system

Abstract

A system is provided.SOLUTION: A system comprising: means for collecting historical case information; means for pre-processing the collected case information; means for training a AI model based on the pre-processed case information; means for providing an insight generation API with the trained AI model; means for communicating with a server to send court information and receive insights; means for sending user-entered court information to the server and displaying insights received from the server; and means for a user to enter court information and view insights.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] The main challenges facing lay judges in the lay judge system are anxiety due to a lack of legal knowledge and experience, and the pressure to make fair judgments. They are also susceptible to being influenced by public opinion and personal preconceptions, which could lead to biased judgments. There is a need to resolve these issues and enable lay judges to participate actively and confidently in the lay judge system. [Means for solving the problem]

[0005] The present invention proposes a system that collects past case data, preprocesses it, trains an AI model, and generates and provides opinions based on the trained AI model. This system includes means for communicating with a server to send court information and receive opinions, means for sending court information entered by a user to the server and displaying opinions received from the server, and means for a user to enter court information and view opinions. Furthermore, it provides an information input screen and an information display screen, and is capable of displaying opinions from an opinion generation API in real time. Furthermore, by using a natural language processing model as the AI ​​model, it provides appropriate opinions based on past case data. In this way, it can support ordinary people in making fair and impartial judgments as jurors, reducing their anxiety and burden.

[0006] "Past case data" refers to data that includes records of cases that have been handled in court in the past and the content of judgments.

[0007] "Preprocessing" is the process of cleaning and shaping raw data to convert it into a format that is easier to analyze.

[0008] An "AI model" is an algorithm trained using machine learning techniques that has the ability to make predictions and distinctions for specific tasks.

[0009] An "opinion generation API" is an application programming interface that uses a trained AI model to generate opinions based on input data and provide them to external systems or users.

[0010] A "server" is a computer system that provides various services over a network and stores and processes data.

[0011] A "terminal" is a computer or device that a user directly operates and that communicates with a server.

[0012] "Court information" is information related to a particular trial, such as a summary of the case, evidence, and applicable law.

[0013] A "user interface" is a mechanism that provides a screen and operating means for a user to interact with a system. [Brief explanation of the drawings]

[0014] [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

[0015] 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.

[0016] First, the terms used in the following description will be explained.

[0017] 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).

[0018] 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.

[0019] 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.

[0020] 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.

[0021] 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."

[0022] [First embodiment]

[0023] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0024] 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.

[0025] 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).

[0026] 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.

[0027] 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.

[0028] 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.

[0029] 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.

[0030] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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."

[0035] ---

[0036] This invention relates to a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. This system collects past case data, generates opinions using an AI model trained on that data, and provides these to lay judges to support their judgments.

[0037] System configuration

[0038] Server Configuration

[0039] The server has the following main functions:

[0040] 1. Data Collection and Preprocessing:

[0041] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[0042] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[0043] 2. Training the AI ​​model:

[0044] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[0045] 3. Providing an opinion generation API:

[0046] The server provides an API that uses a trained AI model to generate opinions, which are then returned in real time based on the input court information.

[0047] Device configuration

[0048] The device has the following main features:

[0049] 1. Providing a user interface:

[0050] The terminal provides an information input screen that jurors can use and a screen that displays AI opinions.

[0051] 2. Sending court information and receiving and displaying opinions:

[0052] The terminal sends the trial information entered by the jurors to the server and displays the AI ​​opinion received from the server to the user.

[0053] User operations

[0054] The user uses the system in the following steps:

[0055] 1. Enter your information:

[0056] The user inputs court information (summary of the case and related evidence) through the device's input screen. For example, the user might input "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0057] 2. Transmission of Information:

[0058] The user sends the entered court information to the server via the send button. This operation causes the terminal to send the information to the server.

[0059] 3. Confirming AI's View:

[0060] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may read, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0061] 4. Use in discussions:

[0062] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[0063] Specific examples

[0064] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[0065] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[0066] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[0067] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0068] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[0069] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden they have to make fair judgments, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[0070] The processing flow will be explained below.

[0071] ---

[0072] Server Processing

[0073] Step 1: Collect case law data

[0074] The server collects past case data from a judicial database.

[0075] Data is obtained using APIs and scraping technology and stored in a local database.

[0076] Step 2: Preprocessing the data

[0077] The server cleans the collected case data, specifically by filling in missing values ​​and removing outliers.

[0078] Essential data items (summary of the case, content of the judgment, applicable law, etc.) are extracted and formatted for analysis.

[0079] Step 3: Training the AI ​​model

[0080] The server uses the preprocessed data to train the AI ​​model.

[0081] Using natural language processing techniques (e.g., Transformer or RNN), a model based on case law data is created.

[0082] Evaluate the performance of the trained model and retune it if necessary.

[0083] Step 4: Providing an insight generation API

[0084] The server builds an API that uses the trained AI model to generate opinions.

[0085] This API has the ability to respond in real time based on the court information entered.

[0086] Terminal handling

[0087] Step 1: Provide a user interface

[0088] The terminal provides input and display screens that are easy for jurors to use.

[0089] Specifically, it will have a form for entering court information and a dashboard for displaying opinions.

[0090] Step 2: Enter court information

[0091] Court information entered by the user is aggregated on the terminal.

[0092] For example, enter "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street."

[0093] Step 3: Submit your information

[0094] The terminal transmits the entered trial information to the server.

[0095] Data is transmitted securely using security protocols (such as SSL / TLS).

[0096] Step 4: Receive and display feedback

[0097] The terminal receives the response from the server and displays the view to the user.

[0098] For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was around 5 million yen."

[0099] User Action

[0100] Step 1: Enter your information

[0101] The user enters the court information into the terminal.

[0102] Specifically, you enter a summary of the case and relevant evidence into a text form.

[0103] Step 2: Confirm submission

[0104] The user confirms the input and presses the send button to send the information to the server.

[0105] Step 3: Confirm your views

[0106] The user checks the AI's opinion displayed on the device.

[0107] For example, you may obtain information such as, "Based on past precedents, the average amount of compensation is 5 million yen."

[0108] Step 4: Prepare for the discussion

[0109] Users will use the AI's views as a reference and prepare to debate with other jurors.

[0110] The views expressed will be used as a basis for making fair judgments.

[0111] These are the specific processing steps of the "Lay Judge Helper" system. This system reduces the anxiety and burden on lay judges and enables them to make fair and neutral judgments.

[0112] Example 1

[0113] 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."

[0114] In the current lay judge system, ordinary people can feel anxious and burdened due to their lack of legal knowledge and experience. This makes it difficult to make impartial judgments and can undermine the transparency and fairness of trials. A support system is needed to solve this problem and improve the quality of trials.

[0115] 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.

[0116] In this invention, the server includes means for collecting past judicial data, means for preprocessing the collected judicial data, means for training a generative AI model based on the preprocessed data, means for providing an opinion generation API using the trained generative AI model, means for communicating with a processing device to send court information and receive opinions, means for sending court information entered by a user to the processing device and displaying opinions received from the processing device, and means for a user to input court information and refer to opinions. This enables ordinary people to receive support in making fair judgments as jurors, reducing their anxiety and burden.

[0117] "Past judicial data" refers to data related to past judicial proceedings, such as precedents, rulings, and applicable laws.

[0118] "Preprocessing" refers to a series of processes carried out to improve the quality of collected data, specifically including data cleaning and missing value imputation.

[0119] A "generative AI model" is an artificial intelligence model built using machine learning techniques to extract useful information from data and generate opinions and predictions.

[0120] The "opinion generation API" is an application programming interface that uses a generative AI model to generate opinions based on input information and provide them to external parties.

[0121] A "processing device" is an information processing device, such as a server or client terminal, that manages the input, processing, communication, and output of data.

[0122] "User" is a concept that refers to users who operate and use the system and jurors.

[0123] The "information input screen" is an interface that allows the user to input necessary information into the system.

[0124] The "information display screen" is an interface that displays the results and opinions processed by the system to the user.

[0125] A "natural language processing model" is a machine learning model for understanding and generating human language. Specifically, it includes techniques for tasks such as text analysis, translation, and summarization.

[0126] "Means" is a term that refers to methods, techniques, equipment, processes, etc. used to achieve a particular purpose.

[0127] Server Processing

[0128] Data collection and preprocessing

[0129] The server collects past judicial data to be used in the lay judge system. Specifically, it retrieves case data from the judicial database using an API. The retrieved data is cleaned, and incomplete data and outliers are removed. This cleansing process is carried out until the data is clean and consistent. The specific software used is Python and its libraries (Pandas, NumPy, etc.).

[0130] Training an AI model

[0131] The server trains a generative AI model based on the preprocessed data. This process uses natural language processing (NLP) techniques. Specifically, machine learning libraries such as TensorFlow and PyTorch are used to train the AI ​​model to generate opinions from past case law data. The data is split into training and testing datasets, and the model's accuracy is evaluated and its hyperparameters are adjusted.

[0132] Providing an opinion generation API

[0133] The server designs and implements an API that uses a trained generative AI model to generate opinions. This API is designed in a RESTful format, and when court information is input to a specific endpoint, the generated opinion is returned in real time. The API is deployed using cloud services such as AWS and Google Cloud.

[0134] Terminal handling

[0135] Providing a user interface

[0136] The terminal provides a user interface that allows jurors to input information and confirm opinions. The information input screen contains text boxes and drop-down menus for entering trial information such as a summary of the case and evidence. The information display screen displays the AI's opinions received from the server. Specific front-end frameworks used are React and Vue.js.

[0137] Sending court information and receiving opinions

[0138] The court information entered by the user is sent to the server via the terminal. This operation is performed via an HTTP POST request. The opinion received from the server is displayed on the terminal screen. The terminal sends and receives information in real time according to the user's operations.

[0139] User operations

[0140] Enter information

[0141] The user inputs court information (summary of the case and evidence) through the device's input screen. For example, the user might input, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0142] Sending information

[0143] The user sends the entered information to the server via the send button. This operation causes the terminal to send the court information to the server.

[0144] Confirming AI's View

[0145] The AI's opinion sent from the server is received by the device and displayed to the user. For example, it may say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0146] Use in discussions

[0147] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[0148] Specific examples

[0149] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[0150] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[0151] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[0152] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0153] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[0154] This system will reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and provide support to help them make impartial decisions, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[0155] Prompt Sentence Examples

[0156] Summary of the incident: "Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0157] Sample prompt: "Based on the outline of this case, please tell us your opinion on similar past cases."

[0158] The above is an embodiment of the present invention.

[0159] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0160] Step 1:

[0161] Data collection

[0162] The server collects past judicial data. This data is obtained from the judicial database via API. The API endpoint URL and access key are required as input, and case data is obtained in JSON or CSV format as output. This allows the server to obtain a large amount of case data. Specifically, it periodically runs a script to download the data.

[0163] Step 2:

[0164] Data Preprocessing

[0165] The server preprocesses the collected case data. The input requires the collected case data. The output is a cleaned, consistent dataset. This process involves removing incomplete data and correcting outliers. Specifically, it uses the Python Pandas library to remove unnecessary columns and unify data types.

[0166] Step 3:

[0167] Feature extraction

[0168] The server extracts the necessary features from the preprocessed data. Preprocessed data is required as input. The output is a dataset suitable for training an AI model. Features include a summary of the case, the verdict, and the applicable law. Specific operations include converting the data into lists or vectors using regular expressions and NLP tools.

[0169] Step 4:

[0170] Training an AI model

[0171] The server uses the extracted features to train a generative AI model. A dataset containing the features is required as input. A trained generative AI model is obtained as output. Specifically, the model is trained using machine learning libraries such as TensorFlow and PyTorch. During training, the accuracy of the model is improved by repeatedly learning the dataset.

[0172] Step 5:

[0173] Evaluating and Tuning the Model

[0174] The server evaluates the performance of the generative AI model using a test dataset. A test dataset is required as input. Evaluation results (precision, recall, F-measure, etc.) are obtained as output. If performance is insufficient, hyperparameters are adjusted or the data is reprocessed. Specifically, evaluation is performed using a confusion matrix, and cross-validation is performed if necessary.

[0175] Step 6:

[0176] Providing an opinion generation API

[0177] The server deploys an API that uses a trained generative AI model to generate opinions. Court information is required as input. The generated opinion is output in JSON format. Specifically, the API is built using frameworks such as Flask or FastAPI and deployed on a cloud service.

[0178] Step 7:

[0179] Providing information input screen

[0180] The terminal provides an information input screen where the user can enter court information. The input requires court information to be entered by the user. The output is court information ready to be sent to the server. Specific operation involves building the screen using React and Vue.js, allowing the user to input information intuitively.

[0181] Step 8:

[0182] Sending court information

[0183] The terminal sends the court information entered by the user to the server. The input requires the court information entered by the user. The output is the information accurately sent to the server. Specifically, the information is sent to the server's API using an HTTP POST request.

[0184] Step 9:

[0185] Receiving and displaying AI insights

[0186] The device displays the AI ​​opinion received from the server to the user. The input requires the AI ​​opinion received from the server. The output is the opinion that is displayed to the user. Specifically, it parses the received JSON data and displays it on the screen in an easy-to-read format.

[0187] Step 10:

[0188] Confirming opinions and using them in discussions

[0189] The user checks the AI's opinion displayed on the device and uses it in discussions with other jurors. The input required is the AI's opinion displayed on the device. The output is an opinion that can be used as material for discussion to reach a fair judgment. Specifically, the system operates by advancing discussions based on the opinion and using it to reach a final verdict.

[0190] (Application example 1)

[0191] 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."

[0192] Internal investigations are becoming increasingly important in modern companies and public institutions. However, internal investigators often lack legal knowledge and information on similar cases, making it difficult to conduct fair and prompt investigations. Other issues include inconsistencies in judgment and unclear investigation directions. The present invention aims to solve these problems and provide a system that enables internal investigators to conduct investigations efficiently and fairly.

[0193] 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.

[0194] In this invention, the server includes a means for collecting past case data, a means for preprocessing the collected case data, and a means for training an AI model based on the preprocessed data. This makes it possible to input information about internal investigation cases and send it to the server, and receive and display opinions from the AI ​​model. This allows internal investigators to proceed with investigations based on opinions based on similar past cases, enabling fair and efficient investigations.

[0195] "Past case data" refers to data that includes the results of trials and case information handled by judicial institutions in the past.

[0196] "Preprocessing" refers to the process of complementing missing and outlier values ​​from collected case law data and extracting necessary information.

[0197] An "AI model" is an inference model trained using machine learning algorithms, particularly those that utilize natural language processing techniques.

[0198] An "opinion generation API" is an application program interface that utilizes a trained AI model to generate and provide opinions based on user-entered information.

[0199] "Server" refers to a computer system that receives information from users, communicates with case law databases, and generates and provides opinions based on AI models.

[0200] "Users" are investigators from companies and public institutions who use this system to conduct internal investigations.

[0201] "Information on internal investigation cases" means information containing facts and evidence collected by a company or public institution for the purposes of an internal investigation.

[0202] The "input and transmission means" is an interface that allows users to input information about internal investigation cases into the system and transmit it to the server.

[0203] The "receiving and displaying means" is an interface for displaying to the user the views of the AI ​​model sent from the server.

[0204] A "natural language processing model" is a machine learning model trained to understand text data and respond to questions or requests in natural language.

[0205] This invention is a system that enables companies and public institutions to conduct internal investigations efficiently and fairly. The system collects past case data, uses a trained AI model based on that data to generate opinions, and provides them to internal investigators. The system is primarily composed of three elements: a server, a terminal, and a user.

[0206] System configuration

[0207] Server Configuration

[0208] The server has the following main functions:

[0209] 1. Data Collection and Preprocessing:

[0210] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[0211] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[0212] 2. Training the AI ​​model:

[0213] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[0214] 3. Providing an opinion generation API:

[0215] The server provides an API that uses a trained AI model to generate opinions, which then generate and respond in real time based on the input internal investigation information.

[0216] Device configuration

[0217] The device has the following main features:

[0218] 1. Providing a user interface:

[0219] The device provides an information input screen that users can use and an AI opinion display screen.

[0220] 2. Sending internal investigation information and receiving and displaying opinions:

[0221] The device sends the survey information entered by the user to the server and displays the AI ​​opinion received from the server to the user.

[0222] User operations

[0223] The user uses the system in the following steps:

[0224] 1. Enter your information:

[0225] The user enters internal investigation information (a summary of the case and related evidence) through the device's input screen. For example, the user might enter "Case summary: Employee A is suspected of using company assets for personal purposes."

[0226] 2. Transmission of Information:

[0227] The user sends the entered internal investigation information to the server via the send button. This operation causes the terminal to send the information to the server.

[0228] 3. Confirming AI's View:

[0229] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may be displayed as "In similar cases in the past, dismissal of employees was applied."

[0230] Hardware and software used

[0231] The system of the present invention uses the following hardware and software:

[0232] Hardware:

[0233] Smartphone (Android / iOS)

[0234] Server: A computer system running on the cloud.

[0235] software:

[0236] Programming language: Python

[0237] Library used: requests (sends HTTP requests)

[0238] Natural language processing models: BERT, GPT-3, etc.

[0239] Examples of concrete examples and prompts

[0240] Examples:

[0241] Input information: The user enters, "Employee A is suspected of using company assets for personal purposes."

[0242] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[0243] Display of opinion: The terminal displays to the user the opinion received from the server, "In similar cases in the past, employee dismissal was applied."

[0244] Example prompt sentence:

[0245] "Case summary: An employee is suspected of using company assets for personal use. Relevant evidence: CCTV footage. What are similar cases in the past and what were the results?"

[0246] The above is a specific embodiment for carrying out the present invention. It is expected that this system will improve the efficiency and accuracy of internal investigations and strengthen compliance in companies and public institutions.

[0247] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0248] Step 1:

[0249] The user enters information about the internal investigation case through the input screen on the terminal. Specifically, the user enters "Case summary: Employee A is suspected of using company assets for personal use." The information entered at this time is in text format, and is stored as is as input data on the terminal.

[0250] Step 2:

[0251] The terminal generates an HTTP request to send the survey information entered by the user to the server. Specifically, it uses the Python library requests to convert the input data into JSON format and sends a POST request to the specified API endpoint. Once the request is sent, the input data is sent to the server.

[0252] Step 3:

[0253] The server processes the received investigation information. First, it parses the received JSON data and converts it into structured data. Then it searches for relevant cases in the preprocessed case law database. The search results are used in the next step.

[0254] Step 4:

[0255] The server generates opinions using a trained AI model based on the search results. Specifically, it uses AI models that use natural language processing techniques (e.g., BERT or GPT-3) to derive opinions from past case data. In this process, the AI ​​model uses the input data as prompts and outputs opinions on relevant cases.

[0256] Step 5:

[0257] The server converts the generated opinion into JSON format and prepares it as response data. The prepared response data is sent to the terminal as an HTTP response. This sends the opinion data to the terminal.

[0258] Step 6:

[0259] The device analyzes the opinion data received from the server and displays it in a user-friendly format. Specifically, it converts the received JSON data into structured data and applies it to the UI components for display. This allows the user to visually confirm the opinion.

[0260] Step 7:

[0261] The user discusses the internal investigation based on the opinion received from the server. For example, based on the opinion that "employee dismissal was applied in past similar cases," the user discusses with other investigators and decides the direction of the investigation.

[0262] Through the above steps, the user can proceed with the internal investigation efficiently and fairly.

[0263] 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.

[0264] ---

[0265] This invention is a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. It also reduces the psychological burden on users by incorporating an emotion engine that recognizes the user's emotions. This system collects past case data, uses a trained AI model based on that data to generate opinions, and provides these to lay judges. Furthermore, the emotion engine recognizes the user's emotions and responds accordingly.

[0266] System configuration

[0267] Server Configuration

[0268] 1. Data Collection and Preprocessing:

[0269] The server collects past case data from judicial databases using APIs and scraping technology, and stores it in a local database.

[0270] As preprocessing, the collected data is cleaned (by filling in missing values ​​and removing outliers) and formatted for analysis.

[0271] 2. Training the AI ​​model:

[0272] The server uses the preprocessed data to train an AI model, specifically applying natural language processing techniques to create a model based on past case law data.

[0273] Evaluate the performance of the trained model and retune the model if necessary.

[0274] 3. Providing an opinion generation API:

[0275] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on the input court information.

[0276] Device configuration

[0277] 1. Providing a user interface:

[0278] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[0279] 2. Emotion engine integration:

[0280] The device is equipped with an emotion engine that collects and analyzes the user's facial images and voice data, recognizes the user's emotional state in real time, and transmits the data to a server.

