System

Abstract

A system is provided.SOLUTION: A system, comprising: means for collecting historical incident and risk information; means for pre-processing the collected information and generating a machine-learning dataset; means for providing a generative AI model that learns the pre-processed and transformed dataset; means for analyzing documents related to a new project and predicting risk factors using the generative AI model; and means for reporting analysis results and risk predictions.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In software and web service development, it is difficult to properly identify risks and problems that arise during the requirements definition and design stages of new projects based on past experience and data. In particular, applying accident and near-miss information from past projects to new projects requires a great deal of time and resources. Furthermore, overlooking risk factors and failing to take appropriate countermeasures in advance can often hinder project progress. There is a need for an effective system that can resolve these issues and strengthen risk management from the early stages of development. [Means for solving the problem]

[0005] We propose a system that includes a means for collecting past incident data and risk information, a means for preprocessing the collected information and generating a dataset for machine learning, and a means for providing a generative AI model that trains on the preprocessed and transformed dataset. This system has a means for analyzing documents related to new projects when a user uploads them and predicting risk factors using the generative AI model. It also includes a means for notifying users of the analysis results and risk predictions, making it possible to identify omissions and risk factors in advance during the requirements definition and design stages of new projects. Furthermore, the system aims to improve the accuracy of risk factors by including a means for cleansing the collected information and correcting inconsistencies, and a means for analyzing using natural language processing technology.

[0006] "Past incident data" refers to reports and records of accidents and malfunctions that have occurred within a company, and is information that describes the circumstances under which risks have occurred and countermeasures taken.

[0007] "Risk information" refers to information that contains detailed data and knowledge about potential dangers and problems that may arise in a project.

[0008] "Means of collection" refers to the methods and devices used to obtain relevant data (e.g., past incident data and risk information) from inside and outside the company and incorporate it into the system.

[0009] "Preprocessing" refers to the process of converting collected data into a format suitable for analysis and learning, and specifically includes noise removal and standardization of data formats.

[0010] A "machine learning dataset" is a structured collection of data prepared for use in training a generative AI model.

[0011] A "generative AI model" is an artificial intelligence model that has been trained using machine learning algorithms and has the ability to predict specific patterns and risks.

[0012] "Documents related to new projects" refers to documents that contain detailed specifications and plans for a newly launched project, such as requirements definition documents and design documents.

[0013] "Means for analysis" refers to a method or device for analyzing a document to understand its contents and extract necessary information, and often involves the use of natural language processing technology.

[0014] "Means for predicting risk factors" refers to methods and devices for predicting risks that may occur in new projects based on past incident data and risk information.

[0015] "Notification means" refers to a method or device for notifying a user of analysis results or risk predictions, and includes, for example, a function for sending alerts or notification messages.

[0016] "Cleansing" is a part of data preprocessing to remove inaccurate information and noise from collected data and make it consistent.

[0017] "Means for correcting inconsistencies" refers to a method or device for detecting inconsistencies or errors within data and correcting them to make them consistent.

[0018] "Natural language processing technology" is a general term for methods and technologies that use computer science and artificial intelligence techniques to understand, analyze, and manipulate human language. [Brief explanation of the drawings]

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

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

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

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

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

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

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

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

[0027] [First embodiment]

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

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

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

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

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

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

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

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

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

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

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

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

[0040] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[0041] Overall system configuration

[0042] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0043] Data collection

[0044] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0045] Data Preprocessing

[0046] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[0047] Training and updating AI models

[0048] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[0049] New Project Analysis

[0050] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[0051] Risk factor prediction and notification

[0052] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0053] Information supplementation and feedback

[0054] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[0055] Specific examples

[0056] Example 1: Software development project

[0057] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on the generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures.

[0058] Example 2: Service launch project

[0059] User B is currently working on a project to launch a new web service and uploads the design documents to his terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the terminal also displays best practices from successful past projects for User B to refer to.

[0060] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which enables risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken.

[0061] The processing flow will be explained below.

[0062] Step 1:

[0063] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[0064] Step 2:

[0065] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0066] Step 3:

[0067] The server converts the pre-processed data into a machine learning dataset that the generative AI model uses to learn from past incidents and risk patterns. The format and content of the dataset are optimized based on the requirements of the learning algorithm.

[0068] Step 4:

[0069] The server uses the generative AI model to perform machine learning. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[0070] Step 5:

[0071] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[0072] Step 6:

[0073] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and semantically analyze the text data to extract important, relevant information.

[0074] Step 7:

[0075] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns, thereby predicting potential risk factors that may arise in new projects.

[0076] Step 8:

[0077] The server sends the identified risk factors and proposed countermeasures to the terminal, including detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[0078] Step 9:

[0079] The terminal displays the received risk information on the user interface. Users can use this information to manage project risks and consider and implement necessary countermeasures.

[0080] Step 10:

[0081] Users can input their feedback on the risk information and advice provided by the system into their devices and send it to the server, which uses this feedback to improve the accuracy of the AI ​​model and the system.

[0082] In this way, by performing specific actions at each step, the system utilizes past data to predict and identify risk factors in new projects in advance.

[0083] Example 1

[0084] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0085] Conventional risk management systems are unable to fully utilize past incident data and risk information, making it highly likely that similar risks will reoccur in new projects. Furthermore, the accuracy of predicting risk factors is low, making it difficult to implement appropriate countermeasures. Furthermore, analyzing documents for new projects takes a lot of time and effort, placing a heavy burden on users.

[0086] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0087] In this invention, the server includes means for collecting past incident data and risk information, means for cleansing the collected information and generating a machine learning dataset, means for training a generative AI model using a machine learning algorithm, means for uploading and analyzing documents related to new projects, means for predicting risk factors using the generative AI model and generating analysis results, means for notifying users of the analysis results and risk predictions, and means for collecting feedback from users and using the feedback to improve the accuracy of the AI ​​model. This makes it possible to predict risk factors in new projects with high accuracy by utilizing past data and to quickly take appropriate measures.

[0088] "Past incident data" refers to data on past accidents and problems recorded in accident reports, near-miss information, issue management sheets, etc.

[0089] "Risk information" refers to information about risk factors and risk cases related to projects and operations.

[0090] "Cleansing" is the process of detecting, correcting, or removing inconsistencies and missing values ​​in collected data.

[0091] A "machine learning dataset" is a collection of data that has been preprocessed to train a generative AI model.

[0092] A "generative AI model" is an artificial intelligence model that is trained using machine learning algorithms to extract specific risk patterns and make predictions.

[0093] "Natural Language Processing (NLP)" is a technology for analyzing text data and understanding and processing human language.

[0094] "New project documents" are documents such as requirements definition documents and design documents related to a newly started project.

[0095] "Risk factors" are elements or conditions that could cause problems or obstacles in the progress of a project.

[0096] "User notification" is the process of informing users of analysis results and risk predictions.

[0097] "Feedback" refers to information received from users, such as opinions and evaluations, that is used to improve the system and generative AI models.

[0098] A "machine learning algorithm" is an algorithm that learns patterns and rules from data and makes predictions about new data.

[0099] "Retraining" is the process of adding new data to an existing model to improve its accuracy and performance.

[0100] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[0101] Hardware and software used

[0102] This system mainly consists of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the generative AI model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. The user is responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0103] Data collection

[0104] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0105] Data Preprocessing

[0106] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is fed into a generative AI model as a machine learning dataset.

[0107] Training and updating AI models

[0108] The server uses machine learning algorithms to train a generative AI model, which is designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the model is retrained to improve its accuracy.

[0109] New Project Analysis

[0110] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server receives these documents and uses natural language processing (NLP) technology to convert the text data into an analyzable format.

[0111] Risk factor prediction and notification

[0112] The server applies a generative AI model based on the analyzed new project documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0113] Information supplementation and feedback

[0114] The server extracts relevant information from internal bulletin boards and notifies users of posts discussing similar risks. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[0115] Specific examples

[0116] Example 1: Software development project

[0117] User A starts a new software development project and uploads a requirements specification document to a device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, prompting User A to consider countermeasures.

[0118] Example 2: Service launch project

[0119] User B is currently working on a project to launch a new web service and uploads the design documents to his / her terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out "database schema design risks" to User B and provides specific design improvement proposals. Along with the design improvement proposals, the terminal also displays best practices from successful past projects for User B to refer to.

[0120] Prompt Sentence Examples

[0121] Below are some examples of prompts that users may use when using the system:

[0122] 1. "I've uploaded a requirements document for a new software development project. What are the risk factors?"

[0123] 2. "I have uploaded the design document for a web service launch project. Please tell me about past near misses and risk factors."

[0124] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which makes it possible to detect risks in advance from the early stages of development and take appropriate measures.

[0125] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0126] Step 1: Data collection

[0127] The server collects past incident data and risk information from various sources within the company. Specifically, the server accesses accident reports, near-miss information, issue management tables, and data posted on internal bulletin boards via the internal network. The data obtained from each source is stored in a database. This process is carried out using a crawler and API.

[0128] Input: Historical incident data and risk information from the company's internal network

[0129] Output: Raw data stored in a database

[0130] Step 2: Data cleansing

[0131] The server cleanses the collected data. Specifically, it detects and corrects inconsistencies and missing values. It removes inaccurate data, adds necessary metadata, and standardizes the data format. At this stage, text data is cleaned and numerical data is standardized.

[0132] Input: Raw data stored in a database

[0133] Output: Cleansed data

[0134] Step 3: Dataset generation

[0135] The server then generates a machine learning dataset from the cleansed data, which is then used to train the AI ​​model. This process also involves standardizing the data and extracting features.

[0136] Input: Cleansed data

[0137] Output: Dataset for machine learning

[0138] Step 4: Model training

[0139] The server uses the generated dataset to train the generative AI model. Specifically, it uses machine learning algorithms (e.g., random forest, deep learning) to learn from past incidents and risk information. The model is designed to extract specific risk patterns.

[0140] Input: Machine learning dataset

[0141] Output: A trained generative AI model

[0142] Step 5: Update the model

[0143] As new incident data is added, the server retrains the AI ​​model, which continuously improves its accuracy. The model update process is automated, allowing it to quickly reflect the impact of new data.

[0144] Input: New incident data added

[0145] Output: An updated generative AI model

[0146] Step 6: Upload documents

[0147] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The device supports document uploading by providing an interface that users can easily operate.

[0148] Input: Documents related to the new project (requirements specification and design documents)

[0149] Output: The document uploaded to the server

[0150] Step 7: Document Analysis

[0151] The server receives the uploaded document and uses natural language processing (NLP) techniques to convert the text data into an analyzable format, specifically tokenizing the document content and extracting important keywords and phrases.

[0152] Input: New project documentation

[0153] Output: Parsed text data

[0154] Step 8: Risk prediction

[0155] The server uses a generative AI model to analyze the uploaded new project documents, match them with similar past cases, and identify specific risk factors and omissions. The generative AI generates risk prediction results and creates a risk report containing specific risk factors and countermeasures.

[0156] Input: Analyzed text data, trained generative AI model

[0157] Output: Risk Report

[0158] Step 9: User Notification

[0159] The created risk report is sent to the device and notified to the user. When the device receives the notification, it displays specific risk factors and countermeasures to the user. For example, it notifies the user of security risks and displays a related configuration checklist.

[0160] Input: Risk Report

[0161] Output: Risk factors and countermeasures displayed to the user

[0162] Step 10: Supplementary Information Notification

[0163] The server extracts relevant information from internal bulletin boards and notifies users of posts where similar risks are being discussed, allowing users to refer to past experiences and discussions to strengthen risk management for new projects.

[0164] Input: Risk report, related information on internal bulletin board

[0165] Output: Additional information provided to the user

[0166] Step 11: User Feedback

[0167] Users input feedback on the risk information and advice provided by the system. The feedback is sent to the server via their device. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[0168] Input: User feedback

[0169] Output: A generative AI model that reflects feedback and improves accuracy, and an improved system

[0170] (Application example 1)

[0171] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0172] In modern companies, it is extremely important to consider past risk factors and take proactive measures when launching a new project. However, manually collecting and analyzing risk information is extremely time-consuming and inefficient. Furthermore, project managers lack the means to check risk factors in real time while working in the field or traveling, making it difficult to respond quickly to new security risks. The objective of this invention is to solve these problems and strengthen security risk management within companies.

[0173] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0174] In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the user of the analysis results and risk predictions, means for a user to upload documents related to the new project from the smart glasses, and means for displaying the analysis results and risk predictions on the smart glasses in real time. This makes it possible to efficiently utilize past risk information, grasp project risks in real time using the smart glasses, and respond quickly.

[0175] "Past incident data" refers to recorded information about accidents or problems that have occurred in the past within a company.

[0176] "Risk information" is information about the potential dangers or problems that may arise from a particular situation or action.

[0177] "Means of collection" refers to the methods and tools used to gather the necessary data and information.

[0178] "Preprocessing" refers to data cleansing and transformation performed on raw data to prepare it in an analyzable format.

[0179] A "machine learning dataset" is a set of data used to train and evaluate machine learning models.

[0180] A "generative AI model" is an artificial intelligence model that generates new data or predictions based on given data.

[0181] "Means for analyzing documents" refers to methods and tools for analyzing project documents and extracting necessary information.

[0182] "Means for predicting risk factors" refers to methods and tools that use past data to detect risks that may lurk in new projects in advance.

[0183] "Means of notification" refers to methods and tools for conveying analysis results and important information to users.

[0184] "Means for uploading from smart glasses" refers to methods or tools for using smart glasses to send data or documents to a system such as a server.

[0185] "Real-time display means" refers to methods or tools that provide information or analytical results to the user's view instantly.

[0186] This invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing this invention are described below.

[0187] Overall system configuration

[0188] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0189] Data collection

[0190] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0191] Data Preprocessing

[0192] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[0193] Training and updating AI models

[0194] The server uses machine learning algorithms to train a generative AI model, which is designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the model is retrained to improve its accuracy.

[0195] New Project Analysis

[0196] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[0197] Risk factor prediction and notification

[0198] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0199] Use of smart glasses

[0200] Users can upload documents related to new projects using the smart glasses. The smart glasses have the ability to display analysis results and risk predictions to users in real time, allowing users to check risk information at any time even while the project is in progress and take immediate action.

[0201] Information supplementation and feedback

[0202] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[0203] Specific examples

[0204] Example 1: Security Services Project

[0205] A user starts a new security service project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts the user to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, prompting the user to consider countermeasures.

[0206] Prompt Sentence Examples

[0207] "Based on the requirements specification for the new system, please analyze past security incident data and identify new risks."

[0208] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0209] Step 1:

[0210] The server collects past incident data and risk information from within the company. It uses information sources such as accident reports, near-miss information, issue management tables, and internal bulletin boards as input and stores it in a database. The collected data is obtained as output.

[0211] Step 2:

[0212] The server cleanses the collected data and corrects inconsistencies. It takes the collected data as input, removes inaccurate data, standardizes the data format, and adds necessary metadata. The output is a pre-processed, consistent dataset.

[0213] Step 3:

[0214] The server uses a machine learning algorithm to input the preprocessed dataset into a generative AI model to train the model. Using the preprocessed dataset as input, the server learns from past incidents and risk information to extract specific risk patterns. The output is a trained generative AI model.

[0215] Step 4:

[0216] The user uploads documents such as requirements and design documents for a new project from their terminal. The project documents are used as input and sent to the server via the terminal. The documents uploaded to the server are obtained as output.

[0217] Step 5:

[0218] The server analyzes the uploaded documents of new projects. It receives the uploaded documents as input and uses natural language processing (NLP) technology to convert the text data into an understandable format. The output is the analyzed document data.

[0219] Step 6:

[0220] The server uses a generative AI model to predict risk factors based on the analyzed documents. It uses the analyzed document data and the trained generative AI model as input and matches it with historical risk data. The output is the identified risk factors and omission points.

[0221] Step 7:

[0222] The server notifies the terminal of the identified risk factors and omissions. Using the identified risk information as input, it notifies the user of specific risk factors and proposed countermeasures. The notification content is displayed on the terminal as output.

[0223] Step 8:

[0224] Users use smart glasses to upload documents related to new projects and check analysis results and risk predictions. The documents sent from the smart glasses as input are uploaded to the server, and the analysis results are displayed in real time. Real-time risk information is displayed on the smart glasses as output.

[0225] Step 9:

[0226] The user inputs feedback on the risk information and advice provided into the device and sends it to the server. The feedback on the risk information is used as input and sent to the server. The feedback is saved as output on the server and used to improve the AI ​​model in the future.

[0227] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0228] The present invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, to predict and identify risk factors in new projects in advance. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the quality of notifications and feedback to the user is improved. Specific embodiments for implementing the present invention are described below.

