement is needed.'"

[1102] A generative AI model (e.g., OpenAI's GPT-4) analyzes the prompt received from the server and edits the post, which might look something like, "I feel there are some areas for improvement in the quality of the new product. Further improvements are expected."

[1103] The server receives the edited content from the generative AI model. This edited content is also stored in a database. The server compares the original and edited content and automatically generates a report on the edits. The report contains the following information:

[1104] Which parts were dangerous and why?

[1105] How did you edit it?

[1106] For example, the report states, "Because saying 'poor quality' could damage the company's image, we softened the expression to 'there are some areas for improvement in quality.'"

[1107] The server sends the generated report to the employee's device. The report is displayed on a dedicated interface that the employee can access. The device displays the report received from the server to the employee. The employee can check the report on their own device and learn which parts of the posted content were problematic and how they could be improved. This can be used to educate employees.

[1108] This system automatically determines the risk of posted content and enables appropriate editing, thereby enabling efficient employee training and company feedback collection. Furthermore, a report detailing the reasons for editing and specific areas for improvement is generated, improving employee understanding and the effectiveness of training. The flow of the identification process in the third embodiment will be described with reference to FIG. 15.

[1109] Step 1:

[1110] The user enters the content to post.

[1111] Users use their own devices (PCs or smartphones) to input their posts into the system interface. For example, they might type, "I feel the quality of the new product is poor. It needs further improvement." The entered post is then sent from the device to the server.

[1112] Step 2:

[1113] The server receives the post.

[1114] The server receives the posts sent by the user. The received posts are temporarily stored in a database on the server. The input is the user's post, and the output is the post stored in the database.

[1115] Step 3:

[1116] The server sends a prompt to the generative AI model.

[1117] The server sends the received post content to the generative AI model as a prompt. The prompt is structured as follows: "Please improve the following post content and generate a report: Post content: 'I feel the quality of the new product is poor. It needs further improvement.'" The input is the post content stored in the database, and the output is the prompt sent to the generative AI model.

[1118] Step 4:

[1119] A generative AI model edits posts.

[1120] A generative AI model (e.g., OpenAI's GPT-4) analyzes the prompt received from the server and edits the post content. The edited content might look something like, "I feel there are some areas for improvement in the quality of the new product. Further improvements are expected." The input is the prompt, and the output is the edited post content.

[1121] Step 5:

[1122] The server receives the edited content.

[1123] The server receives the edited content from the generative AI model. The received edited content is also stored in the database. The input is the edited post from the generative AI model, and the output is the edited post stored in the database.

[1124] Step 6:

[1125] The server generates a report of the edits.

[1126] The server compares the content before and after editing and automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. For example, it might say, "Because saying 'poor quality' could damage the company's image, we softened the expression to 'there are some areas for improvement in quality.'" The input is the post content before and after editing, and the output is the generated report.

[1127] Step 7:

[1128] The server sends the report to the device.

[1129] The server sends the generated report to the employee's terminal. The report is displayed in a dedicated interface that the employee can access. The input is the generated report, and the output is the report sent to the terminal.

[1130] Step 8:

[1131] The terminal displays the report to the employee.

[1132] The terminal displays the report received from the server to the employee. The employee can check the report on his / her own terminal. The input is the report sent from the server, and the output is the report displayed on the terminal.

[1133] Step 9:

[1134] Employees review the reports and use them for training purposes.

[1135] Employees can check the displayed report and learn which parts of the post were problematic and how they were improved. This can be used to educate employees. The input is the report displayed on the terminal, and the output is the employee's understanding and the effectiveness of the education.

[1136] (Application example 3)

[1137] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[1138] On online platforms such as social media and blogs, if users post content that is inappropriate or contains dangerous information, it can become public and pose a significant risk to companies and individuals. Manually monitoring and editing these posts is extremely time-consuming and inefficient. Furthermore, the lack of detailed feedback on edited content makes it difficult for users and security personnel to take appropriate action.

[1139] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1140] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having an employee review the report, a means for sharing the report with the company, a means for sharing the report with a security officer, a means for describing in detail which parts were dangerous and why, and how they were edited, when generating the report, and a means for sharing the report with the user. This makes it possible to automatically detect and edit the riskiness of posted content and provide detailed feedback.

[1141] "Generative AI methods" are artificial intelligence technologies that analyze content posted on online platforms such as social media and blogs, detect dangerous information or inappropriate expressions, and edit them as necessary.

[1142] "Posted content" refers to information such as text, images, and videos that a user intends to publish on online platforms such as social media and blogs.

[1143] A "report" is a document that details which parts of a post were edited by generative AI tools, why they were dangerous, and how they were edited.

[1144] An "employee" is an individual who belongs to a company or organization and performs work for that company or organization.

[1145] A "company" is a legal entity established to carry out a specific business.

[1146] A "security officer" is an individual responsible for managing and monitoring information security within a company or organization.

[1147] "User" refers to an individual who creates and publishes content using online platforms such as social media and blogs.

[1148] "Automatic generation means" refers to technology for automatically creating a report on the content of posts edited by the generative AI means.

[1149] "Means of sharing" refers to the technology used to send and provide generated reports to specific individuals or organizations.

[1150] A system for implementing this invention monitors content posted to online platforms such as social networking sites and blogs in real time, automatically edits dangerous information and inappropriate language, and generates and shares detailed reports on the edited content. Specific embodiments of this system are described below.

[1151] System Configuration

[1152] 1. Hardware Configuration

[1153] Server: Use a server equipped with a high-performance processor and large memory capacity. For example, use an EC2 instance from Amazon Web Services (AWS).

[1154] Device: The smartphone or head-mounted display used by the user. For example, we will use an iPhone or Oculus Quest.

[1155] 2. Software Configuration

[1156] Generative AI models: Use natural language processing models such as GPT-4.

[1157] Social Media API: Use APIs to retrieve posts from social media platforms, such as Twitter API or Facebook Graph API.

[1158] Report generation libraries: Use libraries such as ReportLab to generate reports.

[1159] Emailing API: Share reports using email sending services like SendGrid.

[1160] Cloud Storage API: Save and share your reports using cloud storage services such as AWS S3.

[1161] Data processing and calculation

[1162] 1. Obtaining input data

[1163] The server uses the SNS API to obtain the content that the user is about to post in real time.

[1164] 2. Data analysis and editing

[1165] The server uses a generative AI model (e.g., GPT-4) to analyze the acquired posts. Specifically, it uses natural language processing technology to detect risk and inappropriate expressions in the posts.

[1166] The generative AI model automatically edits the detected parts.

[1167] 3. Generate a report

[1168] The server generates a detailed report of the edited post, detailing which parts were dangerous, why they were dangerous, and how they were edited.

[1169] Use a report generation library (e.g. ReportLab) to generate reports in PDF format.

[1170] 4. Share the report

[1171] The server sends the generated report to the user and security officer using an email sending API (e.g., SendGrid).

[1172] It also uses cloud storage APIs (e.g. AWS S3) to store reports in cloud storage and make them accessible when needed.

[1173] Specific examples

[1174] For example, if a user posts on social media, "This product is completely unusable, it's the worst!", the server retrieves this post via the social media API. A generative AI model (e.g., GPT-4) analyzes the post and edits it to say, "This product didn't work for me, but it might work for other people." A detailed report of the edits is then generated and shared with the user and security personnel.

[1175] Prompt Sentence Examples

[1176] Analyze the following content that users are attempting to post on social media, detect and edit any potentially dangerous or inappropriate content, and generate a detailed report of the edits.

[1177] Posting: "This product is completely unusable and awful!"

[1178] In this way, the system can automatically detect and edit potentially dangerous content and provide detailed feedback.

[1179] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1180] Step 1:

[1181] The server uses the SNS API to obtain the content that the user is about to post in real time. The input is the user's post content, and the output is the obtained post content. Specifically, the server sends a request to the SNS API to obtain the user's post content.

[1182] Step 2:

[1183] The server uses a generative AI model (e.g., GPT-4) to analyze the acquired post content. The input is the acquired post content, and the output is the analysis result. Specifically, the server inputs prompt sentences into the generative AI model to detect risk and inappropriate expressions in the post content.

[1184] Step 3:

[1185] The server uses a generative AI model to automatically edit the detected dangerous parts and inappropriate expressions. The input is the analysis result, and the output is the edited post content. Specifically, the server gives editing instructions to the generative AI model and appropriately corrects the post content.

[1186] Step 4:

[1187] The server generates a detailed report of the edited post. The input is the edited post and the output is the generated report. Specifically, the server uses a report generation library (e.g. ReportLab) to create a report detailing which parts were dangerous, why, and how they were edited.

[1188] Step 5:

[1189] The server sends the generated report to the user and security officer using an email sending API (e.g., SendGrid). The input is the generated report, and the output is the sent email. Specifically, the server attaches the report to the email sending API and sends it to the specified email address.

[1190] Step 6:

[1191] The server stores the generated report using a cloud storage API (e.g. AWS S3) and makes it accessible when needed. The input is the generated report and the output is the report stored in cloud storage. Specifically, the server uploads the report to the cloud storage API and obtains the URL of the storage location.

[1192] 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.