[0281] The emotion engine has the ability to provide appropriate support information and alerts if the user is feeling anxious or stressed.

[0282] User operations

[0283] 1. Enter your information:

[0284] The user accesses the device's input screen and enters court information (an outline of the case and related evidence). For example, the user might enter, "Outline of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0285] 2. Transmission of Information:

[0286] The user confirms the input and presses the send button to send the information to the server. Data is sent securely using a security protocol (such as SSL / TLS).

[0287] 3. Confirming AI's View:

[0288] The server generates an opinion based on the received court information and sends it to the terminal. The terminal then displays the opinion to the user. For example, "In similar past cases, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0289] 4. Recognizing and Responding to Emotions:

[0290] The emotion engine analyzes the user's facial image and voice to recognize their emotions while they are entering court information or confirming opinions. Based on the recognized emotions, it provides appropriate information to reduce the user's anxiety and stress.

[0291] For example, if the user is feeling stressed, the system displays advice and support to help them relax.

[0292] 5. Use in discussions:

[0293] Users can use the AI's views and support from the emotion engine to discuss with other jurors and make fair decisions.

[0294] Specific examples

[0295] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[0296] 1. Information input: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[0297] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[0298] 3. Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0299] 4. Emotion Recognition: The emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression. In this case, it displays relaxation techniques and additional support information.

[0300] 5. Use of discussion: Based on this opinion and support, the user will debate as a juror and make a final decision.

[0301] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden of making fair judgments. Furthermore, by integrating an emotion engine, the psychological burden on users can be further reduced, and the transparency and fairness of trials can be improved.

[0302] The processing flow will be explained below.

[0303] Server Processing

[0304] Step 1:

[0305] The server collects past case data from judicial databases, specifically using APIs and scraping technology to retrieve the data and store it in a local database.

[0306] Step 2:

[0307] The server preprocesses the collected case data, specifically by filling in missing values ​​and removing outliers, and extracting necessary information such as the case summary, the content of the judgment, and the applicable law.

[0308] Step 3:

[0309] The server uses the preprocessed data to train an AI model, uses natural language processing technology to generate a model based on past case law data, and evaluates the performance of the trained model and makes adjustments as necessary.

[0310] Step 4:

[0311] The server uses the trained AI model to build an API that generates opinions, which responds in real time based on the input court information.

[0312] Terminal handling

[0313] Step 1:

[0314] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[0315] Step 2:

[0316] The device collects the user's facial images and voice and provides data for emotion recognition by the emotion engine, which analyzes the collected data and recognizes the user's emotional state in real time.

[0317] Step 3:

[0318] The terminal sends the court information entered by the user to the server. The data is transmitted securely using security protocols (such as SSL / TLS).

[0319] Step 4:

[0320] The device displays the AI's opinion received from the server in real time. For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0321] User Action

[0322] Step 1:

[0323] The user inputs court information (summary of the case and related evidence) into the terminal. Specifically, the user inputs "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0324] Step 2:

[0325] The user checks the input and presses the send button to send the information to the server.

[0326] Step 3:

[0327] The user checks the AI's opinion displayed on the device, and receives information such as, "Based on past court decisions, the average amount of compensation is 5 million yen."

[0328] Step 4:

[0329] While the user is inputting or confirming their opinion, the emotion engine analyzes the user's facial expressions and voice in real time to recognize their emotional state, and if necessary, provides appropriate information to reduce the user's anxiety or stress.

[0330] Step 5:

[0331] Users can use the displayed opinions and supporting information from the emotion engine to discuss with other jurors and use the information to make a fair decision.

[0332] Exemplary Processing Steps

[0333] Step 1:

[0334] The user inputs into the terminal, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light, and suffered a broken bone."

[0335] Step 2:

[0336] The device sends the entered information to the server, which uses a security protocol to send the data securely.

[0337] Step 3:

[0338] The server generates a view using a trained AI model based on the received information. For example, it may send a view to the terminal that, according to past court decisions, the driver of vehicle B was found to be grossly negligent and the average amount of compensation for damages was 5 million yen.

[0339] Step 4:

[0340] The terminal displays the received views to the user, who then confirms the views.

[0341] Step 5:

[0342] The emotion engine analyzes the user's facial image and voice to detect anxiety and stress. If stress is detected, it displays relaxation techniques and additional support information.

[0343] Step 6:

[0344] Based on the displayed opinions and supporting information from the emotion engine, users can discuss with other jurors and make a fair decision.

[0345] In this way, the present invention reduces the anxiety and burden that lay judges have due to their lack of legal knowledge and experience, and by integrating an emotion engine, it further reduces the psychological burden on users, thereby realizing fair and transparent trials.

[0346] Example 2

[0347] 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."

[0348] For ordinary people who participate in the lay judge system, there are issues such as anxiety, psychological stress, and the heavy burden of being expected to make impartial judgments. Another issue is anxiety due to a lack of trial experience and legal knowledge. Furthermore, if the stress and anxiety experienced during trials are left unaddressed, it can have a negative impact on the psychological health of lay judges.

[0349] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training a generative AI model based on the preprocessed data. This makes it possible to provide a system that reduces the anxiety and psychological burden of ordinary people when participating in the lay judge system and supports fair judgment. Furthermore, the server includes means for providing an opinion generation API using the trained generative AI model and means for recognizing the user's emotions, and provides support information according to the user's emotional state. This further reduces the user's psychological burden and improves the transparency and fairness of trials.

[0350] "Past case data" is data collected from judicial databases that includes the results and details of various past trials.

[0351] "Preprocessing" is the process of converting collected data into a format suitable for analysis and training, including cleaning the data, shaping the data, filling in missing values, removing outliers, and tokenizing the text.

[0352] A "generative AI model" is an artificial intelligence model that is trained based on past case data and automatically generates opinions on case law based on given input data.

[0353] The "Opinion Generation API" is an application program interface that uses a trained generative AI model to generate opinions in real time based on input court information and respond to external requests.

[0354] A "server" is a computer system that performs functions such as collecting data, pre-processing, training AI models, and providing an insight generation API.

[0355] "Court information" refers to important information in a trial, such as an outline of the case and relevant evidence.

[0356] "User" refers to a person who uses the system to enter court information and view generated opinions.

[0357] The "emotion engine" is part of a system that analyzes the user's facial image and voice data, recognizes the user's emotional state, and provides appropriate support information based on that information.

[0358] The "information input screen" is a user interface for users to input court information.

[0359] The "information display screen" is a user interface for displaying to the user the views received from the server.

[0360] A "natural language processing model" is an artificial intelligence model that analyzes text data and understands the meaning and structure of sentences.

[0361] This invention is a system that reduces the anxiety and psychological burden of ordinary people participating in the lay judge system and supports fair judgments. In particular, by combining it with an emotion engine that recognizes the user's emotions, the system can minimize the user's psychological burden. This system collects past case data and provides jurors with opinions based on a generative AI model trained on that data. Furthermore, the emotion engine recognizes the user's emotions and responds appropriately.

[0362] Server Configuration

[0363] 1. Data Collection and Preprocessing:

[0364] The server collects past case data from judicial databases using APIs and scraping technology, and the collected data is stored in a local database.

[0365] The server cleans the data (filling in missing values ​​and removing outliers) and prepares it for analysis, which involves tokenizing the data and removing unnecessary symbols and spaces.

[0366] 2. Training the AI ​​model:

[0367] The server trains a generative AI model based on the preprocessed data. Specifically, it uses natural language processing techniques to extract important features from past case data and create a classification model (e.g., SVM or Random Forest).

[0368] The performance of the trained model is evaluated using cross-validation and other methods, and the accuracy and F1 score are calculated.

[0369] 3. Providing an opinion generation API:

[0370] The server builds an API that generates insights using a trained generative AI model. It uses frameworks such as Flask and FastAPI to design a RESTful API.

[0371] The server deploys the view generation API in a cloud environment (e.g., AWS or GCP).

[0372] Device configuration

[0373] 1. Providing a user interface:

[0374] The device provides a user-friendly input and display interface, which uses HTML and CSS to provide a form for entering a case summary and related evidence.

[0375] The display screen provides a dashboard that displays the insights returned from the server in the form of tables and graphs.

[0376] 2. Emotion engine integration:

[0377] The device collects the user's facial images and voice data and analyzes them with an emotion engine, which uses publicly available APIs such as Microsoft Azure's Face API and IBM Watson Tone Analyzer.

[0378] The device sends the recognized emotion data to a server, which provides additional support information and alerts based on the user's emotional state.

[0379] User Actions

[0380] 1. Enter your information:

[0381] The user accesses the device's input screen and enters a summary of the incident and related evidence. For example, the user might enter, "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0382] 2. Transmission of Information:

[0383] The user checks the information entered and presses the send button to send the information to the server. At this time, a security protocol such as SSL / TLS is used for communication.

[0384] 3. Confirming AI's View:

[0385] The server generates an opinion based on the received court information and sends the result to the terminal. The user can check the opinion on the terminal. The message displayed is, "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0386] 4. Recognizing and Responding to Emotions:

[0387] The emotion engine analyzes the user's facial images and voice in real time while inputting or confirming opinions, and recognizes their emotions. For example, if the user is feeling anxious, it will provide relaxation techniques and additional support information.

[0388] 5. Use in discussions:

[0389] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[0390] Specific examples

[0391] For example, if a user wants to participate as a juror in a trial regarding a "pedestrian-vehicle accident," the user will use the system as follows:

[0392] 1. Information input: The user inputs a summary of the incident: "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[0393] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[0394] 3. Display of opinion: The terminal displays the opinion received from the server to the user. It displays the following: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0395] 4. Emotion Recognition: If the emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression, it will display relaxation techniques and additional support information.

[0396] 5. Use of discussion: Based on this opinion and support, users will debate as jurors and make a final decision.

[0397] The present invention provides support to reduce the psychological burden on lay judges and ensure fairer trials.

[0398] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0399] Step 1: Data collection

[0400] The server accesses the judicial database and sends API requests to collect past case data. Specifically, the server retrieves data using HTTP requests and receives the response in JSON format.

[0401] Input: API request parameters (e.g., trial year, type).

[0402] Output: Case data in JSON format.

[0403] What happens: The server sends a request to an API endpoint, parses the response, and stores it in a local database.

[0404] Step 2: Data Preprocessing

[0405] The server cleans the collected case data and formats it for analysis: text data is tokenized, unnecessary symbols and spaces are removed, missing values ​​are imputed, and outliers are removed.

[0406] Input: Case law data in JSON format.

[0407] Output: Preprocessed data (e.g. cleaned text data).

[0408] Specific operation: The server tokenizes the text data, completes missing values ​​in each case data, and removes outliers. For example, Python libraries (NLP libraries and data processing libraries) are used for text tokenization and missing value completion.

[0409] Step 3: Feature extraction

[0410] The server extracts features using the preprocessed data and converts text to numerical data using a TF-IDF vectorizer.

[0411] Input: Preprocessed text data.

[0412] Output: Numerical data (TF-IDF vectors).

[0413] Specific operation: The server applies TF-IDF to the preprocessed text data to extract important features and convert them into numerical data.

[0414] Step 4: Training the AI ​​model

[0415] The server uses the features to train a generative AI model: it builds a classification model (e.g., SVM or Random Forest) and trains it on the training data.

[0416] Input: Numerical data (features) and corresponding labels.

[0417] Output: A trained generative AI model.

[0418] Specific operation: The server feeds the features to the classification model and performs the training process. After training, the performance of the model is evaluated, and cross-validation is performed to calculate the accuracy and F1 score.

[0419] Step 5: Providing an insight generation API

[0420] The server uses a trained generative AI model to build an API that generates insights. A RESTful API is designed using frameworks such as Flask or FastAPI.

[0421] Input: Court information (e.g., case summary and evidence).

[0422] Output: AI-generated opinion.

[0423] Specific operation: The server receives a request from a user, inputs court information into the generative AI model, and generates an opinion. The generated opinion is returned to the client as an API response.

[0424] Step 6: Providing a User Interface

[0425] The terminal will be designed and provided with a form for entering court information and a dashboard displaying opinions, and will have an intuitive design that will make it easy for users to operate.

[0426] Input: User input information (court information).

[0427] Output: Input screen and display screen.

[0428] Specific operation: The terminal designs the form using HTML and CSS, receives the API response, and displays the insights on the dashboard.

[0429] Step 7: Integrating the Emotion Engine

[0430] The device collects the user's facial images and voice data and analyzes them with an emotion engine, for example, using Microsoft Azure's Face API in real time.

[0431] Input: Face images and audio data.

[0432] Output: Emotion analysis results.

[0433] Specific operation: The device sends data collected by the camera and microphone to the emotion analysis API, and sends the returned results to the server. Based on the analysis results, appropriate support information is provided to the user.

[0434] Step 8: Enter your information

[0435] The user accesses the input screen and enters court information (summary of the case and evidence). For example, the user might enter, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0436] Input: Court Information.

[0437] Output: The entered court information.

[0438] Specific operation: The user enters information into the input screen of the device and presses the send button to send the data to the server.

[0439] Step 9: Confirm the AI's opinion

[0440] The server generates an opinion based on the received legal information, which is then sent to the terminal and displayed to the user.

[0441] Input: Court Information.

[0442] Output: AI-generated opinion.

[0443] Specific operation: The server analyzes the court information using the generative AI model, generates an opinion, and sends it to the terminal. The user can check the opinion on the dashboard.

[0444] Step 10: Recognize and respond to emotions

[0445] The emotion engine analyzes the user's facial images and voice while they are typing or confirming their opinion, recognizing their emotions. If they are feeling anxious, it will provide relaxation techniques and additional support information.

[0446] Input: Face images and audio data.

[0447] Output: Sentiment analysis results, supporting information.

[0448] Specific operation: The emotion engine analyzes the user's image and voice in real time and provides appropriate support information based on the results.

[0449] Step 11: Use in discussions

[0450] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[0451] Input: AI-generated opinions, emotional support information.

[0452] Output: Discussion as a judge and final decision.

[0453] Specific operation: Based on the information obtained, the user exchanges opinions with other jurors and conducts the final trial fairly.

[0454] This will reduce the psychological burden on users and create a consistent process that supports fair trials.

[0455] (Application example 2)

[0456] 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."

[0457] The problem that this invention aims to solve is to reduce the stress and anxiety that factory workers feel while working, thereby improving work efficiency and safety. In modern factory work environments, long hours of monotonous work and excessive stress can damage workers' mental health. This can also reduce workers' work performance and negatively impact product quality and production efficiency. To solve this problem, a system is needed that can recognize workers' emotional states in real time and provide appropriate support.

[0458] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training an AI model based on the preprocessed data. The user terminal includes means for providing an opinion generation API using the trained AI model, means for communicating with the server to send court case information and receive opinions, means for sending court case information entered by the user to the server and displaying the opinions received from the server, means for recognizing worker emotions, and means for providing support to workers based on the recognized emotions. This makes it possible to recognize the emotional state of workers working in a factory in real time and provide appropriate support and advice. This reduces worker stress and anxiety, contributing to improved work efficiency and safety.

[0459] "Past case data" is data that includes records of past cases and judgments regarding legal decisions.

[0460] A "collection means" is a device or method that performs a process of obtaining specified data from a particular location and storing it.

[0461] "Preprocessing means" refers to devices or methods that perform processes to prepare and clean collected data for analysis and modeling.

[0462] "Means of training an AI model" refers to the process of using collected and preprocessed data to train an artificial intelligence model and build that model with high accuracy.

[0463] The "means for providing an opinion generation API" is an interface for generating legal opinions from a trained AI model and providing them to external systems or users.

[0464] "Court Information" is information containing facts, evidence, and legal documents relating to a particular court case.

[0465] "User-entered case information" is data about a particular case that is entered into the system by a user.

[0466] A "server" is a central computer system that processes and manages data on a network.

[0467] The "means for receiving an opinion" is a process for receiving the generated legal opinion from the server to the user terminal.

[0468] A "means for displaying an opinion" is a device or method for visually displaying a received legal opinion to a user.

[0469] "Means for recognizing worker emotions" refers to the process of using sensors such as cameras and microphones to analyze a worker's facial expressions and voice to identify their emotional state.

[0470] A "means for providing support to workers based on recognized emotions" is a system or method that suggests appropriate advice or assistance based on the worker's emotional state.

[0471] This invention is an emotion recognition system for reducing stress and anxiety among factory workers and improving work efficiency and safety. This system includes a series of processes, starting with collecting past case data, analyzing emotion data, and providing support.

[0472] System configuration

[0473] Server Configuration

[0474] 1. Data Collection and Preprocessing:

[0475] The server collects historical case data from legal databases and other data sources, using APIs and scraping techniques to retrieve the data and store it in a local database.

[0476] To prepare the collected data for analysis, preprocessing such as cleaning (filling in missing values ​​and removing outliers) and tokenization is performed.

[0477] 2. Training the AI ​​model:

[0478] The server uses the pre-processed dataset to train an AI model, applying natural language processing (NLP) models to create a model capable of generating legal opinions based on historical case law data.

[0479] Evaluate the trained model and retune it if necessary.

[0480] 3. Providing an opinion generation API:

[0481] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on input information, allowing the system to quickly provide legal opinions.

[0482] Configuring the user device

[0483] 1. Information input and display:

[0484] The terminal provides an interface where users can input legal information and emotional data, and in particular, build user-friendly forms and dashboards.

[0485] The entered information is sent to the server, and the view returned by the server is displayed.

[0486] 2. Emotion recognition engine integration:

[0487] The device uses a camera and microphone to collect facial images and voice data of the worker, and sends them to an emotion recognition engine, which recognizes the worker's emotional state in real time.

[0488] Depending on the emotion recognized, the device will display appropriate alerts and support information. For example, if stress levels increase, an alert will be displayed suggesting ways to relax.

[0489] Hardware and software used

[0490] Hardware: Cameras, microphones, database servers, user devices (computers, smartphones, tablets, etc.).

[0491] Software: Google Cloud Vision API, Azure Cognitive Services, MySQL.

[0492] Specific examples

[0493] 1. Information input: The user accesses a terminal in the factory and inputs court information and the emotional state of the worker. For example, the court information might be input as follows: "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B. Pedestrian B suffered a broken bone."

[0494] 2. Server processing: The server analyzes the received information and generates a view using the trained AI model. The view is that "previous court decisions have found the driver of vehicle B to be grossly negligent, and the average amount of compensation is 5 million yen."

[0495] 3. View display: The terminal displays the worker's emotional state in real time along with the view received from the server and provides appropriate advice as needed.

[0496] 4. Emotion Recognition: The emotion recognition engine can identify anxiety from a worker's facial expression and display relaxation techniques or additional support information.

[0497] Prompt Sentence Examples

[0498] "Provide emotional awareness to workers and suggest ways to provide appropriate support and advice where necessary."

[0499] "Consider designing a system that measures workers' stress levels in real time and suggests ways to relax."

[0500] In this way, the present invention makes it possible to recognize the emotional state of workers in a factory in real time and provide appropriate support, thereby reducing stress and anxiety among workers and improving work efficiency and safety.

[0501] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0502] Step 1:

[0503] Data collection and preprocessing:

[0504] The server collects past case data from legal databases and other data sources using APIs and scraping technology. The collected data is stored in a local database. The input is case data obtained through APIs and scraping technology, and the output is formatted data. This data undergoes preprocessing such as missing value completion, outlier removal, and tokenization.

[0505] Step 2:

[0506] Training the AI ​​model:

[0507] The server uses the preprocessed dataset to train a natural language processing (NLP) model, with the input being the preprocessed case data and the output being the trained AI model, which is now capable of generating legal opinions based on past case data.

[0508] Step 3:

[0509] Providing an opinion generation API:

[0510] The server builds and provides an API that generates opinions using a trained AI model. The input is the court information entered by the user, and the output is the legal opinion generated by the AI ​​model. This allows the server to provide legal opinions in real time based on the input information.

[0511] Step 4:

[0512] Information entry and display:

[0513] The user inputs the court information using the terminal interface. The input information is sent from the terminal to the server. The input is the court information entered by the user, and the output is the information sent to the server. The opinion returned by the server is displayed on the terminal for the user to visually confirm.

[0514] Step 5:

[0515] Emotion data collection:

[0516] The device collects facial images and voice data of workers using a camera and microphone. The input is the facial images and voice data acquired through the camera and microphone, and the output is stored in the device and sent to an emotion recognition engine.