[0229] Overall system configuration

[0230] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system. The emotion engine also analyzes and recognizes user emotions and provides optimized notifications and feedback.

[0231] Data collection

[0232] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[0233] Data Preprocessing

[0234] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0235] Training and updating AI models

[0236] The server converts the preprocessed data into a machine learning dataset and performs machine learning using a generative AI model. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[0237] New Project Analysis

[0238] A user starts a new project and uploads requirements and design documents from their device. The device then sends these documents to the server. The server then analyzes the new project documents, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and analyze semantics to extract important, relevant information.

[0239] Risk factor prediction and notification

[0240] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns. This predicts potential risk factors for new projects. The server then sends the identified risk factors and proposed countermeasures to the terminal. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[0241] Emotion engine analysis and notification adjustment

[0242] The server uses an emotion engine to analyze the user's input data and behavior to recognize the user's emotions. Based on this information, the server adjusts the content and format of the risk notification to present it in a way that is easier for the user to understand. For example, if the emotion engine determines that the user's stress level is high, the server will simplify the notification content and present only the main points to avoid burdening the user.

[0243] Feedback and System Improvement

[0244] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[0245] Specific examples

[0246] Example 1: Software development project

[0247] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently been subject to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures. Furthermore, the emotion engine recognizes User A's stress level and adjusts the content of notifications in a simple manner.

[0248] Example 2: Service launch project

[0249] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. Furthermore, the emotion engine recognizes User B's emotions and adjusts the feedback content appropriately.

[0250] As described above, this invention provides a concrete system for strengthening risk management in new projects by utilizing past data. This allows risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken. Furthermore, by combining it with an emotion engine, it is possible to improve the quality of notifications and feedback to users.

[0251] The processing flow will be explained below.

[0252] Step 1:

[0253] The server periodically collects past incident data and risk information from within the company, including accident reports, near-miss information, issue management tables, and posts on internal bulletin boards. The collected data is structured and stored in a database.

[0254] Step 2:

[0255] The server preprocesses the collected data. The data cleansing process involves removing inaccurate information, standardizing data formats, and correcting inconsistencies. For example, it standardizes data formats from different sources and fills in missing values.

[0256] Step 3:

[0257] The server converts the preprocessed data into a machine learning dataset, which is then optimized for model training using feature engineering techniques.

[0258] Step 4:

[0259] The server trains the generative AI model. This process uses the training data to apply a learning algorithm to extract and learn from historical risk patterns. It evaluates the model's accuracy and adjusts hyperparameters or retrains it as needed.

[0260] Step 5:

[0261] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[0262] Step 6:

[0263] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag, and semantically analyze the text to extract key information.

[0264] Step 7:

[0265] The server inputs the analyzed project data into the generative AI model and compares it with past risk patterns, thereby predicting potential risk factors for new projects.

[0266] Step 8:

[0267] The server sends the identified risk factors and proposed countermeasures to the terminal, generating a report that includes detailed information about the risk, examples of similar cases in the past, and countermeasures.

[0268] Step 9:

[0269] The terminal displays the received risk information and countermeasures on the user interface, allowing the user to manage project risks based on this information.

[0270] Step 10:

[0271] The server uses an emotion engine to analyze the user's input data and behavioral data to recognize the user's emotions, such as the user's typing speed and word usage patterns.

[0272] Step 11:

[0273] The server adjusts the content of risk notifications based on the perceived user emotion: for example, if the user is feeling stressed, it will briefly summarize the notification and highlight only the important points.

[0274] Step 12:

[0275] Users input feedback on the risk information and countermeasures provided by the system into their terminal and send it to the server.

[0276] Step 13:

[0277] The server analyzes user feedback and sentiment data to help improve the generative AI model and system. The feedback is used to update the AI ​​model's training data, improving the overall accuracy and usability of the system.

[0278] Specific examples

[0279] Example 1: Software development project

[0280] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. Furthermore, the emotion engine recognizes User A's stress level and adjusts the notification content concisely. The device displays the checklist as well as examples of related past projects, urging User A to take appropriate measures.

[0281] Example 2: Service launch project

[0282] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss incidents from past service launches. As a result, it is discovered that there were numerous errors in the database schema design. The server points out "risks in the database schema design" to User B and provides specific design improvement suggestions. Furthermore, the emotion engine recognizes User B's emotions and adjusts appropriate risk notifications. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. In this way, the system strengthens risk management for new projects and provides information that takes user emotions into consideration.

[0283] Example 2

[0284] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0285] When starting a new project within a company, past incident data and risk information are not fully utilized, making it difficult to predict and address risk factors in advance.Furthermore, notifications and feedback that do not take user emotions into consideration place an excessive burden on users.

[0286] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for analyzing user emotions using an emotion analysis engine and adjusting the notification content based on the analysis results, and means for collecting feedback from users and improving the AI ​​model and the overall system. This makes it possible to predict risk factors for new projects in advance and take appropriate measures, as well as provide notifications and feedback that take user emotions into consideration.

[0287] "Incident data" refers to detailed information and records of accidents and troubles that have occurred in the past within a company.

[0288] "Risk information" is information about potential dangers or problems that have been recognized in the past within a company.

[0289] "Data collection means" refers to an element that has the function of periodically collecting relevant data from within the company and from each department and storing it in a database.

[0290] The "data preprocessing means" is an element that has the function of cleansing collected data, converting it into a unified format, and adding metadata to it.

[0291] A "machine learning dataset" is a set of data generated from preprocessed data and used to train machine learning algorithms.

[0292] A "generative AI model" is an artificial intelligence model that has been trained using machine learning algorithms and has the ability to predict risk factors based on new data.

[0293] The "document analysis means" is an element that has the function of analyzing documents such as requirements definition documents and design documents related to new projects and extracting important information.

[0294] "Natural language processing technology" is a technology for analyzing human language using a computer, and includes processes such as tokenization, part-of-speech tagging, and semantic analysis.

[0295] The "risk prediction means" is an element that inputs analyzed document data into a generative AI model and has the function of identifying and predicting risk factors.

[0296] The "notification means" is an element that has the function of informing the user of predicted risks and countermeasures.

[0297] An "emotion analysis engine" is a technology for analyzing and recognizing a user's emotions, and determines the user's emotional state based on the user's input data and behavior.

[0298] A "feedback collection means" is an element that has the function of collecting opinions and reactions from users and using them to improve the system or retrain the AI ​​model.

[0299] This invention is a system that uses a generative AI model to collect and analyze information on past accidents and near misses that have occurred within a company, as well as issue management tables, and to predict and identify risk factors in new projects in advance. Furthermore, by combining this with an emotion analysis engine that recognizes user emotions, the quality of notifications and feedback provided to users can be improved. A specific embodiment of this system is described below.

[0300] Data collection

[0301] The server collects past incident data and risk information from various data sources within the company. Specifically, it periodically scans internal bulletin boards, accident reports from each department, near-miss information, and issue management tables, and stores new data in a database. This allows the collected data to be centrally managed for analysis.

[0302] Data Preprocessing

[0303] The server performs preprocessing on the collected data. For example, it standardizes data in different formats, deletes incomplete data, and adds metadata. For example, if reports from different departments have different formats, standardizing these formats improves the accuracy of analysis.

[0304] Training an AI model

[0305] The server generates a machine learning dataset based on the preprocessed data. Generative AI models are used to extract risk patterns from past data and build learning models. During this process, algorithms such as neural networks are used to analyze the data's characteristics and learn risk factors. The accuracy of the models is regularly evaluated, and they are retrained as necessary.

[0306] Uploading documents for a new project

[0307] When starting a new project, users upload requirements and design documents from their devices, which then send these documents to the server.

[0308] Document Analysis

[0309] The server receives new project documents and uses natural language processing (NLP) techniques to analyze the text data, including tokenization, part-of-speech tagging, and semantic analysis, to extract key relevant information. In the process, a generative AI model is used to identify relevant risk factors.

[0310] Risk prediction and notification

[0311] The server inputs the analyzed data into a generative AI model and compares it with past risk patterns to predict risk factors. The predicted risk is sent to the device along with detailed information. The device then notifies the user based on this information and provides specific countermeasures. For example, a notification such as "There is a high security risk. Here is a specific configuration checklist" may be displayed.

[0312] Use of sentiment analysis engine

[0313] The server uses an emotion analysis engine to analyze the user's input data and behavior to recognize the user's emotions. Based on the analysis results, the server adjusts the content and format of notifications. For example, if the user is in a high stress state, the server will shorten the notification content and present only the main points.

[0314] Gathering feedback and improving the system

[0315] Users can input feedback on the risk information and advice provided into their devices and send it to the server, which analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and the overall system.

[0316] Specific examples

[0317] Example 1: Software development project

[0318] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it using a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, encouraging User A to consider countermeasures. The sentiment analysis engine recognizes User A's stress level and adjusts the notification content in a simple manner.

[0319] Example 2: Service launch project

[0320] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "risks in the database schema design" to User B and provides specific suggestions for improving the design. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. The sentiment analysis engine recognizes User B's emotions and adjusts the feedback content appropriately.

[0321] Prompt Sentence Examples

[0322] "Please upload the requirements specification for the new project and notify us of the results of the risk analysis."

[0323] As described above, this system utilizes past data to strengthen risk management in new projects. By combining it with a sentiment analysis engine, it is possible to improve the quality of notifications and feedback to users.

[0324] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0325] Step 1:

[0326] The server collects incident data and risk information from various data sources within the company. Input data includes data from internal bulletin boards, accident reports from each department, near-miss information, and issue management tables. This data is scanned periodically, and new data is saved in a database. The output after collection is incident data and risk information that is centrally managed in a database. Specifically, for example, the server runs a script every night to collect information.

[0327] Step 2:

[0328] The server preprocesses the data collected. The input is the collected incident data and risk information. Preprocessing involves removing incomplete information, unifying different formats, and adding metadata. For example, if reports from different departments have different formats, they are converted into a unified format. The output is the preprocessed data. Specifically, the data is cleansed using a cleaning script and metadata is added.

[0329] Step 3:

[0330] The server generates a machine learning dataset based on preprocessed data. The input is the preprocessed data. This dataset is fed to a generative AI model, which performs learning to extract risk patterns. The algorithms used include neural networks. The output is a trained generative AI model. Specifically, the machine learning algorithm is executed, and the model parameters are adjusted based on the evaluation results.

[0331] Step 4:

[0332] A user starts a new project and uploads documents such as requirements specifications and design documents to a terminal. The input is documents related to the new project. The terminal sends these to the server. The output is the documents received by the server. Specifically, the user selects a file from the terminal and presses the upload button.

[0333] Step 5:

[0334] The server receives the new project document and analyzes the text data using natural language processing techniques. The input is the document uploaded by the user. This analysis involves tokenization, part-of-speech tagging, and semantic analysis to extract important relevant information. The output is the analyzed text data. Specifically, the NLP engine is used to perform the text analysis task.

[0335] Step 6:

[0336] The server inputs the analysis results into a generative AI model to predict risk factors. The input is the analyzed text data. The generative AI model compares it with past risk patterns to identify risk factors for new projects. The output is the predicted risk factors and their detailed information. Specifically, the data is supplied to the model and an algorithm is executed to extract risk factors.

[0337] Step 7:

[0338] The server notifies the terminal of the predicted risk factors and their proposed countermeasures. The input is the predicted risk factors and their detailed information. The notification includes details of the risk, similar past cases, and recommended countermeasures. The output is a notification that is displayed to the user. Specifically, the server generates a notification message and sends it to the terminal.

[0339] Step 8:

[0340] The server uses an emotion analysis engine to analyze the user's input data and behavior and recognize the user's emotions. The input is the user's operation log and feedback. Based on the analysis results, the server adjusts the notification content and format. The output is the adjusted notification and feedback. Specifically, the server runs an emotion analysis algorithm and dynamically changes the notification content.

[0341] Step 9:

[0342] The user inputs feedback on the risk information and advice provided into the terminal and sends it to the server. The input is feedback from the user. The server analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and improve the overall system. The output is an improved AI model and system. Specifically, it runs a feedback analysis algorithm and adjusts the model and system.

[0343] The above are the processing steps of this program.

[0344] (Application example 2)

[0345] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0346] Current risk management systems often refer to past accidents and issue management tables, making it difficult to identify potential risks in new projects in advance. Furthermore, adopting a uniform notification method without considering user feelings can easily lead to user stress and a lack of understanding. Especially for industrial equipment, where incorrect operation or configuration can lead to serious accidents, more accurate risk prediction and appropriate notification methods are needed.

[0347] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for recognizing user emotions and adjusting the notification content, and means for providing risk information and countermeasures to industrial equipment. This makes it possible to identify risk factors in a new project in advance and provide feedback to the user using an appropriate notification method.

[0348] "Incident data" refers to detailed information about accidents and troubles that have occurred within a company in the past.

[0349] "Risk information" refers to data about potential hazards or issues associated with a particular project or task.

[0350] "Machine Learning Dataset" refers to a collection of processed and pre-processed data used to train a generative AI model.

[0351] "Generative AI model" refers to an AI system that uses machine learning algorithms to learn risk patterns from historical data and predict risks for new projects.

[0352] "Documents related to a new project" refers to documents that contain details of the project, such as requirements specifications and design documents for a new project.

[0353] "Natural language processing technology" refers to computer science techniques for analyzing and processing human language.

[0354] "Means for recognizing user emotions" refers to tools and algorithms that analyze emotions based on user input and behavior and reflect them in the system.

[0355] "Means for providing risk information and countermeasures" refers to methods and tools for notifying users and industrial equipment of effective risk countermeasures based on the analysis results.

[0356] The system for carrying out the present invention includes the following series of means: The system is mainly composed of three main entities: a server, a terminal, and a user.

[0357] System Overview

[0358] The server collects and preprocesses past incident data and risk information. It uses the preprocessed data set to train and update the generative AI model, analyzes new project documents, and predicts risk factors. It then notifies users of risk prediction results and countermeasures via their devices, helping them to smoothly progress with the project. It also uses an emotion engine to analyze users' emotions and adjust the content of notifications.

[0359] The terminal provides an interface for users to upload documents related to new projects and check the analysis results, allowing users to effectively manage projects through this system.

[0360] Program processing description

[0361] Data collection and preprocessing

[0362] The server collects data from past accident reports, near-miss information, and issue management tables stored within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database and undergoes data cleansing. Data cleansing involves deleting incomplete information, standardizing data formats, and removing unnecessary data. Inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0363] Training and updating generative AI models

[0364] The server generates a machine learning dataset based on the preprocessed data and trains a generative AI model. This model uses machine learning algorithms such as neural networks (using TensorFlow or PyTorch) to extract and learn risk patterns from past data. The accuracy of the model is evaluated and retrained as necessary.

[0365] New project analysis and risk prediction

[0366] When a user starts a new project, they use their device to upload documents such as requirements specifications and design documents to the server. The server then analyzes these documents using Natural Language Processing (NLP) technology. Natural language processing libraries such as spaCy and NLTK are used for analysis, tokenizing the text data, tagging parts of speech, and performing semantic analysis. A generative AI model compares the results with past risk patterns and predicts potential risk factors that may arise in the new project.

[0367] Analysis result notification and emotion engine

[0368] Based on the analysis results, the server sends risk factors and proposed countermeasures to the device. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures. The emotion engine analyzes the user's input and behavior to recognize their emotions. Based on this information, the content and format of the risk notification can be adjusted. For example, if the user is feeling stressed, the notification content will be simplified and only the main points will be presented.

[0369] Feedback and System Improvement

[0370] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[0371] Adding specific examples

[0372] Specific prompt examples

[0373] "We have added configuration information for a new machine. Please analyze the risks associated with this."

[0374] "We've added a safety procedure to this project. What risks have you encountered with similar materials in the past?"

[0375] This allows the system to proactively identify risk factors related to industrial equipment and provide feedback to users through appropriate notification methods.

[0376] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0377] Step 1:

[0378] The server collects past accident reports, near-miss information, and issue management tables from within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database. The specific input is raw data from each source, and the output is data stored in the database in a unified format.

[0379] Step 2:

[0380] The server preprocesses the collected data. It performs data cleansing to remove incomplete information, standardize data formats, and remove unnecessary data. For example, it corrects inconsistencies in data formats from different departments. The input is raw data stored in the database, and the output is cleansed data.

[0381] Step 3:

[0382] The server generates a machine learning dataset based on the preprocessed data. This dataset is input to a generative AI model and used to train the AI ​​model to extract and learn risk patterns from past data. The input is cleansed data, and the output is a machine learning dataset.

[0383] Step 4:

[0384] When a user starts a new project, the server provides a function to upload requirements and design documents through the terminal. The user uploads documents related to the project to the terminal. The input is the user's document, and the output is the document sent to the server.