[1193] "Example 1"

[1194] In one embodiment of the present invention, generative AI is used in combination with an emotion engine that recognizes user emotions. Specifically, before a user posts to a social networking site, the content of the post is input into the generative AI. The generative AI then determines whether the content of the post is likely to spark a controversy or contain confidential company information.

[1195] "Example 2"

[1196] The emotion engine extracts emotions from the content of users' social media posts. For example, it extracts the emotion of "joy" from a post such as "Today was fun!", and the emotion of "distress" from a post such as "Today was tough...".

[1197] "Example 3"

[1198] If the generative AI judges the content of a post to be dangerous, or if the emotion engine judges the user's emotions to be negative, the AI ​​will edit the post, specifically deleting or rewriting any parts that could spark outrage or that contain confidential company information.

[1199] "Example 4"

[1200] After the generative AI edits the post, it automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. This report is then reviewed by employees, and used to educate them.

[1201] The processing flow of each embodiment will be described below.

[1202] "Example 1"

[1203] Step 1: Before a user posts on social media, the content of the post is input into the generative AI.

[1204] Step 2: The generative AI determines whether the content of the post is likely to spark an uproar or contain confidential company information.

[1205] "Example 2"

[1206] Step 1: The emotion engine extracts emotions from the user's social media posts.

[1207] Step 2: For example, extract the emotion of "joy" from a post such as "Today was fun!", and the emotion of "distress" from a post such as "Today was tough...".

[1208] "Example 3"

[1209] Step 1: If the generative AI determines that the post is dangerous, or if the emotion engine determines that the user's emotion is negative, the AI ​​will edit the post.

[1210] Step 2: Specifically, delete or rewrite any parts that could cause controversy or that contain confidential company information.

[1211] "Example 4"

[1212] Step 1: After the generative AI edits the post, it automatically generates a report on the edits.

[1213] Step 2: The report will detail which parts were unsafe, why, and how they were edited.

[1214] Step 3: Have employees review this report and use it to educate them.

[1215] Example 1

[1216] Next, a description will be given of Example 1 of Form 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."

[1217] There is a need to reduce the risk of posts on social media causing outrage and leaking confidential company information. However, in conventional systems, post content is often checked manually, which is inefficient. In addition, there is a lack of means to recognize the sentiment of posts and provide appropriate feedback, making it difficult for users to make appropriate decisions.

[1218] 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.

[1219] In this invention, the server includes generative AI means for determining the content of posts to SNS, generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, means for automatically generating a report on the posted content edited by the generative AI, means for having employees review the report, means for sharing the report with the organization, means for checking keywords and phrases included in the posted content, emotion engine means for recognizing emotions from the posted content, and means for notifying the emotion engine of the determination result. This reduces the risk of content posted to SNS causing a flame war or the risk of confidential information being leaked, and enables users to make appropriate decisions.

[1220] "Generative AI methods" refers to artificial intelligence technology that analyzes the content of posts on social media and determines the risk of online outrage or the risk of confidential information being leaked.

[1221] "Generative AI editing means" refers to artificial intelligence technology that automatically corrects and edits posted content that is deemed dangerous by generative AI.

[1222] "Means for automatically generating reports" refers to technology that automatically creates detailed reports about the content of posts edited by generative AI.

[1223] "Means for employees to check" refers to a technique for presenting the generated report to employees and having them check its contents.

[1224] "Means for sharing within the organization" refers to the technology used to share the generated reports within the organization and make them accessible to relevant parties.

[1225] "Keyword and phrase checking measures" refers to technology that detects specific keywords and phrases in posts and assesses risk.

[1226] "Emotion engine means" refers to technology for recognizing a user's emotions from the content of their posts and determining the type and intensity of the emotions.

[1227] "Means for notifying the judgment results" refers to technology for notifying the user of the analysis results of the generative AI and emotion engine and providing appropriate feedback.

[1228] This invention is a system that checks the content of posts to social media in advance to reduce the risk of flame wars and leaks of confidential information. The system includes a generative AI means, an emotion engine means, a report generation means, etc. Specific embodiments of this system are described below.

[1229] Hardware and software used

[1230] Hardware:

[1231] Server: A central processing unit for analyzing submissions and generating reports.

[1232] Device: The device (PC, smartphone, etc.) that a user uses to enter content to post on a social networking site.

[1233] software:

[1234] Generative AI model: An artificial intelligence technology that analyzes the content of posts and determines the risk of flame wars and the risk of confidential information leaks.

[1235] Emotion engine: Technology for recognizing user emotions from posted content and determining the type and intensity of emotions.

[1236] Report generation software: Software for automatically generating reports on posts edited by generative AI.

[1237] Data processing and calculation

[1238] Enter your post content:

[1239] The user uses the device to input the content to post on the social networking site. For example, the user might type, "The new project isn't progressing at all. My boss's instructions are too vague!"

[1240] Generative AI analysis:

[1241] The device sends the input post content to the generative AI, which checks the keywords and phrases contained in the post to determine the risk of a controversy or the leakage of confidential information. For example, it detects keywords such as "project," "instructions," and "ambiguous."

[1242] Emotion Recognition with Emotion Engine:

[1243] The server uses an emotion engine to recognize the user's emotion from the post, for example, determining whether the post expresses anger or frustration. The emotion engine analyzes the context and wording of the post to identify the user's emotion.

[1244] Report generation and notification:

[1245] The server automatically generates a report based on the analysis results of the generative AI and emotion engine. This report explains why the post was deemed dangerous and what emotions it contains. The generated report is sent to employees' devices for review and is then shared within the organization.

[1246] Examples and prompts

[1247] Examples:

[1248] If a user types, "I'm not making any progress on my new project. My boss's instructions are too vague!", the following process will occur:

[1249] Example prompt sentence:

[1250] Check out the post below:

[1251] "The new project isn't progressing at all. My boss's instructions are too vague!"

[1252] Determine whether this post is likely to cause controversy or contain confidential company information.

[1253] The server inputs this prompt into the generative AI, which then analyzes the content of the post and notifies the user that there is a high risk of a controversy or that the post may contain confidential information about the project.

[1254] In this way, users can reduce the risk of a social media firestorm or the leakage of confidential information before posting on social media.

[1255] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1256] Step 1:

[1257] The user enters the post content

[1258] The user uses the device to input the content to be posted on the social networking site. For example, the user might input, "The new project isn't progressing at all. My boss's instructions are too vague!" This input content is then processed in the next step.

[1259] Step 2:

[1260] The device sends the post content to the generative AI

[1261] The device sends the input post content to the generative AI. Specifically, it generates a prompt sentence: "Please check the following post content: 'The new project isn't progressing at all. My boss's instructions are too vague!'" and sends it to the generative AI. This prompt sentence becomes the input data for the generative AI.

[1262] Step 3:

[1263] The server analyzes the content of posts using generative AI

[1264] The server uses generative AI to analyze the content of the post. The generative AI analyzes the prompt and checks for keywords and phrases contained in the post. For example, it detects keywords such as "project," "instructions," and "ambiguous" to determine the risk of a controversy or the risk of confidential information being leaked. The results of this analysis become the input data for the next step.

[1265] Step 4:

[1266] The server recognizes emotions using an emotion engine

[1267] The server uses an emotion engine to recognize the user's emotions from the posted content. The emotion engine analyzes the context and wording of the posted content to identify the user's emotions. For example, it determines whether the posted content expresses anger or frustration. The emotion recognition results become input data for the next step.

[1268] Step 5:

[1269] The server generates a report based on the analysis results.

[1270] The server automatically generates a report based on the analysis results of the generative AI and emotion engine. This report explains why the post was deemed dangerous and what emotions it contains. The generated report serves as input data for the next step.

[1271] Step 6:

[1272] The server notifies the user of the report

[1273] The server notifies the user of the generated report, for example, by displaying a message such as "This post may cause controversy" or "This post contains confidential information." The user can then decide whether to modify the post or refrain from posting.

[1274] Step 7:

[1275] The server shares the report with the organization

[1276] The server shares the generated reports within the organization and makes them accessible to relevant parties, allowing the entire organization to manage the risks of posted content and take appropriate action.

[1277] (Application example 1)

[1278] Next, a description will be given of Application Example 1 of Form 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."

[1279] In today's companies, there is a growing risk that content posted by employees or part-time workers on social media will spark controversy or leak confidential company information. Effective measures to prevent such risks are needed. Furthermore, it is necessary to protect the company's brand image by analyzing the sentiment of the posts and displaying appropriate warnings if negative sentiment is included.

[1280] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means. In this invention, the server includes: a generative AI means for determining the content of posts to SNS; a generative AI means for editing the posted content if the generative AI determines that the posted content is risky; a means for automatically generating a report on the posted content edited by the generative AI; a means for having an employee review the report; a means for sharing the report with the company; an emotion engine means for analyzing the emotion of the posted content; a means for displaying a warning if the emotion engine detects a negative emotion; and a means for determining, by the generative AI, whether the posted content has the potential to cause a controversy or contains confidential information. This makes it possible to evaluate the risk of the posted content to SNS in advance and take appropriate measures.

[1281] "Generative AI methods" are artificial intelligence technologies that analyze the content of posts on social media and determine whether the content has the potential to cause a firestorm or whether it contains confidential information.

[1282] The "emotion engine means" is a technology for analyzing the emotions in the posted content and determining whether or not it contains negative emotions.