[0517] Step 6:

[0518] Emotion recognition engine analysis:

[0519] The emotion recognition engine analyzes the transmitted facial image and voice data to identify the worker's emotional state. The input is the facial image and voice data, and the output is the analyzed emotional state (stress, anxiety, joy, etc.).

[0520] Step 7:

[0521] Providing emotional support:

[0522] The device provides appropriate support and advice to workers based on their recognized emotional state. For example, if stress levels are high, it suggests relaxation techniques or short breaks. The input is the recognized emotional state data, and the output is support information or alerts.

[0523] Step 8:

[0524] Data Storage and Feedback:

[0525] The terminal stores the worker's emotional data and response history, and later analyzes them to propose improvement measures. The input is emotional data and response history, and the output is data stored in a database. This makes it possible to continuously improve the performance of the entire system.

[0526] 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.

[0527] 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.

[0528] 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.

[0529] [Second embodiment]

[0530] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0531] 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.

[0532] 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).

[0533] 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.

[0534] 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.

[0535] 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).

[0536] 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.

[0537] 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.

[0538] 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.

[0539] 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.

[0540] 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.

[0541] 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."

[0542] ---

[0543] This invention relates to a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. This system collects past case data, generates opinions using an AI model trained on that data, and provides these to lay judges to support their judgments.

[0544] System configuration

[0545] Server Configuration

[0546] The server has the following main functions:

[0547] 1. Data Collection and Preprocessing:

[0548] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[0549] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[0550] 2. Training the AI ​​model:

[0551] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[0552] 3. Providing an opinion generation API:

[0553] The server provides an API that uses a trained AI model to generate opinions, which are then returned in real time based on the input court information.

[0554] Device configuration

[0555] The device has the following main features:

[0556] 1. Providing a user interface:

[0557] The terminal provides an information input screen that jurors can use and a screen that displays AI opinions.

[0558] 2. Sending court information and receiving and displaying opinions:

[0559] The terminal sends the trial information entered by the jurors to the server and displays the AI ​​opinion received from the server to the user.

[0560] User operations

[0561] The user uses the system in the following steps:

[0562] 1. Enter your information:

[0563] The user inputs court information (summary of the case and related evidence) through the device's input screen. For example, the user might input "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0564] 2. Transmission of Information:

[0565] The user sends the entered court information to the server via the send button. This operation causes the terminal to send the information to the server.

[0566] 3. Confirming AI's View:

[0567] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may read, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0568] 4. Use in discussions:

[0569] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[0570] Specific examples

[0571] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[0572] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[0573] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[0574] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0575] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[0576] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden they have to make fair judgments, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[0577] The processing flow will be explained below.

[0578] ---

[0579] Server Processing

[0580] Step 1: Collect case law data

[0581] The server collects past case data from a judicial database.

[0582] Data is obtained using APIs and scraping technology and stored in a local database.

[0583] Step 2: Preprocessing the data

[0584] The server cleans the collected case data, specifically by filling in missing values ​​and removing outliers.

[0585] Essential data items (summary of the case, content of the judgment, applicable law, etc.) are extracted and formatted for analysis.

[0586] Step 3: Training the AI ​​model

[0587] The server uses the preprocessed data to train the AI ​​model.

[0588] Using natural language processing techniques (e.g., Transformer or RNN), a model based on case law data is created.

[0589] Evaluate the performance of the trained model and retune it if necessary.

[0590] Step 4: Providing an insight generation API

[0591] The server builds an API that uses the trained AI model to generate opinions.

[0592] This API has the ability to respond in real time based on the court information entered.

[0593] Terminal handling

[0594] Step 1: Provide a user interface

[0595] The terminal provides input and display screens that are easy for jurors to use.

[0596] Specifically, it will have a form for entering court information and a dashboard for displaying opinions.

[0597] Step 2: Enter court information

[0598] Court information entered by the user is aggregated on the terminal.

[0599] For example, enter "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street."

[0600] Step 3: Submit your information

[0601] The terminal transmits the entered trial information to the server.

[0602] Data is transmitted securely using security protocols (such as SSL / TLS).

[0603] Step 4: Receive and display feedback

[0604] The terminal receives the response from the server and displays the view to the user.

[0605] For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was around 5 million yen."

[0606] User Action

[0607] Step 1: Enter your information

[0608] The user enters the court information into the terminal.

[0609] Specifically, you enter a summary of the case and relevant evidence into a text form.

[0610] Step 2: Confirm submission

[0611] The user confirms the input and presses the send button to send the information to the server.

[0612] Step 3: Confirm your views

[0613] The user checks the AI's opinion displayed on the device.

[0614] For example, you may obtain information such as, "Based on past precedents, the average amount of compensation is 5 million yen."

[0615] Step 4: Prepare for the discussion

[0616] Users will use the AI's views as a reference and prepare to debate with other jurors.

[0617] The views expressed will be used as a basis for making fair judgments.

[0618] These are the specific processing steps of the "Lay Judge Helper" system. This system reduces the anxiety and burden on lay judges and enables them to make fair and neutral judgments.

[0619] Example 1

[0620] 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."

[0621] In the current lay judge system, ordinary people can feel anxious and burdened due to their lack of legal knowledge and experience. This makes it difficult to make impartial judgments and can undermine the transparency and fairness of trials. A support system is needed to solve this problem and improve the quality of trials.

[0622] 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.

[0623] In this invention, the server includes means for collecting past judicial data, means for preprocessing the collected judicial data, means for training a generative AI model based on the preprocessed data, means for providing an opinion generation API using the trained generative AI model, means for communicating with a processing device to send court information and receive opinions, means for sending court information entered by a user to the processing device and displaying opinions received from the processing device, and means for a user to input court information and refer to opinions. This enables ordinary people to receive support in making fair judgments as jurors, reducing their anxiety and burden.

[0624] "Past judicial data" refers to data related to past judicial proceedings, such as precedents, rulings, and applicable laws.

[0625] "Preprocessing" refers to a series of processes carried out to improve the quality of collected data, specifically including data cleaning and missing value imputation.

[0626] A "generative AI model" is an artificial intelligence model built using machine learning techniques to extract useful information from data and generate opinions and predictions.

[0627] The "opinion generation API" is an application programming interface that uses a generative AI model to generate opinions based on input information and provide them to external parties.

[0628] A "processing device" is an information processing device, such as a server or client terminal, that manages the input, processing, communication, and output of data.

[0629] "User" is a concept that refers to users who operate and use the system and jurors.

[0630] The "information input screen" is an interface that allows the user to input necessary information into the system.

[0631] The "information display screen" is an interface that displays the results and opinions processed by the system to the user.

[0632] A "natural language processing model" is a machine learning model for understanding and generating human language. Specifically, it includes techniques for tasks such as text analysis, translation, and summarization.

[0633] "Means" is a term that refers to methods, techniques, equipment, processes, etc. used to achieve a particular purpose.

[0634] Server Processing

[0635] Data collection and preprocessing

[0636] The server collects past judicial data to be used in the lay judge system. Specifically, it retrieves case data from the judicial database using an API. The retrieved data is cleaned, and incomplete data and outliers are removed. This cleansing process is carried out until the data is clean and consistent. The specific software used is Python and its libraries (Pandas, NumPy, etc.).

[0637] Training an AI model

[0638] The server trains a generative AI model based on the preprocessed data. This process uses natural language processing (NLP) techniques. Specifically, machine learning libraries such as TensorFlow and PyTorch are used to train the AI ​​model to generate opinions from past case law data. The data is split into training and testing datasets, and the model's accuracy is evaluated and its hyperparameters are adjusted.

[0639] Providing an opinion generation API

[0640] The server designs and implements an API that uses a trained generative AI model to generate opinions. This API is designed in a RESTful format, and when court information is input to a specific endpoint, the generated opinion is returned in real time. The API is deployed using cloud services such as AWS and Google Cloud.

[0641] Terminal handling

[0642] Providing a user interface

[0643] The terminal provides a user interface that allows jurors to input information and confirm opinions. The information input screen contains text boxes and drop-down menus for entering trial information such as a summary of the case and evidence. The information display screen displays the AI's opinions received from the server. Specific front-end frameworks used are React and Vue.js.

[0644] Sending court information and receiving opinions

[0645] The court information entered by the user is sent to the server via the terminal. This operation is performed via an HTTP POST request. The opinion received from the server is displayed on the terminal screen. The terminal sends and receives information in real time according to the user's operations.

[0646] User operations

[0647] Enter information

[0648] The user inputs court information (summary of the case and evidence) through the device's input screen. For example, the user might input, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0649] Sending information

[0650] The user sends the entered information to the server via the send button. This operation causes the terminal to send the court information to the server.

[0651] Confirming AI's View

[0652] The AI's opinion sent from the server is received by the device and displayed to the user. For example, it may say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0653] Use in discussions

[0654] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[0655] Specific examples

[0656] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[0657] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[0658] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[0659] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0660] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[0661] This system will reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and provide support to help them make impartial decisions, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[0662] Prompt Sentence Examples

[0663] Summary of the incident: "Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0664] Sample prompt: "Based on the outline of this case, please tell us your opinion on similar past cases."

[0665] The above is an embodiment of the present invention.

[0666] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0667] Step 1:

[0668] Data collection

[0669] The server collects past judicial data. This data is obtained from the judicial database via API. The API endpoint URL and access key are required as input, and case data is obtained in JSON or CSV format as output. This allows the server to obtain a large amount of case data. Specifically, it periodically runs a script to download the data.

[0670] Step 2:

[0671] Data Preprocessing

[0672] The server preprocesses the collected case data. The input requires the collected case data. The output is a cleaned, consistent dataset. This process involves removing incomplete data and correcting outliers. Specifically, it uses the Python Pandas library to remove unnecessary columns and unify data types.

[0673] Step 3:

[0674] Feature extraction

[0675] The server extracts the necessary features from the preprocessed data. Preprocessed data is required as input. The output is a dataset suitable for training an AI model. Features include a summary of the case, the verdict, and the applicable law. Specific operations include converting the data into lists or vectors using regular expressions and NLP tools.

[0676] Step 4:

[0677] Training an AI model

[0678] The server uses the extracted features to train a generative AI model. A dataset containing the features is required as input. A trained generative AI model is obtained as output. Specifically, the model is trained using machine learning libraries such as TensorFlow and PyTorch. During training, the accuracy of the model is improved by repeatedly learning the dataset.

[0679] Step 5:

[0680] Evaluating and Tuning the Model

[0681] The server evaluates the performance of the generative AI model using a test dataset. A test dataset is required as input. Evaluation results (precision, recall, F-measure, etc.) are obtained as output. If performance is insufficient, hyperparameters are adjusted or the data is reprocessed. Specifically, evaluation is performed using a confusion matrix, and cross-validation is performed if necessary.

[0682] Step 6:

[0683] Providing an opinion generation API

[0684] The server deploys an API that uses a trained generative AI model to generate opinions. Court information is required as input. The generated opinion is output in JSON format. Specifically, the API is built using frameworks such as Flask or FastAPI and deployed on a cloud service.

[0685] Step 7:

[0686] Providing information input screen

[0687] The terminal provides an information input screen where the user can enter court information. The input requires court information to be entered by the user. The output is court information ready to be sent to the server. Specific operation involves building the screen using React and Vue.js, allowing the user to input information intuitively.

[0688] Step 8:

[0689] Sending court information

[0690] The terminal sends the court information entered by the user to the server. The input requires the court information entered by the user. The output is the information accurately sent to the server. Specifically, the information is sent to the server's API using an HTTP POST request.

[0691] Step 9:

[0692] Receiving and displaying AI insights

[0693] The device displays the AI ​​opinion received from the server to the user. The input requires the AI ​​opinion received from the server. The output is the opinion that is displayed to the user. Specifically, it parses the received JSON data and displays it on the screen in an easy-to-read format.

[0694] Step 10:

[0695] Confirming opinions and using them in discussions

[0696] The user checks the AI's opinion displayed on the device and uses it in discussions with other jurors. The input required is the AI's opinion displayed on the device. The output is an opinion that can be used as material for discussion to reach a fair judgment. Specifically, the system operates by advancing discussions based on the opinion and using it to reach a final verdict.

[0697] (Application example 1)

[0698] 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."

[0699] Internal investigations are becoming increasingly important in modern companies and public institutions. However, internal investigators often lack legal knowledge and information on similar cases, making it difficult to conduct fair and prompt investigations. Other issues include inconsistencies in judgment and unclear investigation directions. The present invention aims to solve these problems and provide a system that enables internal investigators to conduct investigations efficiently and fairly.

[0700] 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.

[0701] In this invention, the server includes a means for collecting past case data, a means for preprocessing the collected case data, and a means for training an AI model based on the preprocessed data. This makes it possible to input information about internal investigation cases and send it to the server, and receive and display opinions from the AI ​​model. This allows internal investigators to proceed with investigations based on opinions based on similar past cases, enabling fair and efficient investigations.

[0702] "Past case data" refers to data that includes the results of trials and case information handled by judicial institutions in the past.

[0703] "Preprocessing" refers to the process of complementing missing and outlier values ​​from collected case law data and extracting necessary information.

[0704] An "AI model" is an inference model trained using machine learning algorithms, particularly those that utilize natural language processing techniques.

[0705] An "opinion generation API" is an application program interface that utilizes a trained AI model to generate and provide opinions based on user-entered information.

[0706] "Server" refers to a computer system that receives information from users, communicates with case law databases, and generates and provides opinions based on AI models.

[0707] "Users" are investigators from companies and public institutions who use this system to conduct internal investigations.

[0708] "Information on internal investigation cases" means information containing facts and evidence collected by a company or public institution for the purposes of an internal investigation.

[0709] The "input and transmission means" is an interface that allows users to input information about internal investigation cases into the system and transmit it to the server.

[0710] The "receiving and displaying means" is an interface for displaying to the user the views of the AI ​​model sent from the server.

[0711] A "natural language processing model" is a machine learning model trained to understand text data and respond to questions or requests in natural language.

[0712] This invention is a system that enables companies and public institutions to conduct internal investigations efficiently and fairly. The system collects past case data, uses a trained AI model based on that data to generate opinions, and provides them to internal investigators. The system is primarily composed of three elements: a server, a terminal, and a user.

[0713] System configuration

[0714] Server Configuration

[0715] The server has the following main functions:

[0716] 1. Data Collection and Preprocessing:

[0717] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[0718] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[0719] 2. Training the AI ​​model:

[0720] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[0721] 3. Providing an opinion generation API:

[0722] The server provides an API that uses a trained AI model to generate opinions, which then generate and respond in real time based on the input internal investigation information.

[0723] Device configuration

[0724] The device has the following main features:

[0725] 1. Providing a user interface:

[0726] The device provides an information input screen that users can use and an AI opinion display screen.

[0727] 2. Sending internal investigation information and receiving and displaying opinions:

[0728] The device sends the survey information entered by the user to the server and displays the AI ​​opinion received from the server to the user.

[0729] User operations

[0730] The user uses the system in the following steps:

[0731] 1. Enter your information:

[0732] The user enters internal investigation information (a summary of the case and related evidence) through the device's input screen. For example, the user might enter "Case summary: Employee A is suspected of using company assets for personal purposes."

[0733] 2. Transmission of Information:

[0734] The user sends the entered internal investigation information to the server via the send button. This operation causes the terminal to send the information to the server.

[0735] 3. Confirming AI's View:

[0736] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may be displayed as "In similar cases in the past, dismissal of employees was applied."

[0737] Hardware and software used

[0738] The system of the present invention uses the following hardware and software:

[0739] Hardware:

[0740] Smartphone (Android / iOS)

[0741] Server: A computer system running on the cloud.

[0742] software:

[0743] Programming language: Python

[0744] Library used: requests (sends HTTP requests)

[0745] Natural language processing models: BERT, GPT-3, etc.

[0746] Examples of concrete examples and prompts

[0747] Examples:

[0748] Input information: The user enters, "Employee A is suspected of using company assets for personal purposes."

[0749] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[0750] Display of opinion: The terminal displays to the user the opinion received from the server, "In similar cases in the past, employee dismissal was applied."

[0751] Example prompt sentence:

[0752] "Case summary: An employee is suspected of using company assets for personal use. Relevant evidence: CCTV footage. What are similar cases in the past and what were the results?"

[0753] The above is a specific embodiment for carrying out the present invention. It is expected that this system will improve the efficiency and accuracy of internal investigations and strengthen compliance in companies and public institutions.

[0754] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0755] Step 1:

[0756] The user enters information about the internal investigation case through the input screen on the terminal. Specifically, the user enters "Case summary: Employee A is suspected of using company assets for personal use." The information entered at this time is in text format, and is stored as is as input data on the terminal.

[0757] Step 2:

[0758] The terminal generates an HTTP request to send the survey information entered by the user to the server. Specifically, it uses the Python library requests to convert the input data into JSON format and sends a POST request to the specified API endpoint. Once the request is sent, the input data is sent to the server.

[0759] Step 3:

[0760] The server processes the received investigation information. First, it parses the received JSON data and converts it into structured data. Then it searches for relevant cases in the preprocessed case law database. The search results are used in the next step.

[0761] Step 4:

[0762] The server generates opinions using a trained AI model based on the search results. Specifically, it uses AI models that use natural language processing techniques (e.g., BERT or GPT-3) to derive opinions from past case data. In this process, the AI ​​model uses the input data as prompts and outputs opinions on relevant cases.

[0763] Step 5:

[0764] The server converts the generated opinion into JSON format and prepares it as response data. The prepared response data is sent to the terminal as an HTTP response. This sends the opinion data to the terminal.

[0765] Step 6:

[0766] The device analyzes the opinion data received from the server and displays it in a user-friendly format. Specifically, it converts the received JSON data into structured data and applies it to the UI components for display. This allows the user to visually confirm the opinion.

[0767] Step 7:

[0768] The user discusses the internal investigation based on the opinion received from the server. For example, based on the opinion that "employee dismissal was applied in past similar cases," the user discusses with other investigators and decides the direction of the investigation.

[0769] Through the above steps, the user can proceed with the internal investigation efficiently and fairly.

[0770] 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.

[0771] ---

[0772] This invention is a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. It also reduces the psychological burden on users by incorporating an emotion engine that recognizes the user's emotions. This system collects past case data, uses a trained AI model based on that data to generate opinions, and provides these to lay judges. Furthermore, the emotion engine recognizes the user's emotions and responds accordingly.

[0773] System configuration

[0774] Server Configuration

[0775] 1. Data Collection and Preprocessing:

[0776] The server collects past case data from judicial databases using APIs and scraping technology, and stores it in a local database.

[0777] As preprocessing, the collected data is cleaned (by filling in missing values ​​and removing outliers) and formatted for analysis.

[0778] 2. Training the AI ​​model:

[0779] The server uses the preprocessed data to train an AI model, specifically applying natural language processing techniques to create a model based on past case law data.

[0780] Evaluate the performance of the trained model and retune the model if necessary.

[0781] 3. Providing an opinion generation API:

[0782] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on the input court information.

[0783] Device configuration

[0784] 1. Providing a user interface:

[0785] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[0786] 2. Emotion engine integration:

[0787] The device is equipped with an emotion engine that collects and analyzes the user's facial images and voice data, recognizes the user's emotional state in real time, and transmits the data to a server.

[0788] The emotion engine has the ability to provide appropriate support information and alerts if the user is feeling anxious or stressed.

[0789] User operations

[0790] 1. Enter your information:

[0791] The user accesses the device's input screen and enters court information (an outline of the case and related evidence). For example, the user might enter, "Outline of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0792] 2. Transmission of Information:

[0793] The user confirms the input and presses the send button to send the information to the server. Data is sent securely using a security protocol (such as SSL / TLS).

[0794] 3. Confirming AI's View:

[0795] The server generates an opinion based on the received court information and sends it to the terminal. The terminal then displays the opinion to the user. For example, "In similar past cases, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0796] 4. Recognizing and Responding to Emotions:

[0797] The emotion engine analyzes the user's facial image and voice to recognize their emotions while they are entering court information or confirming opinions. Based on the recognized emotions, it provides appropriate information to reduce the user's anxiety and stress.

[0798] For example, if the user is feeling stressed, the system displays advice and support to help them relax.

[0799] 5. Use in discussions:

[0800] Users can use the AI's views and support from the emotion engine to discuss with other jurors and make fair decisions.

