system

Abstract

Provide a system. [Solution] means for receiving current hazard maps and historical weather data; a means for storing the hazard map and meteorological data; means for verifying the format of the received data and performing any necessary data cleansing; means for converting the cleansed data into a format that is easy to analyze; A means for constructing a machine learning model and predicting future disaster risks using the predictive model; means for visualizing the prediction results and generating explanations in natural language; A means for a user to display and confirm the visualization and generated explanation. Including system.

Description

[Technical Field]

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

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

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

[0004] In recent years, the frequent occurrence of natural disasters has increased the importance of updating hazard maps and predicting damage. However, current hazard maps are based on past data, and the inability to adequately predict future changes in disaster risk is an issue. In addition, there is a lack of systems that can analyze past data such as rainfall and weather data to predict future risks. Against this background, there is a demand for building highly accurate prediction models and providing information that is easy to understand for users. [Means for solving the problem]

[0005] The present invention addresses the above-mentioned problems by providing a system that receives current hazard maps and past weather data and builds a machine learning model based on this data. This system includes a means for checking the format of the received data and performing any necessary data cleansing. It also includes a means for converting the cleansed data into an easily analyzable format, a means for building a machine learning model, and a means for predicting future disaster risk using the predictive model. Furthermore, it includes a means for visualizing the prediction results and generating explanations in natural language, allowing users to view and confirm the results. This makes it possible to predict future disaster risk with high accuracy and provide information in a format that is easy for users to understand.

[0006] A "hazard map" is a map that shows the risk of natural disasters in a specific area.

[0007] "Weather data" refers to data related to weather information such as rainfall, weather charts, temperature, and wind speed.

[0008] "Data cleansing" is the process of correcting errors and omissions in received data and preparing it in a form that is easier to analyze.

[0009] A "machine learning model" is a computer program built to make future predictions based on past data.

[0010] A "predictive model" is a model that uses a constructed machine learning model to predict future risks and events.

[0011] "Visualization" is the process of displaying data and prediction results in graphs and maps in a way that is easy for users to understand.

[0012] "Generating an explanation in natural language" means expressing the prediction results as text in terms that are easy for humans to understand. [Brief explanation of the drawings]

[0013] [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

[0014] 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.

[0015] First, the terms used in the following description will be explained.

[0016] 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).

[0017] 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.

[0018] 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.

[0019] 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.

[0020] 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."

[0021] [First embodiment]

[0022] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0023] 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.

[0024] 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).

[0025] 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.

[0026] 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.

[0027] 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.

[0028] 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.

[0029] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] 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.

[0031] 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.

[0032] 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.

[0033] 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."

[0034] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[0035] Data acquisition and storage

[0036] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves them in the appropriate folder, and records metadata (such as the file name, upload date and time, and data type) in the database.

[0037] Data Preprocessing

[0038] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (for example, changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and converting image data into a format that allows image analysis.

[0039] Data analysis and model building

[0040] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0041] Visualization / verbalization

[0042] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google (registered trademark) Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[0043] Display and save results

[0044] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[0045] Through the above process, the present invention realizes a system that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides this information to users.

[0046] The processing flow will be explained below.

[0047] Step 1:

[0048] The user uploads current hazard maps and historical weather data through a web interface. During this process, the user clicks the file selection button and selects the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, the user presses the upload button to send the data to the server.

[0049] Step 2:

[0050] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[0051] Step 3:

[0052] The server cleanses the incoming data by identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[0053] Step 4:

[0054] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into JSON format and converts weather chart image files into an analyzable format (PNG or JPEG). This ensures that subsequent analysis processes proceed smoothly.

[0055] Step 5:

[0056] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[0057] Step 6:

[0058] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[0059] Step 7:

[0060] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[0061] Step 8:

[0062] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[0063] Step 9:

[0064] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[0065] Step 10:

[0066] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[0067] Step 11:

[0068] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[0069] Example 1

[0070] 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."

[0071] Systems that use current hazard maps and past weather data to predict future disaster risks must address issues such as data format, inconsistencies, missing values, etc. They must also provide users with easy-to-understand forecast results and have a means of storing them for future data improvement and verification.

[0072] 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.

[0073] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing hazard maps and weather data, means for checking the format of the received data and performing necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for users to display and confirm the visualized and generated explanations, and means for saving the disaster risk prediction results in a database and retaining them for future model improvement and verification. This makes it possible to ensure data consistency and integrity, build highly accurate prediction models, provide users with easy-to-understand information, and retain data for future improvements.

[0074] A "hazard map" is a map that visually displays information about risks such as natural disasters.

[0075] "Weather data" refers to data that includes past weather observation results such as rainfall, temperature, and wind speed.

[0076] "Means for receiving" refers to the function for receiving data from the outside and incorporating it into the system.

[0077] "Means for storing" refers to a function for saving received data.

[0078] "Data cleansing" refers to the process of filling in missing data values ​​and correcting format inconsistencies and inconsistencies.

[0079] "Easy-to-analyze format" refers to converting data into a format suitable for data analysis and model building.

[0080] A "machine learning model" refers to an algorithm or mathematical model that makes predictions and classifications based on data analysis.

[0081] "Disaster risk" is an indicator that shows the possibility of a natural disaster occurring and the extent of its impact.

[0082] "Visualization" refers to the process of visually displaying data and prediction results to make information easier to understand intuitively.

[0083] "Natural language generation" refers to a technology that outputs the prediction results of a machine learning model in a text format that humans can understand.

[0084] "Database" refers to a system that manages the storage of structured data.

[0085] MODE FOR CARRYING OUT THE INVENTION

[0086] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[0087] Data acquisition and storage

[0088] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves it in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[0089] Data Preprocessing

[0090] The server checks the format of the uploaded data and performs data cleansing. Specific examples include identifying and filling in missing values ​​in weather data and standardizing the data format (for example, changing all time notation to UTC). The server then converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and image data into a format that allows image analysis.

[0091] Data analysis and model building

[0092] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0093] Visualization / verbalization

[0094] The server creates a predicted transition of the hazard map based on the prediction results. As a specific example, the predicted transition is visualized using a map display library (e.g., Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." The results are output in a format that is easy for humans to understand by utilizing a natural language generation (NLG) library.

[0095] Display and save results

[0096] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[0097] Specific examples

[0098] 1. Example of generating a prompt:

[0099] Prompt: "Please upload weather data and hazard maps for 2020-2022."

[0100] Example: A user uploads rainfall data in CSV format, and the server checks and saves the data.

[0101] 2. Data Preprocessing Example:

[0102] Prompt: "Please cleanse the data and standardize the format."

[0103] Example: The server will impute missing data with the average value and convert all times to UTC.

[0104] 3. Model Building Example:

[0105] Prompt: "Construct a flood risk prediction model using weather data from the past 10 years."

[0106] Example: The server trains a model using the LSTM algorithm and makes future predictions.

[0107] 4. Example of result visualization:

[0108] Prompt: "Visualize your prediction results and explain them in natural language."

[0109] Execution example: The server uses the Google Maps API to display the prediction results on a map and generates the sentence, "The risk of flooding will increase in certain areas in 2025."

[0110] In this way, a system is realized that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides the results to users.

[0111] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0112] Step 1: Acquire and store data

[0113] Users upload current hazard maps and historical weather data through a web interface.

[0114] Input: Hazard map (e.g., GeoJSON, KML), historical weather data (e.g., CSV, JSON)

[0115] Specific operation: A user accesses the system's upload page in a web browser, selects a file, and clicks the upload button.

[0116] Output: The server saves the received data in a specific folder and records the metadata in a database.

[0117] Step 2: Format and cleanse the data

[0118] The server inspects the format of the uploaded data and performs any necessary data cleansing.

[0119] Input: Uploaded hazard map and weather data

[0120] What happens: The server checks the file format as CSV or JSON and performs cleansing, such as imputing missing values. For example, it converts all timestamps to UTC and fixes inconsistencies.

[0121] Output: Cleansed data

[0122] Step 3: Transform the data

[0123] The server converts the cleansed data into a format that is easy to parse.

[0124] Input: Cleansed hazard map and meteorological data

[0125] Specific operation: The server converts time series data into a format that can be used for time series analysis (e.g., Pandas DataFrame), and converts image data into a format that can be used for image analysis (e.g., NumPy array).

[0126] Output: Data in an easy-to-parse format

[0127] Step 4: Building a machine learning model

[0128] The server builds a machine learning model using past weather data and hazard maps.

[0129] Input: Rainfall data for the past 10 years and current hazard map

[0130] Specific operation: The server trains models using algorithms such as random forest, LSTM, and CNN, optimizes hyperparameters, and selects the best model.

[0131] Output: A trained machine learning model

[0132] Step 5: Predict disaster risk

[0133] The server inputs the latest weather data and uses a trained machine learning model to predict future disaster risks.

[0134] Input: Latest weather data and trained model

[0135] How it works: The server inputs the latest weather data into the model and retrieves the forecast results, scaling and transforming them as needed.

[0136] Output: Predicted disaster risk

[0137] Step 6: Visualize and verbalize the prediction results

[0138] The server visualizes the prediction results and generates explanations in natural language.

[0139] Input: Predicted disaster risk

[0140] Specific operation: The server uses a map display library (e.g., Leaflet or Google Maps API) to display the prediction results on a map, and generates explanatory text using a natural language generation (NLG) library.

[0141] Output: Visualized hazard map and natural language description

[0142] Step 7: View and save the results

[0143] The user displays and checks the prediction results on their device, and the server stores the prediction results in a database.

[0144] Input: Visualized and natural language prediction results

[0145] How it works: The user checks the prediction results in a web browser and zooms in and out on the map to see the details. The server stores the prediction results in a database and saves them for future model improvement and verification.

[0146] Output: Prediction information provided to the user and prediction results stored in a database

[0147] Through each processing step, the system uses past weather data and current hazard maps to accurately predict future disaster risks and provides the results to users.

[0148] (Application example 1)

[0149] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0150] In modern logistics centers, risk management due to natural disasters is extremely important, but systems that utilize past weather data and hazard maps are insufficient, and a system that can effectively predict future disaster risks and present specific countermeasures is needed.In addition, there is a lack of means to visualize these predictions and provide them to users in an easy-to-understand manner, which creates challenges in planning and executing logistics.