[1283] The "means for automatically generating reports" refers to a technology that automatically creates detailed reports about the content of posts edited by generative AI.

[1284] The "means for having employees check" is a technique for presenting the generated report to employees and having them check the contents.

[1285] The "means for sharing with the company" is a technique for sharing the generated report with relevant parties in the company.

[1286] The "means for displaying a warning" is a technique for displaying a warning to the user when a negative emotion is detected by the emotion engine.

[1287] In order to implement this invention, it is necessary to build a system that includes a generative AI means for determining the content of posts on SNS, an emotion engine means, a means for automatically generating reports, a means for employees to check the reports, a means for sharing the reports with the company, and a means for displaying warnings.

[1288] System Program

[1289] This system is realized using the following hardware and software.

[1290] Hardware: Smartphones, servers

[1291] Software: Python, OpenAI API, TextBlob

[1292] Processing Description

[1293] Generative AI method for determining the content of posts on social media

[1294] The server receives the content of posts on social media and analyzes it using a generative AI model to determine whether the content has the potential to spark controversy or whether it contains confidential information.

[1295] Emotion Engine Means

[1296] The server analyzes the sentiment of the post using TextBlob and displays a warning to the user if a negative sentiment is detected by the sentiment engine.

[1297] A means of automatically generating reports

[1298] The server automatically generates a detailed report of the post edited by the generative AI, including an explanation of why the post was dangerous.

[1299] How to get employees to check

[1300] The server presents the generated report to the employee for confirmation, and the employee checks the report and corrects the posted content as necessary.

[1301] How to share with the company

[1302] The server then shares the generated report with relevant parties within the company, allowing the entire company to understand the risks of posted content and take appropriate action.

[1303] How to display a warning

[1304] If the emotion engine detects negative emotions, the server displays a warning to the user, giving them an opportunity to reconsider their post.

[1305] Specific examples

[1306] For example, if an employee tries to post, "Today we'll be announcing details about our new product!", the system works as follows:

[1307] 1. Generative AI method: The server inputs the posted content into the generative AI model and analyzes it.

[1308] 2. Sentiment Engine Method: The server uses TextBlob to analyze the sentiment of the post. In this case, the sentiment is positive, so it proceeds to the next step.

[1309] 3. Means for automatically generating reports: The server automatically generates detailed reports about the posts edited by the generative AI.

[1310] 4. Means for employee to check: The server presents the generated report to the employee and asks them to check the contents.

[1311] 5. Means of sharing with the company: The server shares the generated report with the company's stakeholders.

[1312] 6. Means for displaying a warning: If a negative emotion is detected by the emotion engine, the server displays a warning to the user.

[1313] Prompt Sentence Examples

[1314] Please assess whether the following posts have the potential to cause controversy or contain confidential company information:

[1315] Today we're announcing new product details!

[1316] In this way, the system can assess the risks of posts on social media in advance and take appropriate action.

[1317] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1318] Step 1:

[1319] The server receives posts from users to the SNS. The input is the post entered by the user, and the output is the text data of the post.

[1320] Step 2:

[1321] The server inputs the received post content into a generative AI model for analysis. The input is the text data of the post content, and the output is the analysis results. This analysis determines whether the post content has the potential to cause a controversy or whether it contains confidential information.

[1322] Step 3:

[1323] The server analyzes the sentiment of the post using TextBlob. The input is the text data of the post, and the output is the result of the sentiment analysis. If the sentiment engine detects a negative sentiment, the next step is to display a warning.

[1324] Step 4:

[1325] If the emotion engine detects a negative emotion, the server displays a warning to the user. The input is the result of the emotion analysis, and the output is a warning message. The user receives this warning and has the opportunity to reconsider what they posted.

[1326] Step 5:

[1327] The server automatically generates a detailed report of the post edited by the generative AI. The input is the analysis result of the generative AI model, and the output is a detailed report, including an explanation of why the post was dangerous.

[1328] Step 6:

[1329] The server presents the generated report to the employee for review. The input is the detailed report, and the output is the employee's review. The employee reviews the report and corrects the submission as necessary.

[1330] Step 7:

[1331] The server shares the generated report with company stakeholders. The input is a detailed report, and the output is a report shared with company stakeholders. This allows the entire company to understand the risks of posted content and take appropriate measures.

[1332] Example 2

[1333] Next, a description will be given of Example 2 of Form 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."

[1334] Posts on social media often have the potential to spark outrage and pose a risk of leaking confidential company information. This can threaten a company's reputation and security. Furthermore, it is difficult to accurately grasp a user's psychological state because it is difficult to properly extract and analyze emotions from user posts. A system that can solve these issues is needed.

[1335] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1336] In this invention, the server includes generative AI means for determining the content of posts to the SNS, generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, means for returning the edited posted content by the generative AI to the user, emotion engine means for extracting emotions from the posted content, means for analyzing the emotions extracted by the emotion engine, and means for providing the analysis results to the user. This makes it possible to accurately grasp the emotions of users while ensuring the safety of the content posted to the SNS.

[1337] "Generative AI methods" are artificial intelligence technologies that analyze the content of posts on social media, detect parts that could cause outrage or confidential information, and edit them as necessary.

[1338] The "emotion engine means" is a natural language processing technology for extracting user emotions from the content of social media posts.

[1339] The "means of returning to the user" refers to the communication means for sending the posted content edited by the generative AI to the user's device.

[1340] The "means for analyzing emotions" is a technique for analyzing the emotion data extracted by the emotion engine and evaluating the user's psychological state.

[1341] "Means for providing analysis results to users" refers to means for notifying users of the analysis results from the emotion engine and generative AI.

[1342] This invention is a system for safely and appropriately managing content posted to a social networking service. The system includes a generative AI means, an emotion engine means, a means for sending feedback to a user, a means for analyzing emotions, and a means for providing the analysis results to a user.

[1343] The servers provide the hardware and software to run generative AI models and emotion engines, specifically using servers equipped with NVIDIA GPUs and deep learning frameworks such as TensorFlow and PyTorch.

[1344] When a user posts on an SNS, the device sends the content of the post to the server. The server first uses an emotion engine to extract emotions from the content of the post. For example, from a post such as "Today was fun!", the emotion of "joy" is extracted.

[1345] Next, generative AI analyzes the content of posts to detect parts that could spark controversy or touch on confidential company information. For example, if a post says, "Our new product is better than our competitors' products," the generative AI rewrites this part to read, "Our new product has superior performance."

[1346] The server then sends the edited post back to the user, who then makes a final confirmation.

[1347] As a concrete example, suppose a user posts the following:

[1348] "Today was tough... but our new product is better than our competitors' products!"

[1349] When this post is sent to the server, the emotion engine extracts the emotions of "distress" and "happiness." The generative AI then rewrites the part that says, "Our new product is better than our competitors' products," to "Our new product has superior performance."

[1350] Example prompt sentence:

[1351] Please analyze the following social media posts and remove or rewrite any potentially inflammatory or confidential information:

[1352] "Today was tough... but our new product is better than our competitors' products!"

[1353] By inputting this prompt into a generative AI model, appropriate editing is performed.

[1354] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1355] Step 1:

[1356] A user posts to a social networking site.

[1357] A user uses their own device (such as a smartphone or PC) to input content to post on a social networking site and presses the post button. For example, a user might post, "Today was tough... but our new product is better than our competitors' products!" The input is the user's post, and the output is the post displayed on the device.

[1358] Step 2:

[1359] The device sends the post content to the server.

[1360] The terminal sends the content entered by the user to the server. This data is sent securely using the HTTPS protocol. The input is the user's post, and the output is the post sent to the server.

[1361] Step 3:

[1362] The server uses an emotion engine to extract emotions from the posted content.

[1363] The server inputs the received post content into the emotion engine. The emotion engine uses natural language processing technology to extract emotions from the post content. For example, it extracts the emotion of "distress" from the part "Today was tough..." and the emotion of "joy" from the part "But our new product is better than our competitors' products!" The input is the post content sent to the server, and the output is the extracted emotion data.

[1364] Step 4:

[1365] The server uses generative AI to analyze the content of the post and edit it as necessary.

[1366] The server inputs the content of the post into a generative AI model to detect potentially inflammatory parts and confidential information. For example, it detects the part that says, "Our new product is better than our competitors' products," and rewrites it to, "Our new product has superior performance." The generative AI model uses a pre-trained dataset. The input is the extracted emotion data and the post content, and the output is the edited post content.

[1367] Step 5:

[1368] The server sends the edited post back to the user.

[1369] The server sends the edited post back to the user's device, again securely using the HTTPS protocol. The input is the edited post, and the output is the post sent back to the user's device.

[1370] Step 6:

[1371] The user provides final confirmation.

[1372] The user checks the edited post displayed on the device. If necessary, they can make further corrections. Finally, the user confirms the post and publishes it to the SNS. The input is the post sent back to the user's device, and the output is the final confirmed post.

[1373] (Application example 2)

[1374] Next, a description will be given of Application Example 2 of Form 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."

[1375] In today's social networking sites, users often unintentionally post risky content, leading to problems such as online outrage and information leaks. Another issue is the difficulty of properly extracting emotions from user posts and providing feedback. To solve these problems, a system is needed that automatically assesses the riskiness of posts and edits them as necessary. It is also necessary to support users' psychological health by extracting their emotions and providing appropriate feedback.