[0801] Specific examples

[0802] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[0803] 1. Information input: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[0804] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[0805] 3. Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0806] 4. Emotion Recognition: The emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression. In this case, it displays relaxation techniques and additional support information.

[0807] 5. Use of discussion: Based on this opinion and support, the user will debate as a juror and make a final decision.

[0808] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden of making fair judgments. Furthermore, by integrating an emotion engine, the psychological burden on users can be further reduced, and the transparency and fairness of trials can be improved.

[0809] The processing flow will be explained below.

[0810] Server Processing

[0811] Step 1:

[0812] The server collects past case data from judicial databases, specifically using APIs and scraping technology to retrieve the data and store it in a local database.

[0813] Step 2:

[0814] The server preprocesses the collected case data, specifically by filling in missing values ​​and removing outliers, and extracting necessary information such as the case summary, the content of the judgment, and the applicable law.

[0815] Step 3:

[0816] The server uses the preprocessed data to train an AI model, uses natural language processing technology to generate a model based on past case law data, and evaluates the performance of the trained model and makes adjustments as necessary.

[0817] Step 4:

[0818] The server uses the trained AI model to build an API that generates opinions, which responds in real time based on the input court information.

[0819] Terminal handling

[0820] Step 1:

[0821] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[0822] Step 2:

[0823] The device collects the user's facial images and voice and provides data for emotion recognition by the emotion engine, which analyzes the collected data and recognizes the user's emotional state in real time.

[0824] Step 3:

[0825] The terminal sends the court information entered by the user to the server. The data is transmitted securely using security protocols (such as SSL / TLS).

[0826] Step 4:

[0827] The device displays the AI's opinion received from the server in real time. For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[0828] User Action

[0829] Step 1:

[0830] The user inputs court information (summary of the case and related evidence) into the terminal. Specifically, the user inputs "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0831] Step 2:

[0832] The user checks the input and presses the send button to send the information to the server.

[0833] Step 3:

[0834] The user checks the AI's opinion displayed on the device, and receives information such as, "Based on past court decisions, the average amount of compensation is 5 million yen."

[0835] Step 4:

[0836] While the user is inputting or confirming their opinion, the emotion engine analyzes the user's facial expressions and voice in real time to recognize their emotional state, and if necessary, provides appropriate information to reduce the user's anxiety or stress.

[0837] Step 5:

[0838] Users can use the displayed opinions and supporting information from the emotion engine to discuss with other jurors and use the information to make a fair decision.

[0839] Exemplary Processing Steps

[0840] Step 1:

[0841] The user inputs into the terminal, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light, and suffered a broken bone."

[0842] Step 2:

[0843] The device sends the entered information to the server, which uses a security protocol to send the data securely.

[0844] Step 3:

[0845] The server generates a view using a trained AI model based on the received information. For example, it may send a view to the terminal that, according to past court decisions, the driver of vehicle B was found to be grossly negligent and the average amount of compensation for damages was 5 million yen.

[0846] Step 4:

[0847] The terminal displays the received views to the user, who then confirms the views.

[0848] Step 5:

[0849] The emotion engine analyzes the user's facial image and voice to detect anxiety and stress. If stress is detected, it displays relaxation techniques and additional support information.

[0850] Step 6:

[0851] Based on the displayed opinions and supporting information from the emotion engine, users can discuss with other jurors and make a fair decision.

[0852] In this way, the present invention reduces the anxiety and burden that lay judges have due to their lack of legal knowledge and experience, and by integrating an emotion engine, it further reduces the psychological burden on users, thereby realizing fair and transparent trials.

[0853] Example 2

[0854] 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."

[0855] For ordinary people who participate in the lay judge system, there are issues such as anxiety, psychological stress, and the heavy burden of being expected to make impartial judgments. Another issue is anxiety due to a lack of trial experience and legal knowledge. Furthermore, if the stress and anxiety experienced during trials are left unaddressed, it can have a negative impact on the psychological health of lay judges.

[0856] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training a generative AI model based on the preprocessed data. This makes it possible to provide a system that reduces the anxiety and psychological burden of ordinary people when participating in the lay judge system and supports fair judgment. Furthermore, the server includes means for providing an opinion generation API using the trained generative AI model and means for recognizing the user's emotions, and provides support information according to the user's emotional state. This further reduces the user's psychological burden and improves the transparency and fairness of trials.

[0857] "Past case data" is data collected from judicial databases that includes the results and details of various past trials.

[0858] "Preprocessing" is the process of converting collected data into a format suitable for analysis and training, including cleaning the data, shaping the data, filling in missing values, removing outliers, and tokenizing the text.

[0859] A "generative AI model" is an artificial intelligence model that is trained based on past case data and automatically generates opinions on case law based on given input data.

[0860] The "Opinion Generation API" is an application program interface that uses a trained generative AI model to generate opinions in real time based on input court information and respond to external requests.

[0861] A "server" is a computer system that performs functions such as collecting data, pre-processing, training AI models, and providing an insight generation API.

[0862] "Court information" refers to important information in a trial, such as an outline of the case and relevant evidence.

[0863] "User" refers to a person who uses the system to enter court information and view generated opinions.

[0864] The "emotion engine" is part of a system that analyzes the user's facial image and voice data, recognizes the user's emotional state, and provides appropriate support information based on that information.

[0865] The "information input screen" is a user interface for users to input court information.

[0866] The "information display screen" is a user interface for displaying to the user the views received from the server.

[0867] A "natural language processing model" is an artificial intelligence model that analyzes text data and understands the meaning and structure of sentences.

[0868] This invention is a system that reduces the anxiety and psychological burden of ordinary people participating in the lay judge system and supports fair judgments. In particular, by combining it with an emotion engine that recognizes the user's emotions, the system can minimize the user's psychological burden. This system collects past case data and provides jurors with opinions based on a generative AI model trained on that data. Furthermore, the emotion engine recognizes the user's emotions and responds appropriately.

[0869] Server Configuration

[0870] 1. Data Collection and Preprocessing:

[0871] The server collects past case data from judicial databases using APIs and scraping technology, and the collected data is stored in a local database.

[0872] The server cleans the data (filling in missing values ​​and removing outliers) and prepares it for analysis, which involves tokenizing the data and removing unnecessary symbols and spaces.

[0873] 2. Training the AI ​​model:

[0874] The server trains a generative AI model based on the preprocessed data. Specifically, it uses natural language processing techniques to extract important features from past case data and create a classification model (e.g., SVM or Random Forest).

[0875] The performance of the trained model is evaluated using cross-validation and other methods, and the accuracy and F1 score are calculated.

[0876] 3. Providing an opinion generation API:

[0877] The server builds an API that generates insights using a trained generative AI model. It uses frameworks such as Flask and FastAPI to design a RESTful API.

[0878] The server deploys the view generation API in a cloud environment (e.g., AWS or GCP).

[0879] Device configuration

[0880] 1. Providing a user interface:

[0881] The device provides a user-friendly input and display interface, which uses HTML and CSS to provide a form for entering a case summary and related evidence.

[0882] The display screen provides a dashboard that displays the insights returned from the server in the form of tables and graphs.

[0883] 2. Emotion engine integration:

[0884] The device collects the user's facial images and voice data and analyzes them with an emotion engine, which uses publicly available APIs such as Microsoft Azure's Face API and IBM Watson Tone Analyzer.

[0885] The device sends the recognized emotion data to a server, which provides additional support information and alerts based on the user's emotional state.

[0886] User Actions

[0887] 1. Enter your information:

[0888] The user accesses the device's input screen and enters a summary of the incident and related evidence. For example, the user might enter, "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0889] 2. Transmission of Information:

[0890] The user checks the information entered and presses the send button to send the information to the server. At this time, a security protocol such as SSL / TLS is used for communication.

[0891] 3. Confirming AI's View:

[0892] The server generates an opinion based on the received court information and sends the result to the terminal. The user can check the opinion on the terminal. The message displayed is, "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0893] 4. Recognizing and Responding to Emotions:

[0894] The emotion engine analyzes the user's facial images and voice in real time while inputting or confirming opinions, and recognizes their emotions. For example, if the user is feeling anxious, it will provide relaxation techniques and additional support information.

[0895] 5. Use in discussions:

[0896] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[0897] Specific examples

[0898] For example, if a user wants to participate as a juror in a trial regarding a "pedestrian-vehicle accident," the user will use the system as follows:

[0899] 1. Information input: The user inputs a summary of the incident: "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[0900] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[0901] 3. Display of opinion: The terminal displays the opinion received from the server to the user. It displays the following: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[0902] 4. Emotion Recognition: If the emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression, it will display relaxation techniques and additional support information.

[0903] 5. Use of discussion: Based on this opinion and support, users will debate as jurors and make a final decision.

[0904] The present invention provides support to reduce the psychological burden on lay judges and ensure fairer trials.

[0905] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0906] Step 1: Data collection

[0907] The server accesses the judicial database and sends API requests to collect past case data. Specifically, the server retrieves data using HTTP requests and receives the response in JSON format.

[0908] Input: API request parameters (e.g., trial year, type).

[0909] Output: Case data in JSON format.

[0910] What happens: The server sends a request to an API endpoint, parses the response, and stores it in a local database.

[0911] Step 2: Data Preprocessing

[0912] The server cleans the collected case data and formats it for analysis: text data is tokenized, unnecessary symbols and spaces are removed, missing values ​​are imputed, and outliers are removed.

[0913] Input: Case law data in JSON format.

[0914] Output: Preprocessed data (e.g. cleaned text data).

[0915] Specific operation: The server tokenizes the text data, completes missing values ​​in each case data, and removes outliers. For example, Python libraries (NLP libraries and data processing libraries) are used for text tokenization and missing value completion.

[0916] Step 3: Feature extraction

[0917] The server extracts features using the preprocessed data and converts text to numerical data using a TF-IDF vectorizer.

[0918] Input: Preprocessed text data.

[0919] Output: Numerical data (TF-IDF vectors).

[0920] Specific operation: The server applies TF-IDF to the preprocessed text data to extract important features and convert them into numerical data.

[0921] Step 4: Training the AI ​​model

[0922] The server uses the features to train a generative AI model: it builds a classification model (e.g., SVM or Random Forest) and trains it on the training data.

[0923] Input: Numerical data (features) and corresponding labels.

[0924] Output: A trained generative AI model.

[0925] Specific operation: The server feeds the features to the classification model and performs the training process. After training, the performance of the model is evaluated, and cross-validation is performed to calculate the accuracy and F1 score.

[0926] Step 5: Providing an insight generation API

[0927] The server uses a trained generative AI model to build an API that generates insights. A RESTful API is designed using frameworks such as Flask or FastAPI.

[0928] Input: Court information (e.g., case summary and evidence).

[0929] Output: AI-generated opinion.

[0930] Specific operation: The server receives a request from a user, inputs court information into the generative AI model, and generates an opinion. The generated opinion is returned to the client as an API response.

[0931] Step 6: Providing a User Interface

[0932] The terminal will be designed and provided with a form for entering court information and a dashboard displaying opinions, and will have an intuitive design that will make it easy for users to operate.

[0933] Input: User input information (court information).

[0934] Output: Input screen and display screen.

[0935] Specific operation: The terminal designs the form using HTML and CSS, receives the API response, and displays the insights on the dashboard.

[0936] Step 7: Integrating the Emotion Engine

[0937] The device collects the user's facial images and voice data and analyzes them with an emotion engine, for example, using Microsoft Azure's Face API in real time.

[0938] Input: Face images and audio data.

[0939] Output: Emotion analysis results.

[0940] Specific operation: The device sends data collected by the camera and microphone to the emotion analysis API, and sends the returned results to the server. Based on the analysis results, appropriate support information is provided to the user.

[0941] Step 8: Enter your information

[0942] The user accesses the input screen and enters court information (summary of the case and evidence). For example, the user might enter, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[0943] Input: Court Information.

[0944] Output: The entered court information.

[0945] Specific operation: The user enters information into the input screen of the device and presses the send button to send the data to the server.

[0946] Step 9: Confirm the AI's opinion

[0947] The server generates an opinion based on the received legal information, which is then sent to the terminal and displayed to the user.

[0948] Input: Court Information.

[0949] Output: AI-generated opinion.

[0950] Specific operation: The server analyzes the court information using the generative AI model, generates an opinion, and sends it to the terminal. The user can check the opinion on the dashboard.

[0951] Step 10: Recognize and respond to emotions

[0952] The emotion engine analyzes the user's facial images and voice while they are typing or confirming their opinion, recognizing their emotions. If they are feeling anxious, it will provide relaxation techniques and additional support information.

[0953] Input: Face images and audio data.

[0954] Output: Sentiment analysis results, supporting information.

[0955] Specific operation: The emotion engine analyzes the user's image and voice in real time and provides appropriate support information based on the results.

[0956] Step 11: Use in discussions

[0957] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[0958] Input: AI-generated opinions, emotional support information.

[0959] Output: Discussion as a judge and final decision.

[0960] Specific operation: Based on the information obtained, the user exchanges opinions with other jurors and conducts the final trial fairly.

[0961] This will reduce the psychological burden on users and create a consistent process that supports fair trials.

[0962] (Application example 2)

[0963] 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."

[0964] The problem that this invention aims to solve is to reduce the stress and anxiety that factory workers feel while working, thereby improving work efficiency and safety. In modern factory work environments, long hours of monotonous work and excessive stress can damage workers' mental health. This can also reduce workers' work performance and negatively impact product quality and production efficiency. To solve this problem, a system is needed that can recognize workers' emotional states in real time and provide appropriate support.

[0965] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training an AI model based on the preprocessed data. The user terminal includes means for providing an opinion generation API using the trained AI model, means for communicating with the server to send court case information and receive opinions, means for sending court case information entered by the user to the server and displaying the opinions received from the server, means for recognizing worker emotions, and means for providing support to workers based on the recognized emotions. This makes it possible to recognize the emotional state of workers working in a factory in real time and provide appropriate support and advice. This reduces worker stress and anxiety, contributing to improved work efficiency and safety.

[0966] "Past case data" is data that includes records of past cases and judgments regarding legal decisions.

[0967] A "collection means" is a device or method that performs a process of obtaining specified data from a particular location and storing it.

[0968] "Preprocessing means" refers to devices or methods that perform processes to prepare and clean collected data for analysis and modeling.

[0969] "Means of training an AI model" refers to the process of using collected and preprocessed data to train an artificial intelligence model and build that model with high accuracy.

[0970] The "means for providing an opinion generation API" is an interface for generating legal opinions from a trained AI model and providing them to external systems or users.

[0971] "Court Information" is information containing facts, evidence, and legal documents relating to a particular court case.

[0972] "User-entered case information" is data about a particular case that is entered into the system by a user.

[0973] A "server" is a central computer system that processes and manages data on a network.

[0974] The "means for receiving an opinion" is a process for receiving the generated legal opinion from the server to the user terminal.

[0975] A "means for displaying an opinion" is a device or method for visually displaying a received legal opinion to a user.

[0976] "Means for recognizing worker emotions" refers to the process of using sensors such as cameras and microphones to analyze a worker's facial expressions and voice to identify their emotional state.

[0977] A "means for providing support to workers based on recognized emotions" is a system or method that suggests appropriate advice or assistance based on the worker's emotional state.

[0978] This invention is an emotion recognition system for reducing stress and anxiety among factory workers and improving work efficiency and safety. This system includes a series of processes, starting with collecting past case data, analyzing emotion data, and providing support.

[0979] System configuration

[0980] Server Configuration

[0981] 1. Data Collection and Preprocessing:

[0982] The server collects historical case data from legal databases and other data sources, using APIs and scraping techniques to retrieve the data and store it in a local database.

[0983] To prepare the collected data for analysis, preprocessing such as cleaning (filling in missing values ​​and removing outliers) and tokenization is performed.

[0984] 2. Training the AI ​​model:

[0985] The server uses the pre-processed dataset to train an AI model, applying natural language processing (NLP) models to create a model capable of generating legal opinions based on historical case law data.

[0986] Evaluate the trained model and retune it if necessary.

[0987] 3. Providing an opinion generation API:

[0988] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on input information, allowing the system to quickly provide legal opinions.

[0989] Configuring the user device

[0990] 1. Information input and display:

[0991] The terminal provides an interface where users can input legal information and emotional data, and in particular, build user-friendly forms and dashboards.

[0992] The entered information is sent to the server, and the view returned by the server is displayed.

[0993] 2. Emotion recognition engine integration:

[0994] The device uses a camera and microphone to collect facial images and voice data of the worker, and sends them to an emotion recognition engine, which recognizes the worker's emotional state in real time.

[0995] Depending on the emotion recognized, the device will display appropriate alerts and support information. For example, if stress levels increase, an alert will be displayed suggesting ways to relax.

[0996] Hardware and software used

[0997] Hardware: Cameras, microphones, database servers, user devices (computers, smartphones, tablets, etc.).

[0998] Software: Google Cloud Vision API, Azure Cognitive Services, MySQL.

[0999] Specific examples

[1000] 1. Information input: The user accesses a terminal in the factory and inputs court information and the emotional state of the worker. For example, the court information might be input as follows: "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B. Pedestrian B suffered a broken bone."

[1001] 2. Server processing: The server analyzes the received information and generates a view using the trained AI model. The view is that "previous court decisions have found the driver of vehicle B to be grossly negligent, and the average amount of compensation is 5 million yen."

[1002] 3. View display: The terminal displays the worker's emotional state in real time along with the view received from the server and provides appropriate advice as needed.

[1003] 4. Emotion Recognition: The emotion recognition engine can identify anxiety from a worker's facial expression and display relaxation techniques or additional support information.

[1004] Prompt Sentence Examples

[1005] "Provide emotional awareness to workers and suggest ways to provide appropriate support and advice where necessary."

[1006] "Consider designing a system that measures workers' stress levels in real time and suggests ways to relax."

[1007] In this way, the present invention makes it possible to recognize the emotional state of workers in a factory in real time and provide appropriate support, thereby reducing stress and anxiety among workers and improving work efficiency and safety.

[1008] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1009] Step 1:

[1010] Data collection and preprocessing:

[1011] The server collects past case data from legal databases and other data sources using APIs and scraping technology. The collected data is stored in a local database. The input is case data obtained through APIs and scraping technology, and the output is formatted data. This data undergoes preprocessing such as missing value completion, outlier removal, and tokenization.

[1012] Step 2:

[1013] Training the AI ​​model:

[1014] The server uses the preprocessed dataset to train a natural language processing (NLP) model, with the input being the preprocessed case data and the output being the trained AI model, which is now capable of generating legal opinions based on past case data.

[1015] Step 3:

[1016] Providing an opinion generation API:

[1017] The server builds and provides an API that generates opinions using a trained AI model. The input is the court information entered by the user, and the output is the legal opinion generated by the AI ​​model. This allows the server to provide legal opinions in real time based on the input information.

[1018] Step 4:

[1019] Information entry and display:

[1020] The user inputs the court information using the terminal interface. The input information is sent from the terminal to the server. The input is the court information entered by the user, and the output is the information sent to the server. The opinion returned by the server is displayed on the terminal for the user to visually confirm.

[1021] Step 5:

[1022] Emotion data collection:

[1023] The device collects facial images and voice data of workers using a camera and microphone. The input is the facial images and voice data acquired through the camera and microphone, and the output is stored in the device and sent to an emotion recognition engine.

[1024] Step 6:

[1025] Emotion recognition engine analysis:

[1026] The emotion recognition engine analyzes the transmitted facial image and voice data to identify the worker's emotional state. The input is the facial image and voice data, and the output is the analyzed emotional state (stress, anxiety, joy, etc.).

[1027] Step 7:

[1028] Providing emotional support:

[1029] The device provides appropriate support and advice to workers based on their recognized emotional state. For example, if stress levels are high, it suggests relaxation techniques or short breaks. The input is the recognized emotional state data, and the output is support information or alerts.

[1030] Step 8:

[1031] Data Storage and Feedback:

[1032] The terminal stores the worker's emotional data and response history, and later analyzes them to propose improvement measures. The input is emotional data and response history, and the output is data stored in a database. This makes it possible to continuously improve the performance of the entire system.

[1033] 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.

[1034] 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.

[1035] 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.

[1036] [Third embodiment]

[1037] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1038] 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.

[1039] 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).

[1040] 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.

[1041] 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.

[1042] 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).

[1043] 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.

[1044] 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.

[1045] 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.

[1046] 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.

[1047] 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.

[1048] 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."

[1049] ---

[1050] This invention relates to a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. This system collects past case data, generates opinions using an AI model trained on that data, and provides these to lay judges to support their judgments.