[0151] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0152] In this invention, the server includes a means for receiving the current hazard map and past weather data, a means for storing the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to accurately predict future disaster risks and propose specific countermeasures in the operation of a logistics center.

[0153] A "hazard map" is a map that shows the risk of natural disasters in a specific area.

[0154] "Weather data" refers to data on past weather conditions such as rainfall, temperature, and wind speed.

[0155] A "server" is a computer system for receiving, storing, processing, and analyzing data.

[0156] "Data cleansing" is the process of preparing data to fill in missing values ​​and standardize the format.

[0157] A "machine learning model" is an algorithm built using past data to make future predictions.

[0158] A "predictive model" is a tool for predicting future events using a constructed machine learning model.

[0159] "Visualization" is the process of visually displaying the prediction results.

[0160] "Natural language generation" is a technology that outputs prediction results as text in a form that is easy for humans to understand.

[0161] A "logistics center" is a facility that stores, ships, and delivers goods.

[0162] "Specific measures" are detailed action plans that should be implemented in response to predicted disaster risks.

[0163] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides specific countermeasures appropriate for the operation of a logistics center. Specific embodiments of this system are described below.

[0164] Data acquisition and storage

[0165] Users upload current hazard maps and historical weather data (e.g., rainfall, weather charts) to the system through a web interface. The server receives these data and stores them in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in file format (e.g., CSV or JSON). The server checks the format of these files, saves them in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[0166] Data Preprocessing

[0167] The server verifies the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (e.g., changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and image data into a format that allows image analysis.

[0168] Data analysis and model building

[0169] The server uses past weather data and hazard maps to build a machine learning model. For example, it uses rainfall data and hazard maps from the past 10 years as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0170] Visualization / verbalization

[0171] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[0172] Display and save results

[0173] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database for future model improvement and verification.

[0174] Presentation of specific measures

[0175] Based on the prediction results, the server will suggest specific measures to be taken in accordance with the operation of the logistics center. For example, it could suggest a specific action plan to the logistics center manager, such as, "Due to the increased risk of flooding next week, we recommend that you bring forward your delivery schedule." This will enable quick and effective measures to be taken against disaster risks.

[0176] Prompt Sentence Examples

[0177] "The risk levels for the next days in your area are predicted to be high. Suggest appropriate actions to mitigate this risk."

[0178] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0179] Step 1:

[0180] Users upload current hazard maps and historical weather data through a web interface. The input is a CSV or JSON file uploaded by the user, and the server receives the data. The received data is saved in the corresponding folder, and metadata (file name, upload date and time, data type) is recorded in the database.

[0181] Step 2:

[0182] The server checks the format of the received data and performs data cleansing, i.e., imputing missing values ​​and standardizing the data format. The input is the uploaded raw data, and the output is cleansed data with missing values ​​imputed and standardized format.

[0183] Step 3:

[0184] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into a format that allows time series analysis, and image data into a format that allows image analysis. The input is the cleansed data, and the output is the data converted into an analyzable format.

[0185] Step 4:

[0186] The server uses historical weather data and hazard maps to build a machine learning model. Specifically, it trains a risk prediction model using algorithms such as random forest, LSTM, and CNN. The input is weather data and hazard map data from the past 10 years, and the output is the trained predictive model.

[0187] Step 5:

[0188] The server inputs the latest weather data into the model to predict future disaster risks. The input is the latest weather data, and the output is predicted disaster risk data.

[0189] Step 6:

[0190] The server creates a hazard map based on the prediction results. It uses a map display library (such as Leaflet or Google Maps API) to visually display areas of increased risk. The input is the predicted disaster risk data, and the output is an interactive map showing the predicted changes.

[0191] Step 7:

[0192] The server converts the prediction results into natural language and provides them to the user. Specifically, it uses a natural language generation (NLG) library to generate sentences such as "Flood risk will increase in certain areas in 2025." The input is the prediction results, and the output is a natural language explanation presented to the user.

[0193] Step 8:

[0194] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. They can zoom in and out of the map through a web browser interface to view details. The input is an interactive map and natural language explanations provided by the server, and the output is disaster risk information visually confirmed by the user.

[0195] Step 9:

[0196] Based on the prediction results, the server proposes specific measures appropriate to the operation of the logistics center. For example, it proposes a specific action plan such as, "Due to the increased risk of flooding next week, we recommend bringing forward the delivery schedule." The input is the prediction results and past response data, and the output is a specific countermeasure proposal for the logistics center.

[0197] 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.

[0198] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. Furthermore, the present invention also relates to a system that combines an emotion engine that recognizes the user's emotions and displays appropriate information to the user. A specific embodiment of this system is described below.

[0199] Data acquisition and storage

[0200] Users upload current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, users upload weather data from 2020 to 2022 and current hazard maps in file format. The server checks the data format, saves it in the corresponding folder, and records metadata in the database.

[0201] Data Preprocessing

[0202] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in meteorological data and standardizes the data format. It also converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and converting image data into an analyzable format.

[0203] Data analysis and model building

[0204] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data and hazard maps from the past 10 years are used as training data to build a disaster risk prediction model. Algorithms applied include random forest, LSTM, and CNN. Once the model is trained, it inputs the latest weather data and predicts future disaster risks.

[0205] Visualization / verbalization

[0206] The server visualizes the predicted trends on the hazard map based on the prediction results. Using a map display library, the prediction results are displayed in color on the map. The server also converts the prediction results into natural language and provides it to the user. For example, it automatically generates explanatory text such as, "Flood risk will increase in certain areas in 2025."

[0207] Display and save results

[0208] Users can view the predicted hazard map trends and generated explanations on their devices. They can also zoom in and out of the map via a web browser to view detailed prediction results. The server also stores these prediction results in a database for future analysis and model improvement.

[0209] Optimization by Emotion Engine

[0210] The system includes an emotion engine that recognizes the user's emotions. The device analyzes the user's voice, text, and facial expressions to identify the type and intensity of the emotion. For example, while the user is operating the system, the device captures the user's facial expressions and voice using a webcam and a voice input device and analyzes them in real time. The device then evaluates the user's stress level and interest through the emotion engine.

[0211] Based on emotion recognition, the server adjusts the display and explanation of the prediction results. For example, if the user is feeling anxious, the system will gradually display detailed information so that the user can easily understand. The system will gradually display disaster risks, reducing the user's stress level. In this way, the system provides interactive information according to the user's emotions.

[0212] As described above, the present invention uses past weather data and current hazard maps to predict future disaster risks with high accuracy, provides the information in a format that is easy for users to understand, and optimizes the information display according to the user's emotions.

[0213] The processing flow will be explained below.

[0214] Step 1:

[0215] Users upload current hazard maps and historical weather data through a web interface by clicking the file selection button and selecting the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, they press the upload button to send the data to the server.

[0216] Step 2:

[0217] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[0218] Step 3:

[0219] The server cleanses the incoming data, identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[0220] Step 4:

[0221] The server converts the cleansed data into a format that is easy to parse. For example, it converts time series data into JSON format and converts weather chart image files into parseable formats (e.g., PNG or JPEG). This ensures that subsequent analysis can proceed smoothly.

[0222] Step 5:

[0223] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[0224] Step 6:

[0225] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[0226] Step 7:

[0227] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[0228] Step 8:

[0229] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[0230] Step 9:

[0231] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[0232] Step 10:

[0233] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[0234] Step 11:

[0235] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[0236] Step 12:

[0237] The device analyzes voice, text, and facial expressions to recognize the user's emotions. Specifically, it uses a webcam and a voice input device to capture and analyze the user's facial expressions and voice in real time.

[0238] Step 13:

[0239] The server analyzes the user's emotional data via an emotion engine to identify the type and intensity of the emotion. For example, if the user is feeling stressed, the timing and level of detail of risk information displayed can be adjusted based on that emotional data.

[0240] Step 14:

[0241] The server optimizes the way it displays the prediction results based on the user's emotions. Specifically, it provides detailed information in stages to help users understand the information without straining them. It also displays disaster risks gradually, taking measures to reduce the user's stress level.

[0242] As described above, the present invention can predict future disaster risks with high accuracy using past weather data and current hazard maps, and can optimize the information provided to users in a way that is adaptive to the users' emotions.

[0243] Example 2

[0244] 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."

[0245] Although it is extremely important to predict future disaster risks using current hazard maps and past weather data, there is a problem that it is difficult for users to intuitively understand using only such maps. In particular, the lack of flexible information provision based on the user's emotions leads to insufficient user understanding and stress management. Therefore, there is a need for optimal information display that not only uses past data but also reflects the user's emotions.

[0246] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0247] In this invention, the server includes a means for receiving a current hazard map and past weather data, a means for saving the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to predict future disaster risks with high accuracy using past weather data and a current hazard map and provide the predictions in a format that is intuitively easy for users to understand. Furthermore, by recognizing the user's emotions and optimizing the information display accordingly, the server can deepen the user's understanding and reduce stress.

[0248] A "current hazard map" is a map showing the risk of natural disasters created using the latest geographic information systems.

[0249] "Historical weather data" refers to data relating to the past climate and weather in a particular region.

[0250] "Data cleansing" is the process of filling in missing values ​​in data and removing or correcting inappropriate data.

[0251] A "machine learning model" is a mathematical model that uses algorithms to make predictions or classifications based on input data.

[0252] "Means for generating explanations in natural language" is a process that automatically creates explanations of complex data and prediction results in terms that are easy for users to understand.

[0253] "Visualization" refers to the display of data and prediction results in a visual format such as a graph or map.

[0254] "Means for recognizing emotions and optimizing display content" refers to the process of identifying emotions from the user's voice and facial expressions and adjusting the information displayed according to those emotions.

[0255] "Means for users to display and confirm" refers to the means by which users can display and confirm data and prediction results through a web browser or application.

[0256] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. It also recognizes the user's emotions and optimizes the information display to make it easier for the user to intuitively understand. Specific embodiments are described below.