[1376] 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.

[1377] In this invention, the server includes a generative AI means for determining the content of posts to the social networking site, a generative AI means for editing the post if the generative AI determines that the post is dangerous, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, an emotion engine means for extracting emotions from the post, and a means for providing feedback on the extracted emotions to the user. This prevents users from unintentionally posting dangerous content and reduces the risk of online outrage and information leaks. Furthermore, by appropriately extracting and providing feedback on the user's emotions, the user's psychological health can be supported.

[1378] "Generative AI methods" are methods that use artificial intelligence technology to analyze the content of posts on social media and determine whether they are dangerous.

[1379] "Generative AI editing means" refers to means that use artificial intelligence technology to automatically modify or delete posted content that is deemed dangerous by generative AI.

[1380] "Means for automatically generating reports" refers to a means for automatically creating a report that explains the reasons for the danger and the edited content based on the edited content of a post using generative AI.

[1381] The "means for employee confirmation" refers to a means for presenting the automatically generated report to an employee for confirmation and approval.

[1382] "Means for sharing within the organization" refers to the means for sharing confirmed reports within the organization and disseminating information widely.

[1383] The "emotion engine means" is a means that uses technology to extract user emotions from the content posted on SNS.

[1384] The "feedback means" is a means for appropriately conveying the extracted emotion to the user.

[1385] A system for carrying out this invention includes a generative AI means for determining the content of posts to SNS and editing it as necessary, an emotion engine means for extracting emotions from the posted content, and a means for providing feedback of the extracted emotions to the user. A specific embodiment of this system is described below.

[1386] System Configuration

[1387] The system mainly uses the following hardware and software:

[1388] Hardware: Smartphones, servers

[1389] Software: Python, transformers library, sentiment analysis model

[1390] Program processing

[1391] The server receives the content posted to the SNS and analyzes it using the generative AI means. The generative AI means determines whether the content is dangerous, and if it is determined to be dangerous, the editing generative AI means automatically edits or deletes the content. For edited content, the automatic report generation means creates a report explaining the reason for the edit and the edited content.

[1392] Next, the emotion engine means extracts the user's emotion from the posted content. The extracted emotion is conveyed to the user through the feedback means. This allows the user to understand how their posted content is received and supports their psychological well-being.

[1393] Specific examples

[1394] For example, if a user posts the following:

[1395] Our new product is better than our competitors' products. Today was tough...

[1396] This post content is sent to a server and analyzed by the generative AI means. The generative AI means judges the part "better than competitors' products" to be dangerous, and the editing generative AI means rewrites it to "has excellent performance." After that, the emotion engine means extracts the emotion of "distress" from the part "Today was tough..."

[1397] Prompt Sentence Examples

[1398] An example of a prompt to input to a generative AI model is as follows:

[1399] Check your posts, edit out the risky parts, and extract the emotion.

[1400] Post: "Our new product is better than our competitors' products. Today was tough..."

[1401] Using this prompt, the generative AI model can check and edit posts and extract sentiment, helping users to make safe and appropriate posts and understand their own emotions.

[1402] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1403] Step 1:

[1404] The server receives posts from users to the SNS. The input is the user's post, and the output is the text data of the post. Specifically, the user enters the post content using a smartphone and sends it to the server.

[1405] Step 2:

[1406] The server analyzes the received post using a generative AI method. The input is the text data of the post received in step 1, and the output is the analysis result. Specifically, the generative AI model scans the post and detects dangerous keywords and phrases.

[1407] Step 3:

[1408] If the server determines that the posted content is dangerous, it uses the generative AI means to edit or delete the posted content. The input is the analysis result obtained in step 2, and the output is the corrected posted content. Specifically, the dangerous parts are rewritten to safe expressions.

[1409] Step 4:

[1410] The server creates a report using a means for automatically generating a report on the edited post content using the generative AI means. The input is the post content corrected in step 3, and the output is a report. Specifically, the server generates a report that includes content explaining why the part was dangerous and how it was corrected.

[1411] Step 5:

[1412] The server presents the generated report to the employee using a means for the employee to review it. The input is the report generated in step 4, and the output is the employee's review result. Specifically, the employee reviews the report and provides feedback on approval or correction.

[1413] Step 6:

[1414] The server shares the confirmed report within the organization using a means for sharing the report with the organization. The input is the confirmation result obtained in step 5, and the output is the report shared within the organization. Specifically, the server saves the report in the organization's database and notifies the relevant parties.

[1415] Step 7:

[1416] The server uses an emotion engine to extract the user's emotions from the posted content. The input is the text data of the posted content received in step 1, and the output is the extracted emotion data. Specifically, the emotion analysis model analyzes the posted content and identifies emotions such as joy, sadness, and anger.

[1417] Step 8:

[1418] The server communicates the extracted emotion to the user using a means of providing feedback to the user. The input is the emotion data extracted in step 7, and the output is a feedback message to the user. Specifically, the server sends a notification to the user's smartphone, explaining the extracted emotion and the reason for it.

[1419] Example 3

[1420] Next, a description will be given of Example 3 of Form Example 3. 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."

[1421] When content posted on social media poses risks such as flame wars or the leaking of confidential information, it is necessary to appropriately detect and edit it. It is also necessary to generate detailed reports on the edited content to help educate employees and gather information for the company. However, performing these processes manually is time-consuming, labor-intensive, and inefficient. Therefore, there is a need for a system that can automatically detect risks in posted content, edit it, and generate and share reports on the content.

[1422] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1423] In this invention, the server includes means for receiving posted content, means for passing the received posted content to a generative AI, means for the generative AI to analyze the posted content and detect risky parts or negative emotions, means for the generative AI to edit the posted content based on the analysis results, means for the generative AI to automatically generate a report on the edited content, means for sending the generated report to employees to check, and means for sharing the generated report with the company and collecting information on employees' complaints and good points. This makes it possible to automatically detect risks in posted content, edit it, and generate and share reports on the content.

[1424] "Posted content" refers to text or media information that a user sends to a social networking site or other platform.

[1425] "Generative AI" is a system that uses artificial intelligence technology to analyze data and automatically perform specific tasks.

[1426] "Analysis" is the process by which generative AI evaluates the content of posts and detects risky parts or negative emotions.

[1427] "Editing" is the process by which generative AI changes the content of posts based on the analysis results to prevent outrage or the leaking of confidential information.

[1428] A "report" is a document automatically generated by generative AI about the edits, detailing which parts were dangerous, why, and how they were edited.

[1429] "Employees" are individuals employed by a company who review the generated reports and use them for training purposes.

[1430] The "company" refers to an organization or business entity that shares the generated reports and collects information on employee complaints and positive points.

[1431] "Dangerous parts" are parts of the post that could cause an uproar or parts that touch on confidential company information.

[1432] "Negative emotions" are negative emotions of users detected by the emotion engine, which affect the tone and expression of the content of posts.

[1433] The "emotion engine" is software that evaluates the emotional tone of posts and functions as part of a generative AI.

[1434] This invention is a system that automatically analyzes content posted to social networking sites, detects and edits dangerous parts and negative emotions, and generates and shares reports on the edited content. Specific embodiments of this system are described below.

[1435] Hardware and software used

[1436] Server: A central computer system that receives posts and passes them on to the generative AI.

[1437] Device: The device (e.g., computer, smartphone) on which a user enters and submits their post.

[1438] Generative AI models: Artificial intelligence systems that analyze and edit posts (e.g., OpenAI GPT-4).

[1439] Sentiment engine: Software to assess the emotional tone of posts (e.g., IBM Watson Tone Analyzer).

[1440] System Operation

[1441] 1. The user enters the content to post on their device and presses the send button.

[1442] 2. The server receives this post and temporarily stores it in a database.

[1443] 3. The server creates an API request to pass the received post content to the generative AI.

[1444] 4. Generative AI analyzes posts to detect risk and negative sentiment. Specifically, it uses an emotion engine to evaluate the emotional tone of posts.

[1445] 5. The generative AI edits the post based on the analysis results. For example, a post that says, "The new product's functionality is insufficient, and we've received many complaints from customers" can be edited to say, "We are considering improving the functionality of our new product based on feedback from customers."

[1446] 6. The generative AI automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. For example, it might say, "The expression 'many complaints' was negative, so to emphasize customer dissatisfaction, we edited it to 'we are considering improvements based on feedback.'"

[1447] 7. The server sends the generated report to the employee for review. Specifically, the report is sent to the employee's email address as an attachment.

[1448] 8. The server also saves the report to the company database and collects information on employees' dissatisfaction and positive points. For example, the report may reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[1449] Examples of concrete examples and prompts

[1450] Examples:

[1451] A user posts, "The new product's functionality is insufficient, and we're receiving a lot of complaints from customers."

[1452] The generative AI edits the article to say, "We are currently considering improving the functionality of our new product based on feedback from customers."

[1453] The report states, "The expression 'many complaints' was negative, and to emphasize customer dissatisfaction, we edited it to 'we are considering improvements in response to feedback.'"

[1454] Example prompt sentence:

[1455] "Analyze user posts, detect risky or negative sentiment, and edit them. Generate detailed reports on edits."