[0385] Step 5:

[0386] The server analyzes the uploaded document using Natural Language Processing (NLP) techniques. It tokenizes the text data, tags it with parts of speech, and performs semantic analysis to extract important relevant information. The input is the document uploaded by the user, and the output is the analysis results.

[0387] Step 6:

[0388] The server inputs the analysis results into a generative AI model to predict potential risk factors for new projects. This involves comparing past risk patterns with current project data. The input is the analysis results data, and the output is risk prediction information.

[0389] Step 7:

[0390] The server notifies the user based on the risk prediction information. At this time, it uses an emotion engine to analyze the user's emotions and adjusts the notification format and content to an appropriate one according to the emotion. For example, if the user is feeling stressed, it selects a concise notification content. The input is the risk prediction information and the user's emotional data, and the output is the adjusted notification content.

[0391] Step 8:

[0392] The terminal provides the user with risk prediction information and countermeasures sent from the server. The user can use this information to adjust the progress of the project. The input is the risk prediction information sent from the server, and the output is the risk information and countermeasures displayed to the user.

[0393] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0394] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0395] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0396] [Second embodiment]

[0397] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0398] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0399] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0400] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0401] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0402] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0403] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0404] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0405] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0406] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0407] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0408] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0409] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[0410] Overall system configuration

[0411] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0412] Data collection

[0413] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0414] Data Preprocessing

[0415] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[0416] Training and updating AI models

[0417] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[0418] New Project Analysis

[0419] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[0420] Risk factor prediction and notification

[0421] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0422] Information supplementation and feedback

[0423] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[0424] Specific examples

[0425] Example 1: Software development project

[0426] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on the generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures.

[0427] Example 2: Service launch project

[0428] User B is currently working on a project to launch a new web service and uploads the design documents to his terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the terminal also displays best practices from successful past projects for User B to refer to.

[0429] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which enables risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken.

[0430] The processing flow will be explained below.

[0431] Step 1:

[0432] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[0433] Step 2:

[0434] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0435] Step 3:

[0436] The server converts the pre-processed data into a machine learning dataset that the generative AI model uses to learn from past incidents and risk patterns. The format and content of the dataset are optimized based on the requirements of the learning algorithm.

[0437] Step 4:

[0438] The server uses the generative AI model to perform machine learning. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[0439] Step 5:

[0440] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[0441] Step 6:

[0442] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and semantically analyze the text data to extract important, relevant information.

[0443] Step 7:

[0444] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns, thereby predicting potential risk factors that may arise in new projects.

[0445] Step 8:

[0446] The server sends the identified risk factors and proposed countermeasures to the terminal, including detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[0447] Step 9:

[0448] The terminal displays the received risk information on the user interface. Users can use this information to manage project risks and consider and implement necessary countermeasures.

[0449] Step 10:

[0450] Users can input their feedback on the risk information and advice provided by the system into their devices and send it to the server, which uses this feedback to improve the accuracy of the AI ​​model and the system.

[0451] In this way, by performing specific actions at each step, the system utilizes past data to predict and identify risk factors in new projects in advance.

[0452] Example 1

[0453] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0454] Conventional risk management systems are unable to fully utilize past incident data and risk information, making it highly likely that similar risks will reoccur in new projects. Furthermore, the accuracy of predicting risk factors is low, making it difficult to implement appropriate countermeasures. Furthermore, analyzing documents for new projects takes a lot of time and effort, placing a heavy burden on users.

[0455] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0456] In this invention, the server includes means for collecting past incident data and risk information, means for cleansing the collected information and generating a machine learning dataset, means for training a generative AI model using a machine learning algorithm, means for uploading and analyzing documents related to new projects, means for predicting risk factors using the generative AI model and generating analysis results, means for notifying users of the analysis results and risk predictions, and means for collecting feedback from users and using the feedback to improve the accuracy of the AI ​​model. This makes it possible to predict risk factors in new projects with high accuracy by utilizing past data and to quickly take appropriate measures.

[0457] "Past incident data" refers to data on past accidents and problems recorded in accident reports, near-miss information, issue management sheets, etc.

[0458] "Risk information" refers to information about risk factors and risk cases related to projects and operations.

[0459] "Cleansing" is the process of detecting, correcting, or removing inconsistencies and missing values ​​in collected data.

[0460] A "machine learning dataset" is a collection of data that has been preprocessed to train a generative AI model.

[0461] A "generative AI model" is an artificial intelligence model that is trained using machine learning algorithms to extract specific risk patterns and make predictions.

[0462] "Natural Language Processing (NLP)" is a technology for analyzing text data and understanding and processing human language.

[0463] "New project documents" are documents such as requirements definition documents and design documents related to a newly started project.

[0464] "Risk factors" are elements or conditions that could cause problems or obstacles in the progress of a project.

[0465] "User notification" is the process of informing users of analysis results and risk predictions.

[0466] "Feedback" refers to information received from users, such as opinions and evaluations, that is used to improve the system and generative AI models.

[0467] A "machine learning algorithm" is an algorithm that learns patterns and rules from data and makes predictions about new data.

[0468] "Retraining" is the process of adding new data to an existing model to improve its accuracy and performance.

[0469] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[0470] Hardware and software used

[0471] This system mainly consists of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the generative AI model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. The user is responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0472] Data collection

[0473] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0474] Data Preprocessing

[0475] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is fed into a generative AI model as a machine learning dataset.

[0476] Training and updating AI models

[0477] The server uses machine learning algorithms to train a generative AI model, which is designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the model is retrained to improve its accuracy.

[0478] New Project Analysis

[0479] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server receives these documents and uses natural language processing (NLP) technology to convert the text data into an analyzable format.

[0480] Risk factor prediction and notification

[0481] The server applies a generative AI model based on the analyzed new project documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0482] Information supplementation and feedback

[0483] The server extracts relevant information from internal bulletin boards and notifies users of posts discussing similar risks. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[0484] Specific examples

[0485] Example 1: Software development project

[0486] User A starts a new software development project and uploads a requirements specification document to a device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, prompting User A to consider countermeasures.

[0487] Example 2: Service launch project

[0488] User B is currently working on a project to launch a new web service and uploads the design documents to his / her terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out "database schema design risks" to User B and provides specific design improvement proposals. Along with the design improvement proposals, the terminal also displays best practices from successful past projects for User B to refer to.

[0489] Prompt Sentence Examples

[0490] Below are some examples of prompts that users may use when using the system:

[0491] 1. "I've uploaded a requirements document for a new software development project. What are the risk factors?"

[0492] 2. "I have uploaded the design document for a web service launch project. Please tell me about past near misses and risk factors."

[0493] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which makes it possible to detect risks in advance from the early stages of development and take appropriate measures.

[0494] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0495] Step 1: Data collection

[0496] The server collects past incident data and risk information from various sources within the company. Specifically, the server accesses accident reports, near-miss information, issue management tables, and data posted on internal bulletin boards via the internal network. The data obtained from each source is stored in a database. This process is carried out using a crawler and API.

[0497] Input: Historical incident data and risk information from the company's internal network

[0498] Output: Raw data stored in a database

[0499] Step 2: Data cleansing

[0500] The server cleanses the collected data. Specifically, it detects and corrects inconsistencies and missing values. It removes inaccurate data, adds necessary metadata, and standardizes the data format. At this stage, text data is cleaned and numerical data is standardized.

[0501] Input: Raw data stored in a database

[0502] Output: Cleansed data

[0503] Step 3: Dataset generation

[0504] The server then generates a machine learning dataset from the cleansed data, which is then used to train the AI ​​model. This process also involves standardizing the data and extracting features.

[0505] Input: Cleansed data

[0506] Output: Dataset for machine learning

[0507] Step 4: Model training

[0508] The server uses the generated dataset to train the generative AI model. Specifically, it uses machine learning algorithms (e.g., random forest, deep learning) to learn from past incidents and risk information. The model is designed to extract specific risk patterns.

[0509] Input: Machine learning dataset

[0510] Output: A trained generative AI model

[0511] Step 5: Update the model

[0512] As new incident data is added, the server retrains the AI ​​model, which continuously improves its accuracy. The model update process is automated, allowing it to quickly reflect the impact of new data.

[0513] Input: New incident data added

[0514] Output: An updated generative AI model

[0515] Step 6: Upload documents

[0516] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The device supports document uploading by providing an interface that users can easily operate.

[0517] Input: Documents related to the new project (requirements specification and design documents)

[0518] Output: The document uploaded to the server

[0519] Step 7: Document Analysis

[0520] The server receives the uploaded document and uses natural language processing (NLP) techniques to convert the text data into an analyzable format, specifically tokenizing the document content and extracting important keywords and phrases.

[0521] Input: New project documentation

[0522] Output: Parsed text data

[0523] Step 8: Risk prediction

[0524] The server uses a generative AI model to analyze the uploaded new project documents, match them with similar past cases, and identify specific risk factors and omissions. The generative AI generates risk prediction results and creates a risk report containing specific risk factors and countermeasures.

[0525] Input: Analyzed text data, trained generative AI model

[0526] Output: Risk Report

[0527] Step 9: User Notification

[0528] The created risk report is sent to the device and notified to the user. When the device receives the notification, it displays specific risk factors and countermeasures to the user. For example, it notifies the user of security risks and displays a related configuration checklist.

[0529] Input: Risk Report

[0530] Output: Risk factors and countermeasures displayed to the user

[0531] Step 10: Supplementary Information Notification

[0532] The server extracts relevant information from internal bulletin boards and notifies users of posts where similar risks are being discussed, allowing users to refer to past experiences and discussions to strengthen risk management for new projects.

[0533] Input: Risk report, related information on internal bulletin board

[0534] Output: Additional information provided to the user

[0535] Step 11: User Feedback

[0536] Users input feedback on the risk information and advice provided by the system. The feedback is sent to the server via their device. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[0537] Input: User feedback

[0538] Output: A generative AI model that reflects feedback and improves accuracy, and an improved system

[0539] (Application example 1)

[0540] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0541] In modern companies, it is extremely important to consider past risk factors and take proactive measures when launching a new project. However, manually collecting and analyzing risk information is extremely time-consuming and inefficient. Furthermore, project managers lack the means to check risk factors in real time while working in the field or traveling, making it difficult to respond quickly to new security risks. The objective of this invention is to solve these problems and strengthen security risk management within companies.

[0542] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0543] In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the user of the analysis results and risk predictions, means for a user to upload documents related to the new project from the smart glasses, and means for displaying the analysis results and risk predictions on the smart glasses in real time. This makes it possible to efficiently utilize past risk information, grasp project risks in real time using the smart glasses, and respond quickly.

[0544] "Past incident data" refers to recorded information about accidents or problems that have occurred in the past within a company.

[0545] "Risk information" is information about the potential dangers or problems that may arise from a particular situation or action.

[0546] "Means of collection" refers to the methods and tools used to gather the necessary data and information.

[0547] "Preprocessing" refers to data cleansing and transformation performed on raw data to prepare it in an analyzable format.

[0548] A "machine learning dataset" is a set of data used to train and evaluate machine learning models.

[0549] A "generative AI model" is an artificial intelligence model that generates new data or predictions based on given data.

[0550] "Means for analyzing documents" refers to methods and tools for analyzing project documents and extracting necessary information.

[0551] "Means for predicting risk factors" refers to methods and tools that use past data to detect risks that may lurk in new projects in advance.

[0552] "Means of notification" refers to methods and tools for conveying analysis results and important information to users.

[0553] "Means for uploading from smart glasses" refers to methods or tools for using smart glasses to send data or documents to a system such as a server.

[0554] "Real-time display means" refers to methods or tools that provide information or analytical results to the user's view instantly.

[0555] This invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing this invention are described below.

[0556] Overall system configuration

[0557] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0558] Data collection

[0559] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0560] Data Preprocessing

[0561] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[0562] Training and updating AI models

[0563] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[0564] New Project Analysis

[0565] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[0566] Risk factor prediction and notification

[0567] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0568] Use of smart glasses

[0569] Users can upload documents related to new projects using the smart glasses. The smart glasses have the ability to display analysis results and risk predictions to users in real time, allowing users to check risk information at any time even while the project is in progress and take immediate action.

[0570] Information supplementation and feedback

[0571] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[0572] Specific examples

[0573] Example 1: Security Services Project

[0574] A user starts a new security service project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts the user to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, prompting the user to consider countermeasures.

[0575] Prompt Sentence Examples

[0576] "Based on the requirements specification for the new system, please analyze past security incident data and identify new risks."

[0577] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0578] Step 1:

[0579] The server collects past incident data and risk information from within the company. It uses information sources such as accident reports, near-miss information, issue management tables, and internal bulletin boards as input and stores it in a database. The collected data is obtained as output.

[0580] Step 2:

[0581] The server cleanses the collected data and corrects inconsistencies. It takes the collected data as input, removes inaccurate data, standardizes the data format, and adds necessary metadata. The output is a pre-processed, consistent dataset.

[0582] Step 3:

[0583] The server uses a machine learning algorithm to input the preprocessed dataset into a generative AI model to train the model. Using the preprocessed dataset as input, the server learns from past incidents and risk information to extract specific risk patterns. The output is a trained generative AI model.

[0584] Step 4:

[0585] The user uploads documents such as requirements and design documents for a new project from their terminal. The project documents are used as input and sent to the server via the terminal. The documents uploaded to the server are obtained as output.

[0586] Step 5:

[0587] The server analyzes the uploaded documents of new projects. It receives the uploaded documents as input and uses natural language processing (NLP) technology to convert the text data into an understandable format. The output is the analyzed document data.

[0588] Step 6:

[0589] The server uses a generative AI model to predict risk factors based on the analyzed documents. It uses the analyzed document data and the trained generative AI model as input and matches it with historical risk data. The output is the identified risk factors and omission points.

[0590] Step 7:

[0591] The server notifies the terminal of the identified risk factors and omissions. Using the identified risk information as input, it notifies the user of specific risk factors and proposed countermeasures. The notification content is displayed on the terminal as output.

[0592] Step 8:

[0593] Users use smart glasses to upload documents related to new projects and check analysis results and risk predictions. The documents sent from the smart glasses as input are uploaded to the server, and the analysis results are displayed in real time. Real-time risk information is displayed on the smart glasses as output.

[0594] Step 9:

[0595] The user inputs feedback on the risk information and advice provided into the device and sends it to the server. The feedback on the risk information is used as input and sent to the server. The feedback is saved as output on the server and used to improve the AI ​​model in the future.

[0596] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0597] The present invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, to predict and identify risk factors in new projects in advance. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the quality of notifications and feedback to the user is improved. Specific embodiments for implementing the present invention are described below.

[0598] Overall system configuration

[0599] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system. The emotion engine also analyzes and recognizes user emotions and provides optimized notifications and feedback.

[0600] Data collection

[0601] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[0602] Data Preprocessing

[0603] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0604] Training and updating AI models

[0605] The server converts the preprocessed data into a machine learning dataset and performs machine learning using a generative AI model. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[0606] New Project Analysis

[0607] A user starts a new project and uploads requirements and design documents from their device. The device then sends these documents to the server. The server then analyzes the new project documents, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and analyze semantics to extract important, relevant information.

[0608] Risk factor prediction and notification

[0609] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns. This predicts potential risk factors for new projects. The server then sends the identified risk factors and proposed countermeasures to the terminal. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[0610] Emotion engine analysis and notification adjustment

[0611] The server uses an emotion engine to analyze the user's input data and behavior to recognize the user's emotions. Based on this information, the server adjusts the content and format of the risk notification to present it in a way that is easier for the user to understand. For example, if the emotion engine determines that the user's stress level is high, the server will simplify the notification content and present only the main points to avoid burdening the user.

[0612] Feedback and System Improvement

[0613] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[0614] Specific examples

[0615] Example 1: Software development project

[0616] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently been subject to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures. Furthermore, the emotion engine recognizes User A's stress level and adjusts the content of notifications in a simple manner.

[0617] Example 2: Service launch project

[0618] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. Furthermore, the emotion engine recognizes User B's emotions and adjusts the feedback content appropriately.

[0619] As described above, this invention provides a concrete system for strengthening risk management in new projects by utilizing past data. This allows risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken. Furthermore, by combining it with an emotion engine, it is possible to improve the quality of notifications and feedback to users.

[0620] The processing flow will be explained below.

[0621] Step 1:

[0622] The server periodically collects past incident data and risk information from within the company, including accident reports, near-miss information, issue management tables, and posts on internal bulletin boards. The collected data is structured and stored in a database.

[0623] Step 2:

[0624] The server preprocesses the collected data. The data cleansing process involves removing inaccurate information, standardizing data formats, and correcting inconsistencies. For example, it standardizes data formats from different sources and fills in missing values.

