system

Abstract

To provide a system.SOLUTION: A system includes: means for collecting data about harassment from each department in a company; means for preprocessing the collected data, inputting the preprocessed data into a generative AI, and analyzing a trend of harassment; means for using the generative AI to analyze a relationship between feeling information of an employee analyzed using a feeling engine and occurrence of harassment; means for proposing measures and preventive measures against harassment on the basis of an analysis result of the generative AI and an analysis result of the feeling engine.SELECTED DRAWING: Figure 1

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor and includes steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] There is a problem that inquiries and reports related to workplace harassment within the company are manually classified and filtered, resulting in an increase in risks due to delays in response. There is also a problem that there is no effective means to grasp the trends and patterns of harassment and propose countermeasures and preventive measures.

Means for Solving the Problems

[0005] By using generative AI, we automatically classify inquiries and reports related to harassment within the company and filter them according to their importance and urgency, enabling rapid response and providing a means of risk hedging. Furthermore, by utilizing generative AI to identify trends and patterns of harassment, and by visualizing data and generating reports, we provide a means of proposing harassment countermeasures and preventative measures. [Brief explanation of the drawing]

[0006] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Embodiment 1 of Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1 of Form Example 1. [Figure 13] It is a sequence diagram showing the processing flow of the data processing system in Example 2 of Embodiment Example 2. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 of Embodiment Example 2. [Figure 15] It is a sequence diagram showing the processing flow of the data processing system in Example 3 of Embodiment Example 3. [Figure 16] It is a sequence diagram showing the processing flow of the data processing system in Application Example 3 of Embodiment Example 3. [Figure 17] It is a sequence diagram showing the processing flow of the data processing system in Example 1 of Embodiment Example 1 when combined with an emotion engine. [Figure 18] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1 of Embodiment Example 1 when combined with an emotion engine. [Figure 19] It is a sequence diagram showing the processing flow of the data processing system in Example 2 of Embodiment Example 2 when combined with an emotion engine. [Figure 20] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 of Embodiment Example 2 when combined with an emotion engine. [Figure 21] It is a sequence diagram showing the processing flow of the data processing system in Example 3 of Embodiment Example 3 when combined with an emotion engine. [Figure 22] It is a sequence diagram showing the processing flow of the data processing system in Application Example 3 of Embodiment Example 3 when combined with an emotion engine.

Embodiments for Carrying Out the Invention

[0007] Hereinafter, an example of an embodiment of the system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

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

[0009] In the following embodiments, the numbered processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units 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)), etc.

[0010] In the following embodiments, the numbered RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0011] In the following embodiments, the numbered storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0012] In the following embodiments, the numbered communication I / F (Interface) is an interface including a communication processor and 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), or Bluetooth (registered trademark), etc.

[0013] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0014] [First Embodiment]

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

[0016] As shown in Figure 1, the 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 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0018] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0019] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and 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 perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0021] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

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

[0023] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0024] The storage 32 stores the data generation model 58 and the 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 processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

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

[0027] "Example of form 1"

[0028] One embodiment of the present invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company. Specifically, the generative AI analyzes text data collected through an anonymous reporting channel using a chatbot and automatically classifies the content related to harassment. Furthermore, the results can be filtered according to importance and urgency, enabling prompt action to be taken as needed.

[0029] "Example of form 2"

[0030] Another embodiment of the present invention involves a system that utilizes generative AI to identify harassment trends and patterns, and to visualize the data and generate reports. Specifically, the generative AI extracts harassment trends and patterns based on the results of its analysis and visualizes them. Furthermore, it automatically generates a report based on these results and proposes harassment countermeasures and preventative measures.

[0031] "Example of form 3"

[0032] A further embodiment of the present invention is a system that uses generative AI to propose measures and preventative actions against harassment. Specifically, the system proposes measures and preventative actions against harassment based on the results of analysis by the generative AI. For example, it identifies departments and times of day when harassment is frequent, the type of harassment, etc., and proposes specific measures and preventative actions for these.

[0033] The following describes the processing flow for each example of the form.

[0034] "Example of form 1"

[0035] Step 1: Collect text data from an anonymous reporting channel using a chatbot.

[0036] Step 2: Input the collected text data into a generative AI to automatically classify content related to harassment.

[0037] Step 3: Filter the classification results from the generative AI according to importance and urgency, and take necessary actions quickly.

[0038] "Example of form 2"

[0039] Step 1: Based on the results analyzed by the generative AI, trends and patterns of harassment are extracted.

[0040] Step 2: Visualize the extracted trends and patterns.

[0041] Step 3: Based on the visualized results, an automated report is generated, and harassment prevention and countermeasures are proposed.

[0042] "Example of form 3"

[0043] Step 1: Based on the results of the generative AI analysis, the system proposes measures and preventative actions against harassment.

[0044] Step 2: Identify departments, time slots, and types of harassment that are most frequent.

[0045] Step 3: Propose specific countermeasures and preventative measures for the identified information.

[0046] (Example 1)

[0047] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0048] There is a need to efficiently classify inquiries and reports related to harassment within the company and respond quickly according to their importance and urgency. However, traditional manual responses carry the risk of delays in processing reports and the oversight of important reports. Furthermore, it is difficult to identify trends and patterns of harassment and propose countermeasures and preventive measures. To solve these problems, automated classification and filtering of reports, rapid response, data visualization and report generation, and proposals for countermeasures and preventive measures are necessary.

[0049] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0050] This invention includes a server that includes means for collecting anonymous reports using a chatbot, means for transmitting the collected text data to the server, means for analyzing the text data using a generative AI model and automatically classifying content related to harassment, means for filtering the classification results according to importance and urgency, means for notifying the person in charge of the filtered results, means for enabling a quick response and contributing to risk hedging, means for identifying harassment trends and patterns using generative AI, means for visualizing data and generating reports, and means for proposing harassment countermeasures and preventive measures. This makes it possible to efficiently classify inquiries and reports related to harassment within the company and to respond quickly according to importance and urgency.

[0051] A "chatbot" is software designed to automate conversations with users, and it has the ability to collect information and respond through text and voice.

[0052] An "anonymous report" is a report made without revealing the reporter's identity, and is a means of providing information while protecting privacy.

[0053] "Text data" refers to information expressed as a string of characters, including report content and inquiry details.

[0054] A "server" is a computer system that processes and stores data on a network and provides services in response to requests from clients.

[0055] A "generative AI model" is an algorithm or system that uses artificial intelligence technology to analyze data and automatically perform specific tasks.

[0056] "Analysis" is the process of examining data in detail and understanding its structure and meaning.

[0057] Classification is the process of dividing data into specific categories or groups, and it involves grouping data that shares common characteristics.

[0058] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[0059] "Notification" refers to a means of informing relevant parties of specific information, and is typically done through email, dashboards, or other means.

[0060] "Prompt response" refers to taking action to deal with problems quickly when they arise, in order to minimize risks.

[0061] "Risk hedging" is the process of mitigating the impact of risks by predicting potential risks and taking countermeasures against them.

[0062] A "trend" refers to a general movement or pattern in which data or events tend to move in a particular direction.

[0063] A "pattern" refers to a repeating structure or arrangement in data or events.

[0064] "Visualization" is the process of making information easier to understand intuitively by representing data visually.

[0065] A "report" is a document that compiles information on a specific topic, including analysis results and recommendations.

[0066] "Countermeasures" refer to specific actions or plans taken to address a particular problem.

[0067] "Preventive measures" refer to the means or actions taken to prevent a problem from occurring before it happens.

[0068] Modes for carrying out the invention

[0069] This invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company and filters them according to their importance and urgency. Specific embodiments of this system are described below.

[0070] Hardware and software to be used

[0071] This system uses the following hardware and software:

[0072] Server: A computer system that processes and stores data. It also functions as an execution environment for generative AI models.

[0073] Device: The device used by the user (PC, smartphone, tablet, etc.).

[0074] Chatbot: Software that automates user interactions. It operates on communication tools such as Slack and Microsoft Teams.

[0075] Generative AI models: Algorithms that analyze text data and automatically perform specific tasks. For example, GPT-4 (registered trademark) from OpenAI (registered trademark).

[0076] Data processing and data calculation

[0077] This system performs the following data processing and calculations.

[0078] 1. Report Collection: Users can anonymously report harassment through the chatbot. For example, a user might report, "My boss yells at me every day."

[0079] 2. Data transmission: The device sends the text data collected through the chatbot to the server. The server temporarily stores the received text data.

[0080] 3. Text Analysis and Classification: The server inputs the stored text data into the generative AI model. The generative AI model analyzes the text data using the following prompts.

[0081] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[0082] The generative AI model analyzes the text data and classifies it as "power harassment."

[0083] 4. Filtering of Classification Results: The server evaluates the importance and urgency of reports based on the classification results obtained from the generative AI model. For example, a report classified as "power harassment" is judged to be of high importance. The server filters the reports considering factors such as their content, frequency, and the strength of the words used.

[0084] 5. Notification to the responsible party: The server notifies the responsible party based on the filtered results. Notifications are sent, for example, via email or a dedicated management dashboard. The responsible party receives the notification and can take the necessary actions quickly.

[0085] Specific example

[0086] As a concrete example, consider the following scenario.

[0087] scenario:

[0088] A user anonymously reports to a chatbot on Slack that "My boss yells at me every day." This report is sent to a server, where a generative AI model analyzes it.

[0089] Example of a prompt:

[0090] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[0091] The generative AI model receives this prompt and classifies the report as "power harassment." The server then determines the report's importance to be "high" based on this classification and notifies the responsible person via email.

[0092] The above describes specific embodiments for carrying out this invention.

[0093] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0094] Step 1:

[0095] Users can anonymously report harassment through a chatbot. For example, a user might report, "My boss yells at me every day." This report becomes the input data. The chatbot receives this text data and saves the report.

[0096] Step 2:

[0097] The device sends text data collected through the chatbot to the server. The input data is the text data reported by the user. The server temporarily stores the received text data, specifically in a database.

[0098] Step 3:

[0099] The server inputs the stored text data into the generative AI model. The input data is stored text data. The generative AI model parses the text data using the following prompts.

[0100] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[0101] The generative AI model analyzes text data and classifies it as "power harassment." The output data is the classification result.

[0102] Step 4:

[0103] The server evaluates the importance and urgency of reports based on classification results obtained from a generative AI model. The input data consists of classification results from the generative AI model. The server filters the reports, considering factors such as content, frequency, and the strength of the words used. The output data is the filtered result.

[0104] Step 5:

[0105] The server notifies the responsible person based on the filtered results. The input data is the filtered results. Notifications are sent, for example, via email or a dedicated management dashboard. The output data is the notification content. The responsible person receives the notification and can take necessary actions quickly.

[0106] The above is a detailed explanation of the program's processing flow.

[0107] (Application Example 1)

[0108] Next, we will describe 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 device 14 will be referred to as a "terminal."

[0109] Traditional harassment reporting systems often required manual classification and assessment of report importance, making prompt responses difficult. Furthermore, they lacked mechanisms to ensure anonymity while appropriately filtering important reports and immediately notifying responsible personnel. This created a risk of delays in the early detection and intervention of harassment.

[0110] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0111] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency, a means for analyzing text data collected through anonymous reporting channels, and a means for immediately notifying the responsible person when an important report is received. This enables a rapid response and contributes to risk hedging.

[0112] "Generative AI" is an artificial intelligence technology that generates new data and information based on input information such as text data and image data.

[0113] "Workplace harassment" refers to inappropriate behavior such as harassment, bullying, and discriminatory acts that occur within the workplace.

[0114] "Inquiries and reports" refer to the act of an employee communicating issues or questions regarding harassment to the company.

[0115] "Automatic classification" refers to the process by which a generative AI analyzes input data and automatically classifies it based on specific categories or importance levels.

[0116] "Filtering based on importance and urgency" refers to the process of selecting data classified by generative AI based on its importance and urgency.

[0117] An "anonymous reporting channel" refers to a means of communication that allows people to report harassment without revealing their identity.

[0118] "Text data" refers to data expressed as character information.

[0119] "Analyzing" refers to the process by which a generative AI examines the input data in detail, understands its content, and classifies it.

[0120] "Immediate notification to the responsible person" refers to the process of immediately notifying a designated person in the event of an important report.

[0121] "Prompt response" refers to taking appropriate measures without delay when a problem arises.

[0122] "Risk hedging" refers to minimizing risks by identifying potential risks in advance and taking countermeasures against them.

[0123] The system for implementing this invention uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency. Specific embodiments of this system are described below.

[0124] System Configuration

[0125] The system consists of the following main components:

[0126] 1. Server: Runs generative AI models and performs analysis and classification of text data.

[0127] 2. Device: A device (e.g., a smartphone) used by users to report harassment through an anonymous reporting channel.

[0128] 3. Notification System: The responsible person will be immediately notified when there is an important report.

[0129] Hardware and software to be used

[0130] Hardware: Smartphones, servers

[0131] Software: Python, OpenAI API

[0132] Data processing and data calculation

[0133] 1. Text data collection

[0134] Users access an anonymous reporting channel using their smartphones and submit reports of harassment. The reports are sent to the server as text data.

[0135] 2. Generative AI analysis

[0136] The server analyzes the collected text data using the OpenAI API. Specifically, it inputs the text data into the generative AI model using the following prompt messages.

[0137] Example of a prompt:

[0138] Analyze the following text and categorize the content as related to harassment.

[0139] Text: I'm being harassed by my boss. Please help me.

[0140] Classification results:

[0141] The generative AI model analyzes the text data based on this prompt and classifies the content related to harassment.

[0142] 3. Filtering and notifications

[0143] The server determines the importance and urgency of the report based on the analysis results obtained from the generative AI model. If there is an important report, the system immediately notifies the responsible person.

[0144] Specific example

[0145] For example, if a user reports, "My boss is harassing me. Please help," the server inputs this text data into a generative AI model and analyzes it to obtain a classification result such as, "This report is important. The harassment from your boss is ongoing, and urgent action is required." Based on this result, the server immediately notifies the appropriate person, enabling a swift response.

[0146] In this way, a harassment reporting system using generative AI enables a swift and appropriate response, leading to risk hedging.

[0147] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0148] Step 1:

[0149] Users access an anonymous reporting channel using their smartphones to report harassment. Input is the text data entered by the user. Output is the text data sent to the server.

[0150] Step 2:

[0151] The server inputs the received text data into a generative AI model. Specifically, it analyzes the text data using the following prompts. The input is the text data sent by the user. The output is the analysis result by the generative AI model.

[0152] Example of a prompt:

[0153] Analyze the following text and categorize the content as related to harassment.

[0154] Text: I'm being harassed by my boss. Please help me.

[0155] Classification results:

[0156] Step 3:

[0157] The server determines the importance and urgency of the reported content based on the analysis results obtained from the generative AI model. The input is the analysis results from the generative AI model. The output is the filtered results according to importance and urgency.

[0158] Step 4:

[0159] Based on the filtering results, the server immediately notifies the responsible person via the notification system if there is an important report. The input is the filtering results. The output is the notification to the responsible person.

[0160] Step 5:

[0161] The person in charge receives the notification and takes prompt action. The input is the notification from the server. The output is the action taken.

[0162] In this way, a harassment reporting system using generative AI enables a swift and appropriate response, leading to risk hedging.

[0163] (Example 2)

[0164] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0165] Traditional harassment prevention systems struggled to efficiently categorize inquiries and reports related to harassment within the company and to respond quickly according to their importance and urgency. Furthermore, identifying harassment trends and patterns and proposing countermeasures and preventative measures based on them was difficult. This resulted in an inability to properly manage harassment risks, leading to delays in improving the work environment within companies.

[0166] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using a generative AI and filtering them according to importance and urgency, means for collecting data from a database, means for preprocessing the collected data, means for analyzing the data using a generative AI model and identifying harassment trends and patterns, means for visualizing the analysis results, and means for automatically generating a report based on the visualized data. This enables efficient identification of harassment trends and patterns, and allows for rapid response and proposal of appropriate countermeasures and preventive measures.

[0167] "Generative AI" refers to a system that uses artificial intelligence technology to generate, analyze, and classify data.

[0168] A "database" is a system for efficiently storing, managing, and retrieving data.

[0169] "Preprocessing" refers to the initial steps taken to prepare data for easier analysis, and specifically includes data cleaning and tokenization.

[0170] A "generative AI model" is a type of generative AI that analyzes data based on specific prompts and generates results.

[0171] A "prompt statement" is an input statement used to give specific instructions or questions to a generative AI model.

[0172] "Visualization" refers to displaying the results of data analysis in a visual format, such as graphs or charts.

[0173] A "report" is a document that summarizes the results of data analysis and proposals.

[0174] "Harassment" refers to inappropriate actions or words directed at others in the workplace or other environments.

[0175] An "inquiry" is a question or request made to seek specific information or support.

[0176] A "report" is a written or oral communication intended to provide information about a specific event or situation.

[0177] "Classification" is the process of grouping data based on specific criteria.

[0178] "Filtering" is the process of selecting data based on specific criteria.

[0179] A "trend" refers to a pattern or trend that repeatedly appears in data.

[0180] A "pattern" is a collection of elements in data that share a specific regularity or commonality.

[0181] "Measures" refer to specific actions or plans taken to address a particular problem.

[0182] "Preventive measures" are specific actions or plans to prevent a problem from occurring before it happens.

[0183] This invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company and filters them according to their importance and urgency. Furthermore, it collects data from a database, performs preprocessing, analyzes the data using a generative AI model, and identifies trends and patterns of harassment. By visualizing the analysis results and automatically generating reports, it can propose harassment countermeasures and preventative measures.

[0184] Hardware and software to be used

[0185] Hardware:

[0186] Server: Database server (e.g., MySQL (registered trademark))

[0187] Terminal: A computer used for displaying data and reviewing reports.

[0188] software:

[0189] Database management system: MySQL

[0190] Natural language processing libraries: NLTK, spaCy

[0191] Generative AI model: GPT-4

[0192] Data visualization tools: Tableau, Matplotlib

[0193] Report generation tools: LaTeX, Microsoft Word

[0194] Explanation of the program's processing

[0195] Data collection:

[0196] The server collects data such as chat logs, emails, and survey results from within the company. Specifically, it connects to a database server and retrieves the necessary data using queries. For example, it executes a query like "SELECT FROM chat_logs WHERE date BETWEEN(registered trademark) '2023-01-01' AND '2023-01-31'" to retrieve chat logs for January.

[0197] Data preprocessing:

[0198] The server preprocesses the collected data. Specifically, it cleans the text data (removes unnecessary characters and spaces) and tokenizes it (splits it into words and phrases). The preprocessed data is temporarily stored on the terminal. For example, the cleaned text data is saved as a CSV file.

[0199] Analysis using generative AI models:

[0200] The server inputs pre-processed data into a generative AI model (GPT-4) to analyze harassment tendencies and patterns. The user creates the prompts to be input into the generative AI model. For example, the prompt "Identify harassment tendencies from the following chat log." might be used.

[0201] Data visualization:

[0202] The server visualizes harassment trends and patterns based on the results analyzed by the generating AI model. Specifically, it uses Matplotlib to create graphs, displaying time periods where specific keywords frequently appear as bar graphs. The visualized data is then displayed on the terminal.

[0203] Report generation:

[0204] The server automatically generates a "January Harassment Trends Report" using LaTeX based on the visualized data and saves it in PDF format. Users can review the generated report and make corrections or additional comments as needed.

[0205] Examples of specific cases and prompt statements

[0206] Specific example:

[0207] Identify the time periods and departments where specific keywords (e.g., "harassment," "ignoring") frequently appear in company chat logs.

[0208] Based on the survey results, we will extract the names of departments and supervisors that receive the most reports of harassment.

[0209] Example of a prompt:

[0210] "Please identify the harassment tendencies from the following chat logs. In particular, extract frequently occurring keywords and the time periods in which they occurred."

[0211] "Based on these survey results, please identify the departments with the highest number of harassment reports and analyze the causes."

[0212] The above describes the embodiments for carrying out this invention.

[0213] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0214] Step 1: Data Collection

[0215] The server collects data such as chat logs, emails, and survey results from within the company. Specifically, it connects to a database server and retrieves the necessary data using queries. For example, it executes a query like "SELECT FROM chat_logs WHERE date BETWEEN '2023-01-01' AND '2023-01-31'" to retrieve chat logs for January. The input is a database query, and the output is the collected raw data.

[0216] Step 2: Data Preprocessing

[0217] The server preprocesses the collected data. Specifically, it cleans the text data (removes unnecessary characters and spaces) and tokenizes it (splits it into words and phrases). The preprocessed data is temporarily stored on the terminal. For example, the cleaned text data is saved as a CSV file. The input is the collected raw data, and the output is the preprocessed, clean data.

[0218] Step 3: Analysis using a generative AI model

[0219] The server inputs pre-processed data into a generative AI model (GPT-4) to analyze harassment tendencies and patterns. The user creates prompt sentences to input into the generative AI model. For example, the prompt sentence might be, "Identify harassment tendencies from the following chat log." The input consists of pre-processed data and prompt sentences, and the output is the analysis result from the generative AI model.

[0220] Step 4: Visualize the concept

[0221] The server visualizes harassment trends and patterns based on the results analyzed by the generative AI model. Specifically, it uses Matplotlib to create graphs, displaying time periods where specific keywords frequently appear as bar graphs. The visualized data is displayed on the terminal. The input is the analysis results from the generative AI model, and the output is visualized graphs and charts.

[0222] Step 5: Generate the report

[0223] The server automatically generates a "January Harassment Trends Report" using LaTeX based on the visualized data and saves it in PDF format. The user reviews the generated report and makes corrections or additional comments as needed. The input is visualized data, and the output is an automatically generated report.

[0224] (Application Example 2)

[0225] Next, we will describe 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 as a "terminal".

[0226] Traditional harassment prevention systems struggle to efficiently categorize harassment-related inquiries and reports within a company and to respond quickly based on their importance and urgency. Furthermore, identifying harassment trends and patterns, visualizing them, and generating reports is time-consuming. Additionally, the lack of features to propose specific countermeasures and preventative measures hinders effective harassment prevention.

[0227] In Application Example 2, the specific processing performed by the specific processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for identifying trends and patterns of harassment using generative AI; means for visualizing data and generating reports; means for proposing countermeasures and preventive measures against harassment; means for collecting text data such as reports, chat logs, and emails from users through an application installed on a smartphone; means for identifying trends and patterns of harassment from the collected data using a generative AI model; means for visualizing the analysis results in graphs and charts; and means for automatically generating reports based on the analysis results and proposing specific countermeasures and preventive measures. This makes it possible to efficiently classify inquiries and reports related to harassment and respond quickly, identify and visualize trends and patterns of harassment, and propose specific countermeasures and preventive measures.

[0228] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[0229] "Harassment" refers to inappropriate words or actions towards others in the workplace or other environments.

[0230] An "inquiry" refers to a question or request made to obtain specific information or support.

[0231] "Reporting" refers to the act of providing information about a specific event or situation.

[0232] "Classification" refers to grouping data or information based on specific criteria.

[0233] "Filtering" refers to the process of selecting data or information based on specific criteria.

[0234] "Tendency" refers to a movement or flow that moves in a particular direction or pattern.

[0235] A "pattern" refers to a specific form or arrangement that is repeated.

[0236] "Visualization" refers to the visual representation of data and information.

[0237] A "report" refers to a document that compiles specific information or analysis results.

[0238] "Measures" refer to specific actions or means taken to address a particular problem or issue.

[0239] "Preventive measures" refer to specific actions or means taken to prevent problems or issues from occurring before they happen.

[0240] A "smartphone" refers to a type of mobile phone that is capable of connecting to the internet and using applications.