[1456] This system makes it possible to automatically detect risks in posted content, edit it, and generate and share reports on the content. This makes it possible to efficiently educate employees and collect information about the company. The flow of the identification process in Example 3 will be explained using FIG. 21.

[1457] Step 1:

[1458] The user inputs the content to be posted on the terminal and presses the send button.

[1459] Input: User-typed post content (e.g., "Our new product's functionality is insufficient, and customers are complaining about it.").

[1460] Output: The post is sent to the server.

[1461] Specific behavior: The user enters the content to post into the input field on the device and clicks the send button.

[1462] Step 2:

[1463] The server receives the posted content and temporarily stores it in a database.

[1464] Input: The post submitted by the user.

[1465] Output: Post content temporarily saved in the database.

[1466] Specific operation: The server receives the HTTP request and saves the post content to the database.

[1467] Step 3:

[1468] The server creates an API request to pass the received post content to the generative AI.

[1469] Input: Post content saved in database.

[1470] Output: API request to be passed to the generative AI.

[1471] Specific operation: The server retrieves the post content and sends a request to the generative AI's API endpoint.

[1472] Step 4:

[1473] Generative AI analyzes the content of posts and detects dangerous parts and negative emotions.

[1474] Input: The post content sent from the server.

[1475] Output: Analysis results (detection of dangerous parts and negative emotions).

[1476] How it works: Generative AI uses an emotion engine to assess the emotional tone of posts and identify risky areas.

[1477] Step 5:

[1478] The generative AI edits the post content based on the analysis results.

[1479] Input: Analysis results and original post.

[1480] Output: The edited post.

[1481] How it works: Based on the analysis results, the generative AI edits dangerous parts and negative expressions in the post. For example, it edits "The new product's functionality is insufficient, resulting in many customer complaints" to "We are currently considering improving the functionality of the new product based on customer feedback."

[1482] Step 6:

[1483] Generative AI automatically generates a report on the edits.

[1484] Input: Edited post and analysis results.

[1485] Output: A report of your edits.

[1486] What it does: The generative AI generates a report detailing the reasons for the edits and the changes made. For example, it might say, "The expression 'many complaints' is negative, and to emphasize customer dissatisfaction, we edited it to 'we are considering improvements based on feedback.'"

[1487] Step 7:

[1488] The server sends the generated report to the employee for review.

[1489] Input: The generated report.

[1490] Output: Report sent to employee.

[1491] Specific operation: The server attaches the report and sends it to the employee's email address.

[1492] Step 8:

[1493] The server also stores the reports in the company's database, collecting information on employee complaints and positive points.

[1494] Input: The generated report.

[1495] Output: Reports saved in the company database.

[1496] What it does: The server saves the report in the company database and uses it for analysis. For example, the report may reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[1497] (Application example 3)

[1498] Next, a description will be given of Application Example 3 of Form Example 3. 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."

[1499] The content posted by employees and users on social media and other online platforms can have a negative impact on a company's reputation and security. In particular, posts that have the potential to spark controversy or contain confidential information pose a significant risk to companies. Furthermore, there is a lack of efficient ways to collect information about employees' complaints and positive points and use it to educate them. To address these issues, a system is needed that monitors posts in real time, automatically edits dangerous content, and generates detailed reports on the edits.

[1500] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1501] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having employees review the report, a means for sharing the report with the company, a means for monitoring the posted content in real time, a means for displaying the edited content of the posted content, and a means for using the report to help educate employees. This makes it possible to monitor the dangers of posted content in real time and edit it appropriately to reduce corporate reputation and security risks, and also to help educate employees and improve the company.

[1502] "Generative AI methods" refers to artificial intelligence technology that judges the content of posts on social media and edits them as necessary.

[1503] "Posted content" refers to information such as text, images, and videos that users make public on social media and other online platforms.

[1504] "If deemed dangerous" refers to a situation where the generative AI determines that the content of a post could have a negative impact on a company's reputation or security.

[1505] "Edit" means to delete, change, or add any part of a post.

[1506] "Means for automatically generating reports" refers to technology that automatically creates a report on the content of posts edited by generative AI, including the reasons for the edits and details.

[1507] "Means for employees to check" refers to a method or system for presenting the generated report to employees and having them check the contents.

[1508] "Means for sharing within the company" refers to the methods and systems for sharing generated reports throughout the company and efficiently collecting and utilizing information.

[1509] "Real-time monitoring measures" refers to technology that allows for immediate monitoring of posted content and rapid detection of dangerous content.

[1510] "Means for displaying edited content" refers to a method or system for displaying the post content after it has been edited by generative AI to users and employees.

[1511] "Means for employee education" refers to a method or system for educating or training employees using the generated reports.

[1512] A system for implementing this invention monitors content posted to social networking sites in real time, automatically edits dangerous content using generative AI, and generates a detailed report on the edited content. A specific embodiment of this system is described below.

[1513] System Configuration

[1514] The system consists of the following main components:

[1515] 1. Generative AI: Evaluates the content of posts on social media and edits them as necessary.

[1516] 2. Real-time monitoring measures: Immediately monitor posted content and quickly detect dangerous content.

[1517] 3. Automatic report generation: Automatically generate a report for edited posts, detailing the reasons for the edits and other details.

[1518] 4. Employee confirmation method: Present the generated report to employees and have them confirm the contents.

[1519] 5. Company sharing tool: Generated reports can be shared across the company to efficiently collect and utilize information.

[1520] 6. Display of edited content: The post content after being edited by the generative AI is displayed to users and employees.

[1521] 7. Employee education tool: Use the generated reports to educate and train employees.

[1522] Program processing

[1523] The server monitors the content posted on social media in real time and judges the content using a generative AI. If the judged content is dangerous, the generative AI edits it. A detailed report of the edited content is created by an automatic report generation means. This report is presented to employees through an employee confirmation means, and is further shared throughout the company using a company sharing means. The edited content is displayed to users and employees using a post-edit content display means. Finally, the generated report is used for employee education and training through an employee education means.

[1524] Hardware and software used

[1525] Hardware: Servers, user devices (PCs, smartphones, etc.)

[1526] Software: OpenAI API, real-time monitoring system, report generation system

[1527] Specific examples

[1528] For example, consider the following post on a social networking site:

[1529] Example of post content:

[1530] This new product is completely unusable and the company's response is terrible.

[1531] The server monitors the content of this post in real time and uses generative AI to generate the following prompt:

[1532] Example prompt sentence:

[1533] Please analyze the post below and edit out any dangerous parts.

[1534] Post content: This new product is completely unusable. The company's response is terrible.

[1535] Edited content:

[1536] The generative AI analyzes the posts based on these prompts and edits out any dangerous parts. The edited content and a detailed report are then shared with the entire company for review. Ultimately, this report can be used to educate and train employees.

[1537] In this way, the system can reduce reputational and security risks to the company and help educate employees and improve the company.

[1538] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1539] Step 1:

[1540] The server monitors posts on social media in real time. It receives posts obtained from the social media platform as input and passes the monitored posts to the generative AI as output. Specifically, the server connects with the social media platform via an API and obtains the content of each new post.

[1541] Step 2:

[1542] The server uses generative AI to judge the content of the post. As input, it receives the post content obtained in step 1, and as output, it obtains a judgment result as to whether the post content is dangerous. Specifically, the server generates the following prompt sentence for the generative AI and makes a judgment.

[1543] Please analyze the post below and edit out any dangerous parts.

[1544] Post content: [Post content]

[1545] Edited content:

[1546] Step 3:

[1547] If the generative AI determines that the posted content is dangerous, the server edits the posted content. The server receives the judgment result from step 2 and the posted content as input, and obtains the edited posted content as output. Specifically, the generative AI deletes or changes the dangerous parts and edits the content to be safe.

[1548] Step 4:

[1549] The server automatically generates a report on the post content edited by the generative AI. It receives the edited post content as input and generates a detailed report of the edited content as output. Specifically, the server generates the following prompt for the generative AI and creates the report:

[1550] I edited the post below.

[1551] Original post: [Original post]

[1552] Edited content: [Edited content]

[1553] Please provide details about which parts were dangerous, why, and how you edited them.

[1554] Step 5:

[1555] The server allows the employee to review the generated report. It receives the report as input and presents it to the employee as output. Specifically, the server sends the report to the employee's terminal and notifies them to review it.

[1556] Step 6:

[1557] The server shares the generated report with the company. It receives the report as input and generates information to be shared across the company as output. Specifically, the server uploads the report to the company's internal system and notifies relevant parties.

[1558] Step 7:

[1559] The server displays the edited content to users and employees. It receives the edited content as input and generates the display data as output. Specifically, the server updates the edited content on the social media platform or internal systems.

[1560] Step 8:

[1561] The server uses the generated reports to educate and train employees. It receives the reports as input and generates training materials as output. Specifically, the server creates an educational program based on the reports and provides it to employees.

[1562] 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.

[1563] 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> ) 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.

[1564] Another example of generative AI is Gemini (internet search engine). <url: https: gemini.google.com ?hl="ja">) are listed.

[1565] 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.

[1566] [Third embodiment]

[1567] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1568] 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.

[1569] 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).

[1570] 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.

[1571] 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.

[1572] 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).

[1573] 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.

[1574] 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.

[1575] 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.

[1576] 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.

[1577] 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.