[1051] System configuration

[1052] Server Configuration

[1053] The server has the following main functions:

[1054] 1. Data Collection and Preprocessing:

[1055] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[1056] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[1057] 2. Training the AI ​​model:

[1058] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[1059] 3. Providing an opinion generation API:

[1060] The server provides an API that uses a trained AI model to generate opinions, which are then returned in real time based on the input court information.

[1061] Device configuration

[1062] The device has the following main features:

[1063] 1. Providing a user interface:

[1064] The terminal provides an information input screen that jurors can use and a screen that displays AI opinions.

[1065] 2. Sending court information and receiving and displaying opinions:

[1066] The terminal sends the trial information entered by the jurors to the server and displays the AI ​​opinion received from the server to the user.

[1067] User operations

[1068] The user uses the system in the following steps:

[1069] 1. Enter your information:

[1070] The user inputs court information (summary of the case and related evidence) through the device's input screen. For example, the user might input "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1071] 2. Transmission of Information:

[1072] The user sends the entered court information to the server via the send button. This operation causes the terminal to send the information to the server.

[1073] 3. Confirming AI's View:

[1074] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may read, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1075] 4. Use in discussions:

[1076] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[1077] Specific examples

[1078] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[1079] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[1080] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[1081] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1082] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[1083] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden they have to make fair judgments, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[1084] The processing flow will be explained below.

[1085] ---

[1086] Server Processing

[1087] Step 1: Collect case law data

[1088] The server collects past case data from a judicial database.

[1089] Data is obtained using APIs and scraping technology and stored in a local database.

[1090] Step 2: Preprocessing the data

[1091] The server cleans the collected case data, specifically by filling in missing values ​​and removing outliers.

[1092] Essential data items (summary of the case, content of the judgment, applicable law, etc.) are extracted and formatted for analysis.

[1093] Step 3: Training the AI ​​model

[1094] The server uses the preprocessed data to train the AI ​​model.

[1095] Using natural language processing techniques (e.g., Transformer or RNN), a model based on case law data is created.

[1096] Evaluate the performance of the trained model and retune it if necessary.

[1097] Step 4: Providing an insight generation API

[1098] The server builds an API that uses the trained AI model to generate opinions.

[1099] This API has the ability to respond in real time based on the court information entered.

[1100] Terminal handling

[1101] Step 1: Provide a user interface

[1102] The terminal provides input and display screens that are easy for jurors to use.

[1103] Specifically, it will have a form for entering court information and a dashboard for displaying opinions.

[1104] Step 2: Enter court information

[1105] Court information entered by the user is aggregated on the terminal.

[1106] For example, enter "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street."

[1107] Step 3: Submit your information

[1108] The terminal transmits the entered trial information to the server.

[1109] Data is transmitted securely using security protocols (such as SSL / TLS).

[1110] Step 4: Receive and display feedback

[1111] The terminal receives the response from the server and displays the view to the user.

[1112] For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was around 5 million yen."

[1113] User Action

[1114] Step 1: Enter your information

[1115] The user enters the court information into the terminal.

[1116] Specifically, you enter a summary of the case and relevant evidence into a text form.

[1117] Step 2: Confirm submission

[1118] The user confirms the input and presses the send button to send the information to the server.

[1119] Step 3: Confirm your views

[1120] The user checks the AI's opinion displayed on the device.

[1121] For example, you may obtain information such as, "Based on past precedents, the average amount of compensation is 5 million yen."

[1122] Step 4: Prepare for the discussion

[1123] Users will use the AI's views as a reference and prepare to debate with other jurors.

[1124] The views expressed will be used as a basis for making fair judgments.

[1125] These are the specific processing steps of the "Lay Judge Helper" system. This system reduces the anxiety and burden on lay judges and enables them to make fair and neutral judgments.

[1126] Example 1

[1127] 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."

[1128] In the current lay judge system, ordinary people can feel anxious and burdened due to their lack of legal knowledge and experience. This makes it difficult to make impartial judgments and can undermine the transparency and fairness of trials. A support system is needed to solve this problem and improve the quality of trials.

[1129] 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.

[1130] In this invention, the server includes means for collecting past judicial data, means for preprocessing the collected judicial data, means for training a generative AI model based on the preprocessed data, means for providing an opinion generation API using the trained generative AI model, means for communicating with a processing device to send court information and receive opinions, means for sending court information entered by a user to the processing device and displaying opinions received from the processing device, and means for a user to input court information and refer to opinions. This enables ordinary people to receive support in making fair judgments as jurors, reducing their anxiety and burden.

[1131] "Past judicial data" refers to data related to past judicial proceedings, such as precedents, rulings, and applicable laws.

[1132] "Preprocessing" refers to a series of processes carried out to improve the quality of collected data, specifically including data cleaning and missing value imputation.

[1133] A "generative AI model" is an artificial intelligence model built using machine learning techniques to extract useful information from data and generate opinions and predictions.

[1134] The "opinion generation API" is an application programming interface that uses a generative AI model to generate opinions based on input information and provide them to external parties.

[1135] A "processing device" is an information processing device, such as a server or client terminal, that manages the input, processing, communication, and output of data.

[1136] "User" is a concept that refers to users who operate and use the system and jurors.

[1137] The "information input screen" is an interface that allows the user to input necessary information into the system.

[1138] The "information display screen" is an interface that displays the results and opinions processed by the system to the user.

[1139] A "natural language processing model" is a machine learning model for understanding and generating human language. Specifically, it includes techniques for tasks such as text analysis, translation, and summarization.

[1140] "Means" is a term that refers to methods, techniques, equipment, processes, etc. used to achieve a particular purpose.

[1141] Server Processing

[1142] Data collection and preprocessing

[1143] The server collects past judicial data to be used in the lay judge system. Specifically, it retrieves case data from the judicial database using an API. The retrieved data is cleaned, and incomplete data and outliers are removed. This cleansing process is carried out until the data is clean and consistent. The specific software used is Python and its libraries (Pandas, NumPy, etc.).

[1144] Training an AI model

[1145] The server trains a generative AI model based on the preprocessed data. This process uses natural language processing (NLP) techniques. Specifically, machine learning libraries such as TensorFlow and PyTorch are used to train the AI ​​model to generate opinions from past case law data. The data is split into training and testing datasets, and the model's accuracy is evaluated and its hyperparameters are adjusted.

[1146] Providing an opinion generation API

[1147] The server designs and implements an API that uses a trained generative AI model to generate opinions. This API is designed in a RESTful format, and when court information is input to a specific endpoint, the generated opinion is returned in real time. The API is deployed using cloud services such as AWS and Google Cloud.

[1148] Terminal handling

[1149] Providing a user interface

[1150] The terminal provides a user interface that allows jurors to input information and confirm opinions. The information input screen contains text boxes and drop-down menus for entering trial information such as a summary of the case and evidence. The information display screen displays the AI's opinions received from the server. Specific front-end frameworks used are React and Vue.js.

[1151] Sending court information and receiving opinions

[1152] The court information entered by the user is sent to the server via the terminal. This operation is performed via an HTTP POST request. The opinion received from the server is displayed on the terminal screen. The terminal sends and receives information in real time according to the user's operations.

[1153] User operations

[1154] Enter information

[1155] The user inputs court information (summary of the case and evidence) through the device's input screen. For example, the user might input, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1156] Sending information

[1157] The user sends the entered information to the server via the send button. This operation causes the terminal to send the court information to the server.

[1158] Confirming AI's View

[1159] The AI's opinion sent from the server is received by the device and displayed to the user. For example, it may say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1160] Use in discussions

[1161] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[1162] Specific examples

[1163] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[1164] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[1165] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[1166] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1167] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[1168] This system will reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and provide support to help them make impartial decisions, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[1169] Prompt Sentence Examples

[1170] Summary of the incident: "Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1171] Sample prompt: "Based on the outline of this case, please tell us your opinion on similar past cases."

[1172] The above is an embodiment of the present invention.

[1173] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1174] Step 1:

[1175] Data collection

[1176] The server collects past judicial data. This data is obtained from the judicial database via API. The API endpoint URL and access key are required as input, and case data is obtained in JSON or CSV format as output. This allows the server to obtain a large amount of case data. Specifically, it periodically runs a script to download the data.

[1177] Step 2:

[1178] Data Preprocessing

[1179] The server preprocesses the collected case data. The input requires the collected case data. The output is a cleaned, consistent dataset. This process involves removing incomplete data and correcting outliers. Specifically, it uses the Python Pandas library to remove unnecessary columns and unify data types.

[1180] Step 3:

[1181] Feature extraction

[1182] The server extracts the necessary features from the preprocessed data. Preprocessed data is required as input. The output is a dataset suitable for training an AI model. Features include a summary of the case, the verdict, and the applicable law. Specific operations include converting the data into lists or vectors using regular expressions and NLP tools.

[1183] Step 4:

[1184] Training an AI model

[1185] The server uses the extracted features to train a generative AI model. A dataset containing the features is required as input. A trained generative AI model is obtained as output. Specifically, the model is trained using machine learning libraries such as TensorFlow and PyTorch. During training, the accuracy of the model is improved by repeatedly learning the dataset.

[1186] Step 5:

[1187] Evaluating and Tuning the Model

[1188] The server evaluates the performance of the generative AI model using a test dataset. A test dataset is required as input. Evaluation results (precision, recall, F-measure, etc.) are obtained as output. If performance is insufficient, hyperparameters are adjusted or the data is reprocessed. Specifically, evaluation is performed using a confusion matrix, and cross-validation is performed if necessary.

[1189] Step 6:

[1190] Providing an opinion generation API

[1191] The server deploys an API that uses a trained generative AI model to generate opinions. Court information is required as input. The generated opinion is output in JSON format. Specifically, the API is built using frameworks such as Flask or FastAPI and deployed on a cloud service.

[1192] Step 7:

[1193] Providing information input screen

[1194] The terminal provides an information input screen where the user can enter court information. The input requires court information to be entered by the user. The output is court information ready to be sent to the server. Specific operation involves building the screen using React and Vue.js, allowing the user to input information intuitively.

[1195] Step 8:

[1196] Sending court information

[1197] The terminal sends the court information entered by the user to the server. The input requires the court information entered by the user. The output is the information accurately sent to the server. Specifically, the information is sent to the server's API using an HTTP POST request.

[1198] Step 9:

[1199] Receiving and displaying AI insights

[1200] The device displays the AI ​​opinion received from the server to the user. The input requires the AI ​​opinion received from the server. The output is the opinion that is displayed to the user. Specifically, it parses the received JSON data and displays it on the screen in an easy-to-read format.

[1201] Step 10:

[1202] Confirming opinions and using them in discussions

[1203] The user checks the AI's opinion displayed on the device and uses it in discussions with other jurors. The input required is the AI's opinion displayed on the device. The output is an opinion that can be used as material for discussion to reach a fair judgment. Specifically, the system operates by advancing discussions based on the opinion and using it to reach a final verdict.

[1204] (Application example 1)

[1205] 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."

[1206] Internal investigations are becoming increasingly important in modern companies and public institutions. However, internal investigators often lack legal knowledge and information on similar cases, making it difficult to conduct fair and prompt investigations. Other issues include inconsistencies in judgment and unclear investigation directions. The present invention aims to solve these problems and provide a system that enables internal investigators to conduct investigations efficiently and fairly.

[1207] 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.

[1208] In this invention, the server includes a means for collecting past case data, a means for preprocessing the collected case data, and a means for training an AI model based on the preprocessed data. This makes it possible to input information about internal investigation cases and send it to the server, and receive and display opinions from the AI ​​model. This allows internal investigators to proceed with investigations based on opinions based on similar past cases, enabling fair and efficient investigations.

[1209] "Past case data" refers to data that includes the results of trials and case information handled by judicial institutions in the past.

[1210] "Preprocessing" refers to the process of complementing missing and outlier values ​​from collected case law data and extracting necessary information.

[1211] An "AI model" is an inference model trained using machine learning algorithms, particularly those that utilize natural language processing techniques.

[1212] An "opinion generation API" is an application program interface that utilizes a trained AI model to generate and provide opinions based on user-entered information.

[1213] "Server" refers to a computer system that receives information from users, communicates with case law databases, and generates and provides opinions based on AI models.

[1214] "Users" are investigators from companies and public institutions who use this system to conduct internal investigations.

[1215] "Information on internal investigation cases" means information containing facts and evidence collected by a company or public institution for the purposes of an internal investigation.

[1216] The "input and transmission means" is an interface that allows users to input information about internal investigation cases into the system and transmit it to the server.

[1217] The "receiving and displaying means" is an interface for displaying to the user the views of the AI ​​model sent from the server.

[1218] A "natural language processing model" is a machine learning model trained to understand text data and respond to questions or requests in natural language.

[1219] This invention is a system that enables companies and public institutions to conduct internal investigations efficiently and fairly. The system collects past case data, uses a trained AI model based on that data to generate opinions, and provides them to internal investigators. The system is primarily composed of three elements: a server, a terminal, and a user.

[1220] System configuration

[1221] Server Configuration

[1222] The server has the following main functions:

[1223] 1. Data Collection and Preprocessing:

[1224] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[1225] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[1226] 2. Training the AI ​​model:

[1227] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[1228] 3. Providing an opinion generation API:

[1229] The server provides an API that uses a trained AI model to generate opinions, which then generate and respond in real time based on the input internal investigation information.

[1230] Device configuration

[1231] The device has the following main features:

[1232] 1. Providing a user interface:

[1233] The device provides an information input screen that users can use and an AI opinion display screen.

[1234] 2. Sending internal investigation information and receiving and displaying opinions:

[1235] The device sends the survey information entered by the user to the server and displays the AI ​​opinion received from the server to the user.

[1236] User operations

[1237] The user uses the system in the following steps:

[1238] 1. Enter your information:

[1239] The user enters internal investigation information (a summary of the case and related evidence) through the device's input screen. For example, the user might enter "Case summary: Employee A is suspected of using company assets for personal purposes."

[1240] 2. Transmission of Information:

[1241] The user sends the entered internal investigation information to the server via the send button. This operation causes the terminal to send the information to the server.

[1242] 3. Confirming AI's View:

[1243] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may be displayed as "In similar cases in the past, dismissal of employees was applied."

[1244] Hardware and software used

[1245] The system of the present invention uses the following hardware and software:

[1246] Hardware:

[1247] Smartphone (Android / iOS)

[1248] Server: A computer system running on the cloud.

[1249] software:

[1250] Programming language: Python

[1251] Library used: requests (sends HTTP requests)

[1252] Natural language processing models: BERT, GPT-3, etc.

[1253] Examples of concrete examples and prompts

[1254] Examples:

[1255] Input information: The user enters, "Employee A is suspected of using company assets for personal purposes."

[1256] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[1257] Display of opinion: The terminal displays to the user the opinion received from the server, "In similar cases in the past, employee dismissal was applied."

[1258] Example prompt sentence:

[1259] "Case summary: An employee is suspected of using company assets for personal use. Relevant evidence: CCTV footage. What are similar cases in the past and what were the results?"

[1260] The above is a specific embodiment for carrying out the present invention. It is expected that this system will improve the efficiency and accuracy of internal investigations and strengthen compliance in companies and public institutions.

[1261] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1262] Step 1:

[1263] The user enters information about the internal investigation case through the input screen on the terminal. Specifically, the user enters "Case summary: Employee A is suspected of using company assets for personal use." The information entered at this time is in text format, and is stored as is as input data on the terminal.

[1264] Step 2:

[1265] The terminal generates an HTTP request to send the survey information entered by the user to the server. Specifically, it uses the Python library requests to convert the input data into JSON format and sends a POST request to the specified API endpoint. Once the request is sent, the input data is sent to the server.

[1266] Step 3:

[1267] The server processes the received investigation information. First, it parses the received JSON data and converts it into structured data. Then it searches for relevant cases in the preprocessed case law database. The search results are used in the next step.

[1268] Step 4:

[1269] The server generates opinions using a trained AI model based on the search results. Specifically, it uses AI models that use natural language processing techniques (e.g., BERT or GPT-3) to derive opinions from past case data. In this process, the AI ​​model uses the input data as prompts and outputs opinions on relevant cases.

[1270] Step 5:

[1271] The server converts the generated opinion into JSON format and prepares it as response data. The prepared response data is sent to the terminal as an HTTP response. This sends the opinion data to the terminal.

[1272] Step 6:

[1273] The device analyzes the opinion data received from the server and displays it in a user-friendly format. Specifically, it converts the received JSON data into structured data and applies it to the UI components for display. This allows the user to visually confirm the opinion.

[1274] Step 7:

[1275] The user discusses the internal investigation based on the opinion received from the server. For example, based on the opinion that "employee dismissal was applied in past similar cases," the user discusses with other investigators and decides the direction of the investigation.

[1276] Through the above steps, the user can proceed with the internal investigation efficiently and fairly.

[1277] 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.

[1278] ---

[1279] This invention is a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. It also reduces the psychological burden on users by incorporating an emotion engine that recognizes the user's emotions. This system collects past case data, uses a trained AI model based on that data to generate opinions, and provides these to lay judges. Furthermore, the emotion engine recognizes the user's emotions and responds accordingly.

[1280] System configuration

[1281] Server Configuration

[1282] 1. Data Collection and Preprocessing:

[1283] The server collects past case data from judicial databases using APIs and scraping technology, and stores it in a local database.

[1284] As preprocessing, the collected data is cleaned (by filling in missing values ​​and removing outliers) and formatted for analysis.

[1285] 2. Training the AI ​​model:

[1286] The server uses the preprocessed data to train an AI model, specifically applying natural language processing techniques to create a model based on past case law data.

[1287] Evaluate the performance of the trained model and retune the model if necessary.

[1288] 3. Providing an opinion generation API:

[1289] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on the input court information.

[1290] Device configuration

[1291] 1. Providing a user interface:

[1292] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[1293] 2. Emotion engine integration:

[1294] The device is equipped with an emotion engine that collects and analyzes the user's facial images and voice data, recognizes the user's emotional state in real time, and transmits the data to a server.

[1295] The emotion engine has the ability to provide appropriate support information and alerts if the user is feeling anxious or stressed.

[1296] User operations

[1297] 1. Enter your information:

[1298] The user accesses the device's input screen and enters court information (an outline of the case and related evidence). For example, the user might enter, "Outline of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1299] 2. Transmission of Information:

[1300] The user confirms the input and presses the send button to send the information to the server. Data is sent securely using a security protocol (such as SSL / TLS).

[1301] 3. Confirming AI's View:

[1302] The server generates an opinion based on the received court information and sends it to the terminal. The terminal then displays the opinion to the user. For example, "In similar past cases, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1303] 4. Recognizing and Responding to Emotions:

[1304] The emotion engine analyzes the user's facial image and voice to recognize their emotions while they are entering court information or confirming opinions. Based on the recognized emotions, it provides appropriate information to reduce the user's anxiety and stress.

[1305] For example, if the user is feeling stressed, the system displays advice and support to help them relax.

[1306] 5. Use in discussions:

[1307] Users can use the AI's views and support from the emotion engine to discuss with other jurors and make fair decisions.

[1308] Specific examples

[1309] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[1310] 1. Information input: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[1311] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[1312] 3. Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1313] 4. Emotion Recognition: The emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression. In this case, it displays relaxation techniques and additional support information.

[1314] 5. Use of discussion: Based on this opinion and support, the user will debate as a juror and make a final decision.

[1315] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden of making fair judgments. Furthermore, by integrating an emotion engine, the psychological burden on users can be further reduced, and the transparency and fairness of trials can be improved.

[1316] The processing flow will be explained below.

[1317] Server Processing

[1318] Step 1:

[1319] The server collects past case data from judicial databases, specifically using APIs and scraping technology to retrieve the data and store it in a local database.

[1320] Step 2:

[1321] The server preprocesses the collected case data, specifically by filling in missing values ​​and removing outliers, and extracting necessary information such as the case summary, the content of the judgment, and the applicable law.

[1322] Step 3:

[1323] The server uses the preprocessed data to train an AI model, uses natural language processing technology to generate a model based on past case law data, and evaluates the performance of the trained model and makes adjustments as necessary.

[1324] Step 4:

[1325] The server uses the trained AI model to build an API that generates opinions, which responds in real time based on the input court information.

[1326] Terminal handling

[1327] Step 1:

[1328] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[1329] Step 2:

[1330] The device collects the user's facial images and voice and provides data for emotion recognition by the emotion engine, which analyzes the collected data and recognizes the user's emotional state in real time.