[0241] An "application" refers to a software program that provides specific functions or services.

[0242] A "chat log" refers to a record of a conversation conducted in chat format.

[0243] "Email" refers to messages sent and received electronically.

[0244] "Data collection" refers to the act of gathering data for a specific purpose.

[0245] A "graph" refers to a diagram used to visually represent data.

[0246] A "chart" refers to a table or diagram used to visually represent data.

[0247] The system for implementing this invention uses generative AI to automatically classify inquiries and reports related to harassment within the company, and has a function to filter them according to their importance and urgency. It also utilizes generative AI to identify trends and patterns of harassment, visualize the data, and generate reports. Furthermore, it has a function to propose measures and preventative actions against harassment.

[0248] Hardware and software to be used

[0249] Hardware: Smartphones, servers

[0250] Software: Python, OpenAI API, Matplotlib, Pandas

[0251] Data processing and data calculation

[0252] Data collection

[0253] Users input text data such as harassment reports, chat logs, and emails through an application installed on their smartphones. This data is sent to a server and stored in a database.

[0254] AI analysis

[0255] The server inputs the collected data into a generative AI model (e.g., GPT-3®) to identify harassment tendencies and patterns. The following prompt messages are used during this process.

[0256] Example of a prompt

[0257] Identify the trends and patterns of harassment from the following text data.

[0258] Text: My boss makes sarcastic remarks to me every day.

[0259] Text: My colleague ignored me

[0260] Visualization of the process

[0261] The server uses Pandas and Matplotlib to visualize the data in graphs and charts based on the analysis results obtained from generative AI models. This allows users to visually confirm harassment trends and patterns.

[0262] Report generation

[0263] The server automatically generates a report based on the analysis results. This report includes harassment trends and patterns, as well as specific countermeasures and preventative measures. The report is output as a text file and made available for users to download.

[0264] Specific example

[0265] For example, if a user reports that "my boss makes sarcastic remarks every day," the generative AI model identifies the pattern as "sarcasm." Based on this information, the server generates a graph showing that there are many reports of sarcasm, and the report suggests countermeasures such as "regular counseling."

[0266] In this way, this invention enables the efficient classification and rapid response of inquiries and reports related to harassment, the identification and visualization of harassment trends and patterns, and the proposal of specific countermeasures and preventive measures.

[0267] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0268] Step 1:

[0269] The user inputs text data such as reports on harassment, chat logs, and emails through an application installed on the smartphone. The input data is sent by the application to the server. The input data includes the specific content and date / time of the harassment, etc.

[0270] Step 2:

[0271] The server saves the received text data in the database. At this time, it checks the format and content of the data and converts it into an appropriate format. For example, it converts the text data into JSON format and saves it.

[0272] Step 3:

[0273] The server inputs the saved text data into a generative AI model. The generative AI model (e.g., GPT-3) analyzes the data based on the prompt text and identifies the trends and patterns of harassment. Examples of prompt text are as follows.

[0274] Examples of prompt text

[0275] Please identify the trends and patterns of harassment from the following text data.

[0276] Text: The boss makes sarcastic remarks every day

[0277] Text: Colleague ignores

[0278] Step 4:

[0279] The server receives the analysis results obtained from the generative AI model and saves them in the database. The analysis results include the identified trends and patterns of harassment.

[0280] Step 5:

[0281] The server visualizes the data based on the analysis results. Using Pandas and Matplotlib, the analysis results are presented in graphs and charts. For example, a bar graph indicating a large number of reports of bad taste is generated.

[0282] Step 6:

[0283] The server automatically generates a report based on the visualized data. The report includes harassment trends and patterns, as well as specific countermeasures and preventive measures. The generated report is output as a text file.

[0284] Step 7:

[0285] The user can download and view the report generated through the smartphone application. The report describes specific countermeasures such as "regular counseling".

[0286] In this way, the system can efficiently classify inquiries and reports related to harassment, respond quickly, identify and visualize harassment trends and patterns, and propose specific countermeasures and preventive measures.

[0287] (Example 3)

[0288] Next, Example 3 of Embodiment 3 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart device 14 is referred to as the "terminal".

[0289] In conventional harassment countermeasure systems, the classification of harassment reports and inquiries, the filtering of importance and urgency, and the proposal of specific countermeasures and preventive measures are often performed manually, and there is a problem that it is difficult to respond quickly. In addition, there are insufficient means for identifying harassment trends and patterns and visualizing data, making it difficult to take effective countermeasures. <00009​ The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[0291] This invention includes a server that uses generative AI to automatically classify and filter inquiries and reports related to harassment within the company according to their importance and urgency; a server that collects harassment data from each department within the company; a server that preprocesses the collected data, including imputing missing values ​​and correcting outliers; a server that inputs the preprocessed data into a generative AI model to identify departments, time periods, and types of harassment that are frequent; a server that proposes specific countermeasures and preventive measures based on the analysis results of the generative AI model; and a server that notifies the user of the proposed measures. This enables rapid classification and filtering of harassment reports and inquiries, and allows for the proposal of effective countermeasures and preventive measures.

[0292] "Generative AI" is an artificial intelligence technology that generates new information and suggestions based on data.

[0293] "Harassment" refers to inappropriate words or actions directed at others in the workplace or society, and is an act that causes mental or physical distress.

[0294] An "inquiry" refers to a question or consultation made to seek specific information or advice.

[0295] A "report" refers to an official notification or record made to convey information about a specific event or situation.

[0296] "Classification" refers to grouping collected data or information based on specific criteria.

[0297] "Filtering" refers to the process of selecting data and information based on specific criteria and extracting only what is necessary.

[0298] "Data collection" refers to the act of collecting information and data necessary for a specific purpose.

[0299] "Preprocessing" refers to the process of performing operations such as filling in missing data values and correcting outliers before conducting data analysis, and preparing the data in a state where it can be analyzed.

[0300] "Generative AI model" refers to a specific implementation form of generative AI, and refers to algorithms and programs for generating new information and proposals based on data.

[0301] "Analysis" refers to the act of analyzing the collected data to identify specific patterns and trends.

[0302] "Countermeasures" refer to specific actions and measures taken to address specific problems and issues.

[0303] "Preventive measures" refer to specific actions and measures taken to prevent specific problems and issues from occurring.

[0304] "Notification" refers to an official communication or notice made to convey specific information or proposals to relevant parties.

[0305] This invention is a system that utilizes generative AI to propose countermeasures and preventive measures against harassment. The following shows specific embodiments of this system.

[0306] The server first collects data related to harassment from each department within the company. This data includes the number of harassment reports, occurrence time, occurrence location, type of harassment, etc. For example, it collects report data from the sales department and questionnaire results.

[0307] Next, the server preprocesses the collected data. Specifically, it fills in missing data values, detects and corrects outliers. It also standardizes the data and converts it into a format suitable for the generative AI model. For example, it performs operations such as converting text data into numerical data.

[0308] The server inputs the pre-processed data into a generative AI model (e.g., OpenAI's GPT-4) for analysis. The generative AI model analyzes the data to identify departments, time slots, and types of harassment that are frequent. For example, it might identify that harassment is most common in the sales department between 3 PM and 5 PM.

[0309] Subsequently, the server proposes specific countermeasures and preventative measures based on the analysis results of the generated AI model. For example, it might suggest conducting harassment prevention training for the sales department or strengthening the monitoring system between 3 PM and 5 PM.

[0310] Finally, the server notifies the user of the proposed changes. Specifically, it sends the proposed changes to the relevant parties via email or the company's internal messaging system. For example, it might notify sales managers about the implementation of harassment prevention training.

[0311] As a concrete example, consider the following scenario.

[0312] scenario:

[0313] Data was collected from the sales department of a certain company showing a high number of harassment reports during a specific time slot (3 PM to 5 PM). Based on this data, a generative AI model analyzes the data and proposes countermeasures.

[0314] Example of a prompt:

[0315] "In the sales department, there are many reports of harassment between 3 PM and 5 PM. Please propose measures to address and prevent harassment during this time period."

[0316] Example output from a generative AI model:

[0317] "As a measure against harassment during the 3 PM to 5 PM timeframe, we propose the following:

[0318] 1. Strengthen the monitoring system during this time period.

[0319] 2. Conduct harassment prevention training for all members of the sales department.

[0320] 3. Improve the harassment reporting system to enable anonymous reporting.

[0321] 4. Make adjustments to reduce the workload during this time period.

[0322] In this way, the server utilizes a generated AI model to propose harassment countermeasures and preventative measures and provides a system that notifies the user. The flow of specific processing in Example 3 will be explained using Figure 15.

[0323] Step 1:

[0324] The server collects harassment-related data from various departments within the company. Inputs include departmental databases and survey results. Specifically, the server accesses each department's database and retrieves report data such as "October 1, 2023, 3 PM, Sales Department, Verbal Abuse." The output is the collected harassment data.

[0325] Step 2:

[0326] The server preprocesses the collected data. The input is the harassment data collected in step 1. Specifically, the server imputes missing data values, detects and corrects outliers, and standardizes the data, converting it into a format suitable for the generative AI model. For example, for data lacking "occurrence time," it imputes the average occurrence time. The output is the preprocessed data.

[0327] Step 3:

[0328] The server inputs the pre-processed data into a generative AI model and performs analysis. The input is the data pre-processed in step 2. Specifically, the server inputs the data into a generative AI model (for example, OpenAI's GPT-4) to identify departments, time slots, and types of harassment that are frequent. For example, it might identify a pattern such as "verbal abuse is frequent in the sales department between 3 PM and 5 PM." The output is the analysis results.

[0329] Step 4:

[0330] The server proposes specific countermeasures and preventative measures based on the analysis results of the generated AI model. The input is the analysis results obtained in step 3. Specifically, the server proposes "strengthening the monitoring system for the sales department between 3 PM and 5 PM" and "conducting training to prevent harassment." The output is the proposed content.

[0331] Step 5:

[0332] The server notifies the user of the proposed content. The input is the proposed content obtained in step 4. Specifically, the server sends the proposed content to the relevant parties via email or the internal messaging system. For example, it sends an email to the sales department manager saying, "Please consider strengthening the monitoring system during the time period from 3 PM to 5 PM." The output is that the notification is complete.

[0333] (Application Example 3)

[0334] Next, we will explain Application Example 3 of Form Example 3. In the following explanation, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0335] Traditional harassment prevention systems often involved manual reporting and analysis of harassment incidents, making rapid response difficult. Furthermore, data analysis to identify harassment trends and patterns was insufficient, hindering the proposal of specific countermeasures and preventative measures. Additionally, the lack of a function to detect high-risk harassment situations in real time and send appropriate alerts prevented effective risk mitigation.

[0336] In Application Example 3, the identification processing by the identification processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for analyzing the content of reports using a generative AI model to identify the frequency, type, location, and time of occurrence of harassment; means for proposing specific harassment countermeasures and preventive measures based on the analysis results; and means for detecting situations with a high risk of harassment in real time and sending alerts to users. This enables rapid response to harassment, identification of trends and patterns, proposal of specific countermeasures and preventive measures, and real-time risk detection and alert transmission.

[0337] "Generative AI" is a type of artificial intelligence technology that has the ability to generate new data and information based on input data.

[0338] An "inquiry" is a question or request made to obtain specific information or support.

[0339] A "report" is a written or oral communication made to provide information about a specific event or situation.

[0340] "Classification" is the process of grouping data or information based on specific criteria.

[0341] "Filtering" is the process of selecting data and information based on specific criteria and removing unnecessary elements.

[0342] A "generative AI model" is a specific implementation of generative AI, an artificial intelligence model trained to perform a particular task.

[0343] "Analysis" is the process of examining data and information in detail to identify specific patterns and trends.

[0344] "Frequency" refers to the number or proportion of times a particular event or situation occurs.

[0345] A "type" refers to a category classified based on specific characteristics or properties.

[0346] "Location of occurrence" refers to the physical location or environment in which a particular event or situation occurred.

[0347] A "time zone" refers to a specific period or range of time.

[0348] "Countermeasures" refer to specific actions or measures taken to address a particular problem or risk.

[0349] "Preventive measures" refer to specific actions or steps taken to prevent a particular problem or risk from occurring before it happens.

[0350] "Real-time" refers to processing or responding immediately without delay.

[0351] An "alert" is a notification or warning intended to draw attention to a specific situation or risk.

[0352] The following system configuration and processing procedure will be described as embodiments for carrying out this invention.

[0353] System Configuration

[0354] This system consists of the following main components:

[0355] 1. Server: Runs generative AI models to analyze harassment reports, propose countermeasures and preventative measures, and send real-time alerts.

[0356] 2. Device: A device used by users to input harassment reports and receive analysis results, countermeasures, and preventative measures. This includes smartphones and personal computers.

[0357] 3. Network: A communication infrastructure used to connect servers and terminals and to send and receive data.

[0358] Hardware and software to be used

[0359] Hardware: Servers, smartphones, personal computers

[0360] Software: OpenAI API, Python, database management system (e.g., MySQL)

[0361] Data processing and data calculation

[0362] 1. Entering a harassment report:

[0363] Users use their devices to submit harassment reports via text, voice, images, etc.

[0364] The entered data is sent to the server and stored in the database.

[0365] 2. Analysis using generative AI models:

[0366] The server uses OpenAI's generative AI models (e.g., GPT-4) to analyze the reported content.

[0367] The analysis results will identify the frequency, type, location, and time of occurrence of harassment.

[0368] 3. Proposals for countermeasures and preventative measures:

[0369] Based on the analysis results, the server uses a generative AI model to propose specific countermeasures and preventative measures.

[0370] The proposal will be sent to the device and the user will be notified.

[0371] 4. Real-time alerts:

[0372] The server detects situations with a high risk of harassment in real time and sends alerts to users.

[0373] An alert will appear on the device to draw the user's attention.

[0374] Specific example

[0375] Examples of harassment reports

[0376] Report details: "In the sales department, employees are being subjected to excessive reprimands from their superiors every Friday afternoon."

[0377] Example of analysis results

[0378] Analysis result: "Sales department, Friday afternoon, reprimand from superior."

[0379] Examples of proposed countermeasures and preventative measures

[0380] Countermeasures and preventative measures: "A warning message was sent to the sales department on Friday afternoon. Harassment prevention training was conducted for supervisors."

[0381] Example of a prompt

[0382] Analysis prompt: "Analyze the following harassment report to identify frequency, type, location, and time of occurrence. Report content: {report}"

[0383] Prompt for suggesting countermeasures: "Based on the analysis results below, please propose specific harassment prevention and countermeasures.\nAnalysis results: {analysis}"

[0384] The above describes the embodiments for carrying out this invention. This system enables rapid response to harassment, identification of trends and patterns, proposal of specific countermeasures and preventive measures, and real-time risk detection and alert transmission.

[0385] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[0386] Step 1:

[0387] Users use a terminal to enter harassment reports. Input can be in the form of text, audio, or images. The entered data is sent from the terminal to a server and stored in a database. The input data includes details of the harassment, the location, and the time of occurrence.

[0388] Step 2:

[0389] The server retrieves report data from the database and analyzes the report content using a generative AI model (e.g., GPT-4). Specifically, the report content is input to the generative AI model as prompts to identify the frequency, type, location, and time of occurrence of harassment. The input data is the report content, and the output data is the analysis results.

[0390] Step 3:

[0391] The server uses a generative AI model to propose specific countermeasures and preventative measures based on the analysis results. Prompt messages are used to input the analysis results into the generative AI model and obtain suggestions for countermeasures and preventative measures. The input data is the analysis results, and the output data is the suggested countermeasures and preventative measures.

[0392] Step 4:

[0393] The server sends the proposed countermeasures and preventative measures to the terminal and notifies the user. The user can then review the proposed countermeasures and preventative measures through the terminal and take the necessary actions. The input data consists of proposed countermeasures and preventative measures, while the output data is the notification to the user.

[0394] Step 5:

[0395] The server continuously monitors data to detect high-risk harassment situations in real time. If a high-risk situation is detected, the server generates an alert in real time and sends it to the terminal. The user receives the alert through the terminal and takes appropriate action. Input data is real-time monitoring data, and output data is the alert.

[0396] The above outlines the specific processing steps of this system. The specific actions performed at each step enable rapid response to harassment, identification of trends and patterns, proposal of specific countermeasures and preventive measures, and real-time risk detection and alert transmission.

[0397] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0398] "Example of form 1"

[0399] One embodiment of the present invention is a system that combines a generative AI with an emotion engine. This system collects text data from users, and the generative AI analyzes that data. As a result of the analysis, content related to harassment is automatically classified and filtered according to importance and urgency. Furthermore, the emotion engine recognizes the user's emotions and incorporates that information into the generative AI's analysis. Specifically, the emotion engine recognizes the emotional state of the user when they make a report (anger, fear, sadness, etc.), and the generative AI uses this information in its analysis. This makes it possible to more accurately determine the severity and urgency of the harassment.

[0400] "Example of form 2"

[0401] Another embodiment of the present invention is a system that combines a generative AI and an emotion engine to identify harassment tendencies and patterns. In this system, the results analyzed by the generative AI and the emotional information recognized by the emotion engine are combined to identify harassment tendencies and patterns. Specifically, if a particular emotional state is determined to be closely related to the occurrence of harassment, that information is used to identify harassment tendencies and patterns. This makes it possible to formulate harassment prevention measures more effectively.

[0402] "Example of form 3"

[0403] As a further embodiment of the present invention, there is a system that proposes harassment countermeasures and preventive measures by combining a generative AI and an emotion engine. In this system, the results analyzed by the generative AI and the emotional information recognized by the emotion engine are combined to propose harassment countermeasures and preventive measures. Specifically, if a particular emotional state is determined to be closely related to the occurrence of harassment, that information is used to propose harassment countermeasures and preventive measures. This makes it possible to formulate harassment prevention measures more effectively.

[0404] The following describes the processing flow for each example of the form.

[0405] "Example of form 1"

[0406] Step 1: Collect text data from users.

[0407] Step 2: Generative AI analyzes the text data and automatically categorizes content related to harassment.

[0408] Step 3: Filter the classified items according to their importance and urgency.

[0409] Step 4: The emotion engine recognizes the user's emotions.

[0410] Step 5: Incorporate the recognized emotional information into the analysis of the generative AI.

[0411] "Example of form 2"

[0412] Step 1: Collect text data from users.

[0413] Step 2: Generative AI analyzes the text data and automatically categorizes content related to harassment.

[0414] Step 3: The emotion engine recognizes the user's emotions.

[0415] Step 4: Combine the recognized emotional information with the analysis results from the generative AI to identify harassment tendencies and patterns.

[0416] "Example of form 3"

[0417] Step 1: Collect text data from users.

[0418] Step 2: Generative AI analyzes the text data and automatically categorizes content related to harassment.

[0419] Step 3: The emotion engine recognizes the user's emotions.

[0420] Step 4: Combine the recognized emotional information with the analysis results from the generative AI to propose harassment countermeasures and preventative measures.

[0421] (Example 1)

[0422] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0423] Traditional harassment reporting systems often required manual classification and assessment of urgency, making rapid response difficult. Furthermore, they failed to accurately determine the severity of harassment because they processed reports without considering the user's emotional state. Additionally, there was a lack of systems capable of ensuring anonymity while providing appropriate responses.

[0424] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0425] In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI, means for filtering them according to importance and urgency, means for enabling a rapid response and risk hedging, means for recognizing the user's emotional state using an emotion engine and incorporating that information into the generative AI's analysis, means for collecting text data through an anonymous reporting channel using a chatbot, means for transmitting the collected text data to the server in real time, and means for notifying the appropriate person in charge based on the results of the generative AI's analysis. This enables the rapid and accurate processing of harassment-related reports and allows for responses that take into account the user's emotional state.

[0426] "Generative AI" refers to artificial intelligence that uses natural language processing technology to analyze text data and generate information tailored to specific purposes.

[0427] An "emotion engine" is software that recognizes a user's emotional state (e.g., anger, fear, sadness, etc.) from text data and provides that information.

[0428] A "chatbot" is an automated response system that communicates with users in a conversational format and collects specific information.

[0429] An "anonymous reporting channel" is a means of communication that allows users to report information without revealing their identity.

[0430] "Text data" refers to the written information entered by the user, including data such as report content and inquiry details.

[0431] "Classification" is the process of dividing collected text data into categories based on specific criteria.

[0432] "Filtering" is the process of selecting classified data according to its importance and urgency.

[0433] "Prompt response" refers to taking appropriate action quickly based on the information provided in the report.

[0434] "Risk hedging" refers to identifying potential risks in advance and taking measures to minimize their impact.

[0435] "Real-time" refers to the process of data being processed immediately as soon as it is generated.

[0436] "The appropriate person in charge" refers to the individual or department responsible for the tasks that require action based on the report.

[0437] Modes for carrying out the invention

[0438] This invention uses a system that combines generative AI and an emotion engine to automatically classify inquiries and reports related to workplace harassment and filter them according to their importance and urgency. A specific embodiment of this system is shown below.

[0439] 1. Program generation

[0440] The server generates a program for a system that combines generative AI and an emotion engine. This program includes a series of processes to collect, analyze, classify, and filter user reports and to take appropriate action quickly.

[0441] 2. Program Processing Description

[0442] The server will run the program using the following hardware and software.

[0443] Hardware: Servers, terminals

[0444] Software: Generative AI (e.g., OpenAI's GPT-4), emotion engines (e.g., IBM Watson's Tone Analyzer), chatbots

[0445] Data collection:

[0446] Users enter text data about harassment through an anonymous reporting channel using a chatbot. For example, a user might type, "I'm experiencing ongoing harassment from my boss, and it's taking a toll on my mental health."

[0447] Data transmission:

[0448] The terminal sends text data from the user to the server in real time.

[0449] Data analysis:

[0450] The server inputs the collected text data into a generative AI, which then analyzes the content related to harassment. Specifically, the server inputs the text data into the generative AI as a prompt, instructing it to "analyze this report and determine the severity and urgency of the harassment."

[0451] Emotion recognition:

[0452] The server uses an emotion engine to recognize the user's emotional state (e.g., anger, fear, sadness, etc.) and provides that information to the generative AI. The emotion engine analyzes the text data and recognizes the user's emotional state as "sadness."

[0453] Classification and filtering:

[0454] The generative AI classifies text data according to the severity and urgency of the harassment, based on emotional information provided by the emotion engine. The server then filters the data based on importance and urgency. For example, the generative AI might classify the text data as "serious harassment" and determine its urgency to be "high."

[0455] Response notification:

[0456] Based on the filtered results, the server notifies the appropriate person in charge and promptly takes the necessary action. Specifically, the server sends high-priority reports to the person in charge via email or notification system, notifying them that "this report requires immediate attention."

[0457] 3. Specific Examples and Examples of Prompt Statements

[0458] Specific example:

[0459] If a user reports through an anonymous chatbot that "I am being harassed by my boss and it is causing me emotional distress," it will be handled as follows:

[0460] 1. The user enters the following into the chatbot: "I am experiencing ongoing harassment from my boss, and it is taking a toll on my mental health."

[0461] 2. The terminal sends this text data to the server.

[0462] 3. The server inputs text data into the generative AI and instructs it to "analyze this report and determine the severity and urgency of the harassment."

[0463] 4. The server inputs text data into the emotion engine, which recognizes the emotion of "sadness."

[0464] 5. The generative AI classifies the text data as "serious harassment" and determines its urgency to be "high." The server filters this classification result to extract reports with high urgency.

[0465] 6. The server sends urgent reports to the responsible person via email or notification system, notifying them that "this report requires immediate attention."