[1578] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[1579] "Example 1"

[1580] In one embodiment of the present invention, before employees or part-timers post on SNS, the content of the post is input into a generative AI. The generative AI determines whether the content of the post is likely to cause an outrage or whether it contains confidential company information. Specifically, the generative AI determines whether the content of the post contains keywords or phrases that may cause an outrage or keywords or phrases related to confidential company information.

[1581] Check if it is included.

[1582] "Example 2"

[1583] If the generative AI judges a post to be dangerous, it will edit it. Specifically, it will delete or rewrite any parts that could spark outrage or that touch on confidential company information. For example, if a post says, "Our new product is better than our competitors' products," the AI ​​will rewrite it to say, "Our new product has superior performance."

[1584] "Example 3"

[1585] After the generative AI edits the post, it automatically generates a report on the edits. The report details which parts were dangerous, why they were so, and how they were edited. This report is then reviewed by employees, and used to educate them.

[1586] "Example 4"

[1587] The reports created by the generative AI are shared with the company, allowing the company to efficiently collect information on employees' complaints and positive points. For example, the report can reveal that employees have high opinions of a new product, or that some employees are dissatisfied with their work.

[1588] The processing flow of each embodiment will be described below.

[1589] "Example 1"

[1590] Step 1: Before employees or part-time workers post on social media, the content of the post is input into a generative AI.

[1591] Step 2: The generative AI determines whether the post is likely to spark an uproar or contain confidential company information. Specifically, it checks whether the post contains keywords or phrases that could spark an uproar or keywords or phrases related to confidential company information.

[1592] "Example 2"

[1593] Step 1: If the generative AI determines that the post is dangerous, it will edit it.

[1594] Step 2: Specifically, parts that could cause controversy or parts that touch on confidential company information are deleted or rewritten. For example, if a post says, "Our new product is better than our competitors' products," the AI ​​will rewrite it to say, "Our new product has superior performance."

[1595] "Example 3"

[1596] Step 1: After the generative AI edits the post, it automatically generates a report on the edits.

[1597] Step 2: The report will detail which parts were dangerous, why they were, and how they were edited. This report will be reviewed by employees and used to educate them.

[1598] "Example 4"

[1599] Step 1: The report created by the generative AI is shared with the company.

[1600] Step 2: This allows the company to efficiently collect information on employees' dissatisfaction and positive points. For example, the report can reveal that employees have high opinions about a new product, or that some employees are dissatisfied with their work.

[1601] Example 1

[1602] Next, a description will be given of Example 1 of Form 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."

[1603] When employees or part-time workers post on social media, if there is a possibility that the content of the post may spark outrage or may contain confidential company information, it is necessary to assess the risk in advance and take appropriate action. However, manual checking is time-consuming and labor-intensive, and is inefficient. Furthermore, the accuracy of the checks can vary depending on the person. This poses a challenge, as it increases the risk of damaging a company's brand image and leaking confidential information.

[1604] 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.

[1605] In this invention, the server includes a generative AI means for determining the content of posts to the SNS, a generative AI means for editing the post if the post is determined to be risky by the generative AI, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for a user to input the post content before posting it to the SNS, a means for transmitting the input post content to the server, a means for the server to input the post content to a generative AI model, a means for the generative AI model to analyze the post content and return a determination result to the server, a means for the server to return the determination result to a terminal, and a means for the terminal to display the determination result to the user. This makes it possible to automatically determine the risk of the post content in advance and take appropriate measures quickly and efficiently.

[1606] "Generative AI methods" refers to artificial intelligence technology that analyzes the content of posts on social media and determines the possibility of a controversy or the risk of confidential information being leaked.

[1607] "Generative AI means for editing posted content" refers to artificial intelligence technology that automatically corrects and edits posted content that is deemed dangerous.

[1608] "Means for automatically generating reports" refers to technology that automatically creates reports that explain the risks of the posted content and the details of any corrections based on the posted content edited by the generative AI means.

[1609] "Means for employee review" refers to a technique for presenting the generated report to an employee for review or approval.

[1610] "Means for sharing with the organization" refers to the technology for sharing the generated report with relevant parties within the organization and communicating the information.

[1611] "Means for users to input content before posting on SNS" refers to technology that provides an interface for users to input content they intend to post on SNS in advance.

[1612] The term "means for transmitting the input posting content to the server" refers to a communication technology for transmitting the posting content input by the user to the server.

[1613] "Means by which the server inputs the posted content into the generative AI model" refers to the technology by which the server inputs the posted content received into the generative AI model and has it analyzed.

[1614] "Means for the generative AI model to analyze the content of posts and return the results to the server" refers to the technology that enables the generative AI model to analyze the content of posts and return the results to the server.

[1615] "Means by which the server returns the judgment results to the terminal" refers to the communication technology that enables the server to return the judgment results received from the generative AI model to the terminal.

[1616] "Means for the terminal to display the determination result to the user" refers to a technology that provides an interface for the terminal to display the determination result received from the server to the user.

[1617] MODE FOR CARRYING OUT THE INVENTION

[1618] This invention is a system that checks the content of posts made by employees and part-time workers on social media in advance to determine the possibility of them causing controversy or the risk of confidential information being leaked.The system uses a generative AI model to analyze the content of posts and automatically determine the risk.

[1619] Hardware and software used

[1620] Server: Runs the generative AI model and analyzes and judges the content of posts.

[1621] Terminal: Provides an interface for users to input their posting content and check the results of the evaluation.

[1622] Generative AI models: For example, use OpenAI's GPT-4.

[1623] Data processing and calculation

[1624] 1. Users enter their post content before posting to social media:

[1625] Users can use a dedicated application or web interface on their device to input the content they want to post to SNS. For example, a user might input, "I'm sharing details about our new product on SNS! This is still confidential, but I want everyone to know about it soon!"

[1626] 2. The device sends the entered post content to the server:

[1627] The terminal sends the content of the post entered by the user to the server. At this time, the terminal converts the content of the post into an appropriate format (for example, JSON format) and sends it to the server using a secure communication protocol (for example, HTTPS).

[1628] 3. The server inputs the post into the generative AI model:

[1629] The server inputs the received post content into the generative AI model. Specifically, the server generates a prompt like this:

[1630] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[1631] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[1632] This prompt is sent to the generative AI model.

[1633] 4. The generative AI model analyzes the post and returns the results to the server:

[1634] The generative AI model analyzes the prompt text and checks whether the post contains keywords or phrases that could spark controversy or keywords or phrases related to confidential company information. For example, it detects the keyword "confidential" and determines that the post contains confidential company information. The generative AI model then returns this determination result to the server.

[1635] 5. The server returns the result to the device:

[1636] The server receives the results of the generative AI model and sends them back to the device, including information on whether the content of the post is likely to spark an uproar or whether it contains confidential company information.

[1637] 6. The device displays the result to the user:

[1638] The device displays the result of the judgment received from the server to the user. For example, it may display a message saying, "This post contains confidential company information. Please correct the post." Based on this information, the user can either correct the post or cancel it.

[1639] Specific examples

[1640] Suppose a user enters the following post content:

[1641] "We'll be sharing details about our new products on social media! This is still confidential, but we want you all to know soon!"

[1642] Here's an example prompt to feed this post to a generative AI model:

[1643] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[1644] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[1645] The generative AI model analyzes this prompt, detects that the post contains keywords such as "confidential," and determines that it may contain confidential company information.

[1646] The server returns the result of this determination to the terminal, and the terminal displays to the user a message saying, "This post contains confidential company information. Please correct your post."

[1647] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1648] Step 1:

[1649] The user inputs the content to post before posting to the SNS.

[1650] Users use a dedicated application or web interface on their device to input the content they want to post to SNS. For example, a user might input, "I'm sharing details about our new product on SNS! This is still confidential, but I want everyone to know about it soon!" The input content is temporarily stored in the device's memory.

[1651] Step 2:

[1652] The terminal transmits the input post content to the server.

[1653] The terminal converts the content of the post entered by the user into an appropriate format (e.g., JSON format) and sends it to the server using a secure communication protocol (e.g., HTTPS). The input data is the text information of the post, and the output data is the post sent to the server.

[1654] Step 3:

[1655] The server inputs the post content into a generative AI model.

[1656] The server inputs the received post content into the generative AI model. Specifically, the server generates a prompt like this:

[1657] Please assess whether the following posts are likely to be controversial or contain confidential company information:

[1658] "We'll be sharing details about our new products on social media! They're still confidential, but we want everyone to know about them soon!"

[1659] The input data is the text information of the post, and the output data is the generated prompt.

[1660] Step 4:

[1661] The generative AI model analyzes the content of the post and returns the results to the server.

[1662] The generative AI model analyzes the prompt text and checks whether the post contains keywords or phrases that could spark an uproar or keywords or phrases related to confidential company information. For example, it detects the keyword "confidential" and determines that the post contains confidential company information. The input data is the prompt text, and the output data is the judgment result.

[1663] Step 5:

[1664] The server returns the result of the determination to the terminal.

[1665] The server receives the judgment results from the generative AI model and sends them back to the device. The judgment results include information on whether the content of the post has the potential to spark a controversy or whether it contains confidential company information. The input data is the judgment result, and the output data is the judgment result sent to the device.

[1666] Step 6:

[1667] The terminal displays the judgment result to the user.