[0257] Hardware and Software Configuration

[0258] The server is the hardware that primarily handles processing, and it is desirable that it has high-performance computing capabilities. Additionally, a relational database must be used as the database, and it must be capable of storing large amounts of data. The software used is "Python" for data management and analysis, "pandas" for data cleansing, and "scikit-learn" and "TENSORFLOW (registered trademark)" for building machine learning models. "Leaflet.js" is used for map display, and a "natural language processing tool" is used to generate explanations in natural language.

[0259] The device used by the user is a PC or smartphone equipped with a camera and microphone for emotion recognition, which provides the interface. The software used is a web browser or dedicated application.

[0260] Data acquisition and storage

[0261] A user uploads weather data from 2020 to 2022 and the current hazard map in file format through a web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The server receives the data, checks the file format, and saves it to the database. When saving, metadata is also recorded. For example, metadata such as the file name, upload date and time, and file format are recorded.

[0262] Data Preprocessing

[0263] The server first checks the format of the data received and identifies inappropriate data or missing values. If missing values ​​are found, they are filled in with average values ​​or other data. The specific software used is "pandas." After this data cleansing, the data is converted into time series data "timeseries_weather_data.csv" for easier analysis. During this process, the data is converted into a format suitable for analysis.

[0264] Data analysis and model building

[0265] The server trains the machine learning model based on past weather data and the current hazard map. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." After training is complete, the latest weather data is input to predict future disaster risks.

[0266] Visualization / verbalization

[0267] The server visualizes the progress of the hazard map based on the prediction results. For visualization, it uses the map display library "Leaflet.js," color-coding areas with high disaster risk and displaying them on the map. It also uses natural language processing tools to generate explanations of the prediction results. For example, it automatically generates explanations such as "Flood risk will increase in certain areas in 2025."

[0268] Display and save results

[0269] Users can view the visualized hazard map and generated explanations through their device's web browser. Operations such as zooming in and out on the map are also possible. The server also stores the prediction results and generated explanations in a database for future analysis and model improvement.

[0270] Optimization by Emotion Engine

[0271] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display more detailed information and adjust the display to reduce the user's stress.

[0272] Prompt example

[0273] "Please use weather data from the past 10 years and the current hazard map to predict future disaster risks. Please also explain how to present information flexibly according to the user's emotions."

[0274] The above is a specific embodiment of the present invention.

[0275] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0276] Step 1:

[0277] The user uploads weather data from 2020 to 2022 and the current hazard map through the web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The inputs are these files, and the output is the data stored on the server.

[0278] Step 2:

[0279] The server receives the data uploaded by the user and checks the file format. If the format is invalid, it returns an error message. If the format is correct, it saves it to the database and records the metadata. Specifically, it records the file name, upload date and time, file format, etc. The input is the uploaded file and related metadata, and the output is the data and metadata as stored in the database.

[0280] Step 3:

[0281] The server rechecks the format of the received data and identifies improper data or missing values. Specifically, it checks whether there are missing values ​​in the weather data. The input is the weather data stored in the database, and the output is the complete data after missing values ​​are imputed.

[0282] Step 4:

[0283] The server completes and corrects missing values ​​and improper data. Specifically, it uses "pandas" to complete missing values ​​by the mean and reshape the data. The input is data with missing values, and the output is cleansed data after missing value completion.

[0284] Step 5:

[0285] The server converts the cleansed data into a format that is easy to analyze. For example, it converts weather data into time series data "timeseries_weather_data.csv" and hazard maps into "preprocessed_hazard_map.json." The input is cleansed data, and the output is data in an easy-to-analyze format.

[0286] Step 6:

[0287] The server trains a machine learning model using past weather data and current hazard maps. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." The input is past weather data and hazard maps, and the output is a trained machine learning model.

[0288] Step 7:

[0289] The server inputs the latest weather data to predict future disaster risks. Specifically, it feeds new data into the trained model and obtains prediction results. The input is the latest weather data, and the output is a prediction of future disaster risks.

[0290] Step 8:

[0291] The server visualizes the prediction results using the map display library "Leaflet.js." The visualized map displays areas with high disaster risk in different colors. Specifically, high-risk areas are highlighted in red. The input is the disaster risk prediction results, and the output is a color-coded visualized map.

[0292] Step 9:

[0293] The server generates explanatory text for the prediction results using natural language processing tools. For example, it generates an explanatory text such as "Flood risk will increase in certain areas in 2025." The input is the disaster risk prediction results, and the output is an explanatory text generated in natural language.

[0294] Step 10:

[0295] The user can view the visualized hazard map and generated explanation through a web browser on their device. Operations such as zooming in and out on the map are also possible. The input is the visualized map and explanation, and the output is an interface that the user can view and operate.

[0296] Step 11:

[0297] The server stores the prediction results and generated explanations in a database. The stored data is used for future analysis and model improvement. The input is the prediction results and explanations, and the output is the stored data.

[0298] Step 12:

[0299] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display detailed information. The input is voice and facial expression data, and the output is the emotion analysis results and optimized information display.

[0300] (Application example 2)

[0301] 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."

[0302] In recent years, the frequent occurrence of natural disasters has made disaster risk prediction and countermeasures increasingly important. However, current systems have difficulty accurately predicting disaster risks and are unable to provide information that reflects the user's emotions. This can lead to feelings of anxiety among users, which can delay appropriate countermeasures. The present invention aims to solve these problems and enable users to receive information with peace of mind.

[0303] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0304] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing the hazard maps and weather data, means for checking the format of the received data and performing any necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for the user to display and confirm the visualized and generated explanations, emotion recognition means for recognizing the user's emotions in real time, and means for displaying and optimizing information according to the user's emotional state based on the emotion recognition means. This allows the user to intuitively understand the highly accurate disaster risk prediction results and receive appropriate information according to their emotional state.

[0305] A "hazard map" is a map that shows the regional distribution of disaster risk, visually displaying the predicted extent of the impact and degree of danger of a disaster.

[0306] "Weather data" means data relating to past and current weather conditions, including temperature, precipitation, wind speed, wind direction, and the like.

[0307] "Data cleansing" refers to the process of filling in missing data values ​​and correcting inconsistencies and errors, thereby improving data quality.

[0308] A "machine learning model" is a collection of algorithms that are trained to make specific predictions or classifications using past data.

[0309] A "predictive model" is a mathematical model for predicting future events using machine learning models.

[0310] "Visualization" is the process of visually representing data and prediction results and presenting them in a way that is easy for users to understand.

[0311] "Means for generating explanations in natural language" refers to technology that automatically generates explanations in natural language for users based on prediction results and analysis results.

[0312] "Emotion recognition means" is a technology that analyzes and recognizes emotions from a user's facial expressions, voice, and text data.

[0313] "Means for displaying and optimizing information" refers to technology that appropriately adjusts the display method and content of information based on the user's emotional state.

[0314] This invention relates to a system that predicts disaster risks and displays the resulting information optimally. This system receives and stores hazard maps and past weather data, performs data cleansing, and builds machine learning models to predict future disaster risks. Furthermore, it uses emotion recognition to optimize the display of information according to the user's emotional state.

[0315] First, the server receives current hazard maps and historical weather data. This involves uploading user-provided data to the server, which then stores the data in a database for future access. Specifically, users upload weather data in point or grid format through a web interface. The server verifies the data format and records and saves the metadata.

[0316] The server then performs cleansing on the received data. This is the process of filling in missing values ​​and standardizing data formats. During this process, the data is converted into a format that is easier to analyze. For example, time series data is standardized into a format that allows time series analysis, and data in different units is reconciled.

[0317] The server then builds a machine learning model based on the cleansed data. This uses a training dataset of historical weather data and hazard maps. Algorithms used include random forest, LSTM, and CNN. This results in a model that can accurately predict future disaster risks.

[0318] The predictive model uses the latest weather data as input to predict future disaster risks. The prediction results are visualized using a map display library. For example, Leaflet or Mapbox can be used to color-code high-risk areas. In addition, natural language generation technology is used to automatically generate user-friendly explanations. This includes implementations using TensorFlow and Scikit-learn.

[0319] Users can check the prediction results through a smartphone app or web browser. The app uses GPS to provide detailed prediction results based on the user's current location in real time. Furthermore, the server uses emotion recognition to analyze the user's voice and facial expressions. This analysis uses the camera, microphone, and libraries such as Mediapipe and DeepFace. The content and format of the information displayed are optimized according to the user's emotional state. For example, if the user is feeling anxious, Saver will present information in a gradual and intuitive manner, providing a sense of security.

[0320] For example, when a user launches the app on a rainy day, it predicts that a particular area will be at increased risk of flooding over the next week based on past weather data and current hazard maps. When the emotion engine recognizes that the user is feeling anxious, the app will gradually display messages such as, "Caution is required for the next week. Would you like to see routes to the nearest evacuation shelter?" and provide evacuation routes and shelter information.

[0321] An example of a prompt sentence to input to the generative AI model is as follows:

[0322] "Use the past two years of rainfall data and current hazard maps to predict flood risk based on your location."

[0323] "Recognize the user's emotions from their voice and facial expressions, and optimize the information displayed to provide a sense of security."

[0324] The above is an embodiment of the present invention. This system allows users to intuitively understand the highly accurate disaster risk prediction results and to receive appropriate information according to their own emotional state.

[0325] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0326] Step 1:

[0327] Receiving and storing data

[0328] Users provide current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, when a user uploads data, the server checks its format, records the metadata, and saves it in the appropriate folder. The input is the hazard map and weather data, and the output is the uncleansed data stored in the database.

[0329] Step 2:

[0330] Data cleansing and formatting

[0331] The server checks the format of the received data and performs data cleansing. For example, it fills in missing values ​​and standardizes the format. The cleansed data is converted to make it easier to analyze. The input is the data before cleansing, and the output is the cleansed data that can be analyzed. Specific operations include unifying time series data and standardizing image data.

[0332] Step 3:

[0333] Building a machine learning model

[0334] The server builds a machine learning model based on the cleansed data. It trains a training dataset using past weather data and hazard maps to create a model that can predict future disaster risks. The input is the cleansed data, and the output is the constructed predictive model. Algorithms used include random forest, LSTM, and CNN.