[0466] Example of a prompt:

[0467] "I am experiencing ongoing harassment from my superior, and it is causing me significant emotional distress. Please analyze this report and assess the severity and urgency of the harassment."

[0468] In this way, a system is realized in which servers, terminals, and users work together to process harassment-related reports quickly and accurately.

[0469] The flow of the specific processing in Example 1 will be explained using Figure 17.

[0470] Step 1:

[0471] Users enter text data about harassment through an anonymous reporting channel using a chatbot.

[0472] As a concrete example, the user would type "I'm being harassed by my boss and it's taking a toll on my mental health" into the chatbot's interface.

[0473] Input: Text data entered by the user

[0474] Output: Text data sent to the chatbot

[0475] Step 2:

[0476] The terminal sends text data from the user to the server in real time.

[0477] Specifically, the terminal immediately transfers the text data entered by the user to the server.

[0478] Input: Text data sent to the chatbot

[0479] Output: Text data sent to the server

[0480] Step 3:

[0481] The server inputs the collected text data into a generative AI, which then analyzes the content related to harassment.

[0482] In terms of specific actions, the server inputs text data as a prompt to the generative AI, instructing it to "analyze this report and determine the severity and urgency of the harassment."

[0483] Input: Text data sent to the server

[0484] Output: Analysis results by generative AI

[0485] Step 4:

[0486] The server uses an emotion engine to recognize the user's emotional state (e.g., anger, fear, sadness, etc.) and provides that information to the generative AI.

[0487] Specifically, the server inputs text data into the emotion engine, which then recognizes the emotion of "sadness."

[0488] Input: Text data sent to the server

[0489] Output: Emotion recognition results from the emotion engine

[0490] Step 5:

[0491] The generative AI classifies text data according to the severity and urgency of the harassment, based on emotional information provided by the emotion engine.

[0492] In specific terms, the generative AI classifies the text data as "serious harassment" and determines its urgency to be "high."

[0493] Input: Analysis results by generative AI and emotion recognition results by emotion engine

[0494] Output: Severity and urgency of classified harassment

[0495] Step 6:

[0496] The server uses this classification result to filter the data according to its importance and urgency.

[0497] Specifically, the server extracts reports with high urgency.

[0498] Input: Severity and urgency of classified harassment

[0499] Output: Filtered report data

[0500] Step 7:

[0501] Based on the filtered results, the server notifies the appropriate person in charge and takes the necessary action promptly.

[0502] Specifically, the server sends high-priority reports to the responsible person via email or a notification system, notifying them that "a prompt response is required to this report."

[0503] Input: Filtered report data

[0504] Output: Notification and instructions for action to the person in charge.

[0505] (Application Example 1)

[0506] Next, we will describe 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 device 14 will be referred to as a "terminal."

[0507] Traditional harassment reporting systems often require manual classification and assessment of importance, making prompt responses difficult. Furthermore, they fail to consider the reporter's emotional state, making it difficult to accurately determine the severity and urgency of the harassment. Additionally, anonymous reporting is often difficult, leading to hesitation among reporters. A system is needed to address these issues and manage harassment reports more efficiently and accurately.

[0508] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0509] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency; an emotion engine to recognize the user's emotional state and incorporate that information into the generative AI's analysis; a means to receive anonymous reports through an application installed on a smartphone; a means to propose appropriate countermeasures based on the report content; and a means to enable a rapid response and contribute to risk hedging. As a result, the classification of report content and the determination of importance are automated, enabling a rapid and accurate response. Furthermore, by considering the reporter's emotional state, the severity and urgency of the harassment can be judged more accurately. In addition, anonymous reporting becomes easier, allowing reporters to report without hesitation.

[0510] "Generative AI" is an artificial intelligence technology that generates new information or results based on data provided by users.

[0511] An "emotion engine" is a technology that analyzes and recognizes a user's emotional state from data such as text and audio.

[0512] A "smartphone" is a multi-functional mobile device that, in addition to the functions of a mobile phone, is capable of internet connectivity and the execution of applications.

[0513] "Anonymous reporting" is a method of reporting problems or incidents without disclosing the reporter's personal information.

[0514] "Importance" is an indicator that shows the magnitude and priority of the impact of a reported event or problem.

[0515] "Urgency" is an indicator that shows how quickly a response is needed to a reported event or problem.

[0516] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[0517] "Prompt response" refers to taking appropriate measures quickly in response to a problem or incident.

[0518] "Risk hedging" refers to minimizing risk by predicting potential risks and taking countermeasures against them.

[0519] "Appropriate countermeasures" refer to the most effective and appropriate solutions or measures for the reported incident or problem.

[0520] The system for carrying out this invention combines a generative AI and an emotion engine, and operates through an application installed on a smartphone. Specific embodiments of this system are described below.

[0521] System Configuration

[0522] hardware

[0523] Smartphone: A device used by users to enter reports.

[0524] Server: A computer system that runs generative AI and emotion engines and analyzes data.

[0525] software

[0526] Generative AI: Uses the OpenAI API to analyze and classify the reported content.

[0527] Emotion Engine: Uses TextBlob to analyze the user's emotional state.

[0528] Smartphone application: Provides an interface for users to anonymously enter and submit reports.

[0529] Data processing and data calculation

[0530] smartphone

[0531] Users anonymously submit harassment reports through a smartphone application. The entered text data is sent to a server.

[0532] server

[0533] The server processes the received text data in the following steps:

[0534] 1. Sentiment Analysis: TextBlob is used to analyze the emotional state of the reported text. Specifically, the text is scored for positive, negative, and neutral sentiment.

[0535] 2. Classification using generative AI: Using the OpenAI API, report texts are classified according to the severity and urgency of the harassment. Prompt sentences are generated and analyzed by the AI.

[0536] Specific example

[0537] Report content

[0538] "My boss yells at me every day, and I can't concentrate on my work. I'm feeling extremely stressed."

[0539] Emotion analysis results

[0540] Strong negative emotions

[0541] Classification results of generative AI

[0542] Severity: High, Urgency: High

[0543] Example of a prompt

[0544] Please categorize the following text according to the severity and urgency of the harassment:

[0545] My boss yells at me every day, and I can't concentrate on my work. I'm under a lot of stress.

[0546] Classification:

[0547] effect

[0548] This system automates the classification and prioritization of reports, enabling quick and accurate responses. Furthermore, by considering the reporter's emotional state, it allows for a more accurate assessment of the severity and urgency of harassment. Additionally, it facilitates anonymous reporting, encouraging reporters to report incidents without hesitation.

[0549] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[0550] Step 1:

[0551] Users anonymously submit reports of harassment through a smartphone application.

[0552] Input: Text data entered by the user.

[0553] Output: Text data sent from the smartphone to the server.

[0554] Specific action: The user fills in the report details in the application's input form and presses the submit button.

[0555] Step 2:

[0556] The server receives text data and passes it to the emotion engine for sentiment analysis.

[0557] Input: Text data sent from a smartphone.

[0558] Output: Sentiment analysis results (positive, negative, and neutral scores).

[0559] Specific operation: The server uses TextBlob to analyze the sentiment of text data and generate a sentiment score.

[0560] Step 3:

[0561] The server passes the emotion analysis results to a generative AI, which then classifies the reported content.

[0562] Input: Sentiment analysis results and text data.

[0563] Output: Classification results (severity and urgency) generated by a generative AI.

[0564] Specific operation: The server uses the OpenAI API to generate prompt text and classify the report content based on the text data and sentiment analysis results.

[0565] Step 4:

[0566] The server proposes appropriate countermeasures based on the classification results.

[0567] Input: Classification results from a generative AI.

[0568] Output: Proposed countermeasures.

[0569] Specific operation: The server analyzes the classification results and selects and proposes the most suitable countermeasure from a pre-configured database of countermeasures.

[0570] Step 5:

[0571] The server sends the proposed solution to the smartphone application and notifies the user.

[0572] Input: Proposed countermeasures.

[0573] Output: Solutions displayed in the smartphone application.

[0574] Specific operation: The server sends the countermeasure to the smartphone application and generates a message to notify the user.

[0575] Step 6:

[0576] The user reviews the proposed solutions and provides additional reports and feedback as needed.

[0577] Input: Proposed solutions and user feedback.

[0578] Output: Additional reports and feedback data.

[0579] Specific operation: The user reviews the suggested solutions through the application and provides additional reports and feedback as needed.

[0580] (Example 2)

[0581] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0582] Conventional harassment prevention systems often require manual analysis to identify harassment tendencies and patterns, which is time-consuming and labor-intensive. Furthermore, they fail to consider emotional information in their analysis, making it difficult to develop effective preventative measures. Additionally, the lack of automated data visualization and report generation hinders rapid response. To address these challenges, a system combining generative AI and an emotion engine is needed.

[0583] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0584] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to importance and urgency; means for preprocessing the collected data; means for analyzing the data using a generative AI model; means for recognizing emotional information using an emotion engine; means for integrating the analysis results of the generative AI and the emotional information of the emotion engine to visualize the data; means for automatically generating reports based on the visualized data; and means for enabling rapid response and contributing to risk hedging. This makes it possible to effectively identify harassment trends and patterns and to formulate preventive measures that take emotional information into consideration. Furthermore, data visualization and automatic report generation enable rapid response.

[0585] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[0586] "Harassment" refers to inappropriate words or actions towards others in the workplace or other environments.

[0587] An "inquiry" is a question or request made to seek specific information or support.

[0588] "Reporting" is a form of communication conducted to provide information about a specific event or situation.

[0589] "Classification" is the process of grouping data or information based on specific criteria.

[0590] "Filtering" is the process of selecting data or information based on specific criteria.

[0591] "Preprocessing" refers to the initial stages of processing data to convert it into a format suitable for analysis and processing.

[0592] "Generative AI models" refer to specific implementations and algorithms of generative AI.

[0593] An "emotion engine" is a technology that recognizes emotional information from data such as text and audio.

[0594] "Emotional information" refers to information that indicates the emotional state or tendency contained within the data.

[0595] "Integration" is the process of combining multiple data or pieces of information into one.

[0596] "Visualization" is the process of representing data and information in visual forms such as graphs and charts.

[0597] A "report" is a document that compiles specific information or analysis results.

[0598] "Prompt response" refers to taking quick and appropriate action in response to a problem or situation.

[0599] "Risk hedging" refers to taking measures or means to mitigate potential risks.

[0600] This invention is a system that utilizes generative AI and an emotion engine to identify harassment tendencies and patterns, and to visualize the data and generate reports. Specific embodiments of this system are described below.

[0601] Hardware and software to use

[0602] Hardware: Servers, terminals

[0603] Software: Generative AI models (e.g., GPT-4), sentiment engines (e.g., Affectiva), data preprocessing tools (e.g., Python, NLTK, Pandas), visualization tools (e.g., Tableau)

[0604] System Overview

[0605] This system begins with users uploading harassment-related data from their devices to a server. The server preprocesses the collected data and analyzes it using a generative AI model. Furthermore, it uses an emotion engine to recognize emotional information and integrates it with the generative AI analysis results. The integrated data is visualized, and finally, a report is automatically generated.

[0606] Specific examples of operation

[0607] 1. Data Collection

[0608] Users upload files such as internal chat logs, emails, and survey results from their devices to the server.

[0609] Specific actions: The user opens a browser and accesses the system's upload page. A file selection dialog opens, the user selects the file to upload, and clicks the upload button.

[0610] 2. Data preprocessing

[0611] The server analyzes the received data and performs text cleaning, such as removing HTML tags and special characters. Next, the text is tokenized and converted into a data format for sentiment analysis.

[0612] Software used: Python, NLTK, Pandas

[0613] 3. Analysis using generative AI

[0614] The server inputs prompt text into a generative AI model (e.g., GPT-4) to extract harassment tendencies and patterns.

[0615] Specific action: "This chat log contains many offensive words."

[0616] 4. Emotional analysis using an emotional engine

[0617] The server inputs text data into an emotion engine (e.g., Affectiva) to detect specific emotional states (e.g., anger, sadness).

[0618] Specific action: "This email clearly expresses strong feelings of anger."

[0619] 5. Data Integration and Visualization

[0620] The server integrates the analysis results from the generative AI with the emotional information from the emotion engine, inputs it into a visualization tool (e.g., Tableau), and creates graphs and charts.

[0621] Specific action: "Harassment occurs frequently in a particular department."

[0622] 6. Report generation

[0623] The server uses a report generation tool to create a report that includes analysis results and suggested countermeasures.

[0624] Specific action: "Implementing regular emotional check-ins can be expected to prevent harassment."

[0625] Example of a prompt

[0626] "Please analyze this chat log to identify any trends or patterns of harassment."

[0627] "Please acknowledge the emotional state expressed in this email and determine if it is related to the occurrence of harassment."

[0628] This system makes it possible to effectively identify harassment tendencies and patterns, and to develop preventative measures that take emotional information into account. Furthermore, data visualization and automated report generation enable rapid response.

[0629] The flow of the specific processing in Example 2 will be explained using Figure 19.

[0630] Step 1:

[0631] Data collection

[0632] Users upload files such as internal chat logs, emails, and survey results from their devices to the server.

[0633] Input: User-selected file (e.g., chat logs, emails, survey results)

[0634] Specific actions: The user opens a browser and accesses the system's upload page. A file selection dialog opens, the user selects the file to upload, and clicks the upload button.

[0635] Output: Files uploaded to the server

[0636] Step 2:

[0637] Data preprocessing

[0638] The server preprocesses the data it receives.

[0639] Input: Uploaded file

[0640] Specific operation: The server cleans the text data, removing unwanted characters and noise. Then, it tokenizes the text and converts it into a data format for sentiment analysis.

[0641] Output: Preprocessed text data

[0642] Software used: Python, NLTK, Pandas

[0643] Step 3:

[0644] Analysis using generative AI

[0645] The server analyzes the preprocessed data using a generative AI model (e.g., GPT-4).

[0646] Input: Preprocessed text data

[0647] Specific operation: The server inputs prompt text into a generative AI model, which then extracts harassment tendencies and patterns.

[0648] Output: Analysis results regarding harassment trends and patterns

[0649] Example prompt: "Analyze this chat log to identify harassment tendencies and patterns."

[0650] Step 4:

[0651] Emotional analysis using an emotional engine

[0652] The server uses an emotion engine (e.g., Affectiva) to recognize emotional information within the data.

[0653] Input: Preprocessed text data

[0654] Specific operation: The server inputs text data into the emotion engine and detects specific emotional states (e.g., anger, sadness).

[0655] Output: Analysis results regarding emotional information

[0656] Example prompt: "Please recognize the emotional state of this email and determine if it is related to the occurrence of harassment."

[0657] Step 5:

[0658] Data integration and visualization

[0659] The server integrates the analysis results from the generative AI with the emotional information from the emotion engine and visualizes the data.

[0660] Input: Analysis results from generative AI, emotional information from the emotion engine.

[0661] Specific operation: The server inputs the integrated data into a visualization tool (e.g., Tableau) to create graphs and charts.

[0662] Output: Visualized data (graphs and charts)

[0663] Step 6:

[0664] Report generation

[0665] The server automatically generates reports on harassment trends and patterns based on visualized data.

[0666] Input: Visualized data

[0667] Specific operation: The server uses a report generation tool to create a report that includes analysis results and suggested countermeasures.

[0668] Output: Automated report

[0669] Specific example of operation: Generate a report that includes a suggestion that "introducing regular emotional check-ins can be expected to prevent harassment."

[0670] (Application Example 2)

[0671] Next, we will describe 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 as a "terminal".

[0672] Conventional harassment prevention systems have difficulty identifying harassment tendencies and patterns, and have struggled with real-time monitoring and rapid response. As a result, they have often failed to effectively reduce the risk of harassment, leading to delays in victim protection and the development of preventative measures. Furthermore, the lack of sentiment analysis utilizing emotion engines has made it impossible to accurately grasp the circumstances of harassment occurrences. To address these challenges, a system combining generative AI and an emotion engine is necessary.

[0673] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0674] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency; a server that combines generative AI and an emotion engine to identify harassment trends and patterns; and a server that visualizes data and generates reports. This enables real-time monitoring of harassment risks and prompt response.

[0675] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[0676] An "emotion engine" is a technology that analyzes emotions from data such as text and audio, and identifies the emotional state.

[0677] "Harassment" refers to acts that cause mental or physical distress to others through inappropriate words or actions in environments such as the workplace or school.

[0678] A "tendency" is a characteristic or pattern in which a particular phenomenon or behavior moves in a certain direction.

[0679] A "pattern" is a regularity or form in which a particular phenomenon or behavior repeatedly occurs.

[0680] "Data visualization" is a technique that makes data easier to understand by representing it in visual forms such as graphs and charts.

[0681] "Report generation" is the process of automatically creating a report based on analysis results and data.

[0682] "Countermeasures" refer to specific means or methods taken to address a particular problem or risk.

[0683] "Preventive measures" are means or methods to prevent problems or risks from occurring before they happen.

[0684] A "smartphone" is a multi-functional mobile device that, in addition to the functions of a mobile phone, is capable of internet connectivity and the use of applications.

[0685] "Real-time" refers to a state where data and information are processed immediately and provided without delay.

[0686] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[0687] "Text data" refers to digital data that includes character information.

[0688] "Sentiment analysis" is a technology that analyzes emotional states from data such as text and audio, and identifies those emotions.

[0689] The system for implementing this invention combines a generative AI and an emotion engine to identify harassment tendencies and patterns, and to visualize the data and generate reports. Specific embodiments of this system are described below.

[0690] System Configuration

[0691] The system consists of the following main components:

[0692] 1. Generative AI models: Artificial intelligence technology that generates new information or patterns based on data.

[0693] 2. Emotion Engine: A technology that analyzes emotions from data such as text and audio, and identifies the emotional state.

[0694] 3. Smartphone application: An interface for users to monitor harassment risks in real time and input data.

[0695] 4. Data visualization tools: Use software such as Matplotlib or Pandas to visually represent the data.

[0696] 5. Report generation function: A function that automatically creates reports based on the analysis results.

[0697] Program processing

[0698] The server uses a generative AI model to identify harassment tendencies and patterns from the input text. Specifically, it uses the OpenAI API to analyze the text data entered by the user and assess the risk of harassment. The emotion engine performs sentiment analysis on the text data to determine whether a particular emotional state is strongly associated with the occurrence of harassment.

[0699] Data visualization and report generation

[0700] The server visualizes emotional data analyzed by the emotion engine using Matplotlib and Pandas. This allows users to visually understand harassment tendencies and patterns. Furthermore, it automatically generates reports based on the analysis results of the generative AI model and the emotion engine. These reports include harassment tendencies and patterns, emotional states, and suggestions for countermeasures and preventative measures.

[0701] Specific example

[0702] For example, if workplace conversation logs are used as input data, the user would enter text like this:

[0703] "During yesterday's meeting, my boss used very harsh language towards a subordinate. The subordinate was clearly uncomfortable."

[0704] Based on this text data, the generative AI model uses the following prompts to identify harassment tendencies and patterns:

[0705] Identify the tendencies and patterns of harassment from the following text:

[0706] "During yesterday's meeting, my boss used very harsh language towards a subordinate. The subordinate was clearly uncomfortable."

[0707] The emotion engine analyzes the emotional state from this text data, identifying, for example, strong emotions such as "anger" or "sadness." Based on this data, the server visualizes the data and generates a report.

[0708] In this way, users can monitor the risk of harassment in real time and respond quickly.

[0709] The flow of a specific process in Application Example 2 will be explained using Figure 20.

[0710] Step 1:

[0711] Users input text data, such as conversation logs from work or school, through a smartphone application.

[0712] Input: Text data such as conversation logs

[0713] Output: Input text data

[0714] Step 2:

[0715] The server sends the input text data to a generative AI model for analysis to identify harassment tendencies and patterns. Specifically, it uses the OpenAI API to analyze the text data.

[0716] Input: Entered text data

[0717] Output: Analysis results regarding harassment trends and patterns

[0718] Step 3:

[0719] The server uses an emotion engine based on the analysis results to perform sentiment analysis on the text data. The emotion engine identifies emotional states such as "anger" and "sadness" from the text data.

[0720] Input: Input text data, analysis results of a generative AI model

[0721] Output: Data on emotional state

[0722] Step 4:

[0723] The server visualizes the sentiment data analyzed by the sentiment engine using Matplotlib and Pandas. This allows users to visually understand harassment tendencies and patterns.

[0724] Input: Data related to emotional state

[0725] Output: Visualized data (graphs and charts)

[0726] Step 5:

[0727] The server automatically generates a report based on the analysis results of a generative AI model and an emotion engine. This report includes harassment tendencies and patterns, emotional states, and suggestions for countermeasures and preventative measures.

[0728] Input: Analysis results of generative AI models, data on emotional states

[0729] Output: Automated report

[0730] Step 6:

[0731] Users can review generated reports through a smartphone application, monitor harassment risks in real time, and take prompt action.

[0732] Input: Automated report

[0733] Output: User-confirmed report

[0734] In this way, users can monitor the risk of harassment in real time and respond quickly.

[0735] (Example 3)

[0736] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0737] Traditional harassment prevention systems have struggled to effectively propose specific measures and preventative actions to prevent harassment from occurring in the first place. Furthermore, they lacked measures that considered the emotional state of employees, making it difficult to identify the root causes of harassment. In addition, data collection and analysis regarding harassment were often done manually, resulting in insufficient risk mitigation in situations requiring rapid response.

[0738] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for identifying harassment trends and patterns using a generative AI model; means for collecting employee emotional information using an emotion engine and analyzing the relationship between emotional states and the occurrence of harassment; means for proposing harassment countermeasures and preventive measures based on the analysis results of the generative AI model and the recognition results of the emotion engine; means for visualizing data and generating reports; and means for enabling rapid response and leading to risk hedging. This makes it possible to prevent the occurrence of harassment and propose effective countermeasures and preventive measures that take into account the emotional states of employees.

[0739] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[0740] "Harassment" refers to acts of bullying or unfair treatment within the workplace or organization.

[0741] An "inquiry" is a question or request made to seek specific information or support.

[0742] "Reporting" is the act of providing information about a specific event or situation.

[0743] "Classification" is the process of grouping data or information based on specific criteria.

[0744] "Filtering" is the process of selecting data or information based on specific criteria.

[0745] "Generative AI models" refer to specific implementations and algorithms of generative AI.

[0746] A "trend" is the movement of data that shows a specific direction or pattern.

[0747] A "pattern" is a recurring structure or regularity in data or events.

[0748] An "emotional engine" is a technology for analyzing and recognizing the emotional state of employees.

[0749] "Emotional information" refers to data about the emotional state of employees.

[0750] "Measures" refer to specific actions or means taken to address a particular problem or issue.

[0751] "Preventive measures" are specific actions or means taken to prevent problems or issues from occurring before they happen.

[0752] "Visualization" is the process of representing data and information visually.

[0753] A "report" is a document that compiles specific information or analysis results.

[0754] "Prompt response" refers to taking action to address problems or issues quickly.

[0755] "Risk hedging" refers to measures and means taken to mitigate potential risks.

[0756] This invention is a system that proposes harassment countermeasures and preventative measures using generative AI and an emotion engine. This system operates primarily with a server, terminals, and users.

[0757] Hardware and software to be used

[0758] Hardware: Servers, terminals

[0759] Software: Generative AI models (e.g., GPT-4), emotion engines (e.g., Affectiva)

[0760] Specific operation of the system

[0761] Data collection

[0762] The server collects data on harassment from various departments within the company. Specifically, the server accesses each department's management system to obtain data such as the frequency, time of occurrence, and type of harassment. For example, the server automatically collects monthly reports and internal investigation results.