[1668] The device displays the judgment result received from the server to the user. For example, it may display a message such as "This post contains confidential company information. Please correct the post." The input data is the judgment result, and the output data is the judgment result displayed to the user. Based on this information, the user can correct the post or cancel it.

[1669] (Application example 1)

[1670] Next, a description will be given of Application Example 1 of Form 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."

[1671] When employees or part-time workers post on social media, there is a risk that they may unintentionally post content that will spark controversy or include confidential information. Such posts not only damage the company's credibility, but can also lead to legal issues. Furthermore, manually checking and correcting posted content is time-consuming and labor-intensive, and inefficient. To solve these issues, a system is needed that can automatically assess the content of posts and prompt corrections as necessary.

[1672] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1673] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the post if the post is determined to be dangerous by the generative AI, a means for automatically generating a report on the post edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for converting the post into a prompt sentence and generating a determination result using a generative AI model, a means for displaying the determination result to the user, and a means for prompting the user to modify the post as necessary. This makes it possible to automatically check the content posted by employees and part-time workers on SNS, reducing the risk of online outrage and the risk of confidential information being leaked.

[1674] "Generative AI methods" are methods that use artificial intelligence to analyze the content of social media posts and determine the risk of online outrage or the risk of confidential information being leaked.

[1675] "Generative AI means for editing posted content when the determined posted content is dangerous" refers to a means for automatically correcting posted content that the generative AI has determined to be dangerous.

[1676] "Means for automatically generating a report on the content of posts edited by generative AI" means means for automatically creating a detailed report on the content of posts edited by generative AI.

[1677] The "means for having the employee check the report" refers to a means for presenting the generated report to the employee and having the employee check it.

[1678] The "means for sharing the report with the organization" is a means for sharing the generated report within the organization.

[1679] "Means for converting the content of a post into a prompt sentence and generating a judgment result using a generative AI model" refers to means for converting the content of a post entered by a user into a prompt sentence, analyzing the content using a generative AI model, and generating a judgment result.

[1680] "Means for displaying the judgment results to the user" refers to means for displaying the judgment results generated by the generative AI model to the user.

[1681] The "means of encouraging the user to correct the posted content as necessary" refers to a means of encouraging the user to correct the posted content based on the judgment result.

[1682] As an embodiment of the present invention, a system is constructed that automatically judges the content of posts on SNS and prompts users to make corrections as necessary. A specific embodiment of this system is described below.

[1683] First, the server is equipped with a generative AI means for judging the content of posts on social media. This generative AI means analyzes the content of posts entered by users and judges the risk of flame wars and the risk of confidential information leaks. Specifically, it converts the content of posts into prompt statements and generates judgment results using a generative AI model (for example, OpenAI's API).

[1684] Next, the server includes a generative AI means for editing the posted content when the generative AI determines that the posted content is dangerous. This means automatically corrects the posted content that the generative AI determines to be dangerous and provides the user with safe posted content.

[1685] Additionally, the server has a means for automatically generating a report on the posts edited by the generative AI, which will detail why the posts were dangerous and how they were corrected.

[1686] The server also provides a means for employees to review the generated reports and for them to be shared with the organization, allowing employees to understand the risks of what they post and for information to be shared across the organization.

[1687] As a concrete example, consider the case where a user posts something like, "Details of the new product haven't been released yet, but it's going to be great!" This post is converted into a prompt sentence and input into the generative AI model.

[1688] Example prompt sentence:

[1689] Please determine whether the following posts are inflammatory or contain confidential information.

[1690] Posted: Details of the new product have not been released yet, but it's going to be awesome!

[1691] Judgment result:

[1692] The server sends this prompt to the generative AI model and obtains the judgment result. If the post is deemed dangerous based on the judgment result, the server automatically modifies the post and provides safe content to the user. In addition, the generated report can be reviewed by employees and shared with the organization for thorough risk management.

[1693] In this way, this invention automatically checks the content posted on social media by employees and part-time workers, making it possible to reduce the risk of online outrage and the risk of confidential information being leaked.

[1694] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1695] Step 1:

[1696] The user inputs the content to post on social media. The user uses a smartphone or computer to enter the content to post into the application. This input becomes the starting point for the system's processing.

[1697] Step 2:

[1698] The server receives the post content and converts it into a prompt sentence. The server receives the post content entered by the user and converts it into a prompt sentence suitable for the generative AI model. Specifically, the post content is converted into the following format.

[1699] Please determine whether the following posts are inflammatory or contain confidential information.

[1700] Post content: [User post content]

[1701] Judgment result:

[1702] Input: User post

[1703] Output: A prompt suitable for a generative AI model

[1704] Step 3:

[1705] The server generates a judgment result using a generative AI model. The server sends the prompt text to the generative AI model (e.g., OpenAI's API) and receives the judgment result. The generative AI model analyzes the prompt text and determines whether the post content poses a risk of flaming or leaking confidential information.

[1706] Input: prompt statement

[1707] Output: Judgment result

[1708] Step 4:

[1709] The server displays the judgment result to the user. The server displays the judgment result obtained from the generative AI model to the user. The judgment result includes whether the posted content is dangerous and the specific risk content.

[1710] Input: Judgment result

[1711] Output: The result displayed to the user

[1712] Step 5:

[1713] The server prompts the user to modify the posted content as necessary. Based on the judgment result, the server prompts the user to modify the posted content. For example, if the posted content is judged to be dangerous, the server will present the user with specific suggestions for modification.

[1714] Input: Judgment result

[1715] Output: suggested fixes presented to the user

[1716] Step 6:

[1717] The server automatically generates a report on the post content edited by the generative AI. The server automatically creates a detailed report on the post content edited by the generative AI. This report details why the post content was dangerous and how it was corrected.

[1718] Input: Edited post

[1719] Output: Detailed report

[1720] Step 7:

[1721] The server presents the generated report to employees for review and shares it with the organization. The server also shares the report within the organization to ensure thorough risk management.

[1722] Input: Detailed report

[1723] Output: Reports presented to employees, reports shared with the organization

[1724] Example 2

[1725] Next, a description will be given of Example 2 of Form 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."

[1726] On social media and other online platforms, when user posts pose a risk of causing controversy or leaking confidential information, there is a need to properly manage the content and edit it safely. However, manual monitoring and editing takes time and effort, and is inefficient. Therefore, a system is needed that automatically analyzes posted content, identifies risky parts, and edits them.

[1727] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1728] In this invention, the server includes means for receiving posted content entered by a user, means for transmitting the received posted content to a generative AI model, means for the generative AI model to analyze the posted content and identify dangerous parts, means for the generative AI model to edit the identified dangerous parts, and means for returning the edited posted content to the user. This makes it possible to automatically analyze the posted content of users and reduce the risk of flame wars and the leakage of confidential information.

[1729] "User" means an individual or group that uses the system to input posting content.

[1730] "Posted content" refers to information such as text, images, and videos that users post on social media or other online platforms.

[1731] The "server" is a computer system that receives user submissions and sends them to the generative AI model.

[1732] A "generative AI model" is a model that uses artificial intelligence technology to analyze posted content, identify risky parts, and edit them.

[1733] A "prompt sentence" is an input sentence that instructs the generative AI model to analyze and edit the post content.

[1734] "Analysis" is the process by which the generative AI model understands the context and meaning of the post and identifies risky parts.

[1735] "Dangerous parts" are parts of the post that pose a risk of causing an uproar or that contain confidential information.

[1736] "Editing" is the process by which a generative AI model removes or rewrites identified dangerous parts.

[1737] "Return" is the process by which the server sends back to the user the post content edited by the generative AI model.

[1738] This invention relates to a system that automatically analyzes content posted by users on social networking sites and other online platforms, identifies risky parts, and edits them. Specific embodiments of this system are described below.

[1739] The server receives the posting content entered by the user. The user uses their own device (PC, smartphone, etc.) to enter the posting content. For example, they might enter, "Our new product is better than our competitor's product."

[1740] The terminal sends the content of the post entered by the user to the server, using a communication protocol such as an HTTP request.

[1741] The server sends a prompt to the generative AI model to analyze the content of the post received from the device. The prompt includes the content of the post and instructions for analysis. For example, the server sends a prompt such as, "Analyze the content posted by the user, identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'"

[1742] A generative AI model (such as OpenAI's GPT-4) analyzes the content of posts based on prompts received from the server. The model uses natural language processing techniques to understand the context and meaning of the posts and identify parts that may be controversial or contain sensitive information.

[1743] The generative AI model then removes or rewrites the identified risky parts, for example, rewriting "Our new product is better than our competitors' products" to "Our new product has superior performance."

[1744] The server then sends the edited post content received from the generative AI model back to the device, again using a communication protocol such as an HTTP response.

[1745] The terminal displays the edited post content received from the server to the user, who can then check the edited content and revise it again if necessary.

[1746] As a concrete example, if a user posts, "Our new product is better than our competitor's product," the process will proceed as follows:

[1747] 1. The user enters the content to post on the device.

[1748] 2. The device sends the post content to the server.

[1749] 3. The server receives the post and sends the following prompt to the generative AI model: "Analyze the content posted by the user and identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'"

[1750] 4. The generative AI model analyzes the content of the post and identifies any part that claims to be "better than competitors' products" as dangerous.

[1751] 5. The generative AI model rewrites this part to "has excellent performance."