[0335] Step 4:

[0336] Disaster risk prediction

[0337] The server inputs the latest weather data and uses the constructed model to predict future disaster risk. The inputs are the latest weather data and a prediction model, and the output is the predicted disaster risk value. Specifically, it executes the prediction algorithm and outputs the future risk as a numerical value.

[0338] Step 5:

[0339] Visualization of prediction results and natural language generation

[0340] To visually display the prediction results, the server uses a map display library to display the predicted disaster risk on a map in a color-coded manner. It also uses natural language generation technology to generate explanatory text that is easy for users to understand. The input is the prediction results, and the output is a visualized map and explanatory text. Specifically, it generates a map using Leaflet or Mapbox, and automatically generates explanatory text using TensorFlow or Scikit-learn.

[0341] Step 6:

[0342] Viewing and verifying results

[0343] Users can check the forecast results and their explanations via a smartphone app or web browser. The app uses GPS functionality to provide real-time disaster risk information based on the user's current location. The input is a visualized map and explanatory text, and the output is information displayed on the user's device. Specifically, the app displays the results on a user interface, allowing users to zoom in and out on the map and check detailed information.

[0344] Step 7:

[0345] Emotion recognition and information optimization

[0346] The device captures the user's voice and facial expressions and uses emotion recognition means to analyze the user's emotional state. Based on the analysis results, the server adjusts the way information is displayed and provides optimal information according to the user's emotional state. The input is the user's voice and facial expression data, and the output is optimized information display. Specifically, data is collected in real time using a camera and microphone, and analyzed using Mediapipe or DeepFace.

[0347] By executing the above processing steps, the present invention provides the user with highly accurate disaster risk prediction information and realizes optimization of appropriate information display according to the user's emotional state.

[0348] 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.

[0349] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (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.

[0350] 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.

[0351] [Second embodiment]

[0352] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0353] 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.

[0354] 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).

[0355] 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.

[0356] 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.

[0357] 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).

[0358] 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. 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.

[0359] 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.

[0360] 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.

[0361] 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.

[0362] 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.

[0363] 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."

[0364] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[0365] Data acquisition and storage

[0366] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves them in the appropriate folder, and records metadata (such as the file name, upload date and time, and data type) in the database.

[0367] Data Preprocessing

[0368] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (for example, changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and converting image data into a format that allows image analysis.

[0369] Data analysis and model building

[0370] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0371] Visualization / verbalization

[0372] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[0373] Display and save results

[0374] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[0375] Through the above process, the present invention realizes a system that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides this information to users.

[0376] The processing flow will be explained below.

[0377] Step 1:

[0378] The user uploads current hazard maps and historical weather data through a web interface. During this process, the user clicks the file selection button and selects the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, the user presses the upload button to send the data to the server.

[0379] Step 2:

[0380] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[0381] Step 3:

[0382] The server cleanses the incoming data by identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[0383] Step 4:

[0384] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into JSON format and converts weather chart image files into an analyzable format (PNG or JPEG). This ensures that subsequent analysis processes proceed smoothly.

[0385] Step 5:

[0386] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[0387] Step 6:

[0388] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[0389] Step 7:

[0390] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[0391] Step 8:

[0392] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[0393] Step 9:

[0394] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[0395] Step 10:

[0396] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[0397] Step 11:

[0398] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[0399] Example 1

[0400] 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."

[0401] Systems that use current hazard maps and past weather data to predict future disaster risks must address issues such as data format, inconsistencies, missing values, etc. They must also provide users with easy-to-understand forecast results and have a means of storing them for future data improvement and verification.

[0402] 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.

[0403] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing hazard maps and weather data, means for checking the format of the received data and performing necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for users to display and confirm the visualized and generated explanations, and means for saving the disaster risk prediction results in a database and retaining them for future model improvement and verification. This makes it possible to ensure data consistency and integrity, build highly accurate prediction models, provide users with easy-to-understand information, and retain data for future improvements.

[0404] A "hazard map" is a map that visually displays information about risks such as natural disasters.

[0405] "Weather data" refers to data that includes past weather observation results such as rainfall, temperature, and wind speed.

[0406] "Means for receiving" refers to the function for receiving data from the outside and incorporating it into the system.

[0407] "Means for storing" refers to a function for saving received data.

[0408] "Data cleansing" refers to the process of filling in missing data values ​​and correcting format inconsistencies and inconsistencies.

[0409] "Easy-to-analyze format" refers to converting data into a format suitable for data analysis and model building.

[0410] A "machine learning model" refers to an algorithm or mathematical model that makes predictions and classifications based on data analysis.

[0411] "Disaster risk" is an indicator that shows the possibility of a natural disaster occurring and the extent of its impact.

[0412] "Visualization" refers to the process of visually displaying data and prediction results to make information easier to understand intuitively.

[0413] "Natural language generation" refers to a technology that outputs the prediction results of a machine learning model in a text format that humans can understand.

[0414] "Database" refers to a system that manages the storage of structured data.

[0415] MODE FOR CARRYING OUT THE INVENTION

[0416] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[0417] Data acquisition and storage

[0418] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves it in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[0419] Data Preprocessing

[0420] The server checks the format of the uploaded data and performs data cleansing. Specific examples include identifying and filling in missing values ​​in weather data and standardizing the data format (for example, changing all time notation to UTC). The server then converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and image data into a format that allows image analysis.

[0421] Data analysis and model building

[0422] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0423] Visualization / verbalization

[0424] The server creates a predicted transition of the hazard map based on the prediction results. As a specific example, the predicted transition is visualized using a map display library (e.g., Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." The results are output in a format that is easy for humans to understand by utilizing a natural language generation (NLG) library.

[0425] Display and save results

[0426] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[0427] Specific examples

[0428] 1. Example of generating a prompt:

[0429] Prompt: "Please upload weather data and hazard maps for 2020-2022."

[0430] Example: A user uploads rainfall data in CSV format, and the server checks and saves the data.

[0431] 2. Data Preprocessing Example:

[0432] Prompt: "Please cleanse the data and standardize the format."

[0433] Example: The server will impute missing data with the average value and convert all times to UTC.

[0434] 3. Model Building Example:

[0435] Prompt: "Construct a flood risk prediction model using weather data from the past 10 years."

[0436] Example: The server trains a model using the LSTM algorithm and makes future predictions.

[0437] 4. Example of result visualization:

[0438] Prompt: "Visualize your prediction results and explain them in natural language."

[0439] Execution example: The server uses the Google Maps API to display the prediction results on a map and generates the sentence, "The risk of flooding will increase in certain areas in 2025."

[0440] In this way, a system is realized that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides the results to users.

[0441] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0442] Step 1: Acquire and store data

[0443] Users upload current hazard maps and historical weather data through a web interface.

[0444] Input: Hazard map (e.g., GeoJSON, KML), historical weather data (e.g., CSV, JSON)

[0445] Specific operation: A user accesses the system's upload page in a web browser, selects a file, and clicks the upload button.

[0446] Output: The server saves the received data in a specific folder and records the metadata in a database.

[0447] Step 2: Format and cleanse the data

[0448] The server inspects the format of the uploaded data and performs any necessary data cleansing.

[0449] Input: Uploaded hazard map and weather data

[0450] What happens: The server checks the file format as CSV or JSON and performs cleansing, such as imputing missing values. For example, it converts all timestamps to UTC and fixes inconsistencies.

[0451] Output: Cleansed data

[0452] Step 3: Transform the data

[0453] The server converts the cleansed data into a format that is easy to parse.

[0454] Input: Cleansed hazard map and meteorological data

[0455] Specific operation: The server converts time series data into a format that can be used for time series analysis (e.g., Pandas DataFrame), and converts image data into a format that can be used for image analysis (e.g., NumPy array).

[0456] Output: Data in an easy-to-parse format

[0457] Step 4: Building a machine learning model

[0458] The server builds a machine learning model using past weather data and hazard maps.

[0459] Input: Rainfall data for the past 10 years and current hazard map

[0460] Specific operation: The server trains models using algorithms such as random forest, LSTM, and CNN, optimizes hyperparameters, and selects the best model.

[0461] Output: A trained machine learning model

[0462] Step 5: Predict disaster risk

[0463] The server inputs the latest weather data and uses a trained machine learning model to predict future disaster risks.

[0464] Input: Latest weather data and trained model

[0465] How it works: The server inputs the latest weather data into the model and retrieves the forecast results, scaling and transforming them as needed.

[0466] Output: Predicted disaster risk

[0467] Step 6: Visualize and verbalize the prediction results

[0468] The server visualizes the prediction results and generates explanations in natural language.

[0469] Input: Predicted disaster risk

[0470] Specific operation: The server uses a map display library (e.g., Leaflet or Google Maps API) to display the prediction results on a map, and generates explanatory text using a natural language generation (NLG) library.

[0471] Output: Visualized hazard map and natural language description

[0472] Step 7: View and save the results

[0473] The user displays and checks the prediction results on their device, and the server stores the prediction results in a database.

[0474] Input: Visualized and natural language prediction results

[0475] How it works: The user checks the prediction results in a web browser and zooms in and out on the map to see the details. The server stores the prediction results in a database and saves them for future model improvement and verification.

[0476] Output: Prediction information provided to the user and prediction results stored in a database

[0477] Through each processing step, the system uses past weather data and current hazard maps to accurately predict future disaster risks and provides the results to users.

[0478] (Application example 1)

[0479] 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."

[0480] In modern logistics centers, risk management due to natural disasters is extremely important, but systems that utilize past weather data and hazard maps are insufficient, and a system that can effectively predict future disaster risks and present specific countermeasures is needed.In addition, there is a lack of means to visualize these predictions and provide them to users in an easy-to-understand manner, which creates challenges in planning and executing logistics.

[0481] 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.

[0482] In this invention, the server includes a means for receiving the current hazard map and past weather data, a means for storing the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to accurately predict future disaster risks and propose specific countermeasures in the operation of a logistics center.

[0483] A "hazard map" is a map that shows the risk of natural disasters in a specific area.

[0484] "Weather data" refers to data on past weather conditions such as rainfall, temperature, and wind speed.

[0485] A "server" is a computer system for receiving, storing, processing, and analyzing data.