[0763] Data Analysis

[0764] The server inputs the collected data into a generative AI model to analyze harassment trends. Specifically, the server preprocesses the collected data and inputs it into the generative AI model (e.g., GPT-4) using prompt statements. For example, it might use a prompt statement such as, "If harassment is frequent in a particular department, please identify that department."

[0765] Collection of emotional information

[0766] The terminal collects employee emotional information using an emotion engine. Specifically, the terminal analyzes employees' facial expressions and voices in real time, and the emotion engine (e.g., Affectiva) recognizes their emotional state. For example, the terminal captures an employee's facial expression during a meeting using its camera, and the emotion engine analyzes that expression to determine their emotions.

[0767] Analysis of emotional information

[0768] The server inputs emotional information obtained from the emotion engine into a generative AI model to analyze the relationship between emotional states and the occurrence of harassment. Specifically, the server inputs data obtained from the emotion engine into the generative AI model and uses a prompt message such as, "If a particular emotional state is strongly associated with the occurrence of harassment, please identify that information." For example, the server might identify departments where emotional states such as "anger" or "anxiety" are frequently observed.

[0769] Proposals for countermeasures and preventive measures

[0770] The server proposes specific harassment countermeasures and preventative measures based on the analysis results of the generative AI model and the recognition results of the emotion engine. Specifically, the server inputs a prompt message to the generative AI model such as, "If harassment is occurring frequently in a particular department, what countermeasures should be proposed to that department?" and then formulates countermeasures based on the results obtained. For example, the server might propose regular harassment prevention training and counseling for stress management to a particular department.

[0771] Specific example

[0772] Example 1: At the beginning of each month, the server accesses the management system of each department and automatically downloads the harassment reports from the previous month.

[0773] Example 2: The server downloads a report and inputs it into a generative AI model, which then analyzes it using the prompt "In which department is harassment most frequent?".

[0774] Example 3: The device captures the facial expressions of employees during a meeting using its camera, and an emotion engine analyzes those expressions in real time to determine emotions such as "anger" or "anxiety."

[0775] Example 4: The server inputs emotional data obtained from the emotion engine into a generative AI model and analyzes it using the prompt message, "Identify information if a particular emotional state is strongly associated with the occurrence of harassment."

[0776] Example 5: Based on the analysis results of a generative AI model, the server proposes "harassment prevention training" and "stress management counseling" to a specific department.

[0777] The above describes embodiments for carrying out the present invention. The flow of the specific processing in Example 3 will be explained with reference to Figure 21.

[0778] Step 1: Data Collection

[0779] The server collects harassment-related data from various departments within the company. Specifically, the server accesses each department's management system to obtain data such as the frequency, time of occurrence, and type of harassment. The input is data from each department's management system, and the output is the collected harassment-related data. For example, the server automatically downloads monthly reports and internal investigation results.

[0780] Step 2: Data Preprocessing

[0781] The server preprocesses the collected data. Specifically, it performs tasks such as imputing missing values, detecting and correcting outliers, and normalizing the data. The input is the collected harassment-related data, and the output is the preprocessed data. For example, the server imputates missing values ​​with the mean and detects and corrects outliers.

[0782] Step 3: Data Analysis

[0783] The server inputs pre-processed data into a generative AI model to analyze harassment trends. Specifically, the server inputs prompts into the generative AI model (e.g., GPT-4). The input consists of pre-processed data and prompts, while the output is the analysis results regarding harassment trends and patterns. For example, a prompt such as "If harassment is frequent in a particular department, please identify that department" might be used.

[0784] Step 4: Gathering emotional information

[0785] The terminal collects employee emotional information using an emotion engine. Specifically, the terminal analyzes employees' facial expressions and voices in real time, and the emotion engine (e.g., Affectiva) recognizes their emotional state. The input is employee facial expression and voice data, and the output is the recognized emotional information. For example, the terminal captures an employee's facial expression during a meeting using a camera, and the emotion engine analyzes the emotion from that expression.

[0786] Step 5: Analyzing emotional information

[0787] The server inputs emotional information obtained from the emotion engine into a generative AI model to analyze the relationship between emotional states and the occurrence of harassment. Specifically, the server inputs data obtained from the emotion engine into the generative AI model and uses a prompt message such as, "If a particular emotional state is strongly associated with the occurrence of harassment, please identify that information." The input consists of emotional information and the prompt message, and the output is the analysis results regarding the relationship between emotional states and harassment. For example, the server might identify departments where emotional states such as "anger" or "anxiety" are frequently observed.

[0788] Step 6: Propose countermeasures and preventive measures

[0789] The server proposes specific harassment prevention and countermeasures based on the analysis results of the generative AI model and the recognition results of the emotion engine. Specifically, the server inputs a prompt to the generative AI model such as, "If harassment is occurring frequently in a particular department, what measures should be proposed to that department?" and then formulates countermeasures based on the results obtained. The input consists of the analysis results and the prompt, and the output is the proposed countermeasures and preventive measures. For example, the server might propose regular harassment prevention training and stress management counseling to a particular department.

[0790] (Application Example 3)

[0791] Next, we will explain Application Example 3 of Form Example 3. In the following explanation, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0792] Traditional harassment prevention systems have struggled to efficiently categorize and respond quickly to inquiries and reports related to harassment within the company. Furthermore, identifying trends and patterns of harassment and proposing specific countermeasures and preventative measures has been difficult. In particular, early detection and countermeasures for harassment using audio data are needed in employee communication within factories.

[0793] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[0794] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency; means to enable a rapid response and lead to risk hedging; means to convert conversations into text using speech recognition; means to recognize the emotional state of conversations using an emotion engine; means to analyze conversation content using generative AI and detect signs of harassment; and means to propose specific countermeasures and preventive measures against the detected signs of harassment. This enables early detection and countermeasures against harassment in communication among employees within the factory.

[0795] "Generative AI" refers to artificial intelligence technology that generates new information and content based on data.

[0796] "Speech recognition" is a technology that converts speech data into text data.

[0797] An "emotion engine" is a technology that recognizes emotional states from text and audio.

[0798] "Filtering" is the process of selecting data based on specific criteria.

[0799] "Risk hedging" refers to taking measures to mitigate potential risks.

[0800] "Signs of harassment" refer to patterns of behavior and speech that indicate the possibility of harassment occurring.

[0801] "Measures and preventative measures" refer to specific actions and means taken to prevent harassment from occurring.

[0802] "Analyzing conversation content" is the process of analyzing the text data of a conversation to understand its meaning and intent.

[0803] The system for carrying out this invention is configured as an application installed on a robot working in a factory. A specific embodiment is shown below.

[0804] System Configuration

[0805] 1. Speech recognition

[0806] The server uses speech recognition technology to convert conversations within the factory into text data in real time. Specifically, it uses the speech_recognition library.

[0807] 2. Generative AI Models

[0808] The server analyzes text data using a generative AI model to detect signs of harassment. For this purpose, it uses the GPT-3 model from the transformers library.

[0809] 3. Emotional Engine

[0810] The server uses an emotion engine to recognize emotional states from text data. Specifically, it uses the emotion_recognition library.

[0811] 4. Filtering and Risk Hedging

[0812] The server uses generative AI to automatically classify harassment-related inquiries and reports, filtering them according to their importance and urgency. This enables a rapid response and helps mitigate risks.

[0813] 5. Proposals for countermeasures and preventive measures

[0814] The server uses a generative AI model to propose specific countermeasures and preventative measures for detected signs of harassment.

[0815] Processing flow

[0816] 1. Acquisition of audio data

[0817] The terminal (robot) captures conversations within the factory via a microphone.

[0818] 2. Text conversion of audio data

[0819] The server converts the acquired audio data into text data using the speech_recognition library.

[0820] 3. Recognition of emotional state

[0821] The server uses the emotion_recognition library to recognize the emotional state from the text data.

[0822] 4. Detection of signs of harassment

[0823] The server uses a generative AI model (GPT-3) to analyze text data and detect signs of harassment.

[0824] 5. Proposals for countermeasures and preventive measures

[0825] The server uses a generative AI model to propose specific countermeasures and preventative measures for detected signs of harassment.

[0826] Specific example

[0827] For example, suppose the following conversation took place inside the factory:

[0828] Employee A: "You've been working slowly lately. You need to work harder."

[0829] Employee B: "I'm sorry, but I'm not feeling well..."

[0830] This conversation is captured as audio data and input into the program. The server inputs the following prompts to the generative AI model:

[0831] Analyze the following conversation for harassment: "Recently, your work has been slow. You need to work harder." "I'm sorry, I'm not feeling well."

[0832] Based on this prompt, the server analyzes signs of harassment and suggests appropriate countermeasures and preventative measures.

[0833] The flow of the specific processing in Application Example 3 will be explained using Figure 22.

[0834] Step 1:

[0835] The terminal (robot) captures conversations within the factory via a microphone. The input is audio data, and the output is the same audio data. This audio data is processed in the following steps.

[0836] Step 2:

[0837] The server converts the acquired audio data into text data using the speech_recognition library. The input is audio data, and the output is text data. This conversion allows the audio information to be treated as text information.

[0838] Step 3:

[0839] The server uses the emotion_recognition library to recognize emotional states from text data. The input is text data, and the output is an emotional state (e.g., anger, sadness, joy). This process allows for the capture of emotional nuances in a conversation.

[0840] Step 4:

[0841] The server uses a generative AI model (GPT-3) to analyze text data and detect signs of harassment. The input consists of text data and emotional states, and the output is the analysis results regarding signs of harassment. This analysis determines whether a conversation constitutes harassment.

[0842] Step 5:

[0843] The server uses a generative AI model to propose specific countermeasures and preventative measures for detected signs of harassment. The input is the analysis results regarding signs of harassment, and the output is specific countermeasures and preventative measures. This proposal provides appropriate countermeasures.

[0844] Specific example

[0845] For example, suppose the following conversation took place inside the factory:

[0846] Employee A: "You've been working slowly lately. You need to work harder."

[0847] Employee B: "I'm sorry, but I'm not feeling well..."

[0848] This conversation is captured as audio data and input into the program. The server inputs the following prompts to the generative AI model:

[0849] Analyze the following conversation for harassment: "Recently, your work has been slow. You need to work harder." "I'm sorry, I'm not feeling well."

[0850] Based on this prompt, the server analyzes signs of harassment and suggests appropriate countermeasures and preventative measures.

[0851] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0852] Data generation model 58 is a form of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0853] Other examples of generative AI include Gemini® (registered trademark) (Internet search). <url: https: gemini.google.com ?hl="ja">) are some examples.

[0854] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0855] [Second Embodiment]

[0856] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0857] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0858] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0859] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0860] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0861] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0862] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0863] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0864] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0865] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0866] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0867] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[0868] "Example of form 1"

[0869] One embodiment of the present invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company. Specifically, the generative AI analyzes text data collected through an anonymous reporting channel using a chatbot and automatically classifies the content related to harassment. Furthermore, the results can be filtered according to importance and urgency, enabling prompt action to be taken as needed.

[0870] "Example of form 2"

[0871] Another embodiment of the present invention involves a system that utilizes generative AI to identify harassment trends and patterns, and to visualize the data and generate reports. Specifically, the generative AI extracts harassment trends and patterns based on the results of its analysis and visualizes them. Furthermore, it automatically generates a report based on these results and proposes harassment countermeasures and preventative measures.

[0872] "Example of form 3"

[0873] A further embodiment of the present invention is a system that uses generative AI to propose measures and preventative actions against harassment. Specifically, the system proposes measures and preventative actions against harassment based on the results of analysis by the generative AI. For example, it identifies departments and times of day when harassment is frequent, the type of harassment, etc., and proposes specific measures and preventative actions for these.

[0874] The following describes the processing flow for each example of the form.

[0875] "Example of form 1"

[0876] Step 1: Collect text data from an anonymous reporting channel using a chatbot.

[0877] Step 2: Input the collected text data into a generative AI to automatically classify content related to harassment.

[0878] Step 3: Filter the classification results from the generative AI according to importance and urgency, and take necessary actions quickly.

[0879] "Example of form 2"

[0880] Step 1: Based on the results analyzed by the generative AI, trends and patterns of harassment are extracted.

[0881] Step 2: Visualize the extracted trends and patterns.

[0882] Step 3: Based on the visualized results, an automated report is generated, and harassment prevention and countermeasures are proposed.

[0883] "Example of form 3"

[0884] Step 1: Based on the results of the generative AI analysis, the system proposes measures and preventative actions against harassment.

[0885] Step 2: Identify departments, time slots, and types of harassment that are most frequent.

[0886] Step 3: Propose specific countermeasures and preventative measures for the identified information.

[0887] (Example 1)

[0888] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0889] There is a need to efficiently classify inquiries and reports related to harassment within the company and respond quickly according to their importance and urgency. However, traditional manual responses carry the risk of delays in processing reports and the oversight of important reports. Furthermore, it is difficult to identify trends and patterns of harassment and propose countermeasures and preventive measures. To solve these problems, automated classification and filtering of reports, rapid response, data visualization and report generation, and proposals for countermeasures and preventive measures are necessary.

[0890] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0891] This invention includes a server that includes means for collecting anonymous reports using a chatbot, means for transmitting the collected text data to the server, means for analyzing the text data using a generative AI model and automatically classifying content related to harassment, means for filtering the classification results according to importance and urgency, means for notifying the person in charge of the filtered results, means for enabling a quick response and contributing to risk hedging, means for identifying harassment trends and patterns using generative AI, means for visualizing data and generating reports, and means for proposing harassment countermeasures and preventive measures. This makes it possible to efficiently classify inquiries and reports related to harassment within the company and to respond quickly according to importance and urgency.

[0892] A "chatbot" is software designed to automate conversations with users, and it has the ability to collect information and respond through text and voice.

[0893] An "anonymous report" is a report made without revealing the reporter's identity, and is a means of providing information while protecting privacy.

[0894] "Text data" refers to information expressed as a string of characters, including report content and inquiry details.

[0895] A "server" is a computer system that processes and stores data on a network and provides services in response to requests from clients.

[0896] A "generative AI model" is an algorithm or system that uses artificial intelligence technology to analyze data and automatically perform specific tasks.

[0897] "Analysis" is the process of examining data in detail and understanding its structure and meaning.

[0898] Classification is the process of dividing data into specific categories or groups, and it involves grouping data that shares common characteristics.

[0899] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[0900] "Notification" refers to a means of informing relevant parties of specific information, and is typically done through email, dashboards, or other means.

[0901] "Prompt response" refers to taking action to deal with problems quickly when they arise, in order to minimize risks.

[0902] "Risk hedging" is the process of mitigating the impact of risks by predicting potential risks and taking countermeasures against them.

[0903] A "trend" refers to a general movement or pattern in which data or events tend to move in a particular direction.

[0904] A "pattern" refers to a repeating structure or arrangement in data or events.

[0905] "Visualization" is the process of making information easier to understand intuitively by representing data visually.

[0906] A "report" is a document that compiles information on a specific topic, including analysis results and recommendations.

[0907] "Countermeasures" refer to specific actions or plans taken to address a particular problem.

[0908] "Preventive measures" refer to the means or actions taken to prevent a problem from occurring before it happens.

[0909] Modes for carrying out the invention

[0910] This invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company and filters them according to their importance and urgency. Specific embodiments of this system are described below.

[0911] Hardware and software to be used

[0912] This system uses the following hardware and software:

[0913] Server: A computer system that processes and stores data. It also functions as an execution environment for generative AI models.

[0914] Device: The device used by the user (PC, smartphone, tablet, etc.).

[0915] Chatbot: Software that automates user interactions. It operates on communication tools such as Slack and Microsoft Teams.

[0916] Generative AI models: Algorithms that analyze text data and automatically perform specific tasks. For example, OpenAI's GPT-4.

[0917] Data processing and data calculation

[0918] This system performs the following data processing and calculations.

[0919] 1. Report Collection: Users can anonymously report harassment through the chatbot. For example, a user might report, "My boss yells at me every day."

[0920] 2. Data transmission: The device sends the text data collected through the chatbot to the server. The server temporarily stores the received text data.

[0921] 3. Text Analysis and Classification: The server inputs the stored text data into the generative AI model. The generative AI model analyzes the text data using the following prompts.

[0922] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[0923] The generative AI model analyzes the text data and classifies it as "power harassment."

[0924] 4. Filtering of Classification Results: The server evaluates the importance and urgency of reports based on the classification results obtained from the generative AI model. For example, a report classified as "power harassment" is judged to be of high importance. The server filters the reports considering factors such as their content, frequency, and the strength of the words used.

[0925] 5. Notification to the responsible party: The server notifies the responsible party based on the filtered results. Notifications are sent, for example, via email or a dedicated management dashboard. The responsible party receives the notification and can take the necessary actions quickly.

[0926] Specific example

[0927] As a concrete example, consider the following scenario.

[0928] scenario:

[0929] A user anonymously reports to a chatbot on Slack that "My boss yells at me every day." This report is sent to a server, where a generative AI model analyzes it.

[0930] Example of a prompt:

[0931] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[0932] The generative AI model receives this prompt and classifies the report as "power harassment." The server then determines the report's importance to be "high" based on this classification and notifies the responsible person via email.

[0933] The above describes specific embodiments for carrying out this invention.

[0934] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0935] Step 1:

[0936] Users can anonymously report harassment through a chatbot. For example, a user might report, "My boss yells at me every day." This report becomes the input data. The chatbot receives this text data and saves the report.

[0937] Step 2:

[0938] The device sends text data collected through the chatbot to the server. The input data is the text data reported by the user. The server temporarily stores the received text data, specifically in a database.

[0939] Step 3:

[0940] The server inputs the stored text data into the generative AI model. The input data is stored text data. The generative AI model parses the text data using the following prompts.

[0941] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[0942] The generative AI model analyzes text data and classifies it as "power harassment." The output data is the classification result.

[0943] Step 4:

[0944] The server evaluates the importance and urgency of reports based on classification results obtained from a generative AI model. The input data consists of classification results from the generative AI model. The server filters the reports, considering factors such as content, frequency, and the strength of the words used. The output data is the filtered result.

[0945] Step 5:

[0946] The server notifies the responsible person based on the filtered results. The input data is the filtered results. Notifications are sent, for example, via email or a dedicated management dashboard. The output data is the notification content. The responsible person receives the notification and can take necessary actions quickly.

[0947] The above is a detailed explanation of the program's processing flow.

[0948] (Application Example 1)

[0949] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0950] Traditional harassment reporting systems often required manual classification and assessment of report importance, making prompt responses difficult. Furthermore, they lacked mechanisms to ensure anonymity while appropriately filtering important reports and immediately notifying responsible personnel. This created a risk of delays in the early detection and intervention of harassment.

[0951] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0952] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency, a means for analyzing text data collected through anonymous reporting channels, and a means for immediately notifying the responsible person when an important report is received. This enables a rapid response and contributes to risk hedging.

[0953] "Generative AI" is an artificial intelligence technology that generates new data and information based on input information such as text data and image data.

[0954] "Workplace harassment" refers to inappropriate behavior such as harassment, bullying, and discriminatory acts that occur within the workplace.

[0955] "Inquiries and reports" refer to the act of an employee communicating issues or questions regarding harassment to the company.

[0956] "Automatic classification" refers to the process by which a generative AI analyzes input data and automatically classifies it based on specific categories or importance levels.

[0957] "Filtering based on importance and urgency" refers to the process of selecting data classified by generative AI based on its importance and urgency.

[0958] An "anonymous reporting channel" refers to a means of communication that allows people to report harassment without revealing their identity.

[0959] "Text data" refers to data expressed as character information.

[0960] "Analyzing" refers to the process by which a generative AI examines the input data in detail, understands its content, and classifies it.

[0961] "Immediate notification to the responsible person" refers to the process of immediately notifying a designated person in the event of an important report.

[0962] "Prompt response" refers to taking appropriate measures without delay when a problem arises.

[0963] "Risk hedging" refers to minimizing risks by identifying potential risks in advance and taking countermeasures against them.

[0964] The system for implementing this invention uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency. Specific embodiments of this system are described below.

[0965] System Configuration

[0966] The system consists of the following main components:

[0967] 1. Server: Runs generative AI models and performs analysis and classification of text data.

[0968] 2. Device: A device (e.g., a smartphone) used by users to report harassment through an anonymous reporting channel.

[0969] 3. Notification System: The responsible person will be immediately notified when there is an important report.

[0970] Hardware and software to be used

[0971] Hardware: Smartphones, servers

[0972] Software: Python, OpenAI API

[0973] Data processing and data calculation

[0974] 1. Text data collection

[0975] Users access an anonymous reporting channel using their smartphones and submit reports of harassment. The reports are sent to the server as text data.

[0976] 2. Generative AI analysis

[0977] The server analyzes the collected text data using the OpenAI API. Specifically, it inputs the text data into the generative AI model using the following prompt messages.

[0978] Example of a prompt:

[0979] Analyze the following text and categorize the content as related to harassment.

[0980] Text: I'm being harassed by my boss. Please help me.

[0981] Classification results:

[0982] The generative AI model analyzes the text data based on this prompt and classifies the content related to harassment.

[0983] 3. Filtering and notifications

[0984] The server determines the importance and urgency of the report based on the analysis results obtained from the generative AI model. If there is an important report, the system immediately notifies the responsible person.

[0985] Specific example

[0986] For example, if a user reports, "My boss is harassing me. Please help," the server inputs this text data into a generative AI model and analyzes it to obtain a classification result such as, "This report is important. The harassment from your boss is ongoing, and urgent action is required." Based on this result, the server immediately notifies the appropriate person, enabling a swift response.

[0987] In this way, a harassment reporting system using generative AI enables a swift and appropriate response, leading to risk hedging.

[0988] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0989] Step 1:

[0990] Users access an anonymous reporting channel using their smartphones to report harassment. Input is the text data entered by the user. Output is the text data sent to the server.

[0991] Step 2:

[0992] The server inputs the received text data into a generative AI model. Specifically, it analyzes the text data using the following prompts. The input is the text data sent by the user. The output is the analysis result by the generative AI model.

[0993] Example of a prompt:

[0994] Analyze the following text and categorize the content as related to harassment.

[0995] Text: I'm being harassed by my boss. Please help me.

[0996] Classification results:

[0997] Step 3:

[0998] The server determines the importance and urgency of the reported content based on the analysis results obtained from the generative AI model. The input is the analysis results from the generative AI model. The output is the filtered results according to importance and urgency.

[0999] Step 4:

[1000] Based on the filtering results, the server immediately notifies the responsible person via the notification system if there is an important report. The input is the filtering results. The output is the notification to the responsible person.

[1001] Step 5:

[1002] The person in charge receives the notification and takes prompt action. The input is the notification from the server. The output is the action taken.

[1003] In this way, a harassment reporting system using generative AI enables a swift and appropriate response, leading to risk hedging.

[1004] (Example 2)

[1005] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1006] Traditional harassment prevention systems struggled to efficiently categorize inquiries and reports related to harassment within the company and to respond quickly according to their importance and urgency. Furthermore, identifying harassment trends and patterns and proposing countermeasures and preventative measures based on them was difficult. This resulted in an inability to properly manage harassment risks, leading to delays in improving the work environment within companies.

[1007] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using a generative AI and filtering them according to importance and urgency, means for collecting data from a database, means for preprocessing the collected data, means for analyzing the data using a generative AI model and identifying harassment trends and patterns, means for visualizing the analysis results, and means for automatically generating a report based on the visualized data. This enables efficient identification of harassment trends and patterns, and allows for rapid response and proposal of appropriate countermeasures and preventive measures.

[1008] "Generative AI" refers to a system that uses artificial intelligence technology to generate, analyze, and classify data.