[1752] 6. The server sends the edited post back to the device.

[1753] 7. The device displays the edited post "Our new product has excellent performance" to the user.

[1754] In this way, the system can safely edit user posts.

[1755] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1756] Step 1:

[1757] The user enters the content to post.

[1758] Users use their own devices (PCs, smartphones, etc.) to enter their post content. For example, they might enter, "Our new product is better than our competitors' products." The entered post content is saved on the device.

[1759] Step 2:

[1760] The device sends the post content to the server.

[1761] The terminal sends the content posted by the user to the server. This transmission uses a communication protocol such as an HTTP request. The input is the content posted by the user, and the output is the content sent to the server.

[1762] Step 3:

[1763] The server receives the post and sends a prompt to the generative AI model.

[1764] The server sends a prompt to the generative AI model to analyze the content of the post received from the device. The prompt includes the content of the post and instructions for analysis. For example, it generates a prompt such as, "Analyze the content posted by the user, identify and edit any parts that may pose a risk of controversy or contain confidential information. Post content: 'Our new product is better than our competitors' products.'" The input is the content of the post received from the device, and the output is the prompt sent to the generative AI model.

[1765] Step 4:

[1766] A generative AI model analyzes the content of posts and identifies risky parts.

[1767] A generative AI model (e.g., OpenAI's GPT-4) analyzes the content of posts based on prompts received from the server. The model uses natural language processing techniques to understand the context and meaning of the posts and identify parts that pose a risk of flaming or contain confidential information. The input is the prompt, and the output is the identified risky parts.

[1768] Step 5:

[1769] The generative AI model edits out dangerous parts.

[1770] The generative AI model removes or rewrites the identified risky parts. For example, it rewrites "Our new product is better than our competitors' products" to "Our new product has superior performance." The input is the identified risky parts, and the output is the edited post.

[1771] Step 6:

[1772] The server sends the edited post back to the device.

[1773] The server returns the edited post content received from the generative AI model to the device. This return also uses a communication protocol such as an HTTP response. The input is the edited post content received from the generative AI model, and the output is transmission to the device.

[1774] Step 7:

[1775] The device displays the edited post content to the user.

[1776] The terminal displays the edited post content received from the server to the user. The user can check the edited content and revise the post content again if necessary. The input is the edited post content received from the server, and the output is the display to the user.

[1777] (Application example 2)

[1778] Next, a description will be given of Application Example 2 of Form 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."

[1779] On online platforms such as social media and blogs, if a user posts content that could spark controversy or leak confidential information, the post could cause serious damage to a company or individual. To prevent such risks, a system is needed to check posts in advance and edit them as necessary. However, manual checking and editing takes time and effort, and there are challenges in that it is difficult to respond in real time.

[1780] 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.

[1781] In this invention, the server includes a generative AI means for determining the content of posts to SNS, a generative AI means for editing the posted content if the posted content is determined to be dangerous by the generative AI, a means for automatically generating a report on the posted content edited by the generative AI, a means for having employees review the report, a means for sharing the report with the organization, a means for analyzing the posted content in real time, a means for detecting expressions that pose a risk of causing an uproar, a means for detecting confidential information, and a means for automatically rewriting the posted content to safe expressions. This makes it possible to detect the risk of posted content in real time and automatically rewrite it to safe expressions.

[1782] "Generative AI means" refers to means that use artificial intelligence technology to analyze input data and generate output according to a specific purpose.

[1783] "Posted content" refers to information such as text, images, and videos that users make public on online platforms such as social media and blogs.

[1784] "If judged to be dangerous" means that the generative AI means analyzes the content of the post and determines that there is a risk of it causing a firestorm or leaking confidential information.

[1785] "Generative editing AI means" refers to a means that uses artificial intelligence technology to rewrite posted content that is deemed dangerous into safer expression.

[1786] "Means for automatically generating reports" refers to means for automatically creating a report summarizing the edits and reasons for edits made to posted content by generative AI means.

[1787] The "means for having employees check" refers to a means for presenting the automatically generated report to employees and having them check the contents.

[1788] The "means for sharing with the organization" is a means for sharing the generated report with the people involved in the organization.

[1789] "Means of real-time analysis" refers to means for instantly analyzing the content of posts and determining whether they are dangerous.

[1790] "Means for detecting expressions that pose a risk of causing an outrage" refers to means for identifying expressions contained in posted content that could cause an outrage.

[1791] "Means for detecting confidential information" refers to means for identifying confidential information of companies or individuals contained in posted content.

[1792] "Means for automatically rewriting content into safe expressions" refers to a means for automatically rewriting content that is deemed dangerous into safe expressions that do not pose a risk of online outrage or the leakage of confidential information.

[1793] As an embodiment of the present invention, a system is provided that analyzes content posted on online platforms such as social networking sites and blogs in real time and automatically edits any dangerous parts. Specific embodiments of this system are described below.

[1794] System Configuration

[1795] The system consists of the following main components:

[1796] 1. Generative AI means: Evaluates the content of posts and edits them if deemed dangerous.

[1797] 2. Automatic report generation method: Automatically generate reports on edited post content.

[1798] 3. Employee verification method: Have employees verify the generated reports.

[1799] 4. Organizational sharing means: Share reports within the organization.

[1800] 5. Real-time analysis: Analyze posted content instantly.

[1801] 6. Flaming risk detection method: Detect expressions that pose a risk of flaming.

[1802] 7. Confidential information detection methods: Detect confidential information.

[1803] 8. Automatic rewriting method: Automatically rewrite dangerous posts into safe expressions.

[1804] Program processing

[1805] The server does the following:

[1806] 1. Receiving posted content: Receives content that a user intends to post on social media or a blog.

[1807] 2. Analysis by a generative AI model: The received post content is sent to a generative AI model for analysis. The generative AI model used is, for example, OpenAI's "text-davinci-003."

[1808] 3. Risk assessment: The generative AI model analyzes the content of the post and determines whether there is a risk of a controversy or the leakage of confidential information.

[1809] 4. Automatic editing: If a post is deemed dangerous, the generative AI model automatically rewrites it to make it safer.

[1810] 5. Automatic report generation: Automatically generate a report for edited posts, summarizing the edits made and the reasons for them.

[1811] 6. Notify employees: Notify employees of the generated reports and have them review them.

[1812] 7. Share within your organization: Share the reviewed report within your organization.

[1813] Hardware and software used

[1814] Hardware: Servers, user devices (smartphones, PCs, etc.)

[1815] Software: Generative AI model (OpenAI API), real-time analysis software, report generation software

[1816] Specific examples

[1817] For example, if a user enters the following post:

[1818] Original post: "Our new product is better than our competitors' products."

[1819] In this case, the prompt sent to the generative AI model would be:

[1820] Prompt: "Please reword the following post to make it safer:\n\nOur new product is better than our competitors' products.\n\nPlease remove or rewrite any content that is inflammatory or contains confidential information."

[1821] The generative AI model analyzes the post based on this prompt and rewrites it to a safer expression. For example, the rewritten content might read, "Our new product has excellent performance."

[1822] In this way, users can check the content before posting and, if necessary, rewrite it to safer language, thereby preventing the risk of flame wars and the leakage of confidential information.

[1823] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1824] Step 1:

[1825] The user enters the content they want to post on social media or a blog.

[1826] Input: User-typed post content (e.g., "Our new product is better than our competitors' products.")

[1827] Output: The post is sent to the server.

[1828] Specific operation: A user uses a smartphone or PC to enter text into the posting screen of a social networking site or blog and presses the "Post" button.

[1829] Step 2:

[1830] The server receives the posted content and sends it to the generative AI model.

[1831] Input: User submitted post

[1832] Output: The prompt sent to the generative AI model

[1833] What happens: The server receives the post and generates a prompt, such as "Please rewrite the following post to make it safer:\n\nOur new product is superior to our competitors' products.\n\nPlease remove or rewrite any parts that may be controversial or contain confidential information."

[1834] Step 3:

[1835] A generative AI model analyzes the content of posts and determines whether they are dangerous.

[1836] Input: A prompt sent to the generative AI model

[1837] Output: Analysis results (presence or absence of danger, identification of dangerous parts)

[1838] How it works: A generative AI model (e.g., OpenAI's "text-davinci-003") analyzes the prompt text and determines whether the post content poses a risk of inciting controversy or leaking confidential information.

[1839] Step 4:

[1840] The server receives the analysis results of the generative AI model and edits the post content as necessary.

[1841] Input: Analysis results of generative AI model

[1842] Output: Edited post

[1843] Specific operation: The server receives the analysis results of the generative AI mode...

Claims

1. A first generation AI means for determining the content of posts to an SNS; A second generation AI means for editing the posted content when the posted content determined by the first generation AI means is determined to be dangerous; A means for automatically generating a report on the posted content edited by the second generating AI means; a means for having said report reviewed by an employee; a means for sharing said report with an organization; A means for checking keywords or phrases contained in the posted content; an emotion engine means for recognizing an emotion of a poster from the posted content; a means for notifying a result of the determination by the emotion engine means; A system including:

2. 2. The system of claim 1, wherein the report on the posting compiled by the second generating AI means includes content explaining why the posting was dangerous.

3. The system according to claim 1 , wherein the report includes information that collects information on dissatisfaction or merits of employees.

Images