[0486] "Data cleansing" is the process of preparing data to fill in missing values ​​and standardize the format.

[0487] A "machine learning model" is an algorithm built using past data to make future predictions.

[0488] A "predictive model" is a tool for predicting future events using a constructed machine learning model.

[0489] "Visualization" is the process of visually displaying the prediction results.

[0490] "Natural language generation" is a technology that outputs prediction results as text in a form that is easy for humans to understand.

[0491] A "logistics center" is a facility that stores, ships, and delivers goods.

[0492] "Specific measures" are detailed action plans that should be implemented in response to predicted disaster risks.

[0493] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides specific countermeasures appropriate for the operation of a logistics center. Specific embodiments of this system are described below.

[0494] Data acquisition and storage

[0495] Users upload current hazard maps and historical weather data (e.g., rainfall, weather charts) to the system through a web interface. The server receives these data and stores them in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in file format (e.g., CSV or JSON). The server checks the format of these files, saves them in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[0496] Data Preprocessing

[0497] The server verifies the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (e.g., changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and image data into a format that allows image analysis.

[0498] Data analysis and model building

[0499] The server uses past weather data and hazard maps to build a machine learning model. For example, it uses rainfall data and hazard maps from the past 10 years as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0500] Visualization / verbalization

[0501] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[0502] Display and save results

[0503] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database for future model improvement and verification.

[0504] Presentation of specific measures

[0505] Based on the prediction results, the server will suggest specific measures to be taken in accordance with the operation of the logistics center. For example, it could suggest a specific action plan to the logistics center manager, such as, "Due to the increased risk of flooding next week, we recommend that you bring forward your delivery schedule." This will enable quick and effective measures to be taken against disaster risks.

[0506] Prompt Sentence Examples

[0507] "The risk levels for the next days in your area are predicted to be high. Suggest appropriate actions to mitigate this risk."

[0508] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0509] Step 1:

[0510] Users upload current hazard maps and historical weather data through a web interface. The input is a CSV or JSON file uploaded by the user, and the server receives the data. The received data is saved in the corresponding folder, and metadata (file name, upload date and time, data type) is recorded in the database.

[0511] Step 2:

[0512] The server checks the format of the received data and performs data cleansing, i.e., imputing missing values ​​and standardizing the data format. The input is the uploaded raw data, and the output is cleansed data with missing values ​​imputed and standardized format.

[0513] Step 3:

[0514] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into a format that allows time series analysis, and image data into a format that allows image analysis. The input is the cleansed data, and the output is the data converted into an analyzable format.

[0515] Step 4:

[0516] The server uses historical weather data and hazard maps to build a machine learning model. Specifically, it trains a risk prediction model using algorithms such as random forest, LSTM, and CNN. The input is weather data and hazard map data from the past 10 years, and the output is the trained predictive model.

[0517] Step 5:

[0518] The server inputs the latest weather data into the model to predict future disaster risks. The input is the latest weather data, and the output is predicted disaster risk data.

[0519] Step 6:

[0520] The server creates a hazard map based on the prediction results. It uses a map display library (such as Leaflet or Google Maps API) to visually display areas of increased risk. The input is the predicted disaster risk data, and the output is an interactive map showing the predicted changes.

[0521] Step 7:

[0522] The server converts the prediction results into natural language and provides them to the user. Specifically, it uses a natural language generation (NLG) library to generate sentences such as "Flood risk will increase in certain areas in 2025." The input is the prediction results, and the output is a natural language explanation presented to the user.

[0523] Step 8:

[0524] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. They can zoom in and out of the map through a web browser interface to view details. The input is an interactive map and natural language explanations provided by the server, and the output is disaster risk information visually confirmed by the user.

[0525] Step 9:

[0526] Based on the prediction results, the server proposes specific measures appropriate to the operation of the logistics center. For example, it proposes a specific action plan such as, "Due to the increased risk of flooding next week, we recommend bringing forward the delivery schedule." The input is the prediction results and past response data, and the output is a specific countermeasure proposal for the logistics center.

[0527] 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.

[0528] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. Furthermore, the present invention also relates to a system that combines an emotion engine that recognizes the user's emotions and displays appropriate information to the user. A specific embodiment of this system is described below.

[0529] Data acquisition and storage

[0530] Users upload current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, users upload weather data from 2020 to 2022 and current hazard maps in file format. The server checks the data format, saves it in the corresponding folder, and records metadata in the database.

[0531] Data Preprocessing

[0532] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in meteorological data and standardizes the data format. It also converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and converting image data into an analyzable format.

[0533] Data analysis and model building

[0534] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data and hazard maps from the past 10 years are used as training data to build a disaster risk prediction model. Algorithms applied include random forest, LSTM, and CNN. Once the model is trained, it inputs the latest weather data and predicts future disaster risks.

[0535] Visualization / verbalization

[0536] The server visualizes the predicted trends on the hazard map based on the prediction results. Using a map display library, the prediction results are displayed in color on the map. The server also converts the prediction results into natural language and provides it to the user. For example, it automatically generates explanatory text such as, "Flood risk will increase in certain areas in 2025."

[0537] Display and save results

[0538] Users can view the predicted hazard map trends and generated explanations on their devices. They can also zoom in and out of the map via a web browser to view detailed prediction results. The server also stores these prediction results in a database for future analysis and model improvement.

[0539] Optimization by Emotion Engine

[0540] The system includes an emotion engine that recognizes the user's emotions. The device analyzes the user's voice, text, and facial expressions to identify the type and intensity of the emotion. For example, while the user is operating the system, the device captures the user's facial expressions and voice using a webcam and a voice input device and analyzes them in real time. The device then evaluates the user's stress level and interest through the emotion engine.

[0541] Based on emotion recognition, the server adjusts the display and explanation of the prediction results. For example, if the user is feeling anxious, the system will gradually display detailed information so that the user can easily understand. The system will gradually display disaster risks, reducing the user's stress level. In this way, the system provides interactive information according to the user's emotions.

[0542] As described above, the present invention uses past weather data and current hazard maps to predict future disaster risks with high accuracy, provides the information in a format that is easy for users to understand, and optimizes the information display according to the user's emotions.

[0543] The processing flow will be explained below.

[0544] Step 1:

[0545] Users upload current hazard maps and historical weather data through a web interface by clicking the file selection button and selecting the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, they press the upload button to send the data to the server.

[0546] Step 2:

[0547] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[0548] Step 3:

[0549] The server cleanses the incoming data, identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[0550] Step 4:

[0551] The server converts the cleansed data into a format that is easy to parse. For example, it converts time series data into JSON format and converts weather chart image files into parseable formats (e.g., PNG or JPEG). This ensures that subsequent analysis can proceed smoothly.

[0552] Step 5:

[0553] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[0554] Step 6:

[0555] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[0556] Step 7:

[0557] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[0558] Step 8:

[0559] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[0560] Step 9:

[0561] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[0562] Step 10:

[0563] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[0564] Step 11:

[0565] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[0566] Step 12:

[0567] The device analyzes voice, text, and facial expressions to recognize the user's emotions. Specifically, it uses a webcam and a voice input device to capture and analyze the user's facial expressions and voice in real time.

[0568] Step 13:

[0569] The server analyzes the user's emotional data via an emotion engine to identify the type and intensity of the emotion. For example, if the user is feeling stressed, the timing and level of detail of risk information displayed can be adjusted based on that emotional data.

[0570] Step 14:

[0571] The server optimizes the way it displays the prediction results based on the user's emotions. Specifically, it provides detailed information in stages to help users understand the information without straining them. It also displays disaster risks gradually, taking measures to reduce the user's stress level.

[0572] As described above, the present invention can predict future disaster risks with high accuracy using past weather data and current hazard maps, and can optimize the information provided to users in a way that is adaptive to the users' emotions.

[0573] Example 2

[0574] 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."

[0575] Although it is extremely important to predict future disaster risks using current hazard maps and past weather data, there is a problem that it is difficult for users to intuitively understand using only such maps. In particular, the lack of flexible information provision based on the user's emotions leads to insufficient user understanding and stress management. Therefore, there is a need for optimal information display that not only uses past data but also reflects the user's emotions.

[0576] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0577] In this invention, the server includes a means for receiving a current hazard map and past weather data, a means for saving the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to predict future disaster risks with high accuracy using past weather data and a current hazard map and provide the predictions in a format that is intuitively easy for users to understand. Furthermore, by recognizing the user's emotions and optimizing the information display accordingly, the server can deepen the user's understanding and reduce stress.

[0578] A "current hazard map" is a map showing the risk of natural disasters created using the latest geographic information systems.

[0579] "Historical weather data" refers to data relating to the past climate and weather in a particular region.

[0580] "Data cleansing" is the process of filling in missing values ​​in data and removing or correcting inappropriate data.

[0581] A "machine learning model" is a mathematical model that uses algorithms to make predictions or classifications based on input data.

[0582] "Means for generating explanations in natural language" is a process that automatically creates explanations of complex data and prediction results in terms that are easy for users to understand.

[0583] "Visualization" refers to the display of data and prediction results in a visual format such as a graph or map.

[0584] "Means for recognizing emotions and optimizing display content" refers to the process of identifying emotions from the user's voice and facial expressions and adjusting the information displayed according to those emotions.

[0585] "Means for users to display and confirm" refers to the means by which users can display and confirm data and prediction results through a web browser or application.

[0586] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. It also recognizes the user's emotions and optimizes the information display to make it easier for the user to intuitively understand. Specific embodiments are described below.

[0587] Hardware and Software Configuration

[0588] The server is the hardware that primarily performs processing, and it is desirable that it has high-performance computing capabilities. In addition, a relational database must be used as the database, and it must be capable of storing large amounts of data. The software used is "Python" for data management and analysis, "pandas" for data cleansing, and "scikit-learn" and "TensorFlow" for building machine learning models. "Leaflet.js" is used for map display, and a "natural language processing tool" is used to generate explanations in natural language.

[0589] The device used by the user is a PC or smartphone equipped with a camera and microphone for emotion recognition, which provides the interface. The software used is a web browser or dedicated application.