[1009] A "database" is a system for efficiently storing, managing, and retrieving data.

[1010] "Preprocessing" refers to the initial steps taken to prepare data for easier analysis, and specifically includes data cleaning and tokenization.

[1011] A "generative AI model" is a type of generative AI that analyzes data based on specific prompts and generates results.

[1012] A "prompt statement" is an input statement used to give specific instructions or questions to a generative AI model.

[1013] "Visualization" refers to displaying the results of data analysis in a visual format, such as graphs or charts.

[1014] A "report" is a document that summarizes the results of data analysis and proposals.

[1015] "Harassment" refers to inappropriate actions or words directed at others in the workplace or other environments.

[1016] An "inquiry" is a question or request made to seek specific information or support.

[1017] A "report" is a written or oral communication intended to provide information about a specific event or situation.

[1018] "Classification" is the process of grouping data based on specific criteria.

[1019] "Filtering" is the process of selecting data based on specific criteria.

[1020] A "trend" refers to a pattern or trend that repeatedly appears in data.

[1021] A "pattern" is a collection of elements in data that share a specific regularity or commonality.

[1022] "Measures" refer to specific actions or plans taken to address a particular problem.

[1023] "Preventive measures" are specific actions or plans to prevent a problem from occurring before it happens.

[1024] This invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company and filters them according to their importance and urgency. Furthermore, it collects data from a database, performs preprocessing, analyzes the data using a generative AI model, and identifies trends and patterns of harassment. By visualizing the analysis results and automatically generating reports, it can propose harassment countermeasures and preventative measures.

[1025] Hardware and software to be used

[1026] Hardware:

[1027] Server: Database server (e.g., MySQL)

[1028] Terminal: A computer used for displaying data and reviewing reports.

[1029] software:

[1030] Database management system: MySQL

[1031] Natural language processing libraries: NLTK, spaCy

[1032] Generative AI model: GPT-4

[1033] Data visualization tools: Tableau, Matplotlib

[1034] Report generation tools: LaTeX, Microsoft Word

[1035] Explanation of the program's processing

[1036] Data collection:

[1037] The server collects data such as chat logs, emails, and survey results within the company. Specifically, it connects to a database server and retrieves the necessary data using queries. For example, it executes a query like "SELECT FROM chat_logs WHERE date BETWEEN '2023-01-01' AND '2023-01-31'" to retrieve chat logs for January.

[1038] Data preprocessing:

[1039] The server preprocesses the collected data. Specifically, it cleans the text data (removes unnecessary characters and spaces) and tokenizes it (splits it into words and phrases). The preprocessed data is temporarily stored on the terminal. For example, the cleaned text data is saved as a CSV file.

[1040] Analysis using generative AI models:

[1041] The server inputs pre-processed data into a generative AI model (GPT-4) to analyze harassment tendencies and patterns. The user creates the prompts to be input into the generative AI model. For example, the prompt "Identify harassment tendencies from the following chat log." might be used.

[1042] Data visualization:

[1043] The server visualizes harassment trends and patterns based on the results analyzed by the generating AI model. Specifically, it uses Matplotlib to create graphs, displaying time periods where specific keywords frequently appear as bar graphs. The visualized data is then displayed on the terminal.

[1044] Report generation:

[1045] The server automatically generates a "January Harassment Trends Report" using LaTeX based on the visualized data and saves it in PDF format. Users can review the generated report and make corrections or additional comments as needed.

[1046] Examples of specific cases and prompt statements

[1047] Specific example:

[1048] Identify the time periods and departments where specific keywords (e.g., "harassment," "ignoring") frequently appear in company chat logs.

[1049] Based on the survey results, we will extract the names of departments and supervisors that receive the most reports of harassment.

[1050] Example of a prompt:

[1051] "Please identify the harassment tendencies from the following chat logs. In particular, extract frequently occurring keywords and the time periods in which they occurred."

[1052] "Based on these survey results, please identify the departments with the highest number of harassment reports and analyze the causes."

[1053] The above describes the embodiments for carrying out this invention.

[1054] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1055] Step 1: Data Collection

[1056] The server collects data such as chat logs, emails, and survey results from within the company. Specifically, it connects to a database server and retrieves the necessary data using queries. For example, it executes a query like "SELECT FROM chat_logs WHERE date BETWEEN '2023-01-01' AND '2023-01-31'" to retrieve chat logs for January. The input is a database query, and the output is the collected raw data.

[1057] Step 2: Data Preprocessing

[1058] The server preprocesses the collected data. Specifically, it cleans the text data (removes unnecessary characters and spaces) and tokenizes it (splits it into words and phrases). The preprocessed data is temporarily stored on the terminal. For example, the cleaned text data is saved as a CSV file. The input is the collected raw data, and the output is the preprocessed, clean data.

[1059] Step 3: Analysis using a generative AI model

[1060] The server inputs pre-processed data into a generative AI model (GPT-4) to analyze harassment tendencies and patterns. The user creates prompt sentences to input into the generative AI model. For example, the prompt sentence might be, "Identify harassment tendencies from the following chat log." The input consists of pre-processed data and prompt sentences, and the output is the analysis result from the generative AI model.

[1061] Step 4: Visualize the concept

[1062] The server visualizes harassment trends and patterns based on the results analyzed by the generative AI model. Specifically, it uses Matplotlib to create graphs, displaying time periods where specific keywords frequently appear as bar graphs. The visualized data is displayed on the terminal. The input is the analysis results from the generative AI model, and the output is visualized graphs and charts.

[1063] Step 5: Generate the report

[1064] The server automatically generates a "January Harassment Trends Report" using LaTeX based on the visualized data and saves it in PDF format. The user reviews the generated report and makes corrections or additional comments as needed. The input is visualized data, and the output is an automatically generated report.

[1065] (Application Example 2)

[1066] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1067] Traditional harassment prevention systems struggle to efficiently categorize harassment-related inquiries and reports within a company and to respond quickly based on their importance and urgency. Furthermore, identifying harassment trends and patterns, visualizing them, and generating reports is time-consuming. Additionally, the lack of features to propose specific countermeasures and preventative measures hinders effective harassment prevention.

[1068] In Application Example 2, the specific processing performed by the specific processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for identifying trends and patterns of harassment using generative AI; means for visualizing data and generating reports; means for proposing countermeasures and preventive measures against harassment; means for collecting text data such as reports, chat logs, and emails from users through an application installed on a smartphone; means for identifying trends and patterns of harassment from the collected data using a generative AI model; means for visualizing the analysis results in graphs and charts; and means for automatically generating reports based on the analysis results and proposing specific countermeasures and preventive measures. This makes it possible to efficiently classify inquiries and reports related to harassment and respond quickly, identify and visualize trends and patterns of harassment, and propose specific countermeasures and preventive measures.

[1069] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[1070] "Harassment" refers to inappropriate words or actions towards others in the workplace or other environments.

[1071] An "inquiry" refers to a question or request made to obtain specific information or support.

[1072] "Reporting" refers to the act of providing information about a specific event or situation.

[1073] "Classification" refers to grouping data or information based on specific criteria.

[1074] "Filtering" refers to the process of selecting data or information based on specific criteria.

[1075] "Tendency" refers to a movement or flow that moves in a particular direction or pattern.

[1076] A "pattern" refers to a specific form or arrangement that is repeated.

[1077] "Visualization" refers to the visual representation of data and information.

[1078] A "report" refers to a document that compiles specific information or analysis results.

[1079] "Measures" refer to specific actions or means taken to address a particular problem or issue.

[1080] "Preventive measures" refer to specific actions or means taken to prevent problems or issues from occurring before they happen.

[1081] A "smartphone" refers to a type of mobile phone that is capable of connecting to the internet and using applications.

[1082] An "application" refers to a software program that provides specific functions or services.

[1083] A "chat log" refers to a record of a conversation conducted in chat format.

[1084] "Email" refers to messages sent and received electronically.

[1085] "Data collection" refers to the act of gathering data for a specific purpose.

[1086] A "graph" refers to a diagram used to visually represent data.

[1087] A "chart" refers to a table or diagram used to visually represent data.

[1088] The system for implementing this invention uses generative AI to automatically classify inquiries and reports related to harassment within the company, and has a function to filter them according to their importance and urgency. It also utilizes generative AI to identify trends and patterns of harassment, visualize the data, and generate reports. Furthermore, it has a function to propose measures and preventative actions against harassment.

[1089] Hardware and software to be used

[1090] Hardware: Smartphones, servers

[1091] Software: Python, OpenAI API, Matplotlib, Pandas

[1092] Data processing and data calculation

[1093] Data collection

[1094] Users input text data such as harassment reports, chat logs, and emails through an application installed on their smartphones. This data is sent to a server and stored in a database.

[1095] AI analysis

[1096] The server inputs the collected data into a generative AI model (e.g., GPT-3) to identify harassment tendencies and patterns. The following prompt messages are used during this process.

[1097] Example of a prompt

[1098] Identify the trends and patterns of harassment from the following text data.

[1099] Text: My boss makes sarcastic remarks to me every day.

[1100] Text: My colleague ignored me

[1101] Visualization of the process

[1102] The server uses Pandas and Matplotlib to visualize the data in graphs and charts based on the analysis results obtained from generative AI models. This allows users to visually confirm harassment trends and patterns.

[1103] Report generation

[1104] The server automatically generates a report based on the analysis results. This report includes harassment trends and patterns, as well as specific countermeasures and preventative measures. The report is output as a text file and made available for users to download.

[1105] Specific example

[1106] For example, if a user reports that "my boss makes sarcastic remarks every day," the generative AI model identifies the pattern as "sarcasm." Based on this information, the server generates a graph showing that there are many reports of sarcasm, and the report suggests countermeasures such as "regular counseling."

[1107] In this way, this invention enables the efficient classification and rapid response of inquiries and reports related to harassment, the identification and visualization of harassment trends and patterns, and the proposal of specific countermeasures and preventive measures.

[1108] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1109] Step 1:

[1110] Users input text data such as harassment reports, chat logs, and emails through an application installed on their smartphones. The input data is then sent to a server by the application. The input data includes specific details of the harassment, as well as the date and time.

[1111] Step 2:

[1112] The server saves the received text data to the database. During this process, it checks the data's format and content and converts it to an appropriate format. For example, it might convert text data to JSON format before saving.

[1113] Step 3:

[1114] The server inputs the stored text data into a generative AI model. The generative AI model (e.g., GPT-3) analyzes the data based on the prompt text and identifies harassment tendencies and patterns. Examples of prompt texts are as follows:

[1115] Example of a prompt

[1116] Identify the trends and patterns of harassment from the following text data.

[1117] Text: My boss makes sarcastic remarks to me every day.

[1118] Text: My colleague ignored me

[1119] Step 4:

[1120] The server receives the analysis results obtained from the generative AI model and stores them in a database. The analysis results include identified harassment tendencies and patterns.

[1121] Step 5:

[1122] The server visualizes the data based on the analysis results. It uses Pandas and Matplotlib to represent the analysis results in graphs and charts. For example, it generates a bar graph showing that there are many sarcastic reports.

[1123] Step 6:

[1124] The server automatically generates reports based on the visualized data. These reports include harassment trends and patterns, as well as specific countermeasures and preventative measures. The generated reports are output as text files.

[1125] Step 7:

[1126] Users can download and view reports generated through a smartphone application. The reports include specific measures, such as "regular counseling."

[1127] In this way, the system can efficiently classify and respond quickly to inquiries and reports related to harassment, identify and visualize harassment trends and patterns, and propose specific countermeasures and preventative measures.

[1128] (Example 3)

[1129] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[1130] Traditional harassment prevention systems often involved manual processes for classifying harassment reports and inquiries, filtering them by importance and urgency, and proposing specific countermeasures and preventative measures, making rapid response difficult. Furthermore, there was a lack of means to identify harassment trends and patterns and visualize the data, making it challenging to implement effective countermeasures.

[1131] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[1132] This invention includes a server that uses generative AI to automatically classify and filter inquiries and reports related to harassment within the company according to their importance and urgency; a server that collects harassment data from each department within the company; a server that preprocesses the collected data, including imputing missing values ​​and correcting outliers; a server that inputs the preprocessed data into a generative AI model to identify departments, time periods, and types of harassment that are frequent; a server that proposes specific countermeasures and preventive measures based on the analysis results of the generative AI model; and a server that notifies the user of the proposed measures. This enables rapid classification and filtering of harassment reports and inquiries, and allows for the proposal of effective countermeasures and preventive measures.

[1133] "Generative AI" is an artificial intelligence technology that generates new information and suggestions based on data.

[1134] "Harassment" refers to inappropriate words or actions directed at others in the workplace or society, and is an act that causes mental or physical distress.

[1135] An "inquiry" refers to a question or consultation made to seek specific information or advice.

[1136] A "report" refers to an official notification or record made to convey information about a specific event or situation.

[1137] "Classification" refers to grouping collected data or information based on specific criteria.

[1138] "Filtering" refers to the process of selecting data and information based on specific criteria and extracting only what is necessary.

[1139] "Data collection" refers to the act of gathering information and data necessary for a specific purpose.

[1140] "Preprocessing" refers to the process of preparing data for analysis by performing actions such as imputing missing values ​​and correcting outliers before data analysis is carried out.

[1141] A "generative AI model" refers to a specific implementation of generative AI, specifically an algorithm or program used to generate new information or suggestions based on data.

[1142] "Analysis" refers to the act of examining collected data to identify specific patterns or trends.

[1143] "Countermeasures" refer to specific actions or measures taken in response to a particular problem or issue.

[1144] "Preventive measures" refer to specific actions or steps taken to prevent a particular problem or issue from occurring before it happens.

[1145] "Notification" refers to an official communication or announcement made to convey specific information or proposals to relevant parties.

[1146] This invention is a system that uses generative AI to propose countermeasures and preventative measures against harassment. A specific embodiment of this system is shown below.

[1147] The server first collects data on harassment from various departments within the company. This data includes the number of harassment reports, the time of occurrence, the location, and the type of harassment. For example, it collects data from sales department reports and survey results.

[1148] Next, the server preprocesses the collected data. Specifically, it imputes missing values ​​and detects and corrects outliers. It also standardizes the data and converts it into a format suitable for generative AI models. For example, it may convert text data into numerical data.

[1149] The server inputs the pre-processed data into a generative AI model (e.g., OpenAI's GPT-4) for analysis. The generative AI model analyzes the data to identify departments, time slots, and types of harassment that are frequent. For example, it might identify that harassment is most common in the sales department between 3 PM and 5 PM.

[1150] Subsequently, the server proposes specific countermeasures and preventative measures based on the analysis results of the generated AI model. For example, it might suggest conducting harassment prevention training for the sales department or strengthening the monitoring system between 3 PM and 5 PM.

[1151] Finally, the server notifies the user of the proposed changes. Specifically, it sends the proposed changes to the relevant parties via email or the company's internal messaging system. For example, it might notify sales managers about the implementation of harassment prevention training.

[1152] As a concrete example, consider the following scenario.

[1153] scenario:

[1154] Data was collected from the sales department of a certain company showing a high number of harassment reports during a specific time slot (3 PM to 5 PM). Based on this data, a generative AI model analyzes the data and proposes countermeasures.

[1155] Example of a prompt:

[1156] "In the sales department, there are many reports of harassment between 3 PM and 5 PM. Please propose measures to address and prevent harassment during this time period."

[1157] Example output from a generative AI model:

[1158] "As a measure against harassment during the 3 PM to 5 PM timeframe, we propose the following:

[1159] 1. Strengthen the monitoring system during this time period.

[1160] 2. Conduct harassment prevention training for all members of the sales department.

[1161] 3. Improve the harassment reporting system to enable anonymous reporting.

[1162] 4. Make adjustments to reduce the workload during this time period.

[1163] In this way, the server utilizes a generated AI model to propose harassment countermeasures and preventative measures and provides a system that notifies the user. The flow of specific processing in Example 3 will be explained using Figure 15.

[1164] Step 1:

[1165] The server collects harassment-related data from various departments within the company. Inputs include departmental databases and survey results. Specifically, the server accesses each department's database and retrieves report data such as "October 1, 2023, 3 PM, Sales Department, Verbal Abuse." The output is the collected harassment data.

[1166] Step 2:

[1167] The server preprocesses the collected data. The input is the harassment data collected in step 1. Specifically, the server imputes missing data values, detects and corrects outliers, and standardizes the data, converting it into a format suitable for the generative AI model. For example, for data lacking "occurrence time," it imputes the average occurrence time. The output is the preprocessed data.

[1168] Step 3:

[1169] The server inputs the pre-processed data into a generative AI model and performs analysis. The input is the data pre-processed in step 2. Specifically, the server inputs the data into a generative AI model (for example, OpenAI's GPT-4) to identify departments, time slots, and types of harassment that are frequent. For example, it might identify a pattern such as "verbal abuse is frequent in the sales department between 3 PM and 5 PM." The output is the analysis results.

[1170] Step 4:

[1171] The server proposes specific countermeasures and preventative measures based on the analysis results of the generated AI model. The input is the analysis results obtained in step 3. Specifically, the server proposes "strengthening the monitoring system for the sales department between 3 PM and 5 PM" and "conducting training to prevent harassment." The output is the proposed content.

[1172] Step 5:

[1173] The server notifies the user of the proposed content. The input is the proposed content obtained in step 4. Specifically, the server sends the proposed content to the relevant parties via email or the internal messaging system. For example, it sends an email to the sales department manager saying, "Please consider strengthening the monitoring system during the time period from 3 PM to 5 PM." The output is that the notification is complete.

[1174] (Application Example 3)

[1175] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1176] Traditional harassment prevention systems often involved manual reporting and analysis of harassment incidents, making rapid response difficult. Furthermore, data analysis to identify harassment trends and patterns was insufficient, hindering the proposal of specific countermeasures and preventative measures. Additionally, the lack of a function to detect high-risk harassment situations in real time and send appropriate alerts prevented effective risk mitigation.

[1177] In Application Example 3, the identification processing by the identification processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for analyzing the content of reports using a generative AI model to identify the frequency, type, location, and time of occurrence of harassment; means for proposing specific harassment countermeasures and preventive measures based on the analysis results; and means for detecting situations with a high risk of harassment in real time and sending alerts to users. This enables rapid response to harassment, identification of trends and patterns, proposal of specific countermeasures and preventive measures, and real-time risk detection and alert transmission.

[1178] "Generative AI" is a type of artificial intelligence technology that has the ability to generate new data and information based on input data.

[1179] An "inquiry" is a question or request made to obtain specific information or support.

[1180] A "report" is a written or oral communication made to provide information about a specific event or situation.

[1181] "Classification" is the process of grouping data or information based on specific criteria.

[1182] "Filtering" is the process of selecting data and information based on specific criteria and removing unnecessary elements.

[1183] A "generative AI model" is a specific implementation of generative AI, an artificial intelligence model trained to perform a particular task.

[1184] "Analysis" is the process of examining data and information in detail to identify specific patterns and trends.

[1185] "Frequency" refers to the number or proportion of times a particular event or situation occurs.

[1186] A "type" refers to a category classified based on specific characteristics or properties.

[1187] "Location of occurrence" refers to the physical location or environment in which a particular event or situation occurred.

[1188] A "time zone" refers to a specific period or range of time.

[1189] "Countermeasures" refer to specific actions or measures taken to address a particular problem or risk.

[1190] "Preventive measures" refer to specific actions or steps taken to prevent a particular problem or risk from occurring before it happens.

[1191] "Real-time" refers to processing or responding immediately without delay.

[1192] An "alert" is a notification or warning intended to draw attention to a specific situation or risk.

[1193] The following system configuration and processing procedure will be described as embodiments for carrying out this invention.

[1194] System Configuration

[1195] This system consists of the following main components:

[1196] 1. Server: Runs generative AI models to analyze harassment reports, propose countermeasures and preventative measures, and send real-time alerts.

[1197] 2. Device: A device used by users to input harassment reports and receive analysis results, countermeasures, and preventative measures. This includes smartphones and personal computers.

[1198] 3. Network: A communication infrastructure used to connect servers and terminals and to send and receive data.

[1199] Hardware and software to be used

[1200] Hardware: Servers, smartphones, personal computers

[1201] Software: OpenAI API, Python, database management system (e.g., MySQL)

[1202] Data processing and data calculation

[1203] 1. Entering a harassment report:

[1204] Users use their devices to submit harassment reports via text, voice, images, etc.

[1205] The entered data is sent to the server and stored in the database.

[1206] 2. Analysis using generative AI models:

[1207] The server uses OpenAI's generative AI models (e.g., GPT-4) to analyze the reported content.

[1208] The analysis results will identify the frequency, type, location, and time of occurrence of harassment.

[1209] 3. Proposals for countermeasures and preventative measures:

[1210] Based on the analysis results, the server uses a generative AI model to propose specific countermeasures and preventative measures.

[1211] The proposal will be sent to the device and the user will be notified.

[1212] 4. Real-time alerts:

[1213] The server detects situations with a high risk of harassment in real time and sends alerts to users.

[1214] An alert will appear on the device to draw the user's attention.

[1215] Specific example

[1216] Examples of harassment reports

[1217] Report details: "In the sales department, employees are being subjected to excessive reprimands from their superiors every Friday afternoon."

[1218] Example of analysis results

[1219] Analysis result: "Sales department, Friday afternoon, reprimand from superior."

[1220] Examples of proposed countermeasures and preventative measures

[1221] Countermeasures and preventative measures: "A warning message was sent to the sales department on Friday afternoon. Harassment prevention training was conducted for supervisors."

[1222] Example of a prompt

[1223] Analysis prompt: "Analyze the following harassment report to identify frequency, type, location, and time of occurrence. Report content: {report}"

[1224] Prompt for suggesting countermeasures: "Based on the analysis results below, please propose specific harassment prevention and countermeasures.\nAnalysis results: {analysis}"

[1225] The above describes the embodiments for carrying out this invention. This system enables rapid response to harassment, identification of trends and patterns, proposal of specific countermeasures and preventive measures, and real-time risk detection and alert transmission.

[1226] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[1227] Step 1:

[1228] Users use a terminal to enter harassment reports. Input can be in the form of text, audio, or images. The entered data is sent from the terminal to a server and stored in a database. The input data includes details of the harassment, the location, and the time of occurrence.

[1229] Step 2:

[1230] The server retrieves report data from the database and analyzes the report content using a generative AI model (e.g., GPT-4). Specifically, the report content is input to the generative AI model as prompts to identify the frequency, type, location, and time of occurrence of harassment. The input data is the report content, and the output data is the analysis results.

[1231] Step 3:

[1232] The server uses a generative AI model to propose specific countermeasures and preventative measures based on the analysis results. Prompt messages are used to input the analysis results into the generative AI model and obtain suggestions for countermeasures and preventative measures. The input data is the analysis results, and the output data is the suggested countermeasures and preventative measures.

[1233] Step 4:

[1234] The server sends the proposed countermeasures and preventative measures to the terminal and notifies the user. The user can then review the proposed countermeasures and preventative measures through the terminal and take the necessary actions. The input data consists of proposed countermeasures and preventative measures, while the output data is the notification to the user.

[1235] Step 5:

[1236] The server continuously monitors data to detect high-risk harassment situations in real time. If a high-risk situation is detected, the server generates an alert in real time and sends it to the terminal. The user receives the alert through the terminal and takes appropriate action. Input data is real-time monitoring data, and output data is the alert.

[1237] The above outlines the specific processing steps of this system. The specific actions performed at each step enable rapid response to harassment, identification of trends and patterns, proposal of specific countermeasures and preventive measures, and real-time risk detection and alert transmission.