[0625] Step 3:

[0626] The server converts the preprocessed data into a machine learning dataset, which is then optimized for model training using feature engineering techniques.

[0627] Step 4:

[0628] The server trains the generative AI model. This process uses the training data to apply a learning algorithm to extract and learn from historical risk patterns. It evaluates the model's accuracy and adjusts hyperparameters or retrains it as needed.

[0629] Step 5:

[0630] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[0631] Step 6:

[0632] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag, and semantically analyze the text to extract key information.

[0633] Step 7:

[0634] The server inputs the analyzed project data into the generative AI model and compares it with past risk patterns, thereby predicting potential risk factors for new projects.

[0635] Step 8:

[0636] The server sends the identified risk factors and proposed countermeasures to the terminal, generating a report that includes detailed information about the risk, examples of similar cases in the past, and countermeasures.

[0637] Step 9:

[0638] The terminal displays the received risk information and countermeasures on the user interface, allowing the user to manage project risks based on this information.

[0639] Step 10:

[0640] The server uses an emotion engine to analyze the user's input data and behavioral data to recognize the user's emotions, such as the user's typing speed and word usage patterns.

[0641] Step 11:

[0642] The server adjusts the content of risk notifications based on the perceived user emotion: for example, if the user is feeling stressed, it will briefly summarize the notification and highlight only the important points.

[0643] Step 12:

[0644] Users input feedback on the risk information and countermeasures provided by the system into their terminal and send it to the server.

[0645] Step 13:

[0646] The server analyzes user feedback and sentiment data to help improve the generative AI model and system. The feedback is used to update the AI ​​model's training data, improving the overall accuracy and usability of the system.

[0647] Specific examples

[0648] Example 1: Software development project

[0649] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. Furthermore, the emotion engine recognizes User A's stress level and adjusts the notification content concisely. The device displays the checklist as well as examples of related past projects, urging User A to take appropriate measures.

[0650] Example 2: Service launch project

[0651] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss incidents from past service launches. As a result, it is discovered that there were numerous errors in the database schema design. The server points out "risks in the database schema design" to User B and provides specific design improvement suggestions. Furthermore, the emotion engine recognizes User B's emotions and adjusts appropriate risk notifications. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. In this way, the system strengthens risk management for new projects and provides information that takes user emotions into consideration.

[0652] Example 2

[0653] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0654] When starting a new project within a company, past incident data and risk information are not fully utilized, making it difficult to predict and address risk factors in advance.Furthermore, notifications and feedback that do not take user emotions into consideration place an excessive burden on users.

[0655] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for analyzing user emotions using an emotion analysis engine and adjusting the notification content based on the analysis results, and means for collecting feedback from users and improving the AI ​​model and the overall system. This makes it possible to predict risk factors for new projects in advance and take appropriate measures, as well as provide notifications and feedback that take user emotions into consideration.

[0656] "Incident data" refers to detailed information and records of accidents and troubles that have occurred in the past within a company.

[0657] "Risk information" is information about potential dangers or problems that have been recognized in the past within a company.

[0658] "Data collection means" refers to an element that has the function of periodically collecting relevant data from within the company and from each department and storing it in a database.

[0659] The "data preprocessing means" is an element that has the function of cleansing collected data, converting it into a unified format, and adding metadata to it.

[0660] A "machine learning dataset" is a set of data generated from preprocessed data and used to train machine learning algorithms.

[0661] A "generative AI model" is an artificial intelligence model that has been trained using machine learning algorithms and has the ability to predict risk factors based on new data.

[0662] The "document analysis means" is an element that has the function of analyzing documents such as requirements definition documents and design documents related to new projects and extracting important information.

[0663] "Natural language processing technology" is a technology for analyzing human language using a computer, and includes processes such as tokenization, part-of-speech tagging, and semantic analysis.

[0664] The "risk prediction means" is an element that inputs analyzed document data into a generative AI model and has the function of identifying and predicting risk factors.

[0665] The "notification means" is an element that has the function of informing the user of predicted risks and countermeasures.

[0666] An "emotion analysis engine" is a technology for analyzing and recognizing a user's emotions, and determines the user's emotional state based on the user's input data and behavior.

[0667] A "feedback collection means" is an element that has the function of collecting opinions and reactions from users and using them to improve the system or retrain the AI ​​model.

[0668] This invention is a system that uses a generative AI model to collect and analyze information on past accidents and near misses that have occurred within a company, as well as issue management tables, and to predict and identify risk factors in new projects in advance. Furthermore, by combining this with an emotion analysis engine that recognizes user emotions, the quality of notifications and feedback provided to users can be improved. A specific embodiment of this system is described below.

[0669] Data collection

[0670] The server collects past incident data and risk information from various data sources within the company. Specifically, it periodically scans internal bulletin boards, accident reports from each department, near-miss information, and issue management tables, and stores new data in a database. This allows the collected data to be centrally managed for analysis.

[0671] Data Preprocessing

[0672] The server performs preprocessing on the collected data. For example, it standardizes data in different formats, deletes incomplete data, and adds metadata. For example, if reports from different departments have different formats, standardizing these formats improves the accuracy of analysis.

[0673] Training an AI model

[0674] The server generates a machine learning dataset based on the preprocessed data. Generative AI models are used to extract risk patterns from past data and build learning models. During this process, algorithms such as neural networks are used to analyze the data's characteristics and learn risk factors. The accuracy of the models is regularly evaluated, and they are retrained as necessary.

[0675] Uploading documents for a new project

[0676] When starting a new project, users upload requirements and design documents from their devices, which then send these documents to the server.

[0677] Document Analysis

[0678] The server receives new project documents and uses natural language processing (NLP) techniques to analyze the text data, including tokenization, part-of-speech tagging, and semantic analysis, to extract key relevant information. In the process, a generative AI model is used to identify relevant risk factors.

[0679] Risk prediction and notification

[0680] The server inputs the analyzed data into a generative AI model and compares it with past risk patterns to predict risk factors. The predicted risk is sent to the device along with detailed information. The device then notifies the user based on this information and provides specific countermeasures. For example, a notification such as "There is a high security risk. Here is a specific configuration checklist" may be displayed.

[0681] Use of sentiment analysis engine

[0682] The server uses an emotion analysis engine to analyze the user's input data and behavior to recognize the user's emotions. Based on the analysis results, the server adjusts the content and format of notifications. For example, if the user is in a high stress state, the server will shorten the notification content and present only the main points.

[0683] Gathering feedback and improving the system

[0684] Users can input feedback on the risk information and advice provided into their devices and send it to the server, which analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and the overall system.

[0685] Specific examples

[0686] Example 1: Software development project

[0687] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it using a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, encouraging User A to consider countermeasures. The sentiment analysis engine recognizes User A's stress level and adjusts the notification content in a simple manner.

[0688] Example 2: Service launch project

[0689] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "risks in the database schema design" to User B and provides specific suggestions for improving the design. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. The sentiment analysis engine recognizes User B's emotions and adjusts the feedback content appropriately.

[0690] Prompt Sentence Examples

[0691] "Please upload the requirements specification for the new project and notify us of the results of the risk analysis."

[0692] As described above, this system utilizes past data to strengthen risk management in new projects. By combining it with a sentiment analysis engine, it is possible to improve the quality of notifications and feedback to users.

[0693] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0694] Step 1:

[0695] The server collects incident data and risk information from various data sources within the company. Input data includes data from internal bulletin boards, accident reports from each department, near-miss information, and issue management tables. This data is scanned periodically, and new data is saved in a database. The output after collection is incident data and risk information that is centrally managed in a database. Specifically, for example, the server runs a script every night to collect information.

[0696] Step 2:

[0697] The server preprocesses the data collected. The input is the collected incident data and risk information. Preprocessing involves removing incomplete information, unifying different formats, and adding metadata. For example, if reports from different departments have different formats, they are converted into a unified format. The output is the preprocessed data. Specifically, the data is cleansed using a cleaning script and metadata is added.

[0698] Step 3:

[0699] The server generates a machine learning dataset based on preprocessed data. The input is the preprocessed data. This dataset is fed to a generative AI model, which performs learning to extract risk patterns. The algorithms used include neural networks. The output is a trained generative AI model. Specifically, the machine learning algorithm is executed, and the model parameters are adjusted based on the evaluation results.

[0700] Step 4:

[0701] A user starts a new project and uploads documents such as requirements specifications and design documents to a terminal. The input is documents related to the new project. The terminal sends these to the server. The output is the documents received by the server. Specifically, the user selects a file from the terminal and presses the upload button.

[0702] Step 5:

[0703] The server receives the new project document and analyzes the text data using natural language processing techniques. The input is the document uploaded by the user. This analysis involves tokenization, part-of-speech tagging, and semantic analysis to extract important relevant information. The output is the analyzed text data. Specifically, the NLP engine is used to perform the text analysis task.

[0704] Step 6:

[0705] The server inputs the analysis results into a generative AI model to predict risk factors. The input is the analyzed text data. The generative AI model compares it with past risk patterns to identify risk factors for new projects. The output is the predicted risk factors and their detailed information. Specifically, the data is supplied to the model and an algorithm is executed to extract risk factors.

[0706] Step 7:

[0707] The server notifies the terminal of the predicted risk factors and their proposed countermeasures. The input is the predicted risk factors and their detailed information. The notification includes details of the risk, similar past cases, and recommended countermeasures. The output is a notification that is displayed to the user. Specifically, the server generates a notification message and sends it to the terminal.

[0708] Step 8:

[0709] The server uses an emotion analysis engine to analyze the user's input data and behavior and recognize the user's emotions. The input is the user's operation log and feedback. Based on the analysis results, the server adjusts the notification content and format. The output is the adjusted notification and feedback. Specifically, the server runs an emotion analysis algorithm and dynamically changes the notification content.

[0710] Step 9:

[0711] The user inputs feedback on the risk information and advice provided into the terminal and sends it to the server. The input is feedback from the user. The server analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and improve the overall system. The output is an improved AI model and system. Specifically, it runs a feedback analysis algorithm and adjusts the model and system.

[0712] The above are the processing steps of this program.

[0713] (Application example 2)

[0714] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0715] Current risk management systems often refer to past accidents and issue management tables, making it difficult to identify potential risks in new projects in advance. Furthermore, adopting a uniform notification method without considering user feelings can easily lead to user stress and a lack of understanding. Especially for industrial equipment, where incorrect operation or configuration can lead to serious accidents, more accurate risk prediction and appropriate notification methods are needed.

[0716] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for recognizing user emotions and adjusting the notification content, and means for providing risk information and countermeasures to industrial equipment. This makes it possible to identify risk factors in a new project in advance and provide feedback to the user using an appropriate notification method.

[0717] "Incident data" refers to detailed information about accidents and troubles that have occurred within a company in the past.

[0718] "Risk information" refers to data about potential hazards or issues associated with a particular project or task.

[0719] "Machine Learning Dataset" refers to a collection of processed and pre-processed data used to train a generative AI model.

[0720] "Generative AI model" refers to an AI system that uses machine learning algorithms to learn risk patterns from historical data and predict risks for new projects.

[0721] "Documents related to a new project" refers to documents that contain details of the project, such as requirements specifications and design documents for a new project.

[0722] "Natural language processing technology" refers to computer science techniques for analyzing and processing human language.

[0723] "Means for recognizing user emotions" refers to tools and algorithms that analyze emotions based on user input and behavior and reflect them in the system.

[0724] "Means for providing risk information and countermeasures" refers to methods and tools for notifying users and industrial equipment of effective risk countermeasures based on the analysis results.

[0725] The system for carrying out the present invention includes the following series of means: The system is mainly composed of three main entities: a server, a terminal, and a user.

[0726] System Overview

[0727] The server collects and preprocesses past incident data and risk information. It uses the preprocessed data set to train and update the generative AI model, analyzes new project documents, and predicts risk factors. It then notifies users of risk prediction results and countermeasures via their devices, helping them to smoothly progress with the project. It also uses an emotion engine to analyze users' emotions and adjust the content of notifications.

[0728] The terminal provides an interface for users to upload documents related to new projects and check the analysis results, allowing users to effectively manage projects through this system.

[0729] Program processing description

[0730] Data collection and preprocessing

[0731] The server collects data from past accident reports, near-miss information, and issue management tables stored within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database and undergoes data cleansing. Data cleansing involves deleting incomplete information, standardizing data formats, and removing unnecessary data. Inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0732] Training and updating generative AI models

[0733] The server generates a machine learning dataset based on the preprocessed data and trains a generative AI model. This model uses machine learning algorithms such as neural networks (using TensorFlow or PyTorch) to extract and learn risk patterns from past data. The accuracy of the model is evaluated and retrained as necessary.

[0734] New project analysis and risk prediction

[0735] When a user starts a new project, they use their device to upload documents such as requirements specifications and design documents to the server. The server then analyzes these documents using Natural Language Processing (NLP) technology. Natural language processing libraries such as spaCy and NLTK are used for analysis, tokenizing the text data, tagging parts of speech, and performing semantic analysis. A generative AI model compares the results with past risk patterns and predicts potential risk factors that may arise in the new project.

[0736] Analysis result notification and emotion engine

[0737] Based on the analysis results, the server sends risk factors and proposed countermeasures to the device. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures. The emotion engine analyzes the user's input and behavior to recognize their emotions. Based on this information, the content and format of the risk notification can be adjusted. For example, if the user is feeling stressed, the notification content will be simplified and only the main points will be presented.

[0738] Feedback and System Improvement

[0739] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[0740] Adding specific examples

[0741] Specific prompt examples

[0742] "We have added configuration information for a new machine. Please analyze the risks associated with this."

[0743] "We've added a safety procedure to this project. What risks have you encountered with similar materials in the past?"

[0744] This allows the system to proactively identify risk factors related to industrial equipment and provide feedback to users through appropriate notification methods.

[0745] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0746] Step 1:

[0747] The server collects past accident reports, near-miss information, and issue management tables from within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database. The specific input is raw data from each source, and the output is data stored in the database in a unified format.

[0748] Step 2:

[0749] The server preprocesses the collected data. It performs data cleansing to remove incomplete information, standardize data formats, and remove unnecessary data. For example, it corrects inconsistencies in data formats from different departments. The input is raw data stored in the database, and the output is cleansed data.

[0750] Step 3:

[0751] The server generates a machine learning dataset based on the preprocessed data. This dataset is input to a generative AI model and used to train the AI ​​model to extract and learn risk patterns from past data. The input is cleansed data, and the output is a machine learning dataset.

[0752] Step 4:

[0753] When a user starts a new project, the server provides a function to upload requirements and design documents through the terminal. The user uploads documents related to the project to the terminal. The input is the user's document, and the output is the document sent to the server.

[0754] Step 5:

[0755] The server analyzes the uploaded document using Natural Language Processing (NLP) techniques. It tokenizes the text data, tags it with parts of speech, and performs semantic analysis to extract important relevant information. The input is the document uploaded by the user, and the output is the analysis results.

[0756] Step 6:

[0757] The server inputs the analysis results into a generative AI model to predict potential risk factors for new projects. This involves comparing past risk patterns with current project data. The input is the analysis results data, and the output is risk prediction information.

[0758] Step 7:

[0759] The server notifies the user based on the risk prediction information. At this time, it uses an emotion engine to analyze the user's emotions and adjusts the notification format and content to an appropriate one according to the emotion. For example, if the user is feeling stressed, it selects a concise notification content. The input is the risk prediction information and the user's emotional data, and the output is the adjusted notification content.

[0760] Step 8:

[0761] The terminal provides the user with risk prediction information and countermeasures sent from the server. The user can use this information to adjust the progress of the project. The input is the risk prediction information sent from the server, and the output is the risk information and countermeasures displayed to the user.

[0762] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0763] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0764] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0765] [Third embodiment]

[0766] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0767] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0768] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0769] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0770] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0771] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0772] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0773] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0774] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0775] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0776] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0777] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0778] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[0779] Overall system configuration

[0780] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0781] Data collection

[0782] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0783] Data Preprocessing

[0784] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[0785] Training and updating AI models

[0786] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[0787] New Project Analysis

[0788] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[0789] Risk factor prediction and notification

[0790] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0791] Information supplementation and feedback

[0792] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[0793] Specific examples

[0794] Example 1: Software development project

[0795] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on the generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures.

[0796] Example 2: Service launch project

[0797] User B is currently working on a project to launch a new web service and uploads the design documents to his terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the terminal also displays best practices from successful past projects for User B to refer to.

[0798] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which enables risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken.

[0799] The processing flow will be explained below.