[0590] Data acquisition and storage

[0591] A user uploads weather data from 2020 to 2022 and the current hazard map in file format through a web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The server receives the data, checks the file format, and saves it to the database. When saving, metadata is also recorded. For example, metadata such as the file name, upload date and time, and file format are recorded.

[0592] Data Preprocessing

[0593] The server first checks the format of the data received and identifies inappropriate data or missing values. If missing values ​​are found, they are filled in with average values ​​or other data. The specific software used is "pandas." After this data cleansing, the data is converted into time series data "timeseries_weather_data.csv" for easier analysis. During this process, the data is converted into a format suitable for analysis.

[0594] Data analysis and model building

[0595] The server trains the machine learning model based on past weather data and the current hazard map. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." After training is complete, the latest weather data is input to predict future disaster risks.

[0596] Visualization / verbalization

[0597] The server visualizes the progress of the hazard map based on the prediction results. For visualization, it uses the map display library "Leaflet.js," color-coding areas with high disaster risk and displaying them on the map. It also uses natural language processing tools to generate explanations of the prediction results. For example, it automatically generates explanations such as "Flood risk will increase in certain areas in 2025."

[0598] Display and save results

[0599] Users can view the visualized hazard map and generated explanations through their device's web browser. Operations such as zooming in and out on the map are also possible. The server also stores the prediction results and generated explanations in a database for future analysis and model improvement.

[0600] Optimization by Emotion Engine

[0601] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display more detailed information and adjust the display to reduce the user's stress.

[0602] Prompt example

[0603] "Please use weather data from the past 10 years and the current hazard map to predict future disaster risks. Please also explain how to present information flexibly according to the user's emotions."

[0604] The above is a specific embodiment of the present invention.

[0605] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0606] Step 1:

[0607] The user uploads weather data from 2020 to 2022 and the current hazard map through the web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The inputs are these files, and the output is the data stored on the server.

[0608] Step 2:

[0609] The server receives the data uploaded by the user and checks the file format. If the format is invalid, it returns an error message. If the format is correct, it saves it to the database and records the metadata. Specifically, it records the file name, upload date and time, file format, etc. The input is the uploaded file and related metadata, and the output is the data and metadata as stored in the database.

[0610] Step 3:

[0611] The server rechecks the format of the received data and identifies improper data or missing values. Specifically, it checks whether there are missing values ​​in the weather data. The input is the weather data stored in the database, and the output is the complete data after missing values ​​are imputed.

[0612] Step 4:

[0613] The server completes and corrects missing values ​​and improper data. Specifically, it uses "pandas" to complete missing values ​​by the mean and reshape the data. The input is data with missing values, and the output is cleansed data after missing value completion.

[0614] Step 5:

[0615] The server converts the cleansed data into a format that is easy to analyze. For example, it converts weather data into time series data "timeseries_weather_data.csv" and hazard maps into "preprocessed_hazard_map.json." The input is cleansed data, and the output is data in an easy-to-analyze format.

[0616] Step 6:

[0617] The server trains a machine learning model using past weather data and current hazard maps. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." The input is past weather data and hazard maps, and the output is a trained machine learning model.

[0618] Step 7:

[0619] The server inputs the latest weather data to predict future disaster risks. Specifically, it feeds new data into the trained model and obtains prediction results. The input is the latest weather data, and the output is a prediction of future disaster risks.

[0620] Step 8:

[0621] The server visualizes the prediction results using the map display library "Leaflet.js." The visualized map displays areas with high disaster risk in different colors. Specifically, high-risk areas are highlighted in red. The input is the disaster risk prediction results, and the output is a color-coded visualized map.

[0622] Step 9:

[0623] The server generates explanatory text for the prediction results using natural language processing tools. For example, it generates an explanatory text such as "Flood risk will increase in certain areas in 2025." The input is the disaster risk prediction results, and the output is an explanatory text generated in natural language.

[0624] Step 10:

[0625] The user can view the visualized hazard map and generated explanation through a web browser on their device. Operations such as zooming in and out on the map are also possible. The input is the visualized map and explanation, and the output is an interface that the user can view and operate.

[0626] Step 11:

[0627] The server stores the prediction results and generated explanations in a database. The stored data is used for future analysis and model improvement. The input is the prediction results and explanations, and the output is the stored data.

[0628] Step 12:

[0629] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display detailed information. The input is voice and facial expression data, and the output is the emotion analysis results and optimized information display.

[0630] (Application example 2)

[0631] 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."

[0632] In recent years, the frequent occurrence of natural disasters has made disaster risk prediction and countermeasures increasingly important. However, current systems have difficulty accurately predicting disaster risks and are unable to provide information that reflects the user's emotions. This can lead to feelings of anxiety among users, which can delay appropriate countermeasures. The present invention aims to solve these problems and enable users to receive information with peace of mind.

[0633] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0634] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing the hazard maps and weather data, means for checking the format of the received data and performing any necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for the user to display and confirm the visualized and generated explanations, emotion recognition means for recognizing the user's emotions in real time, and means for displaying and optimizing information according to the user's emotional state based on the emotion recognition means. This allows the user to intuitively understand the highly accurate disaster risk prediction results and receive appropriate information according to their emotional state.

[0635] A "hazard map" is a map that shows the regional distribution of disaster risk, visually displaying the predicted extent of the impact and degree of danger of a disaster.

[0636] "Weather data" means data relating to past and current weather conditions, including temperature, precipitation, wind speed, wind direction, and the like.

[0637] "Data cleansing" refers to the process of filling in missing data values ​​and correcting inconsistencies and errors, thereby improving data quality.

[0638] A "machine learning model" is a collection of algorithms that are trained to make specific predictions or classifications using past data.

[0639] A "predictive model" is a mathematical model for predicting future events using machine learning models.

[0640] "Visualization" is the process of visually representing data and prediction results and presenting them in a way that is easy for users to understand.

[0641] "Means for generating explanations in natural language" refers to technology that automatically generates explanations in natural language for users based on prediction results and analysis results.

[0642] "Emotion recognition means" is a technology that analyzes and recognizes emotions from a user's facial expressions, voice, and text data.

[0643] "Means for displaying and optimizing information" refers to technology that appropriately adjusts the display method and content of information based on the user's emotional state.

[0644] This invention relates to a system that predicts disaster risks and displays the resulting information optimally. This system receives and stores hazard maps and past weather data, performs data cleansing, and builds machine learning models to predict future disaster risks. Furthermore, it uses emotion recognition to optimize the display of information according to the user's emotional state.

[0645] First, the server receives current hazard maps and historical weather data. This involves uploading user-provided data to the server, which then stores the data in a database for future access. Specifically, users upload weather data in point or grid format through a web interface. The server verifies the data format and records and saves the metadata.

[0646] The server then performs cleansing on the received data. This is the process of filling in missing values ​​and standardizing data formats. During this process, the data is converted into a format that is easier to analyze. For example, time series data is standardized into a format that allows time series analysis, and data in different units is reconciled.

[0647] The server then builds a machine learning model based on the cleansed data. This uses a training dataset of historical weather data and hazard maps. Algorithms used include random forest, LSTM, and CNN. This results in a model that can accurately predict future disaster risks.

[0648] The predictive model uses the latest weather data as input to predict future disaster risks. The prediction results are visualized using a map display library. For example, Leaflet or Mapbox can be used to color-code high-risk areas. In addition, natural language generation technology is used to automatically generate user-friendly explanations. This includes implementations using TensorFlow and Scikit-learn.

[0649] Users can check the prediction results through a smartphone app or web browser. The app uses GPS to provide detailed prediction results based on the user's current location in real time. Furthermore, the server uses emotion recognition to analyze the user's voice and facial expressions. This analysis uses the camera, microphone, and libraries such as Mediapipe and DeepFace. The content and format of the information displayed are optimized according to the user's emotional state. For example, if the user is feeling anxious, Saver will present information in a gradual and intuitive manner, providing a sense of security.

[0650] For example, when a user launches the app on a rainy day, it predicts that a particular area will be at increased risk of flooding over the next week based on past weather data and current hazard maps. When the emotion engine recognizes that the user is feeling anxious, the app will gradually display messages such as, "Caution is required for the next week. Would you like to see routes to the nearest evacuation shelter?" and provide evacuation routes and shelter information.

[0651] An example of a prompt sentence to input to the generative AI model is as follows:

[0652] "Use the past two years of rainfall data and current hazard maps to predict flood risk based on your location."

[0653] "Recognize the user's emotions from their voice and facial expressions, and optimize the information displayed to provide a sense of security."

[0654] The above is an embodiment of the present invention. This system allows users to intuitively understand the highly accurate disaster risk prediction results and to receive appropriate information according to their own emotional state.

[0655] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0656] Step 1:

[0657] Receiving and storing data

[0658] Users provide current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, when a user uploads data, the server checks its format, records the metadata, and saves it in the appropriate folder. The input is the hazard map and weather data, and the output is the uncleansed data stored in the database.

[0659] Step 2:

[0660] Data cleansing and formatting

[0661] The server checks the format of the received data and performs data cleansing. For example, it fills in missing values ​​and standardizes the format. The cleansed data is converted to make it easier to analyze. The input is the data before cleansing, and the output is the cleansed data that can be analyzed. Specific operations include unifying time series data and standardizing image data.

[0662] Step 3:

[0663] Building a machine learning model

[0664] The server builds a machine learning model based on the cleansed data. It trains a training dataset using past weather data and hazard maps to create a model that can predict future disaster risks. The input is the cleansed data, and the output is the constructed predictive model. Algorithms used include random forest, LSTM, and CNN.

[0665] Step 4:

[0666] Disaster risk prediction

[0667] The server inputs the latest weather data and uses the constructed model to predict future disaster risk. The inputs are the latest weather data and a prediction model, and the output is the predicted disaster risk value. Specifically, it executes the prediction algorithm and outputs the future risk as a numerical value.

[0668] Step 5:

[0669] Visualization of prediction results and natural language generation

[0670] To visually display the prediction results, the server uses a map display library to display the predicted disaster risk on a map in a color-coded manner. It also uses natural language generation technology to generate explanatory text that is easy for users to understand. The input is the prediction results, and the output is a visualized map and explanatory text. Specifically, it generates a map using Leaflet or Mapbox, and automatically generates explanatory text using TensorFlow or Scikit-learn.