[1238] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1239] "Example of form 1"

[1240] One embodiment of the present invention is a system that combines a generative AI with an emotion engine. This system collects text data from users, and the generative AI analyzes that data. As a result of the analysis, content related to harassment is automatically classified and filtered according to importance and urgency. Furthermore, the emotion engine recognizes the user's emotions and incorporates that information into the generative AI's analysis. Specifically, the emotion engine recognizes the emotional state of the user when they make a report (anger, fear, sadness, etc.), and the generative AI uses this information in its analysis. This makes it possible to more accurately determine the severity and urgency of the harassment.

[1241] "Example of form 2"

[1242] Another embodiment of the present invention is a system that combines a generative AI and an emotion engine to identify harassment tendencies and patterns. In this system, the results analyzed by the generative AI and the emotional information recognized by the emotion engine are combined to identify harassment tendencies and patterns. Specifically, if a particular emotional state is determined to be closely related to the occurrence of harassment, that information is used to identify harassment tendencies and patterns. This makes it possible to formulate harassment prevention measures more effectively.

[1243] "Example of form 3"

[1244] As a further embodiment of the present invention, there is a system that proposes harassment countermeasures and preventive measures by combining a generative AI and an emotion engine. In this system, the results analyzed by the generative AI and the emotional information recognized by the emotion engine are combined to propose harassment countermeasures and preventive measures. Specifically, if a particular emotional state is determined to be closely related to the occurrence of harassment, that information is used to propose harassment countermeasures and preventive measures. This makes it possible to formulate harassment prevention measures more effectively.

[1245] The following describes the processing flow for each example of the form.

[1246] "Example of form 1"

[1247] Step 1: Collect text data from users.

[1248] Step 2: Generative AI analyzes the text data and automatically categorizes content related to harassment.

[1249] Step 3: Filter the classified items according to their importance and urgency.

[1250] Step 4: The emotion engine recognizes the user's emotions.

[1251] Step 5: Incorporate the recognized emotional information into the analysis of the generative AI.

[1252] "Example of form 2"

[1253] Step 1: Collect text data from users.

[1254] Step 2: Generative AI analyzes the text data and automatically categorizes content related to harassment.

[1255] Step 3: The emotion engine recognizes the user's emotions.

[1256] Step 4: Combine the recognized emotional information with the analysis results from the generative AI to identify harassment tendencies and patterns.

[1257] "Example of form 3"

[1258] Step 1: Collect text data from users.

[1259] Step 2: Generative AI analyzes the text data and automatically categorizes content related to harassment.

[1260] Step 3: The emotion engine recognizes the user's emotions.

[1261] Step 4: Combine the recognized emotional information with the analysis results from the generative AI to propose harassment countermeasures and preventative measures.

[1262] (Example 1)

[1263] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1264] Traditional harassment reporting systems often required manual classification and assessment of urgency, making rapid response difficult. Furthermore, they failed to accurately determine the severity of harassment because they processed reports without considering the user's emotional state. Additionally, there was a lack of systems capable of ensuring anonymity while providing appropriate responses.

[1265] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1266] In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI, means for filtering them according to importance and urgency, means for enabling a rapid response and risk hedging, means for recognizing the user's emotional state using an emotion engine and incorporating that information into the generative AI's analysis, means for collecting text data through an anonymous reporting channel using a chatbot, means for transmitting the collected text data to the server in real time, and means for notifying the appropriate person in charge based on the results of the generative AI's analysis. This enables the rapid and accurate processing of harassment-related reports and allows for responses that take into account the user's emotional state.

[1267] "Generative AI" refers to artificial intelligence that uses natural language processing technology to analyze text data and generate information tailored to specific purposes.

[1268] An "emotion engine" is software that recognizes a user's emotional state (e.g., anger, fear, sadness, etc.) from text data and provides that information.

[1269] A "chatbot" is an automated response system that communicates with users in a conversational format and collects specific information.

[1270] An "anonymous reporting channel" is a means of communication that allows users to report information without revealing their identity.

[1271] "Text data" refers to the written information entered by the user, including data such as report content and inquiry details.

[1272] "Classification" is the process of dividing collected text data into categories based on specific criteria.

[1273] "Filtering" is the process of selecting classified data according to its importance and urgency.

[1274] "Prompt response" refers to taking appropriate action quickly based on the information provided in the report.

[1275] "Risk hedging" refers to identifying potential risks in advance and taking measures to minimize their impact.

[1276] "Real-time" refers to the process of data being processed immediately as soon as it is generated.

[1277] "The appropriate person in charge" refers to the individual or department responsible for the tasks that require action based on the report.

[1278] Modes for carrying out the invention

[1279] This invention uses a system that combines generative AI and an emotion engine to automatically classify inquiries and reports related to workplace harassment and filter them according to their importance and urgency. A specific embodiment of this system is shown below.

[1280] 1. Program generation

[1281] The server generates a program for a system that combines generative AI and an emotion engine. This program includes a series of processes to collect, analyze, classify, and filter user reports and to take appropriate action quickly.

[1282] 2. Program Processing Description

[1283] The server will run the program using the following hardware and software.

[1284] Hardware: Servers, terminals

[1285] Software: Generative AI (e.g., OpenAI's GPT-4), emotion engines (e.g., IBM Watson's Tone Analyzer), chatbots

[1286] Data collection:

[1287] Users enter text data about harassment through an anonymous reporting channel using a chatbot. For example, a user might type, "I'm experiencing ongoing harassment from my boss, and it's taking a toll on my mental health."

[1288] Data transmission:

[1289] The terminal sends text data from the user to the server in real time.

[1290] Data analysis:

[1291] The server inputs the collected text data into a generative AI, which then analyzes the content related to harassment. Specifically, the server inputs the text data into the generative AI as a prompt, instructing it to "analyze this report and determine the severity and urgency of the harassment."

[1292] Emotion recognition:

[1293] The server uses an emotion engine to recognize the user's emotional state (e.g., anger, fear, sadness, etc.) and provides that information to the generative AI. The emotion engine analyzes the text data and recognizes the user's emotional state as "sadness."

[1294] Classification and filtering:

[1295] The generative AI classifies text data according to the severity and urgency of the harassment, based on emotional information provided by the emotion engine. The server then filters the data based on importance and urgency. For example, the generative AI might classify the text data as "serious harassment" and determine its urgency to be "high."

[1296] Response notification:

[1297] Based on the filtered results, the server notifies the appropriate person in charge and promptly takes the necessary action. Specifically, the server sends high-priority reports to the person in charge via email or notification system, notifying them that "this report requires immediate attention."

[1298] 3. Specific Examples and Examples of Prompt Statements

[1299] Specific example:

[1300] If a user reports through an anonymous chatbot that "I am being harassed by my boss and it is causing me emotional distress," it will be handled as follows:

[1301] 1. The user enters the following into the chatbot: "I am experiencing ongoing harassment from my boss, and it is taking a toll on my mental health."

[1302] 2. The terminal sends this text data to the server.

[1303] 3. The server inputs text data into the generative AI and instructs it to "analyze this report and determine the severity and urgency of the harassment."

[1304] 4. The server inputs text data into the emotion engine, which recognizes the emotion of "sadness."

[1305] 5. The generative AI classifies the text data as "serious harassment" and determines its urgency to be "high." The server filters this classification result to extract reports with high urgency.

[1306] 6. The server sends urgent reports to the responsible person via email or notification system, notifying them that "this report requires immediate attention."

[1307] Example of a prompt:

[1308] "I am experiencing ongoing harassment from my superior, and it is causing me significant emotional distress. Please analyze this report and assess the severity and urgency of the harassment."

[1309] In this way, a system is realized in which servers, terminals, and users work together to process harassment-related reports quickly and accurately.

[1310] The flow of the specific processing in Example 1 will be explained using Figure 17.

[1311] Step 1:

[1312] Users enter text data about harassment through an anonymous reporting channel using a chatbot.

[1313] As a concrete example, the user would type "I'm being harassed by my boss and it's taking a toll on my mental health" into the chatbot's interface.

[1314] Input: Text data entered by the user

[1315] Output: Text data sent to the chatbot

[1316] Step 2:

[1317] The terminal sends text data from the user to the server in real time.

[1318] Specifically, the terminal immediately transfers the text data entered by the user to the server.

[1319] Input: Text data sent to the chatbot

[1320] Output: Text data sent to the server

[1321] Step 3:

[1322] The server inputs the collected text data into a generative AI, which then analyzes the content related to harassment.

[1323] In terms of specific actions, the server inputs text data as a prompt to the generative AI, instructing it to "analyze this report and determine the severity and urgency of the harassment."

[1324] Input: Text data sent to the server

[1325] Output: Analysis results by generative AI

[1326] Step 4:

[1327] The server uses an emotion engine to recognize the user's emotional state (e.g., anger, fear, sadness, etc.) and provides that information to the generative AI.

[1328] Specifically, the server inputs text data into the emotion engine, which then recognizes the emotion of "sadness."

[1329] Input: Text data sent to the server

[1330] Output: Emotion recognition results from the emotion engine

[1331] Step 5:

[1332] The generative AI classifies text data according to the severity and urgency of the harassment, based on emotional information provided by the emotion engine.

[1333] In specific terms, the generative AI classifies the text data as "serious harassment" and determines its urgency to be "high."

[1334] Input: Analysis results by generative AI and emotion recognition results by emotion engine

[1335] Output: Severity and urgency of classified harassment

[1336] Step 6:

[1337] The server uses this classification result to filter the data according to its importance and urgency.

[1338] Specifically, the server extracts reports with high urgency.

[1339] Input: Severity and urgency of classified harassment

[1340] Output: Filtered report data

[1341] Step 7:

[1342] Based on the filtered results, the server notifies the appropriate person in charge and takes the necessary action promptly.

[1343] Specifically, the server sends high-priority reports to the responsible person via email or a notification system, notifying them that "a prompt response is required to this report."

[1344] Input: Filtered report data

[1345] Output: Notification and instructions for action to the person in charge.

[1346] (Application Example 1)

[1347] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[1348] Traditional harassment reporting systems often require manual classification and assessment of importance, making prompt responses difficult. Furthermore, they fail to consider the reporter's emotional state, making it difficult to accurately determine the severity and urgency of the harassment. Additionally, anonymous reporting is often difficult, leading to hesitation among reporters. A system is needed to address these issues and manage harassment reports more efficiently and accurately.

[1349] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1350] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency; an emotion engine to recognize the user's emotional state and incorporate that information into the generative AI's analysis; a means to receive anonymous reports through an application installed on a smartphone; a means to propose appropriate countermeasures based on the report content; and a means to enable a rapid response and contribute to risk hedging. As a result, the classification of report content and the determination of importance are automated, enabling a rapid and accurate response. Furthermore, by considering the reporter's emotional state, the severity and urgency of the harassment can be judged more accurately. In addition, anonymous reporting becomes easier, allowing reporters to report without hesitation.

[1351] "Generative AI" is an artificial intelligence technology that generates new information or results based on data provided by users.

[1352] An "emotion engine" is a technology that analyzes and recognizes a user's emotional state from data such as text and audio.

[1353] A "smartphone" is a multi-functional mobile device that, in addition to the functions of a mobile phone, is capable of internet connectivity and the execution of applications.

[1354] "Anonymous reporting" is a method of reporting problems or incidents without disclosing the reporter's personal information.

[1355] "Importance" is an indicator that shows the magnitude and priority of the impact of a reported event or problem.

[1356] "Urgency" is an indicator that shows how quickly a response is needed to a reported event or problem.

[1357] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[1358] "Prompt response" refers to taking appropriate measures quickly in response to a problem or incident.

[1359] "Risk hedging" refers to minimizing risk by predicting potential risks and taking countermeasures against them.

[1360] "Appropriate countermeasures" refer to the most effective and appropriate solutions or measures for the reported incident or problem.

[1361] The system for carrying out this invention combines a generative AI and an emotion engine, and operates through an application installed on a smartphone. Specific embodiments of this system are described below.

[1362] System Configuration

[1363] hardware

[1364] Smartphone: A device used by users to enter reports.

[1365] Server: A computer system that runs generative AI and emotion engines and analyzes data.

[1366] software

[1367] Generative AI: Uses the OpenAI API to analyze and classify the reported content.

[1368] Emotion Engine: Uses TextBlob to analyze the user's emotional state.

[1369] Smartphone application: Provides an interface for users to anonymously enter and submit reports.

[1370] Data processing and data calculation

[1371] smartphone

[1372] Users anonymously submit harassment reports through a smartphone application. The entered text data is sent to a server.

[1373] server

[1374] The server processes the received text data in the following steps:

[1375] 1. Sentiment Analysis: TextBlob is used to analyze the emotional state of the reported text. Specifically, the text is scored for positive, negative, and neutral sentiment.

[1376] 2. Classification using generative AI: Using the OpenAI API, report texts are classified according to the severity and urgency of the harassment. Prompt sentences are generated and analyzed by the AI.

[1377] Specific example

[1378] Report content

[1379] "My boss yells at me every day, and I can't concentrate on my work. I'm feeling extremely stressed."

[1380] Emotion analysis results

[1381] Strong negative emotions

[1382] Classification results of generative AI

[1383] Severity: High, Urgency: High

[1384] Example of a prompt

[1385] Please categorize the following text according to the severity and urgency of the harassment:

[1386] My boss yells at me every day, and I can't concentrate on my work. I'm under a lot of stress.

[1387] Classification:

[1388] effect

[1389] This system automates the classification and prioritization of reports, enabling quick and accurate responses. Furthermore, by considering the reporter's emotional state, it allows for a more accurate assessment of the severity and urgency of harassment. Additionally, it facilitates anonymous reporting, encouraging reporters to report incidents without hesitation.

[1390] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[1391] Step 1:

[1392] Users anonymously submit reports of harassment through a smartphone application.

[1393] Input: Text data entered by the user.

[1394] Output: Text data sent from the smartphone to the server.

[1395] Specific action: The user fills in the report details in the application's input form and presses the submit button.

[1396] Step 2:

[1397] The server receives text data and passes it to the emotion engine for sentiment analysis.

[1398] Input: Text data sent from a smartphone.

[1399] Output: Sentiment analysis results (positive, negative, and neutral scores).

[1400] Specific operation: The server uses TextBlob to analyze the sentiment of text data and generate a sentiment score.

[1401] Step 3:

[1402] The server passes the emotion analysis results to a generative AI, which then classifies the reported content.

[1403] Input: Sentiment analysis results and text data.

[1404] Output: Classification results (severity and urgency) generated by a generative AI.

[1405] Specific operation: The server uses the OpenAI API to generate prompt text and classify the report content based on the text data and sentiment analysis results.

[1406] Step 4:

[1407] The server proposes appropriate countermeasures based on the classification results.

[1408] Input: Classification results from a generative AI.

[1409] Output: Proposed countermeasures.

[1410] Specific operation: The server analyzes the classification results and selects and proposes the most suitable countermeasure from a pre-configured database of countermeasures.

[1411] Step 5:

[1412] The server sends the proposed solution to the smartphone application and notifies the user.

[1413] Input: Proposed countermeasures.

[1414] Output: Solutions displayed in the smartphone application.

[1415] Specific operation: The server sends the countermeasure to the smartphone application and generates a message to notify the user.

[1416] Step 6:

[1417] The user reviews the proposed solutions and provides additional reports and feedback as needed.

[1418] Input: Proposed solutions and user feedback.

[1419] Output: Additional reports and feedback data.

[1420] Specific operation: The user reviews the suggested solutions through the application and provides additional reports and feedback as needed.

[1421] (Example 2)

[1422] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1423] Conventional harassment prevention systems often require manual analysis to identify harassment tendencies and patterns, which is time-consuming and labor-intensive. Furthermore, they fail to consider emotional information in their analysis, making it difficult to develop effective preventative measures. Additionally, the lack of automated data visualization and report generation hinders rapid response. To address these challenges, a system combining generative AI and an emotion engine is needed.

[1424] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1425] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to importance and urgency; means for preprocessing the collected data; means for analyzing the data using a generative AI model; means for recognizing emotional information using an emotion engine; means for integrating the analysis results of the generative AI and the emotional information of the emotion engine to visualize the data; means for automatically generating reports based on the visualized data; and means for enabling rapid response and contributing to risk hedging. This makes it possible to effectively identify harassment trends and patterns and to formulate preventive measures that take emotional information into consideration. Furthermore, data visualization and automatic report generation enable rapid response.

[1426] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[1427] "Harassment" refers to inappropriate words or actions towards others in the workplace or other environments.

[1428] An "inquiry" is a question or request made to seek specific information or support.

[1429] "Reporting" is a form of communication conducted to provide information about a specific event or situation.

[1430] "Classification" is the process of grouping data or information based on specific criteria.

[1431] "Filtering" is the process of selecting data or information based on specific criteria.

[1432] "Preprocessing" refers to the initial stages of processing data to convert it into a format suitable for analysis and processing.

[1433] "Generative AI models" refer to specific implementations and algorithms of generative AI.

[1434] An "emotion engine" is a technology that recognizes emotional information from data such as text and audio.

[1435] "Emotional information" refers to information that indicates the emotional state or tendency contained within the data.

[1436] "Integration" is the process of combining multiple data or pieces of information into one.

[1437] "Visualization" is the process of representing data and information in visual forms such as graphs and charts.

[1438] A "report" is a document that compiles specific information or analysis results.

[1439] "Prompt response" refers to taking quick and appropriate action in response to a problem or situation.

[1440] "Risk hedging" refers to taking measures or means to mitigate potential risks.

[1441] This invention is a system that utilizes generative AI and an emotion engine to identify harassment tendencies and patterns, and to visualize the data and generate reports. Specific embodiments of this system are described below.

[1442] Hardware and software to use

[1443] Hardware: Servers, terminals

[1444] Software: Generative AI models (e.g., GPT-4), sentiment engines (e.g., Affectiva), data preprocessing tools (e.g., Python, NLTK, Pandas), visualization tools (e.g., Tableau)

[1445] System Overview

[1446] This system begins with users uploading harassment-related data from their devices to a server. The server preprocesses the collected data and analyzes it using a generative AI model. Furthermore, it uses an emotion engine to recognize emotional information and integrates it with the generative AI analysis results. The integrated data is visualized, and finally, a report is automatically generated.

[1447] Specific examples of operation

[1448] 1. Data Collection

[1449] Users upload files such as internal chat logs, emails, and survey results from their devices to the server.

[1450] Specific actions: The user opens a browser and accesses the system's upload page. A file selection dialog opens, the user selects the file to upload, and clicks the upload button.

[1451] 2. Data preprocessing

[1452] The server analyzes the received data and performs text cleaning, such as removing HTML tags and special characters. Next, the text is tokenized and converted into a data format for sentiment analysis.

[1453] Software used: Python, NLTK, Pandas

[1454] 3. Analysis using generative AI

[1455] The server inputs prompt text into a generative AI model (e.g., GPT-4) to extract harassment tendencies and patterns.

[1456] Specific action: "This chat log contains many offensive words."

[1457] 4. Emotional analysis using an emotional engine

[1458] The server inputs text data into an emotion engine (e.g., Affectiva) to detect specific emotional states (e.g., anger, sadness).

[1459] Specific action: "This email clearly expresses strong feelings of anger."

[1460] 5. Data Integration and Visualization

[1461] The server integrates the analysis results from the generative AI with the emotional information from the emotion engine, inputs it into a visualization tool (e.g., Tableau), and creates graphs and charts.

[1462] Specific action: "Harassment occurs frequently in a particular department."

[1463] 6. Report generation

[1464] The server uses a report generation tool to create a report that includes analysis results and suggested countermeasures.

[1465] Specific action: "Implementing regular emotional check-ins can be expected to prevent harassment."

[1466] Example of a prompt

[1467] "Please analyze this chat log to identify any trends or patterns of harassment."

[1468] "Please acknowledge the emotional state expressed in this email and determine if it is related to the occurrence of harassment."

[1469] This system makes it possible to effectively identify harassment tendencies and patterns, and to develop preventative measures that take emotional information into account. Furthermore, data visualization and automated report generation enable rapid response.

[1470] The flow of the specific processing in Example 2 will be explained using Figure 19.

[1471] Step 1:

[1472] Data collection

[1473] Users upload files such as internal chat logs, emails, and survey results from their devices to the server.

[1474] Input: User-selected file (e.g., chat logs, emails, survey results)

[1475] Specific actions: The user opens a browser and accesses the system's upload page. A file selection dialog opens, the user selects the file to upload, and clicks the upload button.

[1476] Output: Files uploaded to the server

[1477] Step 2:

[1478] Data preprocessing

[1479] The server preprocesses the data it receives.

[1480] Input: Uploaded file

[1481] Specific operation: The server cleans the text data, removing unwanted characters and noise. Then, it tokenizes the text and converts it into a data format for sentiment analysis.

[1482] Output: Preprocessed text data

[1483] Software used: Python, NLTK, Pandas

[1484] Step 3:

[1485] Analysis using generative AI

[1486] The server analyzes the preprocessed data using a generative AI model (e.g., GPT-4).

[1487] Input: Preprocessed text data

[1488] Specific operation: The server inputs prompt text into a generative AI model, which then extracts harassment tendencies and patterns.

[1489] Output: Analysis results regarding harassment trends and patterns

[1490] Example prompt: "Analyze this chat log to identify harassment tendencies and patterns."

[1491] Step 4:

[1492] Emotional analysis using an emotional engine

[1493] The server uses an emotion engine (e.g., Affectiva) to recognize emotional information within the data.

[1494] Input: Preprocessed text data

[1495] Specific operation: The server inputs text data into the emotion engine and detects specific emotional states (e.g., anger, sadness).

[1496] Output: Analysis results regarding emotional information

[1497] Example prompt: "Please recognize the emotional state of this email and determine if it is related to the occurrence of harassment."

[1498] Step 5:

[1499] Data integration and visualization

[1500] The server integrates the analysis results from the generative AI with the emotional information from the emotion engine and visualizes the data.

[1501] Input: Analysis results from generative AI, emotional information from the emotion engine.

[1502] Specific operation: The server inputs the integrated data into a visualization tool (e.g., Tableau) to create graphs and charts.

[1503] Output: Visualized data (graphs and charts)

[1504] Step 6:

[1505] Report generation

[1506] The server automatically generates reports on harassment trends and patterns based on visualized data.

[1507] Input: Visualized data

[1508] Specific operation: The server uses a report generation tool to create a report that includes analysis results and suggested countermeasures.

[1509] Output: Automated report

[1510] Specific example of operation: Generate a report that includes a suggestion that "introducing regular emotional check-ins can be expected to prevent harassment."

[1511] (Application Example 2)

[1512] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1513] Conventional harassment prevention systems have difficulty identifying harassment tendencies and patterns, and have struggled with real-time monitoring and rapid response. As a result, they have often failed to effectively reduce the risk of harassment, leading to delays in victim protection and the development of preventative measures. Furthermore, the lack of sentiment analysis utilizing emotion engines has made it impossible to accurately grasp the circumstances of harassment occurrences. To address these challenges, a system combining generative AI and an emotion engine is necessary.

[1514] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1515] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency; a server that combines generative AI and an emotion engine to identify harassment trends and patterns; and a server that visualizes data and generates reports. This enables real-time monitoring of harassment risks and prompt response.

[1516] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[1517] An "emotion engine" is a technology that analyzes emotions from data such as text and audio, and identifies the emotional state.

[1518] "Harassment" refers to acts that cause mental or physical distress to others through inappropriate words or actions in environments such as the workplace or school.

[1519] A "tendency" is a characteristic or pattern in which a particular phenomenon or behavior moves in a certain direction.

[1520] A "pattern" is a regularity or form in which a particular phenomenon or behavior repeatedly occurs.