[0800] Step 1:

[0801] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[0802] Step 2:

[0803] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0804] Step 3:

[0805] The server converts the pre-processed data into a machine learning dataset that the generative AI model uses to learn from past incidents and risk patterns. The format and content of the dataset are optimized based on the requirements of the learning algorithm.

[0806] Step 4:

[0807] The server uses the generative AI model to perform machine learning. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[0808] Step 5:

[0809] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[0810] Step 6:

[0811] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and semantically analyze the text data to extract important, relevant information.

[0812] Step 7:

[0813] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns, thereby predicting potential risk factors that may arise in new projects.

[0814] Step 8:

[0815] The server sends the identified risk factors and proposed countermeasures to the terminal, including detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[0816] Step 9:

[0817] The terminal displays the received risk information on the user interface. Users can use this information to manage project risks and consider and implement necessary countermeasures.

[0818] Step 10:

[0819] Users can input their feedback on the risk information and advice provided by the system into their devices and send it to the server, which uses this feedback to improve the accuracy of the AI ​​model and the system.

[0820] In this way, by performing specific actions at each step, the system utilizes past data to predict and identify risk factors in new projects in advance.

[0821] Example 1

[0822] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0823] Conventional risk management systems are unable to fully utilize past incident data and risk information, making it highly likely that similar risks will reoccur in new projects. Furthermore, the accuracy of predicting risk factors is low, making it difficult to implement appropriate countermeasures. Furthermore, analyzing documents for new projects takes a lot of time and effort, placing a heavy burden on users.

[0824] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0825] In this invention, the server includes means for collecting past incident data and risk information, means for cleansing the collected information and generating a machine learning dataset, means for training a generative AI model using a machine learning algorithm, means for uploading and analyzing documents related to new projects, means for predicting risk factors using the generative AI model and generating analysis results, means for notifying users of the analysis results and risk predictions, and means for collecting feedback from users and using the feedback to improve the accuracy of the AI ​​model. This makes it possible to predict risk factors in new projects with high accuracy by utilizing past data and to quickly take appropriate measures.

[0826] "Past incident data" refers to data on past accidents and problems recorded in accident reports, near-miss information, issue management sheets, etc.

[0827] "Risk information" refers to information about risk factors and risk cases related to projects and operations.

[0828] "Cleansing" is the process of detecting, correcting, or removing inconsistencies and missing values ​​in collected data.

[0829] A "machine learning dataset" is a collection of data that has been preprocessed to train a generative AI model.

[0830] A "generative AI model" is an artificial intelligence model that is trained using machine learning algorithms to extract specific risk patterns and make predictions.

[0831] "Natural Language Processing (NLP)" is a technology for analyzing text data and understanding and processing human language.

[0832] "New project documents" are documents such as requirements definition documents and design documents related to a newly started project.

[0833] "Risk factors" are elements or conditions that could cause problems or obstacles in the progress of a project.

[0834] "User notification" is the process of informing users of analysis results and risk predictions.

[0835] "Feedback" refers to information received from users, such as opinions and evaluations, that is used to improve the system and generative AI models.

[0836] A "machine learning algorithm" is an algorithm that learns patterns and rules from data and makes predictions about new data.

[0837] "Retraining" is the process of adding new data to an existing model to improve its accuracy and performance.

[0838] The present invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[0839] Hardware and software used

[0840] This system mainly consists of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the generative AI model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. The user is responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0841] Data collection

[0842] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0843] Data Preprocessing

[0844] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is fed into a generative AI model as a machine learning dataset.

[0845] Training and updating AI models

[0846] The server uses machine learning algorithms to train a generative AI model, which is designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the model is retrained to improve its accuracy.

[0847] New Project Analysis

[0848] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server receives these documents and uses natural language processing (NLP) technology to convert the text data into an analyzable format.

[0849] Risk factor prediction and notification

[0850] The server applies a generative AI model based on the analyzed new project documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0851] Information supplementation and feedback

[0852] The server extracts relevant information from internal bulletin boards and notifies users of posts discussing similar risks. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[0853] Specific examples

[0854] Example 1: Software development project

[0855] User A starts a new software development project and uploads a requirements specification document to a device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, prompting User A to consider countermeasures.

[0856] Example 2: Service launch project

[0857] User B is currently working on a project to launch a new web service and uploads the design documents to his / her terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out "database schema design risks" to User B and provides specific design improvement proposals. Along with the design improvement proposals, the terminal also displays best practices from successful past projects for User B to refer to.

[0858] Prompt Sentence Examples

[0859] Below are some examples of prompts that users may use when using the system:

[0860] 1. "I've uploaded a requirements document for a new software development project. What are the risk factors?"

[0861] 2. "I have uploaded the design document for a web service launch project. Please tell me about past near misses and risk factors."

[0862] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which makes it possible to detect risks in advance from the early stages of development and take appropriate measures.

[0863] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0864] Step 1: Data collection

[0865] The server collects past incident data and risk information from various sources within the company. Specifically, the server accesses accident reports, near-miss information, issue management tables, and data posted on internal bulletin boards via the internal network. The data obtained from each source is stored in a database. This process is carried out using a crawler and API.

[0866] Input: Historical incident data and risk information from the company's internal network

[0867] Output: Raw data stored in a database

[0868] Step 2: Data cleansing

[0869] The server cleanses the collected data. Specifically, it detects and corrects inconsistencies and missing values. It removes inaccurate data, adds necessary metadata, and standardizes the data format. At this stage, text data is cleaned and numerical data is standardized.

[0870] Input: Raw data stored in a database

[0871] Output: Cleansed data

[0872] Step 3: Dataset generation

[0873] The server then generates a machine learning dataset from the cleansed data, which is then used to train the AI ​​model. This process also involves standardizing the data and extracting features.

[0874] Input: Cleansed data

[0875] Output: Dataset for machine learning

[0876] Step 4: Model training

[0877] The server uses the generated dataset to train the generative AI model. Specifically, it uses machine learning algorithms (e.g., random forest, deep learning) to learn from past incidents and risk information. The model is designed to extract specific risk patterns.

[0878] Input: Machine learning dataset

[0879] Output: A trained generative AI model

[0880] Step 5: Update the model

[0881] As new incident data is added, the server retrains the AI ​​model, which continuously improves its accuracy. The model update process is automated, allowing it to quickly reflect the impact of new data.

[0882] Input: New incident data added

[0883] Output: An updated generative AI model

[0884] Step 6: Upload documents

[0885] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The device supports document uploading by providing an interface that users can easily operate.

[0886] Input: Documents related to the new project (requirements specification and design documents)

[0887] Output: The document uploaded to the server

[0888] Step 7: Document Analysis

[0889] The server receives the uploaded document and uses natural language processing (NLP) techniques to convert the text data into an analyzable format, specifically tokenizing the document content and extracting important keywords and phrases.

[0890] Input: New project documentation

[0891] Output: Parsed text data

[0892] Step 8: Risk prediction

[0893] The server uses a generative AI model to analyze the uploaded new project documents, match them with similar past cases, and identify specific risk factors and omissions. The generative AI generates risk prediction results and creates a risk report containing specific risk factors and countermeasures.

[0894] Input: Analyzed text data, trained generative AI model

[0895] Output: Risk Report

[0896] Step 9: User Notification

[0897] The created risk report is sent to the device and notified to the user. When the device receives the notification, it displays specific risk factors and countermeasures to the user. For example, it notifies the user of security risks and displays a related configuration checklist.

[0898] Input: Risk Report

[0899] Output: Risk factors and countermeasures displayed to the user

[0900] Step 10: Supplementary Information Notification

[0901] The server extracts relevant information from internal bulletin boards and notifies users of posts where similar risks are being discussed, allowing users to refer to past experiences and discussions to strengthen risk management for new projects.

[0902] Input: Risk report, related information on internal bulletin board

[0903] Output: Additional information provided to the user

[0904] Step 11: User Feedback

[0905] Users input feedback on the risk information and advice provided by the system. The feedback is sent to the server via their device. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[0906] Input: User feedback

[0907] Output: A generative AI model that reflects feedback and improves accuracy, and an improved system

[0908] (Application example 1)

[0909] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0910] In modern companies, it is extremely important to consider past risk factors and take proactive measures when launching a new project. However, manually collecting and analyzing risk information is extremely time-consuming and inefficient. Furthermore, project managers lack the means to check risk factors in real time while working in the field or traveling, making it difficult to respond quickly to new security risks. The objective of this invention is to solve these problems and strengthen security risk management within companies.

[0911] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0912] In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the user of the analysis results and risk predictions, means for a user to upload documents related to the new project from the smart glasses, and means for displaying the analysis results and risk predictions on the smart glasses in real time. This makes it possible to efficiently utilize past risk information, grasp project risks in real time using the smart glasses, and respond quickly.

[0913] "Past incident data" refers to recorded information about accidents or problems that have occurred in the past within a company.

[0914] "Risk information" is information about the potential dangers or problems that may arise from a particular situation or action.

[0915] "Means of collection" refers to the methods and tools used to gather the necessary data and information.

[0916] "Preprocessing" refers to data cleansing and transformation performed on raw data to prepare it in an analyzable format.

[0917] A "machine learning dataset" is a set of data used to train and evaluate machine learning models.

[0918] A "generative AI model" is an artificial intelligence model that generates new data or predictions based on given data.

[0919] "Means for analyzing documents" refers to methods and tools for analyzing project documents and extracting necessary information.

[0920] "Means for predicting risk factors" refers to methods and tools that use past data to detect risks that may lurk in new projects in advance.

[0921] "Means of notification" refers to methods and tools for conveying analysis results and important information to users.

[0922] "Means for uploading from smart glasses" refers to methods or tools for using smart glasses to send data or documents to a system such as a server.

[0923] "Real-time display means" refers to methods or tools that provide information or analytical results to the user's view instantly.

[0924] This invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing this invention are described below.

[0925] Overall system configuration

[0926] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[0927] Data collection

[0928] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[0929] Data Preprocessing

[0930] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[0931] Training and updating AI models

[0932] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[0933] New Project Analysis

[0934] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[0935] Risk factor prediction and notification

[0936] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[0937] Use of smart glasses

[0938] Users can upload documents related to new projects using the smart glasses. The smart glasses have the ability to display analysis results and risk predictions to users in real time, allowing users to check risk information at any time even while the project is in progress and take immediate action.

[0939] Information supplementation and feedback

[0940] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[0941] Specific examples

[0942] Example 1: Security Services Project

[0943] A user starts a new security service project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts the user to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, prompting the user to consider countermeasures.

[0944] Prompt Sentence Examples

[0945] "Based on the requirements specification for the new system, please analyze past security incident data and identify new risks."

[0946] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0947] Step 1:

[0948] The server collects past incident data and risk information from within the company. It uses information sources such as accident reports, near-miss information, issue management tables, and internal bulletin boards as input and stores it in a database. The collected data is obtained as output.

[0949] Step 2:

[0950] The server cleanses the collected data and corrects inconsistencies. It takes the collected data as input, removes inaccurate data, standardizes the data format, and adds necessary metadata. The output is a pre-processed, consistent dataset.

[0951] Step 3:

[0952] The server uses a machine learning algorithm to input the preprocessed dataset into a generative AI model to train the model. Using the preprocessed dataset as input, the server learns from past incidents and risk information to extract specific risk patterns. The output is a trained generative AI model.

[0953] Step 4:

[0954] The user uploads documents such as requirements and design documents for a new project from their terminal. The project documents are used as input and sent to the server via the terminal. The documents uploaded to the server are obtained as output.

[0955] Step 5:

[0956] The server analyzes the uploaded documents of new projects. It receives the uploaded documents as input and uses natural language processing (NLP) technology to convert the text data into an understandable format. The output is the analyzed document data.

[0957] Step 6:

[0958] The server uses a generative AI model to predict risk factors based on the analyzed documents. It uses the analyzed document data and the trained generative AI model as input and matches it with historical risk data. The output is the identified risk factors and omission points.

[0959] Step 7:

[0960] The server notifies the terminal of the identified risk factors and omissions. Using the identified risk information as input, it notifies the user of specific risk factors and proposed countermeasures. The notification content is displayed on the terminal as output.

[0961] Step 8:

[0962] Users use smart glasses to upload documents related to new projects and check analysis results and risk predictions. The documents sent from the smart glasses as input are uploaded to the server, and the analysis results are displayed in real time. Real-time risk information is displayed on the smart glasses as output.

[0963] Step 9:

[0964] The user inputs feedback on the risk information and advice provided into the device and sends it to the server. The feedback on the risk information is used as input and sent to the server. The feedback is saved as output on the server and used to improve the AI ​​model in the future.

[0965] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0966] The present invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, to predict and identify risk factors in new projects in advance. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the quality of notifications and feedback to the user is improved. Specific embodiments for implementing the present invention are described below.

[0967] Overall system configuration

[0968] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system. The emotion engine also analyzes and recognizes user emotions and provides optimized notifications and feedback.

[0969] Data collection

[0970] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[0971] Data Preprocessing

[0972] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[0973] Training and updating AI models

[0974] The server converts the preprocessed data into a machine learning dataset and performs machine learning using a generative AI model. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[0975] New Project Analysis

[0976] A user starts a new project and uploads requirements and design documents from their device. The device then sends these documents to the server. The server then analyzes the new project documents, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and analyze semantics to extract important, relevant information.

[0977] Risk factor prediction and notification

[0978] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns. This predicts potential risk factors for new projects. The server then sends the identified risk factors and proposed countermeasures to the terminal. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[0979] Emotion engine analysis and notification adjustment

[0980] The server uses an emotion engine to analyze the user's input data and behavior to recognize the user's emotions. Based on this information, the server adjusts the content and format of the risk notification to present it in a way that is easier for the user to understand. For example, if the emotion engine determines that the user's stress level is high, the server will simplify the notification content and present only the main points to avoid burdening the user.

[0981] Feedback and System Improvement

[0982] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[0983] Specific examples

[0984] Example 1: Software development project

[0985] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently been subject to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures. Furthermore, the emotion engine recognizes User A's stress level and adjusts the content of notifications in a simple manner.

[0986] Example 2: Service launch project

[0987] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. Furthermore, the emotion engine recognizes User B's emotions and adjusts the feedback content appropriately.

[0988] As described above, this invention provides a concrete system for strengthening risk management in new projects by utilizing past data. This allows risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken. Furthermore, by combining it with an emotion engine, it is possible to improve the quality of notifications and feedback to users.

[0989] The processing flow will be explained below.

[0990] Step 1:

[0991] The server periodically collects past incident data and risk information from within the company, including accident reports, near-miss information, issue management tables, and posts on internal bulletin boards. The collected data is structured and stored in a database.

[0992] Step 2:

[0993] The server preprocesses the collected data. The data cleansing process involves removing inaccurate information, standardizing data formats, and correcting inconsistencies. For example, it standardizes data formats from different sources and fills in missing values.

[0994] Step 3:

[0995] The server converts the preprocessed data into a machine learning dataset, which is then optimized for model training using feature engineering techniques.

[0996] Step 4:

[0997] The server trains the generative AI model. This process uses the training data to apply a learning algorithm to extract and learn from historical risk patterns. It evaluates the model's accuracy and adjusts hyperparameters or retrains it as needed.

[0998] Step 5:

[0999] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[1000] Step 6:

[1001] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag, and semantically analyze the text to extract key information.

[1002] Step 7:

[1003] The server inputs the analyzed project data into the generative AI model and compares it with past risk patterns, thereby predicting potential risk factors for new projects.

[1004] Step 8:

[1005] The server sends the identified risk factors and proposed countermeasures to the terminal, generating a report that includes detailed information about the risk, examples of similar cases in the past, and countermeasures.

[1006] Step 9:

[1007] The terminal displays the received risk information and countermeasures on the user interface, allowing the user to manage project risks based on this information.

[1008] Step 10:

[1009] The server uses an emotion engine to analyze the user's input data and behavioral data to recognize the user's emotions, such as the user's typing speed and word usage patterns.

[1010] Step 11:

[1011] The server adjusts the content of risk notifications based on the perceived user emotion: for example, if the user is feeling stressed, it will briefly summarize the notification and highlight only the important points.

[1012] Step 12:

[1013] Users input feedback on the risk information and countermeasures provided by the system into their terminal and send it to the server.

[1014] Step 13:

[1015] The server analyzes user feedback and sentiment data to help improve the generative AI model and system. The feedback is used to update the AI ​​model's training data, improving the overall accuracy and usability of the system.

[1016] Specific examples

[1017] Example 1: Software development project

[1018] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. Furthermore, the emotion engine recognizes User A's stress level and adjusts the notification content concisely. The device displays the checklist as well as examples of related past projects, urging User A to take appropriate measures.