[1331] Step 3:

[1332] The terminal sends the court information entered by the user to the server. The data is transmitted securely using security protocols (such as SSL / TLS).

[1333] Step 4:

[1334] The device displays the AI's opinion received from the server in real time. For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1335] User Action

[1336] Step 1:

[1337] The user inputs court information (summary of the case and related evidence) into the terminal. Specifically, the user inputs "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1338] Step 2:

[1339] The user checks the input and presses the send button to send the information to the server.

[1340] Step 3:

[1341] The user checks the AI's opinion displayed on the device, and receives information such as, "Based on past court decisions, the average amount of compensation is 5 million yen."

[1342] Step 4:

[1343] While the user is inputting or confirming their opinion, the emotion engine analyzes the user's facial expressions and voice in real time to recognize their emotional state, and if necessary, provides appropriate information to reduce the user's anxiety or stress.

[1344] Step 5:

[1345] Users can use the displayed opinions and supporting information from the emotion engine to discuss with other jurors and use the information to make a fair decision.

[1346] Exemplary Processing Steps

[1347] Step 1:

[1348] The user inputs into the terminal, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light, and suffered a broken bone."

[1349] Step 2:

[1350] The device sends the entered information to the server, which uses a security protocol to send the data securely.

[1351] Step 3:

[1352] The server generates a view using a trained AI model based on the received information. For example, it may send a view to the terminal that, according to past court decisions, the driver of vehicle B was found to be grossly negligent and the average amount of compensation for damages was 5 million yen.

[1353] Step 4:

[1354] The terminal displays the received views to the user, who then confirms the views.

[1355] Step 5:

[1356] The emotion engine analyzes the user's facial image and voice to detect anxiety and stress. If stress is detected, it displays relaxation techniques and additional support information.

[1357] Step 6:

[1358] Based on the displayed opinions and supporting information from the emotion engine, users can discuss with other jurors and make a fair decision.

[1359] In this way, the present invention reduces the anxiety and burden that lay judges have due to their lack of legal knowledge and experience, and by integrating an emotion engine, it further reduces the psychological burden on users, thereby realizing fair and transparent trials.

[1360] Example 2

[1361] 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."

[1362] For ordinary people who participate in the lay judge system, there are issues such as anxiety, psychological stress, and the heavy burden of being expected to make impartial judgments. Another issue is anxiety due to a lack of trial experience and legal knowledge. Furthermore, if the stress and anxiety experienced during trials are left unaddressed, it can have a negative impact on the psychological health of lay judges.

[1363] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training a generative AI model based on the preprocessed data. This makes it possible to provide a system that reduces the anxiety and psychological burden of ordinary people when participating in the lay judge system and supports fair judgment. Furthermore, the server includes means for providing an opinion generation API using the trained generative AI model and means for recognizing the user's emotions, and provides support information according to the user's emotional state. This further reduces the user's psychological burden and improves the transparency and fairness of trials.

[1364] "Past case data" is data collected from judicial databases that includes the results and details of various past trials.

[1365] "Preprocessing" is the process of converting collected data into a format suitable for analysis and training, including cleaning the data, shaping the data, filling in missing values, removing outliers, and tokenizing the text.

[1366] A "generative AI model" is an artificial intelligence model that is trained based on past case data and automatically generates opinions on case law based on given input data.

[1367] The "Opinion Generation API" is an application program interface that uses a trained generative AI model to generate opinions in real time based on input court information and respond to external requests.

[1368] A "server" is a computer system that performs functions such as collecting data, pre-processing, training AI models, and providing an insight generation API.

[1369] "Court information" refers to important information in a trial, such as an outline of the case and relevant evidence.

[1370] "User" refers to a person who uses the system to enter court information and view generated opinions.

[1371] The "emotion engine" is part of a system that analyzes the user's facial image and voice data, recognizes the user's emotional state, and provides appropriate support information based on that information.

[1372] The "information input screen" is a user interface for users to input court information.

[1373] The "information display screen" is a user interface for displaying to the user the views received from the server.

[1374] A "natural language processing model" is an artificial intelligence model that analyzes text data and understands the meaning and structure of sentences.

[1375] This invention is a system that reduces the anxiety and psychological burden of ordinary people participating in the lay judge system and supports fair judgments. In particular, by combining it with an emotion engine that recognizes the user's emotions, the system can minimize the user's psychological burden. This system collects past case data and provides jurors with opinions based on a generative AI model trained on that data. Furthermore, the emotion engine recognizes the user's emotions and responds appropriately.

[1376] Server Configuration

[1377] 1. Data Collection and Preprocessing:

[1378] The server collects past case data from judicial databases using APIs and scraping technology, and the collected data is stored in a local database.

[1379] The server cleans the data (filling in missing values ​​and removing outliers) and prepares it for analysis, which involves tokenizing the data and removing unnecessary symbols and spaces.

[1380] 2. Training the AI ​​model:

[1381] The server trains a generative AI model based on the preprocessed data. Specifically, it uses natural language processing techniques to extract important features from past case data and create a classification model (e.g., SVM or Random Forest).

[1382] The performance of the trained model is evaluated using cross-validation and other methods, and the accuracy and F1 score are calculated.

[1383] 3. Providing an opinion generation API:

[1384] The server builds an API that generates insights using a trained generative AI model. It uses frameworks such as Flask and FastAPI to design a RESTful API.

[1385] The server deploys the view generation API in a cloud environment (e.g., AWS or GCP).

[1386] Device configuration

[1387] 1. Providing a user interface:

[1388] The device provides a user-friendly input and display interface, which uses HTML and CSS to provide a form for entering a case summary and related evidence.

[1389] The display screen provides a dashboard that displays the insights returned from the server in the form of tables and graphs.

[1390] 2. Emotion engine integration:

[1391] The device collects the user's facial images and voice data and analyzes them with an emotion engine, which uses publicly available APIs such as Microsoft Azure's Face API and IBM Watson Tone Analyzer.

[1392] The device sends the recognized emotion data to a server, which provides additional support information and alerts based on the user's emotional state.

[1393] User Actions

[1394] 1. Enter your information:

[1395] The user accesses the device's input screen and enters a summary of the incident and related evidence. For example, the user might enter, "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1396] 2. Transmission of Information:

[1397] The user checks the information entered and presses the send button to send the information to the server. At this time, a security protocol such as SSL / TLS is used for communication.

[1398] 3. Confirming AI's View:

[1399] The server generates an opinion based on the received court information and sends the result to the terminal. The user can check the opinion on the terminal. The message displayed is, "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1400] 4. Recognizing and Responding to Emotions:

[1401] The emotion engine analyzes the user's facial images and voice in real time while inputting or confirming opinions, and recognizes their emotions. For example, if the user is feeling anxious, it will provide relaxation techniques and additional support information.

[1402] 5. Use in discussions:

[1403] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[1404] Specific examples

[1405] For example, if a user wants to participate as a juror in a trial regarding a "pedestrian-vehicle accident," the user will use the system as follows:

[1406] 1. Information input: The user inputs a summary of the incident: "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[1407] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[1408] 3. Display of opinion: The terminal displays the opinion received from the server to the user. It displays the following: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1409] 4. Emotion Recognition: If the emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression, it will display relaxation techniques and additional support information.

[1410] 5. Use of discussion: Based on this opinion and support, users will debate as jurors and make a final decision.

[1411] The present invention provides support to reduce the psychological burden on lay judges and ensure fairer trials.

[1412] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1413] Step 1: Data collection

[1414] The server accesses the judicial database and sends API requests to collect past case data. Specifically, the server retrieves data using HTTP requests and receives the response in JSON format.

[1415] Input: API request parameters (e.g., trial year, type).

[1416] Output: Case data in JSON format.

[1417] What happens: The server sends a request to an API endpoint, parses the response, and stores it in a local database.

[1418] Step 2: Data Preprocessing

[1419] The server cleans the collected case data and formats it for analysis: text data is tokenized, unnecessary symbols and spaces are removed, missing values ​​are imputed, and outliers are removed.

[1420] Input: Case law data in JSON format.

[1421] Output: Preprocessed data (e.g. cleaned text data).

[1422] Specific operation: The server tokenizes the text data, completes missing values ​​in each case data, and removes outliers. For example, Python libraries (NLP libraries and data processing libraries) are used for text tokenization and missing value completion.

[1423] Step 3: Feature extraction

[1424] The server extracts features using the preprocessed data and converts text to numerical data using a TF-IDF vectorizer.

[1425] Input: Preprocessed text data.

[1426] Output: Numerical data (TF-IDF vectors).

[1427] Specific operation: The server applies TF-IDF to the preprocessed text data to extract important features and convert them into numerical data.

[1428] Step 4: Training the AI ​​model

[1429] The server uses the features to train a generative AI model: it builds a classification model (e.g., SVM or Random Forest) and trains it on the training data.

[1430] Input: Numerical data (features) and corresponding labels.

[1431] Output: A trained generative AI model.

[1432] Specific operation: The server feeds the features to the classification model and performs the training process. After training, the performance of the model is evaluated, and cross-validation is performed to calculate the accuracy and F1 score.

[1433] Step 5: Providing an insight generation API

[1434] The server uses a trained generative AI model to build an API that generates insights. A RESTful API is designed using frameworks such as Flask or FastAPI.

[1435] Input: Court information (e.g., case summary and evidence).

[1436] Output: AI-generated opinion.

[1437] Specific operation: The server receives a request from a user, inputs court information into the generative AI model, and generates an opinion. The generated opinion is returned to the client as an API response.

[1438] Step 6: Providing a User Interface

[1439] The terminal will be designed and provided with a form for entering court information and a dashboard displaying opinions, and will have an intuitive design that will make it easy for users to operate.

[1440] Input: User input information (court information).

[1441] Output: Input screen and display screen.

[1442] Specific operation: The terminal designs the form using HTML and CSS, receives the API response, and displays the insights on the dashboard.

[1443] Step 7: Integrating the Emotion Engine

[1444] The device collects the user's facial images and voice data and analyzes them with an emotion engine, for example, using Microsoft Azure's Face API in real time.

[1445] Input: Face images and audio data.

[1446] Output: Emotion analysis results.

[1447] Specific operation: The device sends data collected by the camera and microphone to the emotion analysis API, and sends the returned results to the server. Based on the analysis results, appropriate support information is provided to the user.

[1448] Step 8: Enter your information

[1449] The user accesses the input screen and enters court information (summary of the case and evidence). For example, the user might enter, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1450] Input: Court Information.

[1451] Output: The entered court information.

[1452] Specific operation: The user enters information into the input screen of the device and presses the send button to send the data to the server.

[1453] Step 9: Confirm the AI's opinion

[1454] The server generates an opinion based on the received legal information, which is then sent to the terminal and displayed to the user.

[1455] Input: Court Information.

[1456] Output: AI-generated opinion.

[1457] Specific operation: The server analyzes the court information using the generative AI model, generates an opinion, and sends it to the terminal. The user can check the opinion on the dashboard.

[1458] Step 10: Recognize and respond to emotions

[1459] The emotion engine analyzes the user's facial images and voice while they are typing or confirming their opinion, recognizing their emotions. If they are feeling anxious, it will provide relaxation techniques and additional support information.

[1460] Input: Face images and audio data.

[1461] Output: Sentiment analysis results, supporting information.

[1462] Specific operation: The emotion engine analyzes the user's image and voice in real time and provides appropriate support information based on the results.

[1463] Step 11: Use in discussions

[1464] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[1465] Input: AI-generated opinions, emotional support information.

[1466] Output: Discussion as a judge and final decision.

[1467] Specific operation: Based on the information obtained, the user exchanges opinions with other jurors and conducts the final trial fairly.

[1468] This will reduce the psychological burden on users and create a consistent process that supports fair trials.

[1469] (Application example 2)

[1470] 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."

[1471] The problem that this invention aims to solve is to reduce the stress and anxiety that factory workers feel while working, thereby improving work efficiency and safety. In modern factory work environments, long hours of monotonous work and excessive stress can damage workers' mental health. This can also reduce workers' work performance and negatively impact product quality and production efficiency. To solve this problem, a system is needed that can recognize workers' emotional states in real time and provide appropriate support.

[1472] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training an AI model based on the preprocessed data. The user terminal includes means for providing an opinion generation API using the trained AI model, means for communicating with the server to send court case information and receive opinions, means for sending court case information entered by the user to the server and displaying the opinions received from the server, means for recognizing worker emotions, and means for providing support to workers based on the recognized emotions. This makes it possible to recognize the emotional state of workers working in a factory in real time and provide appropriate support and advice. This reduces worker stress and anxiety, contributing to improved work efficiency and safety.

[1473] "Past case data" is data that includes records of past cases and judgments regarding legal decisions.

[1474] A "collection means" is a device or method that performs a process of obtaining specified data from a particular location and storing it.

[1475] "Preprocessing means" refers to devices or methods that perform processes to prepare and clean collected data for analysis and modeling.

[1476] "Means of training an AI model" refers to the process of using collected and preprocessed data to train an artificial intelligence model and build that model with high accuracy.

[1477] The "means for providing an opinion generation API" is an interface for generating legal opinions from a trained AI model and providing them to external systems or users.

[1478] "Court Information" is information containing facts, evidence, and legal documents relating to a particular court case.

[1479] "User-entered case information" is data about a particular case that is entered into the system by a user.

[1480] A "server" is a central computer system that processes and manages data on a network.

[1481] The "means for receiving an opinion" is a process for receiving the generated legal opinion from the server to the user terminal.

[1482] A "means for displaying an opinion" is a device or method for visually displaying a received legal opinion to a user.

[1483] "Means for recognizing worker emotions" refers to the process of using sensors such as cameras and microphones to analyze a worker's facial expressions and voice to identify their emotional state.

[1484] A "means for providing support to workers based on recognized emotions" is a system or method that suggests appropriate advice or assistance based on the worker's emotional state.

[1485] This invention is an emotion recognition system for reducing stress and anxiety among factory workers and improving work efficiency and safety. This system includes a series of processes, starting with collecting past case data, analyzing emotion data, and providing support.

[1486] System configuration

[1487] Server Configuration

[1488] 1. Data Collection and Preprocessing:

[1489] The server collects historical case data from legal databases and other data sources, using APIs and scraping techniques to retrieve the data and store it in a local database.

[1490] To prepare the collected data for analysis, preprocessing such as cleaning (filling in missing values ​​and removing outliers) and tokenization is performed.

[1491] 2. Training the AI ​​model:

[1492] The server uses the pre-processed dataset to train an AI model, applying natural language processing (NLP) models to create a model capable of generating legal opinions based on historical case law data.

[1493] Evaluate the trained model and retune it if necessary.

[1494] 3. Providing an opinion generation API:

[1495] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on input information, allowing the system to quickly provide legal opinions.

[1496] Configuring the user device

[1497] 1. Information input and display:

[1498] The terminal provides an interface where users can input legal information and emotional data, and in particular, build user-friendly forms and dashboards.

[1499] The entered information is sent to the server, and the view returned by the server is displayed.

[1500] 2. Emotion recognition engine integration:

[1501] The device uses a camera and microphone to collect facial images and voice data of the worker, and sends them to an emotion recognition engine, which recognizes the worker's emotional state in real time.

[1502] Depending on the emotion recognized, the device will display appropriate alerts and support information. For example, if stress levels increase, an alert will be displayed suggesting ways to relax.

[1503] Hardware and software used

[1504] Hardware: Cameras, microphones, database servers, user devices (computers, smartphones, tablets, etc.).

[1505] Software: Google Cloud Vision API, Azure Cognitive Services, MySQL.

[1506] Specific examples

[1507] 1. Information input: The user accesses a terminal in the factory and inputs court information and the emotional state of the worker. For example, the court information might be input as follows: "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B. Pedestrian B suffered a broken bone."

[1508] 2. Server processing: The server analyzes the received information and generates a view using the trained AI model. The view is that "previous court decisions have found the driver of vehicle B to be grossly negligent, and the average amount of compensation is 5 million yen."

[1509] 3. View display: The terminal displays the worker's emotional state in real time along with the view received from the server and provides appropriate advice as needed.

[1510] 4. Emotion Recognition: The emotion recognition engine can identify anxiety from a worker's facial expression and display relaxation techniques or additional support information.

[1511] Prompt Sentence Examples

[1512] "Provide emotional awareness to workers and suggest ways to provide appropriate support and advice where necessary."

[1513] "Consider designing a system that measures workers' stress levels in real time and suggests ways to relax."

[1514] In this way, the present invention makes it possible to recognize the emotional state of workers in a factory in real time and provide appropriate support, thereby reducing stress and anxiety among workers and improving work efficiency and safety.

[1515] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1516] Step 1:

[1517] Data collection and preprocessing:

[1518] The server collects past case data from legal databases and other data sources using APIs and scraping technology. The collected data is stored in a local database. The input is case data obtained through APIs and scraping technology, and the output is formatted data. This data undergoes preprocessing such as missing value completion, outlier removal, and tokenization.

[1519] Step 2:

[1520] Training the AI ​​model:

[1521] The server uses the preprocessed dataset to train a natural language processing (NLP) model, with the input being the preprocessed case data and the output being the trained AI model, which is now capable of generating legal opinions based on past case data.

[1522] Step 3:

[1523] Providing an opinion generation API:

[1524] The server builds and provides an API that generates opinions using a trained AI model. The input is the court information entered by the user, and the output is the legal opinion generated by the AI ​​model. This allows the server to provide legal opinions in real time based on the input information.

[1525] Step 4:

[1526] Information entry and display:

[1527] The user inputs the court information using the terminal interface. The input information is sent from the terminal to the server. The input is the court information entered by the user, and the output is the information sent to the server. The opinion returned by the server is displayed on the terminal for the user to visually confirm.

[1528] Step 5:

[1529] Emotion data collection:

[1530] The device collects facial images and voice data of workers using a camera and microphone. The input is the facial images and voice data acquired through the camera and microphone, and the output is stored in the device and sent to an emotion recognition engine.

[1531] Step 6:

[1532] Emotion recognition engine analysis:

[1533] The emotion recognition engine analyzes the transmitted facial image and voice data to identify the worker's emotional state. The input is the facial image and voice data, and the output is the analyzed emotional state (stress, anxiety, joy, etc.).

[1534] Step 7:

[1535] Providing emotional support:

[1536] The device provides appropriate support and advice to workers based on their recognized emotional state. For example, if stress levels are high, it suggests relaxation techniques or short breaks. The input is the recognized emotional state data, and the output is support information or alerts.

[1537] Step 8:

[1538] Data Storage and Feedback:

[1539] The terminal stores the worker's emotional data and response history, and later analyzes them to propose improvement measures. The input is emotional data and response history, and the output is data stored in a database. This makes it possible to continuously improve the performance of the entire system.

[1540] 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.

[1541] 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.

[1542] 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.

[1543] [Fourth embodiment]

[1544] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1545] 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.

[1546] 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).

[1547] 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.

[1548] 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.

[1549] 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).

[1550] 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.

[1551] 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.

[1552] 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.

[1553] 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.

[1554] 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.

[1555] 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.

[1556] 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."

[1557] ---

[1558] This invention relates to a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. This system collects past case data, generates opinions using an AI model trained on that data, and provides these to lay judges to support their judgments.

[1559] System configuration

[1560] Server Configuration

[1561] The server has the following main functions:

[1562] 1. Data Collection and Preprocessing:

[1563] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[1564] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[1565] 2. Training the AI ​​model:

[1566] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[1567] 3. Providing an opinion generation API:

[1568] The server provides an API that uses a trained AI model to generate opinions, which are then returned in real time based on the input court information.

[1569] Device configuration

[1570] The device has the following main features:

[1571] 1. Providing a user interface:

[1572] The terminal provides an information input screen that jurors can use and a screen that displays AI opinions.

[1573] 2. Sending court information and receiving and displaying opinions:

[1574] The terminal sends the trial information entered by the jurors to the server and displays the AI ​​opinion received from the server to the user.

[1575] User operations

[1576] The user uses the system in the following steps:

[1577] 1. Enter your information:

[1578] The user inputs court information (summary of the case and related evidence) through the device's input screen. For example, the user might input "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1579] 2. Transmission of Information:

[1580] The user sends the entered court information to the server via the send button. This operation causes the terminal to send the information to the server.

[1581] 3. Confirming AI's View:

[1582] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may read, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1583] 4. Use in discussions:

[1584] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[1585] Specific examples

[1586] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[1587] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[1588] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[1589] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1590] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[1591] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden they have to make fair judgments, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[1592] The processing flow will be explained below.