[1521] "Data visualization" is a technique that makes data easier to understand by representing it in visual forms such as graphs and charts.

[1522] "Report generation" is the process of automatically creating a report based on analysis results and data.

[1523] "Countermeasures" refer to specific means or methods taken to address a particular problem or risk.

[1524] "Preventive measures" are means or methods to prevent problems or risks from occurring before they happen.

[1525] A "smartphone" is a multi-functional mobile device that, in addition to the functions of a mobile phone, is capable of internet connectivity and the use of applications.

[1526] "Real-time" refers to a state where data and information are processed immediately and provided without delay.

[1527] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[1528] "Text data" refers to digital data that includes character information.

[1529] "Sentiment analysis" is a technology that analyzes emotional states from data such as text and audio, and identifies those emotions.

[1530] The system for implementing this invention combines a generative AI and an emotion engine to identify harassment tendencies and patterns, and to visualize the data and generate reports. Specific embodiments of this system are described below.

[1531] System Configuration

[1532] The system consists of the following main components:

[1533] 1. Generative AI models: Artificial intelligence technology that generates new information or patterns based on data.

[1534] 2. Emotion Engine: A technology that analyzes emotions from data such as text and audio, and identifies the emotional state.

[1535] 3. Smartphone application: An interface for users to monitor harassment risks in real time and input data.

[1536] 4. Data visualization tools: Use software such as Matplotlib or Pandas to visually represent the data.

[1537] 5. Report generation function: A function that automatically creates reports based on the analysis results.

[1538] Program processing

[1539] The server uses a generative AI model to identify harassment tendencies and patterns from the input text. Specifically, it uses the OpenAI API to analyze the text data entered by the user and assess the risk of harassment. The emotion engine performs sentiment analysis on the text data to determine whether a particular emotional state is strongly associated with the occurrence of harassment.

[1540] Data visualization and report generation

[1541] The server visualizes emotional data analyzed by the emotion engine using Matplotlib and Pandas. This allows users to visually understand harassment tendencies and patterns. Furthermore, it automatically generates reports based on the analysis results of the generative AI model and the emotion engine. These reports include harassment tendencies and patterns, emotional states, and suggestions for countermeasures and preventative measures.

[1542] Specific example

[1543] For example, if workplace conversation logs are used as input data, the user would enter text like this:

[1544] "During yesterday's meeting, my boss used very harsh language towards a subordinate. The subordinate was clearly uncomfortable."

[1545] Based on this text data, the generative AI model uses the following prompts to identify harassment tendencies and patterns:

[1546] Identify the tendencies and patterns of harassment from the following text:

[1547] "During yesterday's meeting, my boss used very harsh language towards a subordinate. The subordinate was clearly uncomfortable."

[1548] The emotion engine analyzes the emotional state from this text data, identifying, for example, strong emotions such as "anger" or "sadness." Based on this data, the server visualizes the data and generates a report.

[1549] In this way, users can monitor the risk of harassment in real time and respond quickly.

[1550] The flow of a specific process in Application Example 2 will be explained using Figure 20.

[1551] Step 1:

[1552] Users input text data, such as conversation logs from work or school, through a smartphone application.

[1553] Input: Text data such as conversation logs

[1554] Output: Input text data

[1555] Step 2:

[1556] The server sends the input text data to a generative AI model for analysis to identify harassment tendencies and patterns. Specifically, it uses the OpenAI API to analyze the text data.

[1557] Input: Entered text data

[1558] Output: Analysis results regarding harassment trends and patterns

[1559] Step 3:

[1560] The server uses an emotion engine based on the analysis results to perform sentiment analysis on the text data. The emotion engine identifies emotional states such as "anger" and "sadness" from the text data.

[1561] Input: Input text data, analysis results of a generative AI model

[1562] Output: Data on emotional state

[1563] Step 4:

[1564] The server visualizes the sentiment data analyzed by the sentiment engine using Matplotlib and Pandas. This allows users to visually understand harassment tendencies and patterns.

[1565] Input: Data related to emotional state

[1566] Output: Visualized data (graphs and charts)

[1567] Step 5:

[1568] The server automatically generates a report based on the analysis results of a generative AI model and an emotion engine. This report includes harassment tendencies and patterns, emotional states, and suggestions for countermeasures and preventative measures.

[1569] Input: Analysis results of generative AI models, data on emotional states

[1570] Output: Automated report

[1571] Step 6:

[1572] Users can review generated reports through a smartphone application, monitor harassment risks in real time, and take prompt action.

[1573] Input: Automated report

[1574] Output: User-confirmed report

[1575] In this way, users can monitor the risk of harassment in real time and respond quickly.

[1576] (Example 3)

[1577] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[1578] Traditional harassment prevention systems have struggled to effectively propose specific measures and preventative actions to prevent harassment from occurring in the first place. Furthermore, they lacked measures that considered the emotional state of employees, making it difficult to identify the root causes of harassment. In addition, data collection and analysis regarding harassment were often done manually, resulting in insufficient risk mitigation in situations requiring rapid response.

[1579] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for identifying harassment trends and patterns using a generative AI model; means for collecting employee emotional information using an emotion engine and analyzing the relationship between emotional states and the occurrence of harassment; means for proposing harassment countermeasures and preventive measures based on the analysis results of the generative AI model and the recognition results of the emotion engine; means for visualizing data and generating reports; and means for enabling rapid response and leading to risk hedging. This makes it possible to prevent the occurrence of harassment and propose effective countermeasures and preventive measures that take into account the emotional states of employees.

[1580] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[1581] "Harassment" refers to acts of bullying or unfair treatment within the workplace or organization.

[1582] An "inquiry" is a question or request made to seek specific information or support.

[1583] "Reporting" is the act of providing information about a specific event or situation.

[1584] "Classification" is the process of grouping data or information based on specific criteria.

[1585] "Filtering" is the process of selecting data or information based on specific criteria.

[1586] "Generative AI models" refer to specific implementations and algorithms of generative AI.

[1587] A "trend" is the movement of data that shows a specific direction or pattern.

[1588] A "pattern" is a recurring structure or regularity in data or events.

[1589] An "emotional engine" is a technology for analyzing and recognizing the emotional state of employees.

[1590] "Emotional information" refers to data about the emotional state of employees.

[1591] "Measures" refer to specific actions or means taken to address a particular problem or issue.

[1592] "Preventive measures" are specific actions or means taken to prevent problems or issues from occurring before they happen.

[1593] "Visualization" is the process of representing data and information visually.

[1594] A "report" is a document that compiles specific information or analysis results.

[1595] "Prompt response" refers to taking action to address problems or issues quickly.

[1596] "Risk hedging" refers to measures and means taken to mitigate potential risks.

[1597] This invention is a system that proposes harassment countermeasures and preventative measures using generative AI and an emotion engine. This system operates primarily with a server, terminals, and users.

[1598] Hardware and software to be used

[1599] Hardware: Servers, terminals

[1600] Software: Generative AI models (e.g., GPT-4), emotion engines (e.g., Affectiva)

[1601] Specific operation of the system

[1602] Data collection

[1603] The server collects data on harassment from various departments within the company. Specifically, the server accesses each department's management system to obtain data such as the frequency, time of occurrence, and type of harassment. For example, the server automatically collects monthly reports and internal investigation results.

[1604] Data Analysis

[1605] The server inputs the collected data into a generative AI model to analyze harassment trends. Specifically, the server preprocesses the collected data and inputs it into the generative AI model (e.g., GPT-4) using prompt statements. For example, it might use a prompt statement such as, "If harassment is frequent in a particular department, please identify that department."

[1606] Collection of emotional information

[1607] The terminal collects employee emotional information using an emotion engine. Specifically, the terminal analyzes employees' facial expressions and voices in real time, and the emotion engine (e.g., Affectiva) recognizes their emotional state. For example, the terminal captures an employee's facial expression during a meeting using its camera, and the emotion engine analyzes that expression to determine their emotions.

[1608] Analysis of emotional information

[1609] The server inputs emotional information obtained from the emotion engine into a generative AI model to analyze the relationship between emotional states and the occurrence of harassment. Specifically, the server inputs data obtained from the emotion engine into the generative AI model and uses a prompt message such as, "If a particular emotional state is strongly associated with the occurrence of harassment, please identify that information." For example, the server might identify departments where emotional states such as "anger" or "anxiety" are frequently observed.

[1610] Proposals for countermeasures and preventive measures

[1611] The server proposes specific harassment countermeasures and preventative measures based on the analysis results of the generative AI model and the recognition results of the emotion engine. Specifically, the server inputs a prompt message to the generative AI model such as, "If harassment is occurring frequently in a particular department, what countermeasures should be proposed to that department?" and then formulates countermeasures based on the results obtained. For example, the server might propose regular harassment prevention training and counseling for stress management to a particular department.

[1612] Specific example

[1613] Example 1: At the beginning of each month, the server accesses the management system of each department and automatically downloads the harassment reports from the previous month.

[1614] Example 2: The server downloads a report and inputs it into a generative AI model, which then analyzes it using the prompt "In which department is harassment most frequent?".

[1615] Example 3: The device captures the facial expressions of employees during a meeting using its camera, and an emotion engine analyzes those expressions in real time to determine emotions such as "anger" or "anxiety."

[1616] Example 4: The server inputs emotional data obtained from the emotion engine into a generative AI model and analyzes it using the prompt message, "Identify information if a particular emotional state is strongly associated with the occurrence of harassment."

[1617] Example 5: Based on the analysis results of a generative AI model, the server proposes "harassment prevention training" and "stress management counseling" to a specific department.

[1618] The above describes embodiments for carrying out the present invention. The flow of the specific processing in Example 3 will be explained with reference to Figure 21.

[1619] Step 1: Data Collection

[1620] The server collects harassment-related data from various departments within the company. Specifically, the server accesses each department's management system to obtain data such as the frequency, time of occurrence, and type of harassment. The input is data from each department's management system, and the output is the collected harassment-related data. For example, the server automatically downloads monthly reports and internal investigation results.

[1621] Step 2: Data Preprocessing

[1622] The server preprocesses the collected data. Specifically, it performs tasks such as imputing missing values, detecting and correcting outliers, and normalizing the data. The input is the collected harassment-related data, and the output is the preprocessed data. For example, the server imputates missing values ​​with the mean and detects and corrects outliers.

[1623] Step 3: Data Analysis

[1624] The server inputs pre-processed data into a generative AI model to analyze harassment trends. Specifically, the server inputs prompts into the generative AI model (e.g., GPT-4). The input consists of pre-processed data and prompts, while the output is the analysis results regarding harassment trends and patterns. For example, a prompt such as "If harassment is frequent in a particular department, please identify that department" might be used.

[1625] Step 4: Gathering emotional information

[1626] The terminal collects employee emotional information using an emotion engine. Specifically, the terminal analyzes employees' facial expressions and voices in real time, and the emotion engine (e.g., Affectiva) recognizes their emotional state. The input is employee facial expression and voice data, and the output is the recognized emotional information. For example, the terminal captures an employee's facial expression during a meeting using a camera, and the emotion engine analyzes the emotion from that expression.

[1627] Step 5: Analyzing emotional information

[1628] The server inputs emotional information obtained from the emotion engine into a generative AI model to analyze the relationship between emotional states and the occurrence of harassment. Specifically, the server inputs data obtained from the emotion engine into the generative AI model and uses a prompt message such as, "If a particular emotional state is strongly associated with the occurrence of harassment, please identify that information." The input consists of emotional information and the prompt message, and the output is the analysis results regarding the relationship between emotional states and harassment. For example, the server might identify departments where emotional states such as "anger" or "anxiety" are frequently observed.

[1629] Step 6: Propose countermeasures and preventive measures

[1630] The server proposes specific harassment prevention and countermeasures based on the analysis results of the generative AI model and the recognition results of the emotion engine. Specifically, the server inputs a prompt to the generative AI model such as, "If harassment is occurring frequently in a particular department, what measures should be proposed to that department?" and then formulates countermeasures based on the results obtained. The input consists of the analysis results and the prompt, and the output is the proposed countermeasures and preventive measures. For example, the server might propose regular harassment prevention training and stress management counseling to a particular department.

[1631] (Application Example 3)

[1632] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1633] Traditional harassment prevention systems have struggled to efficiently categorize and respond quickly to inquiries and reports related to harassment within the company. Furthermore, identifying trends and patterns of harassment and proposing specific countermeasures and preventative measures has been difficult. In particular, early detection and countermeasures for harassment using audio data are needed in employee communication within factories.

[1634] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[1635] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency; means to enable a rapid response and lead to risk hedging; means to convert conversations into text using speech recognition; means to recognize the emotional state of conversations using an emotion engine; means to analyze conversation content using generative AI and detect signs of harassment; and means to propose specific countermeasures and preventive measures against the detected signs of harassment. This enables early detection and countermeasures against harassment in communication among employees within the factory.

[1636] "Generative AI" refers to artificial intelligence technology that generates new information and content based on data.

[1637] "Speech recognition" is a technology that converts speech data into text data.

[1638] An "emotion engine" is a technology that recognizes emotional states from text and audio.

[1639] "Filtering" is the process of selecting data based on specific criteria.

[1640] "Risk hedging" refers to taking measures to mitigate potential risks.

[1641] "Signs of harassment" refer to patterns of behavior and speech that indicate the possibility of harassment occurring.

[1642] "Measures and preventative measures" refer to specific actions and means taken to prevent harassment from occurring.

[1643] "Analyzing conversation content" is the process of analyzing the text data of a conversation to understand its meaning and intent.

[1644] The system for carrying out this invention is configured as an application installed on a robot working in a factory. A specific embodiment is shown below.

[1645] System Configuration

[1646] 1. Speech recognition

[1647] The server uses speech recognition technology to convert conversations within the factory into text data in real time. Specifically, it uses the speech_recognition library.

[1648] 2. Generative AI Models

[1649] The server analyzes text data using a generative AI model to detect signs of harassment. For this purpose, it uses the GPT-3 model from the transformers library.

[1650] 3. Emotional Engine

[1651] The server uses an emotion engine to recognize emotional states from text data. Specifically, it uses the emotion_recognition library.

[1652] 4. Filtering and Risk Hedging

[1653] The server uses generative AI to automatically classify harassment-related inquiries and reports, filtering them according to their importance and urgency. This enables a rapid response and helps mitigate risks.

[1654] 5. Proposals for countermeasures and preventive measures

[1655] The server uses a generative AI model to propose specific countermeasures and preventative measures for detected signs of harassment.

[1656] Processing flow

[1657] 1. Acquisition of audio data

[1658] The terminal (robot) captures conversations within the factory via a microphone.

[1659] 2. Text conversion of audio data

[1660] The server converts the acquired audio data into text data using the speech_recognition library.

[1661] 3. Recognition of emotional state

[1662] The server uses the emotion_recognition library to recognize the emotional state from the text data.

[1663] 4. Detection of signs of harassment

[1664] The server uses a generative AI model (GPT-3) to analyze text data and detect signs of harassment.

[1665] 5. Proposals for countermeasures and preventive measures

[1666] The server uses a generative AI model to propose specific countermeasures and preventative measures for detected signs of harassment.

[1667] Specific example

[1668] For example, suppose the following conversation took place inside the factory:

[1669] Employee A: "You've been working slowly lately. You need to work harder."

[1670] Employee B: "I'm sorry, but I'm not feeling well..."

[1671] This conversation is captured as audio data and input into the program. The server inputs the following prompts to the generative AI model:

[1672] Analyze the following conversation for harassment: "Recently, your work has been slow. You need to work harder." "I'm sorry, I'm not feeling well."

[1673] Based on this prompt, the server analyzes signs of harassment and suggests appropriate countermeasures and preventative measures.

[1674] The flow of the specific processing in Application Example 3 will be explained using Figure 22.

[1675] Step 1:

[1676] The terminal (robot) captures conversations within the factory via a microphone. The input is audio data, and the output is the same audio data. This audio data is processed in the following steps.

[1677] Step 2:

[1678] The server converts the acquired audio data into text data using the speech_recognition library. The input is audio data, and the output is text data. This conversion allows the audio information to be treated as text information.

[1679] Step 3:

[1680] The server uses the emotion_recognition library to recognize emotional states from text data. The input is text data, and the output is an emotional state (e.g., anger, sadness, joy). This process allows for the capture of emotional nuances in a conversation.

[1681] Step 4:

[1682] The server uses a generative AI model (GPT-3) to analyze text data and detect signs of harassment. The input consists of text data and emotional states, and the output is the analysis results regarding signs of harassment. This analysis determines whether a conversation constitutes harassment.

[1683] Step 5:

[1684] The server uses a generative AI model to propose specific countermeasures and preventative measures for detected signs of harassment. The input is the analysis results regarding signs of harassment, and the output is specific countermeasures and preventative measures. This proposal provides appropriate countermeasures.

[1685] Specific example

[1686] For example, suppose the following conversation took place inside the factory:

[1687] Employee A: "You've been working slowly lately. You need to work harder."

[1688] Employee B: "I'm sorry, but I'm not feeling well..."

[1689] This conversation is captured as audio data and input into the program. The server inputs the following prompts to the generative AI model:

[1690] Analyze the following conversation for harassment: "Recently, your work has been slow. You need to work harder." "I'm sorry, I'm not feeling well."

[1691] Based on this prompt, the server analyzes signs of harassment and suggests appropriate countermeasures and preventative measures.

[1692] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1693] The data generation model 58 is a form of so-called generative AI (Artificial Intelligence). One example of the data generation model 58 is ChatGPT (Internet Search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1694] Other examples of generative AI include Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) are some examples.

[1695] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[1696] [Third Embodiment]

[1697] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[1698] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1699] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1700] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[1701] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1702] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1703] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1704] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1705] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[1706] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1707] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1708] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[1709] "Example of form 1"

[1710] One embodiment of the present invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company. Specifically, the generative AI analyzes text data collected through an anonymous reporting channel using a chatbot and automatically classifies the content related to harassment. Furthermore, the results can be filtered according to importance and urgency, enabling prompt action to be taken as needed.

[1711] "Example of form 2"

[1712] Another embodiment of the present invention involves a system that utilizes generative AI to identify harassment trends and patterns, and to visualize the data and generate reports. Specifically, the generative AI extracts harassment trends and patterns based on the results of its analysis and visualizes them. Furthermore, it automatically generates a report based on these results and proposes harassment countermeasures and preventative measures.

[1713] "Example of form 3"

[1714] A further embodiment of the present invention is a system that uses generative AI to propose measures and preventative actions against harassment. Specifically, the system proposes measures and preventative actions against harassment based on the results of analysis by the generative AI. For example, it identifies departments and times of day when harassment is frequent, the type of harassment, etc., and proposes specific measures and preventative actions for these.

[1715] The following describes the processing flow for each example of the form.

[1716] "Example of form 1"

[1717] Step 1: Collect text data from an anonymous reporting channel using a chatbot.

[1718] Step 2: Input the collected text data into a generative AI to automatically classify content related to harassment.

[1719] Step 3: Filter the classification results from the generative AI according to importance and urgency, and take necessary actions quickly.

[1720] "Example of form 2"

[1721] Step 1: Based on the results analyzed by the generative AI, trends and patterns of harassment are extracted.

[1722] Step 2: Visualize the extracted trends and patterns.

[1723] Step 3: Based on the visualized results, an automated report is generated, and harassment prevention and countermeasures are proposed.

[1724] "Example of form 3"

[1725] Step 1: Based on the results of the generative AI analysis, the system proposes measures and preventative actions against harassment.

[1726] Step 2: Identify departments, time slots, and types of harassment that are most frequent.

[1727] Step 3: Propose specific countermeasures and preventative measures for the identified information.

[1728] (Example 1)

[1729] Next, we will describe Embodiment 1 of Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1730] There is a need to efficiently classify inquiries and reports related to harassment within the company and respond quickly according to their importance and urgency. However, traditional manual responses carry the risk of delays in processing reports and the oversight of important reports. Furthermore, it is difficult to identify trends and patterns of harassment and propose countermeasures and preventive measures. To solve these problems, automated classification and filtering of reports, rapid response, data visualization and report generation, and proposals for countermeasures and preventive measures are necessary.

[1731] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1732] This invention includes a server that includes means for collecting anonymous reports using a chatbot, means for transmitting the collected text data to the server, means for analyzing the text data using a generative AI model and automatically classifying content related to harassment, means for filtering the classification results according to importance and urgency, means for notifying the person in charge of the filtered results, means for enabling a quick response and contributing to risk hedging, means for identifying harassment trends and patterns using generative AI, means for visualizing data and generating reports, and means for proposing harassment countermeasures and preventive measures. This makes it possible to efficiently classify inquiries and reports related to harassment within the company and to respond quickly according to importance and urgency.

[1733] A "chatbot" is software designed to automate conversations with users, and it has the ability to collect information and respond through text and voice.

[1734] An "anonymous report" is a report made without revealing the reporter's identity, and is a means of providing information while protecting privacy.

[1735] "Text data" refers to information expressed as a string of characters, including report content and inquiry details.

[1736] A "server" is a computer system that processes and stores data on a network and provides services in response to requests from clients.

[1737] A "generative AI model" is an algorithm or system that uses artificial intelligence technology to analyze data and automatically perform specific tasks.

[1738] "Analysis" is the process of examining data in detail and understanding its structure and meaning.

[1739] Classification is the process of dividing data into specific categories or groups, and it involves grouping data that shares common characteristics.

[1740] "Filtering" is the process of selecting data based on specific criteria and extracting only the necessary information.

[1741] "Notification" refers to a means of informing relevant parties of specific information, and is typically done through email, dashboards, or other means.

[1742] "Prompt response" refers to taking action to deal with problems quickly when they arise, in order to minimize risks.

[1743] "Risk hedging" is the process of mitigating the impact of risks by predicting potential risks and taking countermeasures against them.

[1744] A "trend" refers to a general movement or pattern in which data or events tend to move in a particular direction.

[1745] A "pattern" refers to a repeating structure or arrangement in data or events.

[1746] "Visualization" is the process of making information easier to understand intuitively by representing data visually.

[1747] A "report" is a document that compiles information on a specific topic, including analysis results and recommendations.

[1748] "Countermeasures" refer to specific actions or plans taken to address a particular problem.

[1749] "Preventive measures" refer to the means or actions taken to prevent a problem from occurring before it happens.

[1750] Modes for carrying out the invention

[1751] This invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company and filters them according to their importance and urgency. Specific embodiments of this system are described below.

[1752] Hardware and software to be used

[1753] This system uses the following hardware and software:

[1754] Server: A computer system that processes and stores data. It also functions as an execution environment for generative AI models.

[1755] Device: The device used by the user (PC, smartphone, tablet, etc.).

[1756] Chatbot: Software that automates user interactions. It operates on communication tools such as Slack and Microsoft Teams.

[1757] Generative AI models: Algorithms that analyze text data and automatically perform specific tasks. For example, OpenAI's GPT-4.

[1758] Data processing and data calculation

[1759] This system performs the following data processing and calculations.

[1760] 1. Report Collection: Users can anonymously report harassment through the chatbot. For example, a user might report, "My boss yells at me every day."

[1761] 2. Data transmission: The device sends the text data collected through the chatbot to the server. The server temporarily stores the received text data.

[1762] 3. Text Analysis and Classification: The server inputs the stored text data into the generative AI model. The generative AI model analyzes the text data using the following prompts.

[1763] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[1764] The generative AI model analyzes the text data and classifies it as "power harassment."

[1765] 4. Filtering of Classification Results: The server evaluates the importance and urgency of reports based on the classification results obtained from the generative AI model. For example, a report classified as "power harassment" is judged to be of high importance. The server filters the reports considering factors such as their content, frequency, and the strength of the words used.