[1019] Example 2: Service launch project

[1020] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss incidents from past service launches. As a result, it is discovered that there were numerous errors in the database schema design. The server points out "risks in the database schema design" to User B and provides specific design improvement suggestions. Furthermore, the emotion engine recognizes User B's emotions and adjusts appropriate risk notifications. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. In this way, the system strengthens risk management for new projects and provides information that takes user emotions into consideration.

[1021] Example 2

[1022] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1023] When starting a new project within a company, past incident data and risk information are not fully utilized, making it difficult to predict and address risk factors in advance.Furthermore, notifications and feedback that do not take user emotions into consideration place an excessive burden on users.

[1024] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for analyzing user emotions using an emotion analysis engine and adjusting the notification content based on the analysis results, and means for collecting feedback from users and improving the AI ​​model and the overall system. This makes it possible to predict risk factors for new projects in advance and take appropriate measures, as well as provide notifications and feedback that take user emotions into consideration.

[1025] "Incident data" refers to detailed information and records of accidents and troubles that have occurred in the past within a company.

[1026] "Risk information" is information about potential dangers or problems that have been recognized in the past within a company.

[1027] "Data collection means" refers to an element that has the function of periodically collecting relevant data from within the company and from each department and storing it in a database.

[1028] The "data preprocessing means" is an element that has the function of cleansing collected data, converting it into a unified format, and adding metadata to it.

[1029] A "machine learning dataset" is a set of data generated from preprocessed data and used to train machine learning algorithms.

[1030] A "generative AI model" is an artificial intelligence model that has been trained using machine learning algorithms and has the ability to predict risk factors based on new data.

[1031] The "document analysis means" is an element that has the function of analyzing documents such as requirements definition documents and design documents related to new projects and extracting important information.

[1032] "Natural language processing technology" is a technology for analyzing human language using a computer, and includes processes such as tokenization, part-of-speech tagging, and semantic analysis.

[1033] The "risk prediction means" is an element that inputs analyzed document data into a generative AI model and has the function of identifying and predicting risk factors.

[1034] The "notification means" is an element that has the function of informing the user of predicted risks and countermeasures.

[1035] An "emotion analysis engine" is a technology for analyzing and recognizing a user's emotions, and determines the user's emotional state based on the user's input data and behavior.

[1036] A "feedback collection means" is an element that has the function of collecting opinions and reactions from users and using them to improve the system or retrain the AI ​​model.

[1037] This invention is a system that uses a generative AI model to collect and analyze information on past accidents and near misses that have occurred within a company, as well as issue management tables, and to predict and identify risk factors in new projects in advance. Furthermore, by combining this with an emotion analysis engine that recognizes user emotions, the quality of notifications and feedback provided to users can be improved. A specific embodiment of this system is described below.

[1038] Data collection

[1039] The server collects past incident data and risk information from various data sources within the company. Specifically, it periodically scans internal bulletin boards, accident reports from each department, near-miss information, and issue management tables, and stores new data in a database. This allows the collected data to be centrally managed for analysis.

[1040] Data Preprocessing

[1041] The server performs preprocessing on the collected data. For example, it standardizes data in different formats, deletes incomplete data, and adds metadata. For example, if reports from different departments have different formats, standardizing these formats improves the accuracy of analysis.

[1042] Training an AI model

[1043] The server generates a machine learning dataset based on the preprocessed data. Generative AI models are used to extract risk patterns from past data and build learning models. During this process, algorithms such as neural networks are used to analyze the data's characteristics and learn risk factors. The accuracy of the models is regularly evaluated, and they are retrained as necessary.

[1044] Uploading documents for a new project

[1045] When starting a new project, users upload requirements and design documents from their devices, which then send these documents to the server.

[1046] Document Analysis

[1047] The server receives new project documents and uses natural language processing (NLP) techniques to analyze the text data, including tokenization, part-of-speech tagging, and semantic analysis, to extract key relevant information. In the process, a generative AI model is used to identify relevant risk factors.

[1048] Risk prediction and notification

[1049] The server inputs the analyzed data into a generative AI model and compares it with past risk patterns to predict risk factors. The predicted risk is sent to the device along with detailed information. The device then notifies the user based on this information and provides specific countermeasures. For example, a notification such as "There is a high security risk. Here is a specific configuration checklist" may be displayed.

[1050] Use of sentiment analysis engine

[1051] The server uses an emotion analysis engine to analyze the user's input data and behavior to recognize the user's emotions. Based on the analysis results, the server adjusts the content and format of notifications. For example, if the user is in a high stress state, the server will shorten the notification content and present only the main points.

[1052] Gathering feedback and improving the system

[1053] Users can input feedback on the risk information and advice provided into their devices and send it to the server, which analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and the overall system.

[1054] Specific examples

[1055] Example 1: Software development project

[1056] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it using a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, encouraging User A to consider countermeasures. The sentiment analysis engine recognizes User A's stress level and adjusts the notification content in a simple manner.

[1057] Example 2: Service launch project

[1058] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "risks in the database schema design" to User B and provides specific suggestions for improving the design. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. The sentiment analysis engine recognizes User B's emotions and adjusts the feedback content appropriately.

[1059] Prompt Sentence Examples

[1060] "Please upload the requirements specification for the new project and notify us of the results of the risk analysis."

[1061] As described above, this system utilizes past data to strengthen risk management in new projects. By combining it with a sentiment analysis engine, it is possible to improve the quality of notifications and feedback to users.

[1062] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1063] Step 1:

[1064] The server collects incident data and risk information from various data sources within the company. Input data includes data from internal bulletin boards, accident reports from each department, near-miss information, and issue management tables. This data is scanned periodically, and new data is saved in a database. The output after collection is incident data and risk information that is centrally managed in a database. Specifically, for example, the server runs a script every night to collect information.

[1065] Step 2:

[1066] The server preprocesses the data collected. The input is the collected incident data and risk information. Preprocessing involves removing incomplete information, unifying different formats, and adding metadata. For example, if reports from different departments have different formats, they are converted into a unified format. The output is the preprocessed data. Specifically, the data is cleansed using a cleaning script and metadata is added.

[1067] Step 3:

[1068] The server generates a machine learning dataset based on preprocessed data. The input is the preprocessed data. This dataset is fed to a generative AI model, which performs learning to extract risk patterns. The algorithms used include neural networks. The output is a trained generative AI model. Specifically, the machine learning algorithm is executed, and the model parameters are adjusted based on the evaluation results.

[1069] Step 4:

[1070] A user starts a new project and uploads documents such as requirements specifications and design documents to a terminal. The input is documents related to the new project. The terminal sends these to the server. The output is the documents received by the server. Specifically, the user selects a file from the terminal and presses the upload button.

[1071] Step 5:

[1072] The server receives the new project document and analyzes the text data using natural language processing techniques. The input is the document uploaded by the user. This analysis involves tokenization, part-of-speech tagging, and semantic analysis to extract important relevant information. The output is the analyzed text data. Specifically, the NLP engine is used to perform the text analysis task.

[1073] Step 6:

[1074] The server inputs the analysis results into a generative AI model to predict risk factors. The input is the analyzed text data. The generative AI model compares it with past risk patterns to identify risk factors for new projects. The output is the predicted risk factors and their detailed information. Specifically, the data is supplied to the model and an algorithm is executed to extract risk factors.

[1075] Step 7:

[1076] The server notifies the terminal of the predicted risk factors and their proposed countermeasures. The input is the predicted risk factors and their detailed information. The notification includes details of the risk, similar past cases, and recommended countermeasures. The output is a notification that is displayed to the user. Specifically, the server generates a notification message and sends it to the terminal.

[1077] Step 8:

[1078] The server uses an emotion analysis engine to analyze the user's input data and behavior and recognize the user's emotions. The input is the user's operation log and feedback. Based on the analysis results, the server adjusts the notification content and format. The output is the adjusted notification and feedback. Specifically, the server runs an emotion analysis algorithm and dynamically changes the notification content.

[1079] Step 9:

[1080] The user inputs feedback on the risk information and advice provided into the terminal and sends it to the server. The input is feedback from the user. The server analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and improve the overall system. The output is an improved AI model and system. Specifically, it runs a feedback analysis algorithm and adjusts the model and system.

[1081] The above are the processing steps of this program.

[1082] (Application example 2)

[1083] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1084] Current risk management systems often refer to past accidents and issue management tables, making it difficult to identify potential risks in new projects in advance. Furthermore, adopting a uniform notification method without considering user feelings can easily lead to user stress and a lack of understanding. Especially for industrial equipment, where incorrect operation or configuration can lead to serious accidents, more accurate risk prediction and appropriate notification methods are needed.

[1085] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for recognizing user emotions and adjusting the notification content, and means for providing risk information and countermeasures to industrial equipment. This makes it possible to identify risk factors in a new project in advance and provide feedback to the user using an appropriate notification method.

[1086] "Incident data" refers to detailed information about accidents and troubles that have occurred within a company in the past.

[1087] "Risk information" refers to data about potential hazards or issues associated with a particular project or task.

[1088] "Machine Learning Dataset" refers to a collection of processed and pre-processed data used to train a generative AI model.

[1089] "Generative AI model" refers to an AI system that uses machine learning algorithms to learn risk patterns from historical data and predict risks for new projects.

[1090] "Documents related to a new project" refers to documents that contain details of the project, such as requirements specifications and design documents for a new project.

[1091] "Natural language processing technology" refers to computer science techniques for analyzing and processing human language.

[1092] "Means for recognizing user emotions" refers to tools and algorithms that analyze emotions based on user input and behavior and reflect them in the system.

[1093] "Means for providing risk information and countermeasures" refers to methods and tools for notifying users and industrial equipment of effective risk countermeasures based on the analysis results.

[1094] The system for carrying out the present invention includes the following series of means: The system is mainly composed of three main entities: a server, a terminal, and a user.

[1095] System Overview

[1096] The server collects and preprocesses past incident data and risk information. It uses the preprocessed data set to train and update the generative AI model, analyzes new project documents, and predicts risk factors. It then notifies users of risk prediction results and countermeasures via their devices, helping them to smoothly progress with the project. It also uses an emotion engine to analyze users' emotions and adjust the content of notifications.

[1097] The terminal provides an interface for users to upload documents related to new projects and check the analysis results, allowing users to effectively manage projects through this system.

[1098] Program processing description

[1099] Data collection and preprocessing

[1100] The server collects data from past accident reports, near-miss information, and issue management tables stored within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database and undergoes data cleansing. Data cleansing involves deleting incomplete information, standardizing data formats, and removing unnecessary data. Inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[1101] Training and updating generative AI models

[1102] The server generates a machine learning dataset based on the preprocessed data and trains a generative AI model. This model uses machine learning algorithms such as neural networks (using TensorFlow or PyTorch) to extract and learn risk patterns from past data. The accuracy of the model is evaluated and retrained as necessary.

[1103] New project analysis and risk prediction

[1104] When a user starts a new project, they use their device to upload documents such as requirements specifications and design documents to the server. The server then analyzes these documents using Natural Language Processing (NLP) technology. Natural language processing libraries such as spaCy and NLTK are used for analysis, tokenizing the text data, tagging parts of speech, and performing semantic analysis. A generative AI model compares the results with past risk patterns and predicts potential risk factors that may arise in the new project.

[1105] Analysis result notification and emotion engine

[1106] Based on the analysis results, the server sends risk factors and proposed countermeasures to the device. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures. The emotion engine analyzes the user's input and behavior to recognize their emotions. Based on this information, the content and format of the risk notification can be adjusted. For example, if the user is feeling stressed, the notification content will be simplified and only the main points will be presented.

[1107] Feedback and System Improvement

[1108] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[1109] Adding specific examples

[1110] Specific prompt examples

[1111] "We have added configuration information for a new machine. Please analyze the risks associated with this."

[1112] "We've added a safety procedure to this project. What risks have you encountered with similar materials in the past?"

[1113] This allows the system to proactively identify risk factors related to industrial equipment and provide feedback to users through appropriate notification methods.

[1114] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1115] Step 1:

[1116] The server collects past accident reports, near-miss information, and issue management tables from within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database. The specific input is raw data from each source, and the output is data stored in the database in a unified format.

[1117] Step 2:

[1118] The server preprocesses the collected data. It performs data cleansing to remove incomplete information, standardize data formats, and remove unnecessary data. For example, it corrects inconsistencies in data formats from different departments. The input is raw data stored in the database, and the output is cleansed data.

[1119] Step 3:

[1120] The server generates a machine learning dataset based on the preprocessed data. This dataset is input to a generative AI model and used to train the AI ​​model to extract and learn risk patterns from past data. The input is cleansed data, and the output is a machine learning dataset.

[1121] Step 4:

[1122] When a user starts a new project, the server provides a function to upload requirements and design documents through the terminal. The user uploads documents related to the project to the terminal. The input is the user's document, and the output is the document sent to the server.

[1123] Step 5:

[1124] The server analyzes the uploaded document using Natural Language Processing (NLP) techniques. It tokenizes the text data, tags it with parts of speech, and performs semantic analysis to extract important relevant information. The input is the document uploaded by the user, and the output is the analysis results.

[1125] Step 6:

[1126] The server inputs the analysis results into a generative AI model to predict potential risk factors for new projects. This involves comparing past risk patterns with current project data. The input is the analysis results data, and the output is risk prediction information.

[1127] Step 7:

[1128] The server notifies the user based on the risk prediction information. At this time, it uses an emotion engine to analyze the user's emotions and adjusts the notification format and content to an appropriate one according to the emotion. For example, if the user is feeling stressed, it selects a concise notification content. The input is the risk prediction information and the user's emotional data, and the output is the adjusted notification content.

[1129] Step 8:

[1130] The terminal provides the user with risk prediction information and countermeasures sent from the server. The user can use this information to adjust the progress of the project. The input is the risk prediction information sent from the server, and the output is the risk information and countermeasures displayed to the user.

[1131] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1132] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1133] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1134] [Fourth embodiment]

[1135] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1136] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1137] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1138] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1139] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1140] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1141] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1142] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1143] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1144] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1145] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1146] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1147] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1148] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[1149] Overall system configuration

[1150] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[1151] Data collection

[1152] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[1153] Data Preprocessing

[1154] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[1155] Training and updating AI models

[1156] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[1157] New Project Analysis

[1158] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[1159] Risk factor prediction and notification

[1160] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[1161] Information supplementation and feedback

[1162] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[1163] Specific examples

[1164] Example 1: Software development project

[1165] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on the generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures.

[1166] Example 2: Service launch project

[1167] User B is currently working on a project to launch a new web service and uploads the design documents to his terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the terminal also displays best practices from successful past projects for User B to refer to.

[1168] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which enables risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken.

[1169] The processing flow will be explained below.

[1170] Step 1:

[1171] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[1172] Step 2:

[1173] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[1174] Step 3:

[1175] The server converts the pre-processed data into a machine learning dataset that the generative AI model uses to learn from past incidents and risk patterns. The format and content of the dataset are optimized based on the requirements of the learning algorithm.

[1176] Step 4:

[1177] The server uses the generative AI model to perform machine learning. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[1178] Step 5:

[1179] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[1180] Step 6:

[1181] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and semantically analyze the text data to extract important, relevant information.

[1182] Step 7:

[1183] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns, thereby predicting potential risk factors that may arise in new projects.

[1184] Step 8:

[1185] The server sends the identified risk factors and proposed countermeasures to the terminal, including detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[1186] Step 9:

[1187] The terminal displays the received risk information on the user interface. Users can use this information to manage project risks and consider and implement necessary countermeasures.

[1188] Step 10:

[1189] Users can input their feedback on the risk information and advice provided by the system into their devices and send it to the server, which uses this feedback to improve the accuracy of the AI ​​model and the system.

[1190] In this way, by performing specific actions at each step, the system utilizes past data to predict and identify risk factors in new projects in advance.

[1191] Example 1

[1192] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1193] Conventional risk management systems are unable to fully utilize past incident data and risk information, making it highly likely that similar risks will reoccur in new projects. Furthermore, the accuracy of predicting risk factors is low, making it difficult to implement appropriate countermeasures. Furthermore, analyzing documents for new projects takes a lot of time and effort, placing a heavy burden on users.

[1194] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1195] In this invention, the server includes means for collecting past incident data and risk information, means for cleansing the collected information and generating a machine learning dataset, means for training a generative AI model using a machine learning algorithm, means for uploading and analyzing documents related to new projects, means for predicting risk factors using the generative AI model and generating analysis results, means for notifying users of the analysis results and risk predictions, and means for collecting feedback from users and using the feedback to improve the accuracy of the AI ​​model. This makes it possible to predict risk factors in new projects with high accuracy by utilizing past data and to quickly take appropriate measures.

[1196] "Past incident data" refers to data on past accidents and problems recorded in accident reports, near-miss information, issue management sheets, etc.