[1593] ---

[1594] Server Processing

[1595] Step 1: Collect case law data

[1596] The server collects past case data from a judicial database.

[1597] Data is obtained using APIs and scraping technology and stored in a local database.

[1598] Step 2: Preprocessing the data

[1599] The server cleans the collected case data, specifically by filling in missing values ​​and removing outliers.

[1600] Essential data items (summary of the case, content of the judgment, applicable law, etc.) are extracted and formatted for analysis.

[1601] Step 3: Training the AI ​​model

[1602] The server uses the preprocessed data to train the AI ​​model.

[1603] Using natural language processing techniques (e.g., Transformer or RNN), a model based on case law data is created.

[1604] Evaluate the performance of the trained model and retune it if necessary.

[1605] Step 4: Providing an insight generation API

[1606] The server builds an API that uses the trained AI model to generate opinions.

[1607] This API has the ability to respond in real time based on the court information entered.

[1608] Terminal handling

[1609] Step 1: Provide a user interface

[1610] The terminal provides input and display screens that are easy for jurors to use.

[1611] Specifically, it will have a form for entering court information and a dashboard for displaying opinions.

[1612] Step 2: Enter court information

[1613] Court information entered by the user is aggregated on the terminal.

[1614] For example, enter "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street."

[1615] Step 3: Submit your information

[1616] The terminal transmits the entered trial information to the server.

[1617] Data is transmitted securely using security protocols (such as SSL / TLS).

[1618] Step 4: Receive and display feedback

[1619] The terminal receives the response from the server and displays the view to the user.

[1620] For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was around 5 million yen."

[1621] User Action

[1622] Step 1: Enter your information

[1623] The user enters the court information into the terminal.

[1624] Specifically, you enter a summary of the case and relevant evidence into a text form.

[1625] Step 2: Confirm submission

[1626] The user confirms the input and presses the send button to send the information to the server.

[1627] Step 3: Confirm your views

[1628] The user checks the AI's opinion displayed on the device.

[1629] For example, you may obtain information such as, "Based on past precedents, the average amount of compensation is 5 million yen."

[1630] Step 4: Prepare for the discussion

[1631] Users will use the AI's views as a reference and prepare to debate with other jurors.

[1632] The views expressed will be used as a basis for making fair judgments.

[1633] These are the specific processing steps of the "Lay Judge Helper" system. This system reduces the anxiety and burden on lay judges and enables them to make fair and neutral judgments.

[1634] Example 1

[1635] 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."

[1636] In the current lay judge system, ordinary people can feel anxious and burdened due to their lack of legal knowledge and experience. This makes it difficult to make impartial judgments and can undermine the transparency and fairness of trials. A support system is needed to solve this problem and improve the quality of trials.

[1637] 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.

[1638] In this invention, the server includes means for collecting past judicial data, means for preprocessing the collected judicial data, means for training a generative AI model based on the preprocessed data, means for providing an opinion generation API using the trained generative AI model, means for communicating with a processing device to send court information and receive opinions, means for sending court information entered by a user to the processing device and displaying opinions received from the processing device, and means for a user to input court information and refer to opinions. This enables ordinary people to receive support in making fair judgments as jurors, reducing their anxiety and burden.

[1639] "Past judicial data" refers to data related to past judicial proceedings, such as precedents, rulings, and applicable laws.

[1640] "Preprocessing" refers to a series of processes carried out to improve the quality of collected data, specifically including data cleaning and missing value imputation.

[1641] A "generative AI model" is an artificial intelligence model built using machine learning techniques to extract useful information from data and generate opinions and predictions.

[1642] The "opinion generation API" is an application programming interface that uses a generative AI model to generate opinions based on input information and provide them to external parties.

[1643] A "processing device" is an information processing device, such as a server or client terminal, that manages the input, processing, communication, and output of data.

[1644] "User" is a concept that refers to users who operate and use the system and jurors.

[1645] The "information input screen" is an interface that allows the user to input necessary information into the system.

[1646] The "information display screen" is an interface that displays the results and opinions processed by the system to the user.

[1647] A "natural language processing model" is a machine learning model for understanding and generating human language. Specifically, it includes techniques for tasks such as text analysis, translation, and summarization.

[1648] "Means" is a term that refers to methods, techniques, equipment, processes, etc. used to achieve a particular purpose.

[1649] Server Processing

[1650] Data collection and preprocessing

[1651] The server collects past judicial data to be used in the lay judge system. Specifically, it retrieves case data from the judicial database using an API. The retrieved data is cleaned, and incomplete data and outliers are removed. This cleansing process is carried out until the data is clean and consistent. The specific software used is Python and its libraries (Pandas, NumPy, etc.).

[1652] Training an AI model

[1653] The server trains a generative AI model based on the preprocessed data. This process uses natural language processing (NLP) techniques. Specifically, machine learning libraries such as TensorFlow and PyTorch are used to train the AI ​​model to generate opinions from past case law data. The data is split into training and testing datasets, and the model's accuracy is evaluated and its hyperparameters are adjusted.

[1654] Providing an opinion generation API

[1655] The server designs and implements an API that uses a trained generative AI model to generate opinions. This API is designed in a RESTful format, and when court information is input to a specific endpoint, the generated opinion is returned in real time. The API is deployed using cloud services such as AWS and Google Cloud.

[1656] Terminal handling

[1657] Providing a user interface

[1658] The terminal provides a user interface that allows jurors to input information and confirm opinions. The information input screen contains text boxes and drop-down menus for entering trial information such as a summary of the case and evidence. The information display screen displays the AI's opinions received from the server. Specific front-end frameworks used are React and Vue.js.

[1659] Sending court information and receiving opinions

[1660] The court information entered by the user is sent to the server via the terminal. This operation is performed via an HTTP POST request. The opinion received from the server is displayed on the terminal screen. The terminal sends and receives information in real time according to the user's operations.

[1661] User operations

[1662] Enter information

[1663] The user inputs court information (summary of the case and evidence) through the device's input screen. For example, the user might input, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1664] Sending information

[1665] The user sends the entered information to the server via the send button. This operation causes the terminal to send the court information to the server.

[1666] Confirming AI's View

[1667] The AI's opinion sent from the server is received by the device and displayed to the user. For example, it may say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1668] Use in discussions

[1669] Users can use the AI's opinions to discuss with other jurors and make fair decisions, which will improve the transparency and fairness of trials.

[1670] Specific examples

[1671] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[1672] Input information: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light. Pedestrian A broke his leg."

[1673] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[1674] Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1675] Use in discussion: The user will use this opinion to discuss with other jurors and use it in the final verdict.

[1676] This system will reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and provide support to help them make impartial decisions, thereby improving the transparency and fairness of trials and increasing the credibility of the judicial system.

[1677] Prompt Sentence Examples

[1678] Summary of the incident: "Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1679] Sample prompt: "Based on the outline of this case, please tell us your opinion on similar past cases."

[1680] The above is an embodiment of the present invention.

[1681] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1682] Step 1:

[1683] Data collection

[1684] The server collects past judicial data. This data is obtained from the judicial database via API. The API endpoint URL and access key are required as input, and case data is obtained in JSON or CSV format as output. This allows the server to obtain a large amount of case data. Specifically, it periodically runs a script to download the data.

[1685] Step 2:

[1686] Data Preprocessing

[1687] The server preprocesses the collected case data. The input requires the collected case data. The output is a cleaned, consistent dataset. This process involves removing incomplete data and correcting outliers. Specifically, it uses the Python Pandas library to remove unnecessary columns and unify data types.

[1688] Step 3:

[1689] Feature extraction

[1690] The server extracts the necessary features from the preprocessed data. Preprocessed data is required as input. The output is a dataset suitable for training an AI model. Features include a summary of the case, the verdict, and the applicable law. Specific operations include converting the data into lists or vectors using regular expressions and NLP tools.

[1691] Step 4:

[1692] Training an AI model

[1693] The server uses the extracted features to train a generative AI model. A dataset containing the features is required as input. A trained generative AI model is obtained as output. Specifically, the model is trained using machine learning libraries such as TensorFlow and PyTorch. During training, the accuracy of the model is improved by repeatedly learning the dataset.

[1694] Step 5:

[1695] Evaluating and Tuning the Model

[1696] The server evaluates the performance of the generative AI model using a test dataset. A test dataset is required as input. Evaluation results (precision, recall, F-measure, etc.) are obtained as output. If performance is insufficient, hyperparameters are adjusted or the data is reprocessed. Specifically, evaluation is performed using a confusion matrix, and cross-validation is performed if necessary.

[1697] Step 6:

[1698] Providing an opinion generation API

[1699] The server deploys an API that uses a trained generative AI model to generate opinions. Court information is required as input. The generated opinion is output in JSON format. Specifically, the API is built using frameworks such as Flask or FastAPI and deployed on a cloud service.

[1700] Step 7:

[1701] Providing information input screen

[1702] The terminal provides an information input screen where the user can enter court information. The input requires court information to be entered by the user. The output is court information ready to be sent to the server. Specific operation involves building the screen using React and Vue.js, allowing the user to input information intuitively.

[1703] Step 8:

[1704] Sending court information

[1705] The terminal sends the court information entered by the user to the server. The input requires the court information entered by the user. The output is the information accurately sent to the server. Specifically, the information is sent to the server's API using an HTTP POST request.

[1706] Step 9:

[1707] Receiving and displaying AI insights

[1708] The device displays the AI ​​opinion received from the server to the user. The input requires the AI ​​opinion received from the server. The output is the opinion that is displayed to the user. Specifically, it parses the received JSON data and displays it on the screen in an easy-to-read format.

[1709] Step 10:

[1710] Confirming opinions and using them in discussions

[1711] The user checks the AI's opinion displayed on the device and uses it in discussions with other jurors. The input required is the AI's opinion displayed on the device. The output is an opinion that can be used as material for discussion to reach a fair judgment. Specifically, the system operates by advancing discussions based on the opinion and using it to reach a final verdict.

[1712] (Application example 1)

[1713] 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."

[1714] Internal investigations are becoming increasingly important in modern companies and public institutions. However, internal investigators often lack legal knowledge and information on similar cases, making it difficult to conduct fair and prompt investigations. Other issues include inconsistencies in judgment and unclear investigation directions. The present invention aims to solve these problems and provide a system that enables internal investigators to conduct investigations efficiently and fairly.

[1715] 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.

[1716] In this invention, the server includes a means for collecting past case data, a means for preprocessing the collected case data, and a means for training an AI model based on the preprocessed data. This makes it possible to input information about internal investigation cases and send it to the server, and receive and display opinions from the AI ​​model. This allows internal investigators to proceed with investigations based on opinions based on similar past cases, enabling fair and efficient investigations.

[1717] "Past case data" refers to data that includes the results of trials and case information handled by judicial institutions in the past.

[1718] "Preprocessing" refers to the process of complementing missing and outlier values ​​from collected case law data and extracting necessary information.

[1719] An "AI model" is an inference model trained using machine learning algorithms, particularly those that utilize natural language processing techniques.

[1720] An "opinion generation API" is an application program interface that utilizes a trained AI model to generate and provide opinions based on user-entered information.

[1721] "Server" refers to a computer system that receives information from users, communicates with case law databases, and generates and provides opinions based on AI models.

[1722] "Users" are investigators from companies and public institutions who use this system to conduct internal investigations.

[1723] "Information on internal investigation cases" means information containing facts and evidence collected by a company or public institution for the purposes of an internal investigation.

[1724] The "input and transmission means" is an interface that allows users to input information about internal investigation cases into the system and transmit it to the server.

[1725] The "receiving and displaying means" is an interface for displaying to the user the views of the AI ​​model sent from the server.

[1726] A "natural language processing model" is a machine learning model trained to understand text data and respond to questions or requests in natural language.

[1727] This invention is a system that enables companies and public institutions to conduct internal investigations efficiently and fairly. The system collects past case data, uses a trained AI model based on that data to generate opinions, and provides them to internal investigators. The system is primarily composed of three elements: a server, a terminal, and a user.

[1728] System configuration

[1729] Server Configuration

[1730] The server has the following main functions:

[1731] 1. Data Collection and Preprocessing:

[1732] The server collects past case data from judicial databases, and the collected data is cleansed to remove missing and outlier values.

[1733] The pre-processed data extracts necessary information such as the case summary, the judgment, and the applicable law.

[1734] 2. Training the AI ​​model:

[1735] The server uses the preprocessed data to train an AI model, which uses natural language processing techniques to generate opinions from past case data.

[1736] 3. Providing an opinion generation API:

[1737] The server provides an API that uses a trained AI model to generate opinions, which then generate and respond in real time based on the input internal investigation information.

[1738] Device configuration

[1739] The device has the following main features:

[1740] 1. Providing a user interface:

[1741] The device provides an information input screen that users can use and an AI opinion display screen.

[1742] 2. Sending internal investigation information and receiving and displaying opinions:

[1743] The device sends the survey information entered by the user to the server and displays the AI ​​opinion received from the server to the user.

[1744] User operations

[1745] The user uses the system in the following steps:

[1746] 1. Enter your information:

[1747] The user enters internal investigation information (a summary of the case and related evidence) through the device's input screen. For example, the user might enter "Case summary: Employee A is suspected of using company assets for personal purposes."

[1748] 2. Transmission of Information:

[1749] The user sends the entered internal investigation information to the server via the send button. This operation causes the terminal to send the information to the server.

[1750] 3. Confirming AI's View:

[1751] The AI's opinion sent from the server is received by the device and displayed to the user. For example, the server's opinion may be displayed as "In similar cases in the past, dismissal of employees was applied."

[1752] Hardware and software used

[1753] The system of the present invention uses the following hardware and software:

[1754] Hardware:

[1755] Smartphone (Android / iOS)

[1756] Server: A computer system running on the cloud.

[1757] software:

[1758] Programming language: Python

[1759] Library used: requests (sends HTTP requests)

[1760] Natural language processing models: BERT, GPT-3, etc.

[1761] Examples of concrete examples and prompts

[1762] Examples:

[1763] Input information: The user enters, "Employee A is suspected of using company assets for personal purposes."

[1764] Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[1765] Display of opinion: The terminal displays to the user the opinion received from the server, "In similar cases in the past, employee dismissal was applied."

[1766] Example prompt sentence:

[1767] "Case summary: An employee is suspected of using company assets for personal use. Relevant evidence: CCTV footage. What are similar cases in the past and what were the results?"

[1768] The above is a specific embodiment for carrying out the present invention. It is expected that this system will improve the efficiency and accuracy of internal investigations and strengthen compliance in companies and public institutions.

[1769] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1770] Step 1:

[1771] The user enters information about the internal investigation case through the input screen on the terminal. Specifically, the user enters "Case summary: Employee A is suspected of using company assets for personal use." The information entered at this time is in text format, and is stored as is as input data on the terminal.

[1772] Step 2:

[1773] The terminal generates an HTTP request to send the survey information entered by the user to the server. Specifically, it uses the Python library requests to convert the input data into JSON format and sends a POST request to the specified API endpoint. Once the request is sent, the input data is sent to the server.

[1774] Step 3:

[1775] The server processes the received investigation information. First, it parses the received JSON data and converts it into structured data. Then it searches for relevant cases in the preprocessed case law database. The search results are used in the next step.

[1776] Step 4:

[1777] The server generates opinions using a trained AI model based on the search results. Specifically, it uses AI models that use natural language processing techniques (e.g., BERT or GPT-3) to derive opinions from past case data. In this process, the AI ​​model uses the input data as prompts and outputs opinions on relevant cases.

[1778] Step 5:

[1779] The server converts the generated opinion into JSON format and prepares it as response data. The prepared response data is sent to the terminal as an HTTP response. This sends the opinion data to the terminal.

[1780] Step 6:

[1781] The device analyzes the opinion data received from the server and displays it in a user-friendly format. Specifically, it converts the received JSON data into structured data and applies it to the UI components for display. This allows the user to visually confirm the opinion.

[1782] Step 7:

[1783] The user discusses the internal investigation based on the opinion received from the server. For example, based on the opinion that "employee dismissal was applied in past similar cases," the user discusses with other investigators and decides the direction of the investigation.

[1784] Through the above steps, the user can proceed with the internal investigation efficiently and fairly.

[1785] 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.

[1786] ---

[1787] This invention is a system that reduces the anxiety and burden that ordinary people feel when participating in the lay judge system and supports fair judgments. It also reduces the psychological burden on users by incorporating an emotion engine that recognizes the user's emotions. This system collects past case data, uses a trained AI model based on that data to generate opinions, and provides these to lay judges. Furthermore, the emotion engine recognizes the user's emotions and responds accordingly.

[1788] System configuration

[1789] Server Configuration

[1790] 1. Data Collection and Preprocessing:

[1791] The server collects past case data from judicial databases using APIs and scraping technology, and stores it in a local database.

[1792] As preprocessing, the collected data is cleaned (by filling in missing values ​​and removing outliers) and formatted for analysis.

[1793] 2. Training the AI ​​model:

[1794] The server uses the preprocessed data to train an AI model, specifically applying natural language processing techniques to create a model based on past case law data.

[1795] Evaluate the performance of the trained model and retune the model if necessary.

[1796] 3. Providing an opinion generation API:

[1797] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on the input court information.

[1798] Device configuration

[1799] 1. Providing a user interface:

[1800] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[1801] 2. Emotion engine integration:

[1802] The device is equipped with an emotion engine that collects and analyzes the user's facial images and voice data, recognizes the user's emotional state in real time, and transmits the data to a server.

[1803] The emotion engine has the ability to provide appropriate support information and alerts if the user is feeling anxious or stressed.

[1804] User operations

[1805] 1. Enter your information:

[1806] The user accesses the device's input screen and enters court information (an outline of the case and related evidence). For example, the user might enter, "Outline of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1807] 2. Transmission of Information:

[1808] The user confirms the input and presses the send button to send the information to the server. Data is sent securely using a security protocol (such as SSL / TLS).

[1809] 3. Confirming AI's View:

[1810] The server generates an opinion based on the received court information and sends it to the terminal. The terminal then displays the opinion to the user. For example, "In similar past cases, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1811] 4. Recognizing and Responding to Emotions:

[1812] The emotion engine analyzes the user's facial image and voice to recognize their emotions while they are entering court information or confirming opinions. Based on the recognized emotions, it provides appropriate information to reduce the user's anxiety and stress.

[1813] For example, if the user is feeling stressed, the system displays advice and support to help them relax.

[1814] 5. Use in discussions:

[1815] Users can use the AI's views and support from the emotion engine to discuss with other jurors and make fair decisions.

[1816] Specific examples

[1817] Let us take the example of a user participating as a juror in a trial regarding a "pedestrian-vehicle accident."

[1818] 1. Information input: The user inputs, "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[1819] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained AI model.

[1820] 3. Display of opinion: The terminal displays to the user the opinion received from the server: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1821] 4. Emotion Recognition: The emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression. In this case, it displays relaxation techniques and additional support information.

[1822] 5. Use of discussion: Based on this opinion and support, the user will debate as a juror and make a final decision.

[1823] In this way, this invention can reduce the anxiety that lay judges have due to their lack of legal knowledge and experience, and the burden of making fair judgments. Furthermore, by integrating an emotion engine, the psychological burden on users can be further reduced, and the transparency and fairness of trials can be improved.

[1824] The processing flow will be explained below.

[1825] Server Processing

[1826] Step 1:

[1827] The server collects past case data from judicial databases, specifically using APIs and scraping technology to retrieve the data and store it in a local database.

[1828] Step 2:

[1829] The server preprocesses the collected case data, specifically by filling in missing values ​​and removing outliers, and extracting necessary information such as the case summary, the content of the judgment, and the applicable law.

[1830] Step 3:

[1831] The server uses the preprocessed data to train an AI model, uses natural language processing technology to generate a model based on past case law data, and evaluates the performance of the trained model and makes adjustments as necessary.

[1832] Step 4:

[1833] The server uses the trained AI model to build an API that generates opinions, which responds in real time based on the input court information.

[1834] Terminal handling

[1835] Step 1:

[1836] The terminals will provide jurors with an easy-to-use input and display screen, including a form for entering trial information and a dashboard for displaying their opinions.

[1837] Step 2:

[1838] The device collects the user's facial images and voice and provides data for emotion recognition by the emotion engine, which analyzes the collected data and recognizes the user's emotional state in real time.