[0671] Step 6:

[0672] Viewing and verifying results

[0673] Users can check the forecast results and their explanations via a smartphone app or web browser. The app uses GPS functionality to provide real-time disaster risk information based on the user's current location. The input is a visualized map and explanatory text, and the output is information displayed on the user's device. Specifically, the app displays the results on a user interface, allowing users to zoom in and out on the map and check detailed information.

[0674] Step 7:

[0675] Emotion recognition and information optimization

[0676] The device captures the user's voice and facial expressions and uses emotion recognition means to analyze the user's emotional state. Based on the analysis results, the server adjusts the way information is displayed and provides optimal information according to the user's emotional state. The input is the user's voice and facial expression data, and the output is optimized information display. Specifically, data is collected in real time using a camera and microphone, and analyzed using Mediapipe or DeepFace.

[0677] By executing the above processing steps, the present invention provides the user with highly accurate disaster risk prediction information and realizes optimization of appropriate information display according to the user's emotional state.

[0678] 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.

[0679] 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.

[0680] 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.

[0681] [Third embodiment]

[0682] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0683] 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.

[0684] 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).

[0685] 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.

[0686] 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.

[0687] 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).

[0688] 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. 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.

[0689] 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.

[0690] 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.

[0691] 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.

[0692] 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.

[0693] 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."

[0694] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[0695] Data acquisition and storage

[0696] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves them in the appropriate folder, and records metadata (such as the file name, upload date and time, and data type) in the database.

[0697] Data Preprocessing

[0698] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (for example, changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and converting image data into a format that allows image analysis.

[0699] Data analysis and model building

[0700] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0701] Visualization / verbalization

[0702] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[0703] Display and save results

[0704] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[0705] Through the above process, the present invention realizes a system that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides this information to users.

[0706] The processing flow will be explained below.

[0707] Step 1:

[0708] The user uploads current hazard maps and historical weather data through a web interface. During this process, the user clicks the file selection button and selects the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, the user presses the upload button to send the data to the server.

[0709] Step 2:

[0710] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[0711] Step 3:

[0712] The server cleanses the incoming data by identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[0713] Step 4:

[0714] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into JSON format and converts weather chart image files into an analyzable format (PNG or JPEG). This ensures that subsequent analysis processes proceed smoothly.

[0715] Step 5:

[0716] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[0717] Step 6:

[0718] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[0719] Step 7:

[0720] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[0721] Step 8:

[0722] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[0723] Step 9:

[0724] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[0725] Step 10:

[0726] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[0727] Step 11:

[0728] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[0729] Example 1

[0730] 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."

[0731] Systems that use current hazard maps and past weather data to predict future disaster risks must address issues such as data format, inconsistencies, missing values, etc. They must also provide users with easy-to-understand forecast results and have a means of storing them for future data improvement and verification.

[0732] 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.

[0733] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing hazard maps and weather data, means for checking the format of the received data and performing necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for users to display and confirm the visualized and generated explanations, and means for saving the disaster risk prediction results in a database and retaining them for future model improvement and verification. This makes it possible to ensure data consistency and integrity, build highly accurate prediction models, provide users with easy-to-understand information, and retain data for future improvements.

[0734] A "hazard map" is a map that visually displays information about risks such as natural disasters.

[0735] "Weather data" refers to data that includes past weather observation results such as rainfall, temperature, and wind speed.

[0736] "Means for receiving" refers to the function for receiving data from the outside and incorporating it into the system.

[0737] "Means for storing" refers to a function for saving received data.

[0738] "Data cleansing" refers to the process of filling in missing data values ​​and correcting format inconsistencies and inconsistencies.

[0739] "Easy-to-analyze format" refers to converting data into a format suitable for data analysis and model building.

[0740] A "machine learning model" refers to an algorithm or mathematical model that makes predictions and classifications based on data analysis.

[0741] "Disaster risk" is an indicator that shows the possibility of a natural disaster occurring and the extent of its impact.

[0742] "Visualization" refers to the process of visually displaying data and prediction results to make information easier to understand intuitively.

[0743] "Natural language generation" refers to a technology that outputs the prediction results of a machine learning model in a text format that humans can understand.

[0744] "Database" refers to a system that manages the storage of structured data.

[0745] MODE FOR CARRYING OUT THE INVENTION

[0746] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[0747] Data acquisition and storage

[0748] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves it in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[0749] Data Preprocessing

[0750] The server checks the format of the uploaded data and performs data cleansing. Specific examples include identifying and filling in missing values ​​in weather data and standardizing the data format (for example, changing all time notation to UTC). The server then converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and image data into a format that allows image analysis.

[0751] Data analysis and model building

[0752] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0753] Visualization / verbalization

[0754] The server creates a predicted transition of the hazard map based on the prediction results. As a specific example, the predicted transition is visualized using a map display library (e.g., Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." The results are output in a format that is easy for humans to understand by utilizing a natural language generation (NLG) library.

[0755] Display and save results

[0756] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[0757] Specific examples

[0758] 1. Example of generating a prompt:

[0759] Prompt: "Please upload weather data and hazard maps for 2020-2022."

[0760] Example: A user uploads rainfall data in CSV format, and the server checks and saves the data.

[0761] 2. Data Preprocessing Example:

[0762] Prompt: "Please cleanse the data and standardize the format."

[0763] Example: The server will impute missing data with the average value and convert all times to UTC.

[0764] 3. Model Building Example:

[0765] Prompt: "Construct a flood risk prediction model using weather data from the past 10 years."

[0766] Example: The server trains a model using the LSTM algorithm and makes future predictions.

[0767] 4. Example of result visualization:

[0768] Prompt: "Visualize your prediction results and explain them in natural language."

[0769] Execution example: The server uses the Google Maps API to display the prediction results on a map and generates the sentence, "The risk of flooding will increase in certain areas in 2025."

[0770] In this way, a system is realized that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides the results to users.

[0771] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0772] Step 1: Acquire and store data

[0773] Users upload current hazard maps and historical weather data through a web interface.

[0774] Input: Hazard map (e.g., GeoJSON, KML), historical weather data (e.g., CSV, JSON)

[0775] Specific operation: A user accesses the system's upload page in a web browser, selects a file, and clicks the upload button.

[0776] Output: The server saves the received data in a specific folder and records the metadata in a database.

[0777] Step 2: Format and cleanse the data

[0778] The server inspects the format of the uploaded data and performs any necessary data cleansing.

[0779] Input: Uploaded hazard map and weather data

[0780] What happens: The server checks the file format as CSV or JSON and performs cleansing, such as imputing missing values. For example, it converts all timestamps to UTC and fixes inconsistencies.

[0781] Output: Cleansed data

[0782] Step 3: Transform the data

[0783] The server converts the cleansed data into a format that is easy to parse.

[0784] Input: Cleansed hazard map and meteorological data

[0785] Specific operation: The server converts time series data into a format that can be used for time series analysis (e.g., Pandas DataFrame), and converts image data into a format that can be used for image analysis (e.g., NumPy array).

[0786] Output: Data in an easy-to-parse format

[0787] Step 4: Building a machine learning model

[0788] The server builds a machine learning model using past weather data and hazard maps.

[0789] Input: Rainfall data for the past 10 years and current hazard map

[0790] Specific operation: The server trains models using algorithms such as random forest, LSTM, and CNN, optimizes hyperparameters, and selects the best model.

[0791] Output: A trained machine learning model

[0792] Step 5: Predict disaster risk

[0793] The server inputs the latest weather data and uses a trained machine learning model to predict future disaster risks.

[0794] Input: Latest weather data and trained model

[0795] How it works: The server inputs the latest weather data into the model and retrieves the forecast results, scaling and transforming them as needed.

[0796] Output: Predicted disaster risk

[0797] Step 6: Visualize and verbalize the prediction results

[0798] The server visualizes the prediction results and generates explanations in natural language.

[0799] Input: Predicted disaster risk

[0800] Specific operation: The server uses a map display library (e.g., Leaflet or Google Maps API) to display the prediction results on a map, and generates explanatory text using a natural language generation (NLG) library.

[0801] Output: Visualized hazard map and natural language description

[0802] Step 7: View and save the results

[0803] The user displays and checks the prediction results on their device, and the server stores the prediction results in a database.

[0804] Input: Visualized and natural language prediction results

[0805] How it works: The user checks the prediction results in a web browser and zooms in and out on the map to see the details. The server stores the prediction results in a database and saves them for future model improvement and verification.

[0806] Output: Prediction information provided to the user and prediction results stored in a database

[0807] Through each processing step, the system uses past weather data and current hazard maps to accurately predict future disaster risks and provides the results to users.

[0808] (Application example 1)

[0809] 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."

[0810] In modern logistics centers, risk management due to natural disasters is extremely important, but systems that utilize past weather data and hazard maps are insufficient, and a system that can effectively predict future disaster risks and present specific countermeasures is needed.In addition, there is a lack of means to visualize these predictions and provide them to users in an easy-to-understand manner, which creates challenges in planning and executing logistics.

[0811] 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.

[0812] In this invention, the server includes a means for receiving the current hazard map and past weather data, a means for storing the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to accurately predict future disaster risks and propose specific countermeasures in the operation of a logistics center.

[0813] A "hazard map" is a map that shows the risk of natural disasters in a specific area.

[0814] "Weather data" refers to data on past weather conditions such as rainfall, temperature, and wind speed.

[0815] A "server" is a computer system for receiving, storing, processing, and analyzing data.

[0816] "Data cleansing" is the process of preparing data to fill in missing values ​​and standardize the format.

[0817] A "machine learning model" is an algorithm built using past data to make future predictions.

[0818] A "predictive model" is a tool for predicting future events using a constructed machine learning model.

[0819] "Visualization" is the process of visually displaying the prediction results.

[0820] "Natural language generation" is a technology that outputs prediction results as text in a form that is easy for humans to understand.

[0821] A "logistics center" is a facility that stores, ships, and delivers goods.

[0822] "Specific measures" are detailed action plans that should be implemented in response to predicted disaster risks.