[1197] "Risk information" refers to information about risk factors and risk cases related to projects and operations.

[1198] "Cleansing" is the process of detecting, correcting, or removing inconsistencies and missing values ​​in collected data.

[1199] A "machine learning dataset" is a collection of data that has been preprocessed to train a generative AI model.

[1200] A "generative AI model" is an artificial intelligence model that is trained using machine learning algorithms to extract specific risk patterns and make predictions.

[1201] "Natural Language Processing (NLP)" is a technology for analyzing text data and understanding and processing human language.

[1202] "New project documents" are documents such as requirements definition documents and design documents related to a newly started project.

[1203] "Risk factors" are elements or conditions that could cause problems or obstacles in the progress of a project.

[1204] "User notification" is the process of informing users of analysis results and risk predictions.

[1205] "Feedback" refers to information received from users, such as opinions and evaluations, that is used to improve the system and generative AI models.

[1206] A "machine learning algorithm" is an algorithm that learns patterns and rules from data and makes predictions about new data.

[1207] "Retraining" is the process of adding new data to an existing model to improve its accuracy and performance.

[1208] The present invention is a generative AI-based system that collects and analyzes information on accidents and near misses that have occurred in the past within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing the present invention are described below.

[1209] Hardware and software used

[1210] This system mainly consists of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the generative AI model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. The user is responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[1211] Data collection

[1212] The server periodically collects past incident data and risk information from within the company. This includes information from sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[1213] Data Preprocessing

[1214] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is fed into a generative AI model as a machine learning dataset.

[1215] Training and updating AI models

[1216] The server uses machine learning algorithms to train a generative AI model, which is designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the model is retrained to improve its accuracy.

[1217] New Project Analysis

[1218] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server receives these documents and uses natural language processing (NLP) technology to convert the text data into an analyzable format.

[1219] Risk factor prediction and notification

[1220] The server applies a generative AI model based on the analyzed new project documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[1221] Information supplementation and feedback

[1222] The server extracts relevant information from internal bulletin boards and notifies users of posts discussing similar risks. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[1223] Specific examples

[1224] Example 1: Software development project

[1225] User A starts a new software development project and uploads a requirements specification document to a device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, prompting User A to consider countermeasures.

[1226] Example 2: Service launch project

[1227] User B is currently working on a project to launch a new web service and uploads the design documents to his / her terminal. The server analyzes the design documents and references near-miss incidents from past service launches. The analysis results reveal that there were numerous errors in the database schema design. The server points out "database schema design risks" to User B and provides specific design improvement proposals. Along with the design improvement proposals, the terminal also displays best practices from successful past projects for User B to refer to.

[1228] Prompt Sentence Examples

[1229] Below are some examples of prompts that users may use when using the system:

[1230] 1. "I've uploaded a requirements document for a new software development project. What are the risk factors?"

[1231] 2. "I have uploaded the design document for a web service launch project. Please tell me about past near misses and risk factors."

[1232] As described above, the present invention provides a specific system for strengthening risk management in new projects by utilizing past data, which makes it possible to detect risks in advance from the early stages of development and take appropriate measures.

[1233] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1234] Step 1: Data collection

[1235] The server collects past incident data and risk information from various sources within the company. Specifically, the server accesses accident reports, near-miss information, issue management tables, and data posted on internal bulletin boards via the internal network. The data obtained from each source is stored in a database. This process is carried out using a crawler and API.

[1236] Input: Historical incident data and risk information from the company's internal network

[1237] Output: Raw data stored in a database

[1238] Step 2: Data cleansing

[1239] The server cleanses the collected data. Specifically, it detects and corrects inconsistencies and missing values. It removes inaccurate data, adds necessary metadata, and standardizes the data format. At this stage, text data is cleaned and numerical data is standardized.

[1240] Input: Raw data stored in a database

[1241] Output: Cleansed data

[1242] Step 3: Dataset generation

[1243] The server then generates a machine learning dataset from the cleansed data, which is then used to train the AI ​​model. This process also involves standardizing the data and extracting features.

[1244] Input: Cleansed data

[1245] Output: Dataset for machine learning

[1246] Step 4: Model training

[1247] The server uses the generated dataset to train the generative AI model. Specifically, it uses machine learning algorithms (e.g., random forest, deep learning) to learn from past incidents and risk information. The model is designed to extract specific risk patterns.

[1248] Input: Machine learning dataset

[1249] Output: A trained generative AI model

[1250] Step 5: Update the model

[1251] As new incident data is added, the server retrains the AI ​​model, which continuously improves its accuracy. The model update process is automated, allowing it to quickly reflect the impact of new data.

[1252] Input: New incident data added

[1253] Output: An updated generative AI model

[1254] Step 6: Upload documents

[1255] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The device supports document uploading by providing an interface that users can easily operate.

[1256] Input: Documents related to the new project (requirements specification and design documents)

[1257] Output: The document uploaded to the server

[1258] Step 7: Document Analysis

[1259] The server receives the uploaded document and uses natural language processing (NLP) techniques to convert the text data into an analyzable format, specifically tokenizing the document content and extracting important keywords and phrases.

[1260] Input: New project documentation

[1261] Output: Parsed text data

[1262] Step 8: Risk prediction

[1263] The server uses a generative AI model to analyze the uploaded new project documents, match them with similar past cases, and identify specific risk factors and omissions. The generative AI generates risk prediction results and creates a risk report containing specific risk factors and countermeasures.

[1264] Input: Analyzed text data, trained generative AI model

[1265] Output: Risk Report

[1266] Step 9: User Notification

[1267] The created risk report is sent to the device and notified to the user. When the device receives the notification, it displays specific risk factors and countermeasures to the user. For example, it notifies the user of security risks and displays a related configuration checklist.

[1268] Input: Risk Report

[1269] Output: Risk factors and countermeasures displayed to the user

[1270] Step 10: Supplementary Information Notification

[1271] The server extracts relevant information from internal bulletin boards and notifies users of posts where similar risks are being discussed, allowing users to refer to past experiences and discussions to strengthen risk management for new projects.

[1272] Input: Risk report, related information on internal bulletin board

[1273] Output: Additional information provided to the user

[1274] Step 11: User Feedback

[1275] Users input feedback on the risk information and advice provided by the system. The feedback is sent to the server via their device. The server uses this feedback to improve the accuracy of the generative AI model and the system.

[1276] Input: User feedback

[1277] Output: A generative AI model that reflects feedback and improves accuracy, and an improved system

[1278] (Application example 1)

[1279] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1280] In modern companies, it is extremely important to consider past risk factors and take proactive measures when launching a new project. However, manually collecting and analyzing risk information is extremely time-consuming and inefficient. Furthermore, project managers lack the means to check risk factors in real time while working in the field or traveling, making it difficult to respond quickly to new security risks. The objective of this invention is to solve these problems and strengthen security risk management within companies.

[1281] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1282] In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the user of the analysis results and risk predictions, means for a user to upload documents related to the new project from the smart glasses, and means for displaying the analysis results and risk predictions on the smart glasses in real time. This makes it possible to efficiently utilize past risk information, grasp project risks in real time using the smart glasses, and respond quickly.

[1283] "Past incident data" refers to recorded information about accidents or problems that have occurred in the past within a company.

[1284] "Risk information" is information about the potential dangers or problems that may arise from a particular situation or action.

[1285] "Means of collection" refers to the methods and tools used to gather the necessary data and information.

[1286] "Preprocessing" refers to data cleansing and transformation performed on raw data to prepare it in an analyzable format.

[1287] A "machine learning dataset" is a set of data used to train and evaluate machine learning models.

[1288] A "generative AI model" is an artificial intelligence model that generates new data or predictions based on given data.

[1289] "Means for analyzing documents" refers to methods and tools for analyzing project documents and extracting necessary information.

[1290] "Means for predicting risk factors" refers to methods and tools that use past data to detect risks that may lurk in new projects in advance.

[1291] "Means of notification" refers to methods and tools for conveying analysis results and important information to users.

[1292] "Means for uploading from smart glasses" refers to methods or tools for using smart glasses to send data or documents to a system such as a server.

[1293] "Real-time display means" refers to methods or tools that provide information or analytical results to the user's view instantly.

[1294] This invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, and predicts and identifies risk factors in new projects in advance. Specific embodiments for implementing this invention are described below.

[1295] Overall system configuration

[1296] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system.

[1297] Data collection

[1298] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports, near miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database and is subject to preprocessing.

[1299] Data Preprocessing

[1300] The server cleanses the collected data and corrects inconsistencies. This process includes removing inaccurate data, standardizing data formats, and adding necessary metadata. Once preprocessed, the data is input into the generative AI as a dataset for machine learning.

[1301] Training and updating AI models

[1302] The server uses machine learning algorithms to train generative AI models, which are designed to learn from past incidents and risk information and extract specific risk patterns. As new data is added, the models are retrained to improve their accuracy.

[1303] New Project Analysis

[1304] When a user starts a new project, they upload documents such as requirements specifications and design documents from their device. The server analyzes these documents and uses natural language processing (NLP) technology to convert the text data into an understandable format.

[1305] Risk factor prediction and notification

[1306] The server applies a generative AI model to the new project's documents and matches them with similar past risk data. As a result, specific risk factors and omissions are identified. This information is sent to the device and notified to the user. The notification includes specific risk factors and suggested countermeasures.

[1307] Use of smart glasses

[1308] Users can upload documents related to new projects using the smart glasses. The smart glasses have the ability to display analysis results and risk predictions to users in real time, allowing users to check risk information at any time even while the project is in progress and take immediate action.

[1309] Information supplementation and feedback

[1310] The server extracts relevant information from internal bulletin boards and notifies users of posts in which similar risks are being discussed. Users refer to this information to strengthen risk management in new projects. Users also enter feedback on the risk information and advice provided by the system into their devices and send it to the server. The server uses this feedback to improve the accuracy of the AI ​​model and the system.

[1311] Specific examples

[1312] Example 1: Security Services Project

[1313] A user starts a new security service project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that security configuration errors have frequently occurred in the past with similar requirements. The server alerts the user to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, prompting the user to consider countermeasures.

[1314] Prompt Sentence Examples

[1315] "Based on the requirements specification for the new system, please analyze past security incident data and identify new risks."

[1316] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1317] Step 1:

[1318] The server collects past incident data and risk information from within the company. It uses information sources such as accident reports, near-miss information, issue management tables, and internal bulletin boards as input and stores it in a database. The collected data is obtained as output.

[1319] Step 2:

[1320] The server cleanses the collected data and corrects inconsistencies. It takes the collected data as input, removes inaccurate data, standardizes the data format, and adds necessary metadata. The output is a pre-processed, consistent dataset.

[1321] Step 3:

[1322] The server uses a machine learning algorithm to input the preprocessed dataset into a generative AI model to train the model. Using the preprocessed dataset as input, the server learns from past incidents and risk information to extract specific risk patterns. The output is a trained generative AI model.

[1323] Step 4:

[1324] The user uploads documents such as requirements and design documents for a new project from their terminal. The project documents are used as input and sent to the server via the terminal. The documents uploaded to the server are obtained as output.

[1325] Step 5:

[1326] The server analyzes the uploaded documents of new projects. It receives the uploaded documents as input and uses natural language processing (NLP) technology to convert the text data into an understandable format. The output is the analyzed document data.

[1327] Step 6:

[1328] The server uses a generative AI model to predict risk factors based on the analyzed documents. It uses the analyzed document data and the trained generative AI model as input and matches it with historical risk data. The output is the identified risk factors and omission points.

[1329] Step 7:

[1330] The server notifies the terminal of the identified risk factors and omissions. Using the identified risk information as input, it notifies the user of specific risk factors and proposed countermeasures. The notification content is displayed on the terminal as output.

[1331] Step 8:

[1332] Users use smart glasses to upload documents related to new projects and check analysis results and risk predictions. The documents sent from the smart glasses as input are uploaded to the server, and the analysis results are displayed in real time. Real-time risk information is displayed on the smart glasses as output.

[1333] Step 9:

[1334] The user inputs feedback on the risk information and advice provided into the device and sends it to the server. The feedback on the risk information is used as input and sent to the server. The feedback is saved as output on the server and used to improve the AI ​​model in the future.

[1335] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1336] The present invention is a generative AI-based system that collects and analyzes information on past accidents and near misses that have occurred within a company, as well as issue management tables, to predict and identify risk factors in new projects in advance. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the quality of notifications and feedback to the user is improved. Specific embodiments for implementing the present invention are described below.

[1337] Overall system configuration

[1338] The system is primarily composed of three components: a server, a terminal, and a user. The server collects, preprocesses, and analyzes data, and trains and updates the AI ​​model. The terminal provides an interface for users to upload documents related to new projects and receive analysis results. Users are responsible for creating requirements definitions and design documents for new projects and inputting them into the system. The emotion engine also analyzes and recognizes user emotions and provides optimized notifications and feedback.

[1339] Data collection

[1340] The server periodically collects past incident data and risk information from within the company. This includes information sources such as accident reports from each department, near-miss information, issue management tables, and internal bulletin boards. The collected data is stored in a database.

[1341] Data Preprocessing

[1342] The server preprocesses the collected data. During the data cleansing process, incomplete information is removed, formats are standardized, and unnecessary data is eliminated. For example, inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[1343] Training and updating AI models

[1344] The server converts the preprocessed data into a machine learning dataset and performs machine learning using a generative AI model. During this process, algorithms such as neural networks are used to extract and learn risk patterns from past data. The accuracy of the AI ​​model is evaluated and the model is retrained as necessary.

[1345] New Project Analysis

[1346] A user starts a new project and uploads requirements and design documents from their device. The device then sends these documents to the server. The server then analyzes the new project documents, using natural language processing (NLP) techniques to tokenize, tag parts of speech, and analyze semantics to extract important, relevant information.

[1347] Risk factor prediction and notification

[1348] The server inputs the analyzed project data into a generative AI model and compares it with past risk patterns. This predicts potential risk factors for new projects. The server then sends the identified risk factors and proposed countermeasures to the terminal. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures.

[1349] Emotion engine analysis and notification adjustment

[1350] The server uses an emotion engine to analyze the user's input data and behavior to recognize the user's emotions. Based on this information, the server adjusts the content and format of the risk notification to present it in a way that is easier for the user to understand. For example, if the emotion engine determines that the user's stress level is high, the server will simplify the notification content and present only the main points to avoid burdening the user.

[1351] Feedback and System Improvement

[1352] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[1353] Specific examples

[1354] Example 1: Software development project

[1355] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently been subject to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples of related past projects, encouraging User A to consider countermeasures. Furthermore, the emotion engine recognizes User A's stress level and adjusts the content of notifications in a simple manner.

[1356] Example 2: Service launch project

[1357] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "database schema design risks" to User B and provides specific design improvement suggestions. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. Furthermore, the emotion engine recognizes User B's emotions and adjusts the feedback content appropriately.

[1358] As described above, this invention provides a concrete system for strengthening risk management in new projects by utilizing past data. This allows risks to be detected in advance from the early stages of development and appropriate countermeasures to be taken. Furthermore, by combining it with an emotion engine, it is possible to improve the quality of notifications and feedback to users.

[1359] The processing flow will be explained below.

[1360] Step 1:

[1361] The server periodically collects past incident data and risk information from within the company, including accident reports, near-miss information, issue management tables, and posts on internal bulletin boards. The collected data is structured and stored in a database.

[1362] Step 2:

[1363] The server preprocesses the collected data. The data cleansing process involves removing inaccurate information, standardizing data formats, and correcting inconsistencies. For example, it standardizes data formats from different sources and fills in missing values.

[1364] Step 3:

[1365] The server converts the preprocessed data into a machine learning dataset, which is then optimized for model training using feature engineering techniques.

[1366] Step 4:

[1367] The server trains the generative AI model. This process uses the training data to apply a learning algorithm to extract and learn from historical risk patterns. It evaluates the model's accuracy and adjusts hyperparameters or retrains it as needed.

[1368] Step 5:

[1369] A user starts a new project and uploads requirements and design documents from their device, which then sends these documents to the server.

[1370] Step 6:

[1371] The server analyzes the uploaded documents for new projects, using natural language processing (NLP) techniques to tokenize, tag, and semantically analyze the text to extract key information.

[1372] Step 7:

[1373] The server inputs the analyzed project data into the generative AI model and compares it with past risk patterns, thereby predicting potential risk factors for new projects.

[1374] Step 8:

[1375] The server sends the identified risk factors and proposed countermeasures to the terminal, generating a report that includes detailed information about the risk, examples of similar cases in the past, and countermeasures.

[1376] Step 9:

[1377] The terminal displays the received risk information and countermeasures on the user interface, allowing the user to manage project risks based on this information.