[1766] 5. Notification to the responsible party: The server notifies the responsible party based on the filtered results. Notifications are sent, for example, via email or a dedicated management dashboard. The responsible party receives the notification and can take the necessary actions quickly.

[1767] Specific example

[1768] As a concrete example, consider the following scenario.

[1769] scenario:

[1770] A user anonymously reports to a chatbot on Slack that "My boss yells at me every day." This report is sent to a server, where a generative AI model analyzes it.

[1771] Example of a prompt:

[1772] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[1773] The generative AI model receives this prompt and classifies the report as "power harassment." The server then determines the report's importance to be "high" based on this classification and notifies the responsible person via email.

[1774] The above describes specific embodiments for carrying out this invention.

[1775] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1776] Step 1:

[1777] Users can anonymously report harassment through a chatbot. For example, a user might report, "My boss yells at me every day." This report becomes the input data. The chatbot receives this text data and saves the report.

[1778] Step 2:

[1779] The device sends text data collected through the chatbot to the server. The input data is the text data reported by the user. The server temporarily stores the received text data, specifically in a database.

[1780] Step 3:

[1781] The server inputs the stored text data into the generative AI model. The input data is stored text data. The generative AI model parses the text data using the following prompts.

[1782] We received a report stating, "My boss yells at me every day." Please analyze this report and categorize it appropriately.

[1783] The generative AI model analyzes text data and classifies it as "power harassment." The output data is the classification result.

[1784] Step 4:

[1785] The server evaluates the importance and urgency of reports based on classification results obtained from a generative AI model. The input data consists of classification results from the generative AI model. The server filters the reports, considering factors such as content, frequency, and the strength of the words used. The output data is the filtered result.

[1786] Step 5:

[1787] The server notifies the responsible person based on the filtered results. The input data is the filtered results. Notifications are sent, for example, via email or a dedicated management dashboard. The output data is the notification content. The responsible person receives the notification and can take necessary actions quickly.

[1788] The above is a detailed explanation of the program's processing flow.

[1789] (Application Example 1)

[1790] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1791] Traditional harassment reporting systems often required manual classification and assessment of report importance, making prompt responses difficult. Furthermore, they lacked mechanisms to ensure anonymity while appropriately filtering important reports and immediately notifying responsible personnel. This created a risk of delays in the early detection and intervention of harassment.

[1792] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1793] This invention includes a server that uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency, a means for analyzing text data collected through anonymous reporting channels, and a means for immediately notifying the responsible person when an important report is received. This enables a rapid response and contributes to risk hedging.

[1794] "Generative AI" is an artificial intelligence technology that generates new data and information based on input information such as text data and image data.

[1795] "Workplace harassment" refers to inappropriate behavior such as harassment, bullying, and discriminatory acts that occur within the workplace.

[1796] "Inquiries and reports" refer to the act of an employee communicating issues or questions regarding harassment to the company.

[1797] "Automatic classification" refers to the process by which a generative AI analyzes input data and automatically classifies it based on specific categories or importance levels.

[1798] "Filtering based on importance and urgency" refers to the process of selecting data classified by generative AI based on its importance and urgency.

[1799] An "anonymous reporting channel" refers to a means of communication that allows people to report harassment without revealing their identity.

[1800] "Text data" refers to data expressed as character information.

[1801] "Analyzing" refers to the process by which a generative AI examines the input data in detail, understands its content, and classifies it.

[1802] "Immediate notification to the responsible person" refers to the process of immediately notifying a designated person in the event of an important report.

[1803] "Prompt response" refers to taking appropriate measures without delay when a problem arises.

[1804] "Risk hedging" refers to minimizing risks by identifying potential risks in advance and taking countermeasures against them.

[1805] The system for implementing this invention uses generative AI to automatically classify inquiries and reports related to harassment within the company and filters them according to their importance and urgency. Specific embodiments of this system are described below.

[1806] System Configuration

[1807] The system consists of the following main components:

[1808] 1. Server: Runs generative AI models and performs analysis and classification of text data.

[1809] 2. Device: A device (e.g., a smartphone) used by users to report harassment through an anonymous reporting channel.

[1810] 3. Notification System: The responsible person will be immediately notified when there is an important report.

[1811] Hardware and software to be used

[1812] Hardware: Smartphones, servers

[1813] Software: Python, OpenAI API

[1814] Data processing and data calculation

[1815] 1. Text data collection

[1816] Users access an anonymous reporting channel using their smartphones and submit reports of harassment. The reports are sent to the server as text data.

[1817] 2. Generative AI analysis

[1818] The server analyzes the collected text data using the OpenAI API. Specifically, it inputs the text data into the generative AI model using the following prompt messages.

[1819] Example of a prompt:

[1820] Analyze the following text and categorize the content as related to harassment.

[1821] Text: I'm being harassed by my boss. Please help me.

[1822] Classification results:

[1823] The generative AI model analyzes the text data based on this prompt and classifies the content related to harassment.

[1824] 3. Filtering and notifications

[1825] The server determines the importance and urgency of the report based on the analysis results obtained from the generative AI model. If there is an important report, the system immediately notifies the responsible person.

[1826] Specific example

[1827] For example, if a user reports, "My boss is harassing me. Please help," the server inputs this text data into a generative AI model and analyzes it to obtain a classification result such as, "This report is important. The harassment from your boss is ongoing, and urgent action is required." Based on this result, the server immediately notifies the appropriate person, enabling a swift response.

[1828] In this way, a harassment reporting system using generative AI enables a swift and appropriate response, leading to risk hedging.

[1829] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1830] Step 1:

[1831] Users access an anonymous reporting channel using their smartphones to report harassment. Input is the text data entered by the user. Output is the text data sent to the server.

[1832] Step 2:

[1833] The server inputs the received text data into a generative AI model. Specifically, it analyzes the text data using the following prompts. The input is the text data sent by the user. The output is the analysis result by the generative AI model.

[1834] Example of a prompt:

[1835] Analyze the following text and categorize the content as related to harassment.

[1836] Text: I'm being harassed by my boss. Please help me.

[1837] Classification results:

[1838] Step 3:

[1839] The server determines the importance and urgency of the reported content based on the analysis results obtained from the generative AI model. The input is the analysis results from the generative AI model. The output is the filtered results according to importance and urgency.

[1840] Step 4:

[1841] Based on the filtering results, the server immediately notifies the responsible person via the notification system if there is an important report. The input is the filtering results. The output is the notification to the responsible person.

[1842] Step 5:

[1843] The person in charge receives the notification and takes prompt action. The input is the notification from the server. The output is the action taken.

[1844] In this way, a harassment reporting system using generative AI enables a swift and appropriate response, leading to risk hedging.

[1845] (Example 2)

[1846] Next, we will describe Example 2 of the Form Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1847] Traditional harassment prevention systems struggled to efficiently categorize inquiries and reports related to harassment within the company and to respond quickly according to their importance and urgency. Furthermore, identifying harassment trends and patterns and proposing countermeasures and preventative measures based on them was difficult. This resulted in an inability to properly manage harassment risks, leading to delays in improving the work environment within companies.

[1848] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using a generative AI and filtering them according to importance and urgency, means for collecting data from a database, means for preprocessing the collected data, means for analyzing the data using a generative AI model and identifying harassment trends and patterns, means for visualizing the analysis results, and means for automatically generating a report based on the visualized data. This enables efficient identification of harassment trends and patterns, and allows for rapid response and proposal of appropriate countermeasures and preventive measures.

[1849] "Generative AI" refers to a system that uses artificial intelligence technology to generate, analyze, and classify data.

[1850] A "database" is a system for efficiently storing, managing, and retrieving data.

[1851] "Preprocessing" refers to the initial steps taken to prepare data for easier analysis, and specifically includes data cleaning and tokenization.

[1852] A "generative AI model" is a type of generative AI that analyzes data based on specific prompts and generates results.

[1853] A "prompt statement" is an input statement used to give specific instructions or questions to a generative AI model.

[1854] "Visualization" refers to displaying the results of data analysis in a visual format, such as graphs or charts.

[1855] A "report" is a document that summarizes the results of data analysis and proposals.

[1856] "Harassment" refers to inappropriate actions or words directed at others in the workplace or other environments.

[1857] An "inquiry" is a question or request made to seek specific information or support.

[1858] A "report" is a written or oral communication intended to provide information about a specific event or situation.

[1859] "Classification" is the process of grouping data based on specific criteria.

[1860] "Filtering" is the process of selecting data based on specific criteria.

[1861] A "trend" refers to a pattern or trend that repeatedly appears in data.

[1862] A "pattern" is a collection of elements in data that share a specific regularity or commonality.

[1863] "Measures" refer to specific actions or plans taken to address a particular problem.

[1864] "Preventive measures" are specific actions or plans to prevent a problem from occurring before it happens.

[1865] This invention is a system that uses generative AI to automatically classify inquiries and reports related to harassment within a company and filters them according to their importance and urgency. Furthermore, it collects data from a database, performs preprocessing, analyzes the data using a generative AI model, and identifies trends and patterns of harassment. By visualizing the analysis results and automatically generating reports, it can propose harassment countermeasures and preventative measures.

[1866] Hardware and software to be used

[1867] Hardware:

[1868] Server: Database server (e.g., MySQL)

[1869] Terminal: A computer used for displaying data and reviewing reports.

[1870] software:

[1871] Database management system: MySQL

[1872] Natural language processing libraries: NLTK, spaCy

[1873] Generative AI model: GPT-4

[1874] Data visualization tools: Tableau, Matplotlib

[1875] Report generation tools: LaTeX, Microsoft Word

[1876] Explanation of the program's processing

[1877] Data collection:

[1878] The server collects data such as chat logs, emails, and survey results within the company. Specifically, it connects to a database server and retrieves the necessary data using queries. For example, it executes a query like "SELECT FROM chat_logs WHERE date BETWEEN '2023-01-01' AND '2023-01-31'" to retrieve chat logs for January.

[1879] Data preprocessing:

[1880] The server preprocesses the collected data. Specifically, it cleans the text data (removes unnecessary characters and spaces) and tokenizes it (splits it into words and phrases). The preprocessed data is temporarily stored on the terminal. For example, the cleaned text data is saved as a CSV file.

[1881] Analysis using generative AI models:

[1882] The server inputs pre-processed data into a generative AI model (GPT-4) to analyze harassment tendencies and patterns. The user creates the prompts to be input into the generative AI model. For example, the prompt "Identify harassment tendencies from the following chat log." might be used.

[1883] Data visualization:

[1884] The server visualizes harassment trends and patterns based on the results analyzed by the generating AI model. Specifically, it uses Matplotlib to create graphs, displaying time periods where specific keywords frequently appear as bar graphs. The visualized data is then displayed on the terminal.

[1885] Report generation:

[1886] The server automatically generates a "January Harassment Trends Report" using LaTeX based on the visualized data and saves it in PDF format. Users can review the generated report and make corrections or additional comments as needed.

[1887] Examples of specific cases and prompt statements

[1888] Specific example:

[1889] Identify the time periods and departments where specific keywords (e.g., "harassment," "ignoring") frequently appear in company chat logs.

[1890] Based on the survey results, we will extract the names of departments and supervisors that receive the most reports of harassment.

[1891] Example of a prompt:

[1892] "Please identify the harassment tendencies from the following chat logs. In particular, extract frequently occurring keywords and the time periods in which they occurred."

[1893] "Based on these survey results, please identify the departments with the highest number of harassment reports and analyze the causes."

[1894] The above describes the embodiments for carrying out this invention.

[1895] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1896] Step 1: Data Collection

[1897] The server collects data such as chat logs, emails, and survey results from within the company. Specifically, it connects to a database server and retrieves the necessary data using queries. For example, it executes a query like "SELECT FROM chat_logs WHERE date BETWEEN '2023-01-01' AND '2023-01-31'" to retrieve chat logs for January. The input is a database query, and the output is the collected raw data.

[1898] Step 2: Data Preprocessing

[1899] The server preprocesses the collected data. Specifically, it cleans the text data (removes unnecessary characters and spaces) and tokenizes it (splits it into words and phrases). The preprocessed data is temporarily stored on the terminal. For example, the cleaned text data is saved as a CSV file. The input is the collected raw data, and the output is the preprocessed, clean data.

[1900] Step 3: Analysis using a generative AI model

[1901] The server inputs pre-processed data into a generative AI model (GPT-4) to analyze harassment tendencies and patterns. The user creates prompt sentences to input into the generative AI model. For example, the prompt sentence might be, "Identify harassment tendencies from the following chat log." The input consists of pre-processed data and prompt sentences, and the output is the analysis result from the generative AI model.

[1902] Step 4: Visualize the concept

[1903] The server visualizes harassment trends and patterns based on the results analyzed by the generative AI model. Specifically, it uses Matplotlib to create graphs, displaying time periods where specific keywords frequently appear as bar graphs. The visualized data is displayed on the terminal. The input is the analysis results from the generative AI model, and the output is visualized graphs and charts.

[1904] Step 5: Generate the report

[1905] The server automatically generates a "January Harassment Trends Report" using LaTeX based on the visualized data and saves it in PDF format. The user reviews the generated report and makes corrections or additional comments as needed. The input is visualized data, and the output is an automatically generated report.

[1906] (Application Example 2)

[1907] Next, we will describe 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."

[1908] Traditional harassment prevention systems struggle to efficiently categorize harassment-related inquiries and reports within a company and to respond quickly based on their importance and urgency. Furthermore, identifying harassment trends and patterns, visualizing them, and generating reports is time-consuming. Additionally, the lack of features to propose specific countermeasures and preventative measures hinders effective harassment prevention.

[1909] In Application Example 2, the specific processing performed by the specific processing unit 290 of the data processing device 12 is realized by the following means. In this invention, the server includes means for automatically classifying inquiries and reports related to harassment within the company using generative AI and filtering them according to importance and urgency; means for identifying trends and patterns of harassment using generative AI; means for visualizing data and generating reports; means for proposing countermeasures and preventive measures against harassment; means for collecting text data such as reports, chat logs, and emails from users through an application installed on a smartphone; means for identifying trends and patterns of harassment from the collected data using a generative AI model; means for visualizing the analysis results in graphs and charts; and means for automatically generating reports based on the analysis results and proposing specific countermeasures and preventive measures. This makes it possible to efficiently classify inquiries and reports related to harassment and respond quickly, identify and visualize trends and patterns of harassment, and propose specific countermeasures and preventive measures.

[1910] "Generative AI" is an artificial intelligence technology that generates new information and patterns based on data.

[1911] "Harassment" refers to inappropriate words or actions towards others in the workplace or other environments.

[1912] An "inquiry" refers to a question or request made to obtain specific information or support.

[1913] "Reporting" refers to the act of providing information about a specific event or situation.

[1914] "Classification" refers to grouping data or information based on specific criteria.

[1915] "Filtering" refers to the process of selecting data or information based on specific criteria.

[1916] "Tendency" refers to a movement or flow that moves in a particular direction or pattern.

[1917] A "pattern" refers to a specific form or arrangement that is repeated.

[1918] "Visualization" refers to the visual representation of data and information.

[1919] A "report" refers to a document that compiles specific information or analysis results.

[1920] "Measures" refer to specific actions or means taken to address a particular problem or issue.

[1921] "Preventive measures" refer to specific actions or means taken to prevent problems or issues from occurring before they happen.

[1922] A "smartphone" refers to a type of mobile phone that is capable of connecting to the internet and using applications.

[1923] An "application" refers to a software program that provides specific functions or services.

[1924] A "chat log" refers to a record of a conversation conducted in chat format.

[1925] "Email" refers to messages sent and received electronically.

[1926] "Data collection" refers to the act of gathering data for a specific purpose.

[1927] A "graph" refers to a diagram used to visually represent data.

[1928] A "chart" refers to a table or diagram used to visually represent data.

[1929] The system for implementing this invention uses generative AI to automatically classify inquiries and reports related to harassment within the company, and has a function to filter them according to their importance and urgency. It also utilizes generative AI to identify trends and patterns of harassment, visualize the data, and generate reports. Furthermore, it has a function to propose measures and preventative actions against harassment.

[1930] Hardware and software to be used

[1931] Hardware: Smartphones, servers

[1932] Software: Python, OpenAI API, Matplotlib, Pandas

[1933] Data processing and data calculation

[1934] Data collection

[1935] Users input text data such as harassment reports, chat logs, and emails through an application installed on their smartphones. This data is sent to a server and stored in a database.

[1936] AI analysis

[1937] The server inputs the collected data into a generative AI model (e.g., GPT-3) to identify harassment tendencies and patterns. The following prompt messages are used during this process.

[1938] Example of a prompt

[1939] Identify the trends and patterns of harassment from the following text data.

[1940] Text: My boss makes sarcastic remarks to me every day.

[1941] Text: My colleague ignored me

[1942] Visualization of the process

[1943] The server uses Pandas and Matplotlib to visualize the data in graphs and charts based on the analysis results obtained from generative AI models. This allows users to visually confirm harassment trends and patterns.

[1944] Report generation

[1945] The server automatically generates a report based on the analysis results. This report includes harassment trends and patterns, as well as specific countermeasures and preventative measures. The report is output as a text file and made available for users to download.

[1946] Specific example

[1947] For example, if a user reports that "my boss makes sarcastic remarks every day," the generative AI model identifies the pattern as "sarcasm." Based on this information, the server generates a graph showing that there are many reports of sarcasm, and the report suggests countermeasures such as "regular counseling."

[1948] In this way, this invention enables the efficient classification and rapid response of inquiries and reports related to harassment, the identification and visualization of harassment trends and patterns, and the proposal of specific countermeasures and preventive measures.

[1949] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1950] Step 1:

[1951] Users input text data such as harassment reports, chat logs, and emails through an application installed on their smartphones. The input data is then sent to a server by the application. The input data includes specific details of the harassment, as well as the date and time.

[1952] Step 2:

[1953] The server saves the received text data to the database. During this process, it checks the data's format and content and converts it to an appropriate format. For example, it might convert text data to JSON format before saving.

[1954] Step 3:

[1955] The server inputs the stored text data into a generative AI model. The generative AI model (e.g., GPT-3) analyzes the data based on the prompt text and identifies harassment tendencies and patterns. Examples of prompt texts are as follows:

[1956] Example of a prompt

[1957] Identify the trends and patterns of harassment from the following text data.

[1958] Text: My boss makes sarcastic remarks to me every day.

[1959] Text: My colleague ignored me

[1960] Step 4:

[1961] The server receives the analysis results obtained from the generative AI model and stores them in a database. The analysis results include identified harassment tendencies and patterns.

[1962] Step 5:

[1963] The server visualizes the data based on the analysis results. It uses Pandas and Matplotlib to represent the analysis results in graphs and charts. For example, it generates a bar graph showing that there are many sarcastic reports.

[1964] Step 6:

[1965] The server automatically generates reports based on the visualized data. These reports include harassment trends and patterns, as well as specific countermeasures and preventative measures. The generated reports are output as text files.

[1966] Step 7:

[1967] Users can download and view reports generated through a smartphone application. The reports include specific measures, such as "regular counseling."

[1968] In this way, the system can efficiently classify and respond quickly to inquiries and reports related to harassment, identify and visualize harassment trends and patterns, and propose specific countermeasures and preventative measures.

[1969] (Example 3)

[1970] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1971] Traditional harassment prevention systems often involved manual processes for classifying harassment reports and inquiries, filtering them by importance and urgency, and proposing specific countermeasures and preventative measures, making rapid response difficult. Furthermore, there was a lack of means to identify harassment trends and patterns and visualize the data, making it challenging to implement effective countermeasures.

[1972] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[1973] This invention includes a server that uses generative AI to automatically classify and filter inquiries and reports related to harassment within the company according to their importance and urgency; a server that collects harassment data from each department within the company; a server that preprocesses the collected data, including imputing missing values ​​and correcting outliers; a server that inputs the preprocessed data into a generative AI model to identify departments, time periods, and types of harassment that are frequent; a server that proposes specific countermeasures and preventive measures based on the analysis results of the generative AI model; and a server that notifies the user of the proposed measures. This enables rapid classification and filtering of harassment reports and inquiries, and allows for the proposal of effective countermeasures and preventive measures.

[1974] "Generative AI" is an artificial intelligence technology that generates new information and suggestions based on data.

[1975] "Harassment" refers to inappropriate words or actions directed at others in the workplace or society, and is an act that causes mental or physical distress.

[1976] An "inquiry" refers to a question or consultation made to seek specific information or advice.

[1977] A "report" refers to an official notification or record made to convey information about a specific event or situation.

[1978] "Classification" refers to grouping collected data or information based on specific criteria.

[1979] "Filtering" refers to the process of selecting data and information based on specific criteria and extracting only what is necessary.

[1980] "Data collection" refers to the act of gathering information and data necessary for a specific purpose.

[1981] "Preprocessing" refers to the process of preparing data for analysis by performing actions such as imputing missing values ​​and correcting outliers before data analysis is carried out.

[1982] A "generative AI model" refers to a specific implementation of generative AI, specifically an algorithm or program used to generate new information or suggestions based on data.

[1983] "Analysis" refers to the act of examining collected data to identify specific patterns or trends.

[1984] "Countermeasures" refer to specific actions or measures taken in response to a particular problem or issue.

[1985] "Preventive measures" refer to specific actions or steps taken to prevent a particular problem or issue from occurring before it happens.

[1986] "Notification" refers to an official communication or announcement made to convey specific information or proposals to relevant parties.

[1987] This invention is a system that uses generative AI to propose countermeasures and preventative measures against harassment. A specific embodiment of this system is shown below.

[1988] The server first collects data on harassment from various departments within the company. This data includes the number of harassment reports, the time of occurrence, the location, and the type of harassment. For example, it collects data from sales department reports and survey results.

[1989] Next, the server preprocesses the collected data. Specifically, it imputes missing values ​​and detects and co...

Claims

[Claim 1] A means of collecting diverse text data from various departments within a company regarding the frequency, time of occurrence, location, and type of harassment, The collected text data is subjected to preprocessing, including imputation of missing values, detection of outliers, and normalization of the data, and the preprocessed text data is converted into a format suitable for a generative AI. A means for inputting the text data, which has been converted into a format suitable for the generative AI, into the generative AI using specific prompt sentences for analyzing harassment trends, including departments with a high frequency of harassment, time periods with a high frequency of harassment, and types of harassment, to analyze the harassment trends and output the analysis results regarding the trends. A means for inputting emotional information recognized by an emotion engine that analyzes the facial expressions and voices of employees in each department within a company in real time and recognizes emotional information, and data indicating the date, time and location of harassment identified from the collected text data, into a generative AI using specific prompt sentences to analyze whether a particular emotional state of the employee is closely related to the occurrence of harassment, and for analyzing the relationship between the specific emotional state and the occurrence of harassment, and outputting data indicating the relationship, A means for proposing harassment countermeasures and preventive measures by inputting specific prompt sentences into the generation AI, based on the analysis results of the aforementioned trends in harassment and data showing the correlation between the aforementioned specific emotional states and the occurrence of harassment, to propose harassment countermeasures and preventive measures, including at least one of harassment prevention training, counseling for stress management, strengthening of the monitoring system, reduction of workload, and improvement of the harassment reporting system. A means for generating a report that includes the visualized text data, harassment countermeasures, and preventive measures, based on the analysis results regarding the aforementioned trends and data showing the correlation between the aforementioned specific emotional states and the occurrence of harassment, using graphs or charts based on the analysis results regarding the aforementioned trends and the emotional information, and based on the data, based on the visualization results regarding the aforementioned trends and data showing the correlation between the aforementioned specific emotional states and the occurrence of harassment. A system that includes this.

Images