[1839] Step 3:

[1840] The terminal sends the court information entered by the user to the server. The data is transmitted securely using security protocols (such as SSL / TLS).

[1841] Step 4:

[1842] The device displays the AI's opinion received from the server in real time. For example, it might say, "In similar cases in the past, the perpetrator was found to be grossly negligent, and the average amount of compensation was 5 million yen."

[1843] User Action

[1844] Step 1:

[1845] The user inputs court information (summary of the case and related evidence) into the terminal. Specifically, the user inputs "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1846] Step 2:

[1847] The user checks the input and presses the send button to send the information to the server.

[1848] Step 3:

[1849] The user checks the AI's opinion displayed on the device, and receives information such as, "Based on past court decisions, the average amount of compensation is 5 million yen."

[1850] Step 4:

[1851] While the user is inputting or confirming their opinion, the emotion engine analyzes the user's facial expressions and voice in real time to recognize their emotional state, and if necessary, provides appropriate information to reduce the user's anxiety or stress.

[1852] Step 5:

[1853] Users can use the displayed opinions and supporting information from the emotion engine to discuss with other jurors and use the information to make a fair decision.

[1854] Exemplary Processing Steps

[1855] Step 1:

[1856] The user inputs into the terminal, "Pedestrian A was crossing the street when he was hit by vehicle B, which ignored the green light, and suffered a broken bone."

[1857] Step 2:

[1858] The device sends the entered information to the server, which uses a security protocol to send the data securely.

[1859] Step 3:

[1860] The server generates a view using a trained AI model based on the received information. For example, it may send a view to the terminal that, according to past court decisions, the driver of vehicle B was found to be grossly negligent and the average amount of compensation for damages was 5 million yen.

[1861] Step 4:

[1862] The terminal displays the received views to the user, who then confirms the views.

[1863] Step 5:

[1864] The emotion engine analyzes the user's facial image and voice to detect anxiety and stress. If stress is detected, it displays relaxation techniques and additional support information.

[1865] Step 6:

[1866] Based on the displayed opinions and supporting information from the emotion engine, users can discuss with other jurors and make a fair decision.

[1867] In this way, the present invention reduces the anxiety and burden that lay judges have due to their lack of legal knowledge and experience, and by integrating an emotion engine, it further reduces the psychological burden on users, thereby realizing fair and transparent trials.

[1868] Example 2

[1869] 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."

[1870] For ordinary people who participate in the lay judge system, there are issues such as anxiety, psychological stress, and the heavy burden of being expected to make impartial judgments. Another issue is anxiety due to a lack of trial experience and legal knowledge. Furthermore, if the stress and anxiety experienced during trials are left unaddressed, it can have a negative impact on the psychological health of lay judges.

[1871] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training a generative AI model based on the preprocessed data. This makes it possible to provide a system that reduces the anxiety and psychological burden of ordinary people when participating in the lay judge system and supports fair judgment. Furthermore, the server includes means for providing an opinion generation API using the trained generative AI model and means for recognizing the user's emotions, and provides support information according to the user's emotional state. This further reduces the user's psychological burden and improves the transparency and fairness of trials.

[1872] "Past case data" is data collected from judicial databases that includes the results and details of various past trials.

[1873] "Preprocessing" is the process of converting collected data into a format suitable for analysis and training, including cleaning the data, shaping the data, filling in missing values, removing outliers, and tokenizing the text.

[1874] A "generative AI model" is an artificial intelligence model that is trained based on past case data and automatically generates opinions on case law based on given input data.

[1875] The "Opinion Generation API" is an application program interface that uses a trained generative AI model to generate opinions in real time based on input court information and respond to external requests.

[1876] A "server" is a computer system that performs functions such as collecting data, pre-processing, training AI models, and providing an insight generation API.

[1877] "Court information" refers to important information in a trial, such as an outline of the case and relevant evidence.

[1878] "User" refers to a person who uses the system to enter court information and view generated opinions.

[1879] The "emotion engine" is part of a system that analyzes the user's facial image and voice data, recognizes the user's emotional state, and provides appropriate support information based on that information.

[1880] The "information input screen" is a user interface for users to input court information.

[1881] The "information display screen" is a user interface for displaying to the user the views received from the server.

[1882] A "natural language processing model" is an artificial intelligence model that analyzes text data and understands the meaning and structure of sentences.

[1883] This invention is a system that reduces the anxiety and psychological burden of ordinary people participating in the lay judge system and supports fair judgments. In particular, by combining it with an emotion engine that recognizes the user's emotions, the system can minimize the user's psychological burden. This system collects past case data and provides jurors with opinions based on a generative AI model trained on that data. Furthermore, the emotion engine recognizes the user's emotions and responds appropriately.

[1884] Server Configuration

[1885] 1. Data Collection and Preprocessing:

[1886] The server collects past case data from judicial databases using APIs and scraping technology, and the collected data is stored in a local database.

[1887] The server cleans the data (filling in missing values ​​and removing outliers) and prepares it for analysis, which involves tokenizing the data and removing unnecessary symbols and spaces.

[1888] 2. Training the AI ​​model:

[1889] The server trains a generative AI model based on the preprocessed data. Specifically, it uses natural language processing techniques to extract important features from past case data and create a classification model (e.g., SVM or Random Forest).

[1890] The performance of the trained model is evaluated using cross-validation and other methods, and the accuracy and F1 score are calculated.

[1891] 3. Providing an opinion generation API:

[1892] The server builds an API that generates insights using a trained generative AI model. It uses frameworks such as Flask and FastAPI to design a RESTful API.

[1893] The server deploys the view generation API in a cloud environment (e.g., AWS or GCP).

[1894] Device configuration

[1895] 1. Providing a user interface:

[1896] The device provides a user-friendly input and display interface, which uses HTML and CSS to provide a form for entering a case summary and related evidence.

[1897] The display screen provides a dashboard that displays the insights returned from the server in the form of tables and graphs.

[1898] 2. Emotion engine integration:

[1899] The device collects the user's facial images and voice data and analyzes them with an emotion engine, which uses publicly available APIs such as Microsoft Azure's Face API and IBM Watson Tone Analyzer.

[1900] The device sends the recognized emotion data to a server, which provides additional support information and alerts based on the user's emotional state.

[1901] User Actions

[1902] 1. Enter your information:

[1903] The user accesses the device's input screen and enters a summary of the incident and related evidence. For example, the user might enter, "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1904] 2. Transmission of Information:

[1905] The user checks the information entered and presses the send button to send the information to the server. At this time, a security protocol such as SSL / TLS is used for communication.

[1906] 3. Confirming AI's View:

[1907] The server generates an opinion based on the received court information and sends the result to the terminal. The user can check the opinion on the terminal. The message displayed is, "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1908] 4. Recognizing and Responding to Emotions:

[1909] The emotion engine analyzes the user's facial images and voice in real time while inputting or confirming opinions, and recognizes their emotions. For example, if the user is feeling anxious, it will provide relaxation techniques and additional support information.

[1910] 5. Use in discussions:

[1911] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[1912] Specific examples

[1913] For example, if a user wants to participate as a juror in a trial regarding a "pedestrian-vehicle accident," the user will use the system as follows:

[1914] 1. Information input: The user inputs a summary of the incident: "Pedestrian A was crossing the street when he was hit by vehicle B, which ran a green light, and suffered a broken bone."

[1915] 2. Server processing: Based on the received information, the server searches for similar past cases and generates an opinion using a trained generative AI model.

[1916] 3. Display of opinion: The terminal displays the opinion received from the server to the user. It displays the following: "In past court cases, the driver of vehicle B was found to be grossly negligent, and the average amount of compensation was approximately 5 million yen."

[1917] 4. Emotion Recognition: If the emotion engine analyzes the user's facial image and recognizes anxiety from their facial expression, it will display relaxation techniques and additional support information.

[1918] 5. Use of discussion: Based on this opinion and support, users will debate as jurors and make a final decision.

[1919] The present invention provides support to reduce the psychological burden on lay judges and ensure fairer trials.

[1920] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1921] Step 1: Data collection

[1922] The server accesses the judicial database and sends API requests to collect past case data. Specifically, the server retrieves data using HTTP requests and receives the response in JSON format.

[1923] Input: API request parameters (e.g., trial year, type).

[1924] Output: Case data in JSON format.

[1925] What happens: The server sends a request to an API endpoint, parses the response, and stores it in a local database.

[1926] Step 2: Data Preprocessing

[1927] The server cleans the collected case data and formats it for analysis: text data is tokenized, unnecessary symbols and spaces are removed, missing values ​​are imputed, and outliers are removed.

[1928] Input: Case law data in JSON format.

[1929] Output: Preprocessed data (e.g. cleaned text data).

[1930] Specific operation: The server tokenizes the text data, completes missing values ​​in each case data, and removes outliers. For example, Python libraries (NLP libraries and data processing libraries) are used for text tokenization and missing value completion.

[1931] Step 3: Feature extraction

[1932] The server extracts features using the preprocessed data and converts text to numerical data using a TF-IDF vectorizer.

[1933] Input: Preprocessed text data.

[1934] Output: Numerical data (TF-IDF vectors).

[1935] Specific operation: The server applies TF-IDF to the preprocessed text data to extract important features and convert them into numerical data.

[1936] Step 4: Training the AI ​​model

[1937] The server uses the features to train a generative AI model: it builds a classification model (e.g., SVM or Random Forest) and trains it on the training data.

[1938] Input: Numerical data (features) and corresponding labels.

[1939] Output: A trained generative AI model.

[1940] Specific operation: The server feeds the features to the classification model and performs the training process. After training, the performance of the model is evaluated, and cross-validation is performed to calculate the accuracy and F1 score.

[1941] Step 5: Providing an insight generation API

[1942] The server uses a trained generative AI model to build an API that generates insights. A RESTful API is designed using frameworks such as Flask or FastAPI.

[1943] Input: Court information (e.g., case summary and evidence).

[1944] Output: AI-generated opinion.

[1945] Specific operation: The server receives a request from a user, inputs court information into the generative AI model, and generates an opinion. The generated opinion is returned to the client as an API response.

[1946] Step 6: Providing a User Interface

[1947] The terminal will be designed and provided with a form for entering court information and a dashboard displaying opinions, and will have an intuitive design that will make it easy for users to operate.

[1948] Input: User input information (court information).

[1949] Output: Input screen and display screen.

[1950] Specific operation: The terminal designs the form using HTML and CSS, receives the API response, and displays the insights on the dashboard.

[1951] Step 7: Integrating the Emotion Engine

[1952] The device collects the user's facial images and voice data and analyzes them with an emotion engine, for example, using Microsoft Azure's Face API in real time.

[1953] Input: Face images and audio data.

[1954] Output: Emotion analysis results.

[1955] Specific operation: The device sends data collected by the camera and microphone to the emotion analysis API, and sends the returned results to the server. Based on the analysis results, appropriate support information is provided to the user.

[1956] Step 8: Enter your information

[1957] The user accesses the input screen and enters court information (summary of the case and evidence). For example, the user might enter, "Summary of the case: Vehicle A ran a red light and collided with pedestrian B who was crossing the street. Pedestrian B suffered a broken bone."

[1958] Input: Court Information.

[1959] Output: The entered court information.

[1960] Specific operation: The user enters information into the input screen of the device and presses the send button to send the data to the server.

[1961] Step 9: Confirm the AI's opinion

[1962] The server generates an opinion based on the received legal information, which is then sent to the terminal and displayed to the user.

[1963] Input: Court Information.

[1964] Output: AI-generated opinion.

[1965] Specific operation: The server analyzes the court information using the generative AI model, generates an opinion, and sends it to the terminal. The user can check the opinion on the dashboard.

[1966] Step 10: Recognize and respond to emotions

[1967] The emotion engine analyzes the user's facial images and voice while they are typing or confirming their opinion, recognizing their emotions. If they are feeling anxious, it will provide relaxation techniques and additional support information.

[1968] Input: Face images and audio data.

[1969] Output: Sentiment analysis results, supporting information.

[1970] Specific operation: The emotion engine analyzes the user's image and voice in real time and provides appropriate support information based on the results.

[1971] Step 11: Use in discussions

[1972] Users can discuss with other jurors based on the AI's views and the support of the emotion engine, and use this information to make fair decisions.

[1973] Input: AI-generated opinions, emotional support information.

[1974] Output: Discussion as a judge and final decision.

[1975] Specific operation: Based on the information obtained, the user exchanges opinions with other jurors and conducts the final trial fairly.

[1976] This will reduce the psychological burden on users and create a consistent process that supports fair trials.

[1977] (Application example 2)

[1978] 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."

[1979] The problem that this invention aims to solve is to reduce the stress and anxiety that factory workers feel while working, thereby improving work efficiency and safety. In modern factory work environments, long hours of monotonous work and excessive stress can damage workers' mental health. This can also reduce workers' work performance and negatively impact product quality and production efficiency. To solve this problem, a system is needed that can recognize workers' emotional states in real time and provide appropriate support.

[1980] The specific processing by the specific 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 case data, means for preprocessing the collected case data, and means for training an AI model based on the preprocessed data. The user terminal includes means for providing an opinion generation API using the trained AI model, means for communicating with the server to send court case information and receive opinions, means for sending court case information entered by the user to the server and displaying the opinions received from the server, means for recognizing worker emotions, and means for providing support to workers based on the recognized emotions. This makes it possible to recognize the emotional state of workers working in a factory in real time and provide appropriate support and advice. This reduces worker stress and anxiety, contributing to improved work efficiency and safety.

[1981] "Past case data" is data that includes records of past cases and judgments regarding legal decisions.

[1982] A "collection means" is a device or method that performs a process of obtaining specified data from a particular location and storing it.

[1983] "Preprocessing means" refers to devices or methods that perform processes to prepare and clean collected data for analysis and modeling.

[1984] "Means of training an AI model" refers to the process of using collected and preprocessed data to train an artificial intelligence model and build that model with high accuracy.

[1985] The "means for providing an opinion generation API" is an interface for generating legal opinions from a trained AI model and providing them to external systems or users.

[1986] "Court Information" is information containing facts, evidence, and legal documents relating to a particular court case.

[1987] "User-entered case information" is data about a particular case that is entered into the system by a user.

[1988] A "server" is a central computer system that processes and manages data on a network.

[1989] The "means for receiving an opinion" is a process for receiving the generated legal opinion from the server to the user terminal.

[1990] A "means for displaying an opinion" is a device or method for visually displaying a received legal opinion to a user.

[1991] "Means for recognizing worker emotions" refers to the process of using sensors such as cameras and microphones to analyze a worker's facial expressions and voice to identify their emotional state.

[1992] A "means for providing support to workers based on recognized emotions" is a system or method that suggests appropriate advice or assistance based on the worker's emotional state.

[1993] This invention is an emotion recognition system for reducing stress and anxiety among factory workers and improving work efficiency and safety. This system includes a series of processes, starting with collecting past case data, analyzing emotion data, and providing support.

[1994] System configuration

[1995] Server Configuration

[1996] 1. Data Collection and Preprocessing:

[1997] The server collects historical case data from legal databases and other data sources, using APIs and scraping techniques to retrieve the data and store it in a local database.

[1998] To prepare the collected data for analysis, preprocessing such as cleaning (filling in missing values ​​and removing outliers) and tokenization is performed.

[1999] 2. Training the AI ​​model:

[2000] The server uses the pre-processed dataset to train an AI model, applying natural language processing (NLP) models to create a model capable of generating legal opinions based on historical case law data.

[2001] Evaluate the trained model and retune it if necessary.

[2002] 3. Providing an opinion generation API:

[2003] The server builds an API that uses a trained AI model to generate opinions and responds in real time based on input information, allowing the system to quickly provide legal opinions.

[2004] Configuring the user device

[2005] 1. Information input and display:

[2006] The terminal provides an interface where users can input legal information and emotional data, and in particular, build user-friendly forms and dashboards.

[2007] The entered information is sent to the server, and the view returned by the server is displayed.

[2008] 2. Emotion recognition engine integration:

[2009] The device uses a camera and microphone to collect facial images and voice data of the worker, and sends them to an emotion recognition engine, which recognizes the worker's emotional state in real time.

[2010] Depending on the emotion recognized, the device will display appropriate alerts and support information. For example, if stress levels increase, an alert will be displayed suggesting ways to relax.

[2011] Hardware and software used

[2012] Hardware: Cameras, microphones, database servers, user devices (computers, smartphones, tablets, etc.).

[2013] Software: Google Cloud Vision API, Azure Cognitive Services, MySQL.

[2014] Specific examples

[2015] 1. Information input: The user accesses a terminal in the factory and inputs court information and the emotional state of the worker. For example, the court information might be input as follows: "Summary of the incident: Vehicle A ran a red light and collided with pedestrian B. Pedestrian B suffered a broken bone."

[2016] 2. Server processing: The server analyzes the received information and generates a view using the trained AI model. The view is that "previous court decisions have found the driver of vehicle B to be grossly negligent, and the average amount of compensation is 5 million yen."

[2017] 3. View display: The terminal displays the worker's emotional state in real time along with the view received from the server and provides appropriate advice as needed.

[2018] 4. Emotion Recognition: The emotion recognition engine can identify anxiety from a worker's facial expression and display relaxation techniques or additional support information.

[2019] Prompt Sentence Examples

[2020] "Provide emotional awareness to workers and suggest ways to provide appropriate support and advice where necessary."

[2021] "Consider designing a system that measures workers' stress levels in real time and suggests ways to relax."

[2022] In this way, the present invention makes it possible to recognize the emotional state of workers in a factory in real time and provide appropriate support, thereby reducing stress and anxiety among workers and improving work efficiency and safety.

[2023] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2024] Step 1:

[2025] Data collection and preprocessing:

[2026] The server collects past case data from legal databases and other data sources using APIs and scraping technology. The collected data is stored in a local database. The input is case data obtained through APIs and scraping technology, and the output is formatted data. This data undergoes preprocessing such as missing value completion, outlier removal, and tokenization.

[2027] Step 2:

[2028] Training the AI ​​model:

[2029] The server uses the preprocessed dataset to train a natural language processing (NLP) model, with the input being the preprocessed case data and the output being the trained AI model, which is now capable of generating legal opinions based on past case data.

[2030] Step 3:

[2031] Providing an opinion generation API:

[2032] The server builds and provides an API that generates opinions using a trained AI model. The input is the court information entered by the user, and the output is the legal opinion generated by the AI ​​model. This allows the server to provide legal opinions in real time based on the input information.

[2033] Step 4:

[2034] Information entry and display:

[2035] The user inputs the court information using the terminal interface. The input information is sent from the terminal to the server. The input is the court information entered by the user, and the output is the information sent to the server. The opinion returned by the server is displayed on the terminal for the user to visually confirm.

[2036] Step 5:

[2037] Emotion data collection:

[2038] The device collects facial images and voice data of workers using a camera and microphone. The input is the facial images and voice data acquired through the camera and microphone, and the output is stored in the device and sent to an emotion recognition engine.

[2039] Step 6:

[2040] Emotion recognition engine analysis:

[2041] The emotion recognition engine analyzes the transmitted facial image and voice data to identify the worker's emotional state. The input is the facial image and voice data, and the output is the analyzed emotional state (stress, anxiety, joy, etc.).

[2042] Step 7:

[2043] Providing emotional support:

[2044] The device provides appropriate support and advice to workers based on their recognized emotional state. For example, if stress levels are high, it suggests relaxation techniques or short breaks. The input is the recognized emotional state data, and the output is support information or alerts.

[2045] Step 8:

[2046] Data Storage and Feedback:

[2047] The terminal stores the worker's emotional data and response history, and later analyzes them to propose improvement measures. The input is emotional data and response history, and the output is data stored in a database. This makes it possible to continuously improve the performance of the entire system.

[2048] 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.

[2049] 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.

[2050] 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.

[2051] 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.

[2052] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[2053] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[2054] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[2055] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[2056] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[2057] ...

Claims

1. A means of collecting past case data; a means for pre-processing the collected case law data; A means to train an AI model based on the preprocessed data; and a means for providing an opinion generation API using the trained AI model; means for communicating with the server to transmit court information and receive opinions; a means for transmitting the court decision information input by the user to a server and displaying the opinion received from the server; A means for users to input court information and view opinions; A system including:

2. means for providing an information input screen and an information display screen; means for displaying insights from said insight generation API in real time; The system of claim 1 further comprising:

3. The system of claim 1 , wherein the AI ​​model is a natural language processing model. The above is the draft of the patent claims, which show the technical scope of the invention.

Images