[1378] Step 10:

[1379] The server uses an emotion engine to analyze the user's input data and behavioral data to recognize the user's emotions, such as the user's typing speed and word usage patterns.

[1380] Step 11:

[1381] The server adjusts the content of risk notifications based on the perceived user emotion: for example, if the user is feeling stressed, it will briefly summarize the notification and highlight only the important points.

[1382] Step 12:

[1383] Users input feedback on the risk information and countermeasures provided by the system into their terminal and send it to the server.

[1384] Step 13:

[1385] The server analyzes user feedback and sentiment data to help improve the generative AI model and system. The feedback is used to update the AI ​​model's training data, improving the overall accuracy and usability of the system.

[1386] Specific examples

[1387] Example 1: Software development project

[1388] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it based on a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. Furthermore, the emotion engine recognizes User A's stress level and adjusts the notification content concisely. The device displays the checklist as well as examples of related past projects, urging User A to take appropriate measures.

[1389] Example 2: Service launch project

[1390] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss incidents from past service launches. As a result, it is discovered that there were numerous errors in the database schema design. The server points out "risks in the database schema design" to User B and provides specific design improvement suggestions. Furthermore, the emotion engine recognizes User B's emotions and adjusts appropriate risk notifications. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. In this way, the system strengthens risk management for new projects and provides information that takes user emotions into consideration.

[1391] Example 2

[1392] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1393] When starting a new project within a company, past incident data and risk information are not fully utilized, making it difficult to predict and address risk factors in advance.Furthermore, notifications and feedback that do not take user emotions into consideration place an excessive burden on users.

[1394] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for analyzing user emotions using an emotion analysis engine and adjusting the notification content based on the analysis results, and means for collecting feedback from users and improving the AI ​​model and the overall system. This makes it possible to predict risk factors for new projects in advance and take appropriate measures, as well as provide notifications and feedback that take user emotions into consideration.

[1395] "Incident data" refers to detailed information and records of accidents and troubles that have occurred in the past within a company.

[1396] "Risk information" is information about potential dangers or problems that have been recognized in the past within a company.

[1397] "Data collection means" refers to an element that has the function of periodically collecting relevant data from within the company and from each department and storing it in a database.

[1398] The "data preprocessing means" is an element that has the function of cleansing collected data, converting it into a unified format, and adding metadata to it.

[1399] A "machine learning dataset" is a set of data generated from preprocessed data and used to train machine learning algorithms.

[1400] A "generative AI model" is an artificial intelligence model that has been trained using machine learning algorithms and has the ability to predict risk factors based on new data.

[1401] The "document analysis means" is an element that has the function of analyzing documents such as requirements definition documents and design documents related to new projects and extracting important information.

[1402] "Natural language processing technology" is a technology for analyzing human language using a computer, and includes processes such as tokenization, part-of-speech tagging, and semantic analysis.

[1403] The "risk prediction means" is an element that inputs analyzed document data into a generative AI model and has the function of identifying and predicting risk factors.

[1404] The "notification means" is an element that has the function of informing the user of predicted risks and countermeasures.

[1405] An "emotion analysis engine" is a technology for analyzing and recognizing a user's emotions, and determines the user's emotional state based on the user's input data and behavior.

[1406] A "feedback collection means" is an element that has the function of collecting opinions and reactions from users and using them to improve the system or retrain the AI ​​model.

[1407] This invention is a system that uses a generative AI model to collect and analyze information on past accidents and near misses that have occurred within a company, as well as issue management tables, and to predict and identify risk factors in new projects in advance. Furthermore, by combining this with an emotion analysis engine that recognizes user emotions, the quality of notifications and feedback provided to users can be improved. A specific embodiment of this system is described below.

[1408] Data collection

[1409] The server collects past incident data and risk information from various data sources within the company. Specifically, it periodically scans internal bulletin boards, accident reports from each department, near-miss information, and issue management tables, and stores new data in a database. This allows the collected data to be centrally managed for analysis.

[1410] Data Preprocessing

[1411] The server performs preprocessing on the collected data. For example, it standardizes data in different formats, deletes incomplete data, and adds metadata. For example, if reports from different departments have different formats, standardizing these formats improves the accuracy of analysis.

[1412] Training an AI model

[1413] The server generates a machine learning dataset based on the preprocessed data. Generative AI models are used to extract risk patterns from past data and build learning models. During this process, algorithms such as neural networks are used to analyze the data's characteristics and learn risk factors. The accuracy of the models is regularly evaluated, and they are retrained as necessary.

[1414] Uploading documents for a new project

[1415] When starting a new project, users upload requirements and design documents from their devices, which then send these documents to the server.

[1416] Document Analysis

[1417] The server receives new project documents and uses natural language processing (NLP) techniques to analyze the text data, including tokenization, part-of-speech tagging, and semantic analysis, to extract key relevant information. In the process, a generative AI model is used to identify relevant risk factors.

[1418] Risk prediction and notification

[1419] The server inputs the analyzed data into a generative AI model and compares it with past risk patterns to predict risk factors. The predicted risk is sent to the device along with detailed information. The device then notifies the user based on this information and provides specific countermeasures. For example, a notification such as "There is a high security risk. Here is a specific configuration checklist" may be displayed.

[1420] Use of sentiment analysis engine

[1421] The server uses an emotion analysis engine to analyze the user's input data and behavior to recognize the user's emotions. Based on the analysis results, the server adjusts the content and format of notifications. For example, if the user is in a high stress state, the server will shorten the notification content and present only the main points.

[1422] Gathering feedback and improving the system

[1423] Users can input feedback on the risk information and advice provided into their devices and send it to the server, which analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and the overall system.

[1424] Specific examples

[1425] Example 1: Software development project

[1426] User A starts a new software development project and uploads a requirements specification document to the device. The server receives the requirements specification document and analyzes it using a generative AI model. The analysis reveals that similar requirements have frequently led to security configuration errors in the past. The server alerts User A to the "security risks caused by configuration errors" and provides a specific configuration checklist. The device displays the checklist along with examples from related past projects, encouraging User A to consider countermeasures. The sentiment analysis engine recognizes User A's stress level and adjusts the notification content in a simple manner.

[1427] Example 2: Service launch project

[1428] User B is currently working on a project to launch a new web service and uploads the design documents to his device. The server analyzes the design documents and references near-miss information from past service launches. The analysis reveals that there were numerous errors in the database schema design. The server points out the "risks in the database schema design" to User B and provides specific suggestions for improving the design. Along with the design improvement suggestions, the device also displays best practices from successful past projects for User B to refer to. The sentiment analysis engine recognizes User B's emotions and adjusts the feedback content appropriately.

[1429] Prompt Sentence Examples

[1430] "Please upload the requirements specification for the new project and notify us of the results of the risk analysis."

[1431] As described above, this system utilizes past data to strengthen risk management in new projects. By combining it with a sentiment analysis engine, it is possible to improve the quality of notifications and feedback to users.

[1432] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1433] Step 1:

[1434] The server collects incident data and risk information from various data sources within the company. Input data includes data from internal bulletin boards, accident reports from each department, near-miss information, and issue management tables. This data is scanned periodically, and new data is saved in a database. The output after collection is incident data and risk information that is centrally managed in a database. Specifically, for example, the server runs a script every night to collect information.

[1435] Step 2:

[1436] The server preprocesses the data collected. The input is the collected incident data and risk information. Preprocessing involves removing incomplete information, unifying different formats, and adding metadata. For example, if reports from different departments have different formats, they are converted into a unified format. The output is the preprocessed data. Specifically, the data is cleansed using a cleaning script and metadata is added.

[1437] Step 3:

[1438] The server generates a machine learning dataset based on preprocessed data. The input is the preprocessed data. This dataset is fed to a generative AI model, which performs learning to extract risk patterns. The algorithms used include neural networks. The output is a trained generative AI model. Specifically, the machine learning algorithm is executed, and the model parameters are adjusted based on the evaluation results.

[1439] Step 4:

[1440] A user starts a new project and uploads documents such as requirements specifications and design documents to a terminal. The input is documents related to the new project. The terminal sends these to the server. The output is the documents received by the server. Specifically, the user selects a file from the terminal and presses the upload button.

[1441] Step 5:

[1442] The server receives the new project document and analyzes the text data using natural language processing techniques. The input is the document uploaded by the user. This analysis involves tokenization, part-of-speech tagging, and semantic analysis to extract important relevant information. The output is the analyzed text data. Specifically, the NLP engine is used to perform the text analysis task.

[1443] Step 6:

[1444] The server inputs the analysis results into a generative AI model to predict risk factors. The input is the analyzed text data. The generative AI model compares it with past risk patterns to identify risk factors for new projects. The output is the predicted risk factors and their detailed information. Specifically, the data is supplied to the model and an algorithm is executed to extract risk factors.

[1445] Step 7:

[1446] The server notifies the terminal of the predicted risk factors and their proposed countermeasures. The input is the predicted risk factors and their detailed information. The notification includes details of the risk, similar past cases, and recommended countermeasures. The output is a notification that is displayed to the user. Specifically, the server generates a notification message and sends it to the terminal.

[1447] Step 8:

[1448] The server uses an emotion analysis engine to analyze the user's input data and behavior and recognize the user's emotions. The input is the user's operation log and feedback. Based on the analysis results, the server adjusts the notification content and format. The output is the adjusted notification and feedback. Specifically, the server runs an emotion analysis algorithm and dynamically changes the notification content.

[1449] Step 9:

[1450] The user inputs feedback on the risk information and advice provided into the terminal and sends it to the server. The input is feedback from the user. The server analyzes this feedback and combines it with data from the sentiment analysis engine to improve the accuracy of the AI ​​model and improve the overall system. The output is an improved AI model and system. Specifically, it runs a feedback analysis algorithm and adjusts the model and system.

[1451] The above are the processing steps of this program.

[1452] (Application example 2)

[1453] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1454] Current risk management systems often refer to past accidents and issue management tables, making it difficult to identify potential risks in new projects in advance. Furthermore, adopting a uniform notification method without considering user feelings can easily lead to user stress and a lack of understanding. Especially for industrial equipment, where incorrect operation or configuration can lead to serious accidents, more accurate risk prediction and appropriate notification methods are needed.

[1455] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting past incident data and risk information, means for preprocessing the collected information and generating a machine learning dataset, means for providing a generative AI model that learns from the preprocessed and converted dataset, means for analyzing documents related to a new project and predicting risk factors using the generative AI model, means for notifying the analysis results and risk predictions, means for recognizing user emotions and adjusting the notification content, and means for providing risk information and countermeasures to industrial equipment. This makes it possible to identify risk factors in a new project in advance and provide feedback to the user using an appropriate notification method.

[1456] "Incident data" refers to detailed information about accidents and troubles that have occurred within a company in the past.

[1457] "Risk information" refers to data about potential hazards or issues associated with a particular project or task.

[1458] "Machine Learning Dataset" refers to a collection of processed and pre-processed data used to train a generative AI model.

[1459] "Generative AI model" refers to an AI system that uses machine learning algorithms to learn risk patterns from historical data and predict risks for new projects.

[1460] "Documents related to a new project" refers to documents that contain details of the project, such as requirements specifications and design documents for a new project.

[1461] "Natural language processing technology" refers to computer science techniques for analyzing and processing human language.

[1462] "Means for recognizing user emotions" refers to tools and algorithms that analyze emotions based on user input and behavior and reflect them in the system.

[1463] "Means for providing risk information and countermeasures" refers to methods and tools for notifying users and industrial equipment of effective risk countermeasures based on the analysis results.

[1464] The system for carrying out the present invention includes the following series of means: The system is mainly composed of three main entities: a server, a terminal, and a user.

[1465] System Overview

[1466] The server collects and preprocesses past incident data and risk information. It uses the preprocessed data set to train and update the generative AI model, analyzes new project documents, and predicts risk factors. It then notifies users of risk prediction results and countermeasures via their devices, helping them to smoothly progress with the project. It also uses an emotion engine to analyze users' emotions and adjust the content of notifications.

[1467] The terminal provides an interface for users to upload documents related to new projects and check the analysis results, allowing users to effectively manage projects through this system.

[1468] Program processing description

[1469] Data collection and preprocessing

[1470] The server collects data from past accident reports, near-miss information, and issue management tables stored within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database and undergoes data cleansing. Data cleansing involves deleting incomplete information, standardizing data formats, and removing unnecessary data. Inconsistencies in data formats from different departments are corrected and necessary metadata is added.

[1471] Training and updating generative AI models

[1472] The server generates a machine learning dataset based on the preprocessed data and trains a generative AI model. This model uses machine learning algorithms such as neural networks (using TensorFlow or PyTorch) to extract and learn risk patterns from past data. The accuracy of the model is evaluated and retrained as necessary.

[1473] New project analysis and risk prediction

[1474] When a user starts a new project, they use their device to upload documents such as requirements specifications and design documents to the server. The server then analyzes these documents using Natural Language Processing (NLP) technology. Natural language processing libraries such as spaCy and NLTK are used for analysis, tokenizing the text data, tagging parts of speech, and performing semantic analysis. A generative AI model compares the results with past risk patterns and predicts potential risk factors that may arise in the new project.

[1475] Analysis result notification and emotion engine

[1476] Based on the analysis results, the server sends risk factors and proposed countermeasures to the device. Specifically, this includes detailed information about the risk, examples of similar cases in the past, and guidelines for countermeasures. The emotion engine analyzes the user's input and behavior to recognize their emotions. Based on this information, the content and format of the risk notification can be adjusted. For example, if the user is feeling stressed, the notification content will be simplified and only the main points will be presented.

[1477] Feedback and System Improvement

[1478] Users can input their feedback on the risk information and advice provided by the system into their device and send it to the server, which analyzes this feedback and combines it with data from the emotion engine to improve the accuracy of the AI ​​model and the overall system.

[1479] Adding specific examples

[1480] Specific prompt examples

[1481] "We have added configuration information for a new machine. Please analyze the risks associated with this."

[1482] "We've added a safety procedure to this project. What risks have you encountered with similar materials in the past?"

[1483] This allows the system to proactively identify risk factors related to industrial equipment and provide feedback to users through appropriate notification methods.

[1484] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1485] Step 1:

[1486] The server collects past accident reports, near-miss information, and issue management tables from within the company. This includes information from each department and content from internal bulletin boards. The collected data is stored in a database. The specific input is raw data from each source, and the output is data stored in the database in a unified format.

[1487] Step 2:

[1488] The server preprocesses the collected data. It performs data cleansing to remove incomplete information, standardize data formats, and remove unnecessary data. For example, it corrects inconsistencies in data formats from different departments. The input is raw data stored in the database, and the output is cleansed data.

[1489] Step 3:

[1490] The server generates a machine learning dataset based on the preprocessed data. This dataset is input to a generative AI model and used to train the AI ​​model to extract and learn risk patterns from past data. The input is cleansed data, and the output is a machine learning dataset.

[1491] Step 4:

[1492] When a user starts a new project, the server provides a function to upload requirements and design documents through the terminal. The user uploads documents related to the project to the terminal. The input is the user's document, and the output is the document sent to the server.

[1493] Step 5:

[1494] The server analyzes the uploaded document using Natural Language Processing (NLP) techniques. It tokenizes the text data, tags it with parts of speech, and performs semantic analysis to extract important relevant information. The input is the document uploaded by the user, and the output is the analysis results.

[1495] Step 6:

[1496] The server inputs the analysis results into a generative AI model to predict potential risk factors for new projects. This involves comparing past risk patterns with current project data. The input is the analysis results data, and the output is risk prediction information.

[1497] Step 7:

[1498] The server notifies the user based on the risk prediction information. At this time, it uses an emotion engine to analyze the user's emotions and adjusts the notification format and content to an appropriate one according to the emotion. For example, if the user is feeling stressed, it selects a concise notification content. The input is the risk prediction information and the user's emotional data, and the output is the adjusted notification content.

[1499] Step 8:

[1500] The terminal provides the user with risk prediction information and countermeasures sent from the server. The user can use this information to adjust the progress of the project. The input is the risk prediction information sent from the server, and the output is the risk information and countermeasures displayed to the user.

[1501] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1502] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1503] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1504] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1505] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left sid...

Claims

1. a means of collecting past incident data and risk information; a means for preprocessing the collected information to generate a machine learning dataset; a means for providing a generative AI model that learns the preprocessed and transformed dataset; and A means of analyzing documents about new projects and predicting risk factors using generative AI models; A system that includes a means for communicating analysis results and risk predictions.

2. The system of claim 1 , further comprising means for cleansing the collected information and correcting inconsistencies.

3. The system of claim 1 , further comprising: means for analyzing the new project documentation using natural language processing techniques.

Images