[0823] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides specific countermeasures appropriate for the operation of a logistics center. Specific embodiments of this system are described below.

[0824] Data acquisition and storage

[0825] Users upload current hazard maps and historical weather data (e.g., rainfall, weather charts) to the system through a web interface. The server receives these data and stores them in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in file format (e.g., CSV or JSON). The server checks the format of these files, saves them in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[0826] Data Preprocessing

[0827] The server verifies the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (e.g., changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and image data into a format that allows image analysis.

[0828] Data analysis and model building

[0829] The server uses past weather data and hazard maps to build a machine learning model. For example, it uses rainfall data and hazard maps from the past 10 years as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[0830] Visualization / verbalization

[0831] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[0832] Display and save results

[0833] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database for future model improvement and verification.

[0834] Presentation of specific measures

[0835] Based on the prediction results, the server will suggest specific measures to be taken in accordance with the operation of the logistics center. For example, it could suggest a specific action plan to the logistics center manager, such as, "Due to the increased risk of flooding next week, we recommend that you bring forward your delivery schedule." This will enable quick and effective measures to be taken against disaster risks.

[0836] Prompt Sentence Examples

[0837] "The risk levels for the next days in your area are predicted to be high. Suggest appropriate actions to mitigate this risk."

[0838] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0839] Step 1:

[0840] Users upload current hazard maps and historical weather data through a web interface. The input is a CSV or JSON file uploaded by the user, and the server receives the data. The received data is saved in the corresponding folder, and metadata (file name, upload date and time, data type) is recorded in the database.

[0841] Step 2:

[0842] The server checks the format of the received data and performs data cleansing, i.e., imputing missing values ​​and standardizing the data format. The input is the uploaded raw data, and the output is cleansed data with missing values ​​imputed and standardized format.

[0843] Step 3:

[0844] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into a format that allows time series analysis, and image data into a format that allows image analysis. The input is the cleansed data, and the output is the data converted into an analyzable format.

[0845] Step 4:

[0846] The server uses historical weather data and hazard maps to build a machine learning model. Specifically, it trains a risk prediction model using algorithms such as random forest, LSTM, and CNN. The input is weather data and hazard map data from the past 10 years, and the output is the trained predictive model.

[0847] Step 5:

[0848] The server inputs the latest weather data into the model to predict future disaster risks. The input is the latest weather data, and the output is predicted disaster risk data.

[0849] Step 6:

[0850] The server creates a hazard map based on the prediction results. It uses a map display library (such as Leaflet or Google Maps API) to visually display areas of increased risk. The input is the predicted disaster risk data, and the output is an interactive map showing the predicted changes.

[0851] Step 7:

[0852] The server converts the prediction results into natural language and provides them to the user. Specifically, it uses a natural language generation (NLG) library to generate sentences such as "Flood risk will increase in certain areas in 2025." The input is the prediction results, and the output is a natural language explanation presented to the user.

[0853] Step 8:

[0854] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. They can zoom in and out of the map through a web browser interface to view details. The input is an interactive map and natural language explanations provided by the server, and the output is disaster risk information visually confirmed by the user.

[0855] Step 9:

[0856] Based on the prediction results, the server proposes specific measures appropriate to the operation of the logistics center. For example, it proposes a specific action plan such as, "Due to the increased risk of flooding next week, we recommend bringing forward the delivery schedule." The input is the prediction results and past response data, and the output is a specific countermeasure proposal for the logistics center.

[0857] 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.

[0858] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. Furthermore, the present invention also relates to a system that combines an emotion engine that recognizes the user's emotions and displays appropriate information to the user. A specific embodiment of this system is described below.

[0859] Data acquisition and storage

[0860] Users upload current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, users upload weather data from 2020 to 2022 and current hazard maps in file format. The server checks the data format, saves it in the corresponding folder, and records metadata in the database.

[0861] Data Preprocessing

[0862] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in meteorological data and standardizes the data format. It also converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and converting image data into an analyzable format.

[0863] Data analysis and model building

[0864] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data and hazard maps from the past 10 years are used as training data to build a disaster risk prediction model. Algorithms applied include random forest, LSTM, and CNN. Once the model is trained, it inputs the latest weather data and predicts future disaster risks.

[0865] Visualization / verbalization

[0866] The server visualizes the predicted trends on the hazard map based on the prediction results. Using a map display library, the prediction results are displayed in color on the map. The server also converts the prediction results into natural language and provides it to the user. For example, it automatically generates explanatory text such as, "Flood risk will increase in certain areas in 2025."

[0867] Display and save results

[0868] Users can view the predicted hazard map trends and generated explanations on their devices. They can also zoom in and out of the map via a web browser to view detailed prediction results. The server also stores these prediction results in a database for future analysis and model improvement.

[0869] Optimization by Emotion Engine

[0870] The system includes an emotion engine that recognizes the user's emotions. The device analyzes the user's voice, text, and facial expressions to identify the type and intensity of the emotion. For example, while the user is operating the system, the device captures the user's facial expressions and voice using a webcam and a voice input device and analyzes them in real time. The device then evaluates the user's stress level and interest through the emotion engine.

[0871] Based on emotion recognition, the server adjusts the display and explanation of the prediction results. For example, if the user is feeling anxious, the system will gradually display detailed information so that the user can easily understand. The system will gradually display disaster risks, reducing the user's stress level. In this way, the system provides interactive information according to the user's emotions.

[0872] As described above, the present invention uses past weather data and current hazard maps to predict future disaster risks with high accuracy, provides the information in a format that is easy for users to understand, and optimizes the information display according to the user's emotions.

[0873] The processing flow will be explained below.

[0874] Step 1:

[0875] Users upload current hazard maps and historical weather data through a web interface by clicking the file selection button and selecting the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, they press the upload button to send the data to the server.

[0876] Step 2:

[0877] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[0878] Step 3:

[0879] The server cleanses the incoming data, identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[0880] Step 4:

[0881] The server converts the cleansed data into a format that is easy to parse. For example, it converts time series data into JSON format and converts weather chart image files into parseable formats (e.g., PNG or JPEG). This ensures that subsequent analysis can proceed smoothly.

[0882] Step 5:

[0883] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[0884] Step 6:

[0885] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[0886] Step 7:

[0887] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[0888] Step 8:

[0889] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[0890] Step 9:

[0891] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[0892] Step 10:

[0893] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[0894] Step 11:

[0895] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[0896] Step 12:

[0897] The device analyzes voice, text, and facial expressions to recognize the user's emotions. Specifically, it uses a webcam and a voice input device to capture and analyze the user's facial expressions and voice in real time.

[0898] Step 13:

[0899] The server analyzes the user's emotional data via an emotion engine to identify the type and intensity of the emotion. For example, if the user is feeling stressed, the timing and level of detail of risk information displayed can be adjusted based on that emotional data.

[0900] Step 14:

[0901] The server optimizes the way it displays the prediction results based on the user's emotions. Specifically, it provides detailed information in stages to help users understand the information without straining them. It also displays disaster risks gradually, taking measures to reduce the user's stress level.

[0902] As described above, the present invention can predict future disaster risks with high accuracy using past weather data and current hazard maps, and can optimize the information provided to users in a way that is adaptive to the users' emotions.

[0903] Example 2

[0904] 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."

[0905] Although it is extremely important to predict future disaster risks using current hazard maps and past weather data, there is a problem that it is difficult for users to intuitively understand using only such maps. In particular, the lack of flexible information provision based on the user's emotions leads to insufficient user understanding and stress management. Therefore, there is a need for optimal information display that not only uses past data but also reflects the user's emotions.

[0906] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0907] In this invention, the server includes a means for receiving a current hazard map and past weather data, a means for saving the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to predict future disaster risks with high accuracy using past weather data and a current hazard map and provide the predictions in a format that is intuitively easy for users to understand. Furthermore, by recognizing the user's emotions and optimizing the information display accordingly, the server can deepen the user's understanding and reduce stress.

[0908] A "current hazard map" is a map showing the risk of natural disasters created using the latest geographic information systems.

[0909] "Historical weather data" refers to data relating to the past climate and weather in a particular region.

[0910] "Data cleansing" is the process of filling in missing values ​​in data and removing or correcting inappropriate data.

[0911] A "machine learning model" is a mathematical model that uses algorithms to make predictions or classifications based on input data.

[0912] "Means for generating explanations in natural language" is a process that automatically creates explanations of complex data and prediction results in terms that are easy for users to understand.

[0913] "Visualization" refers to the display of data and prediction results in a visual format such as a graph or map.

[0914] "Means for recognizing emotions and optimizing display content" refers to the process of identifying emotions from the user's voice and facial expressions and adjusting the information displayed according to those emotions.

[0915] "Means for users to display and confirm" refers to the means by which users can display and confirm data and prediction results through a web browser or application.

[0916] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. It also recognizes the user's emotions and optimizes the information display to make it easier for the user to intuitively understand. Specific embodiments are described below.

[0917] Hardware and Software Configuration

[0918] The server is the hardware that primarily performs processing, and it is desirable that it has high-performance computing capabilities. In addition, a relational database must be used as the database, and it must be capable of storing large amounts of data. The software used is "Python" for data management and analysis, "pandas" for data cleansing, and "scikit-learn" and "TensorFlow" for building machine learning models. "Leaflet.js" is used for map display, and a "natural language processing tool" is used to generate explanations in natural language.

[0919] The device used by the user is a PC or smartphone equipped with a camera and microphone for emotion recognition, which provides the interface. The software used is a web browser or dedicated application.

[0920] Data acquisition and storage

[0921] A user uploads weather data from 2020 to 2022 and the current hazard map in file format through a web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The server receives the data, checks the file format, and saves it to the database. When saving, metadata is also recorded. For example, metadata such as the file name, upload date and time, and file format are recorded.

[0922] Data Preprocessing

[0923] The server first checks the format of the data received and identifies inappropriate data or missing values. If missing values ​​are found, they are filled in with average values ​​or other data. The specific software used is "pandas." After this data cleansing, the data is converted into time series data "timeseries_weather_data.csv" for easier analysis. During this process, the data is converted into a format suitable for analysis.

[0924] Data analysis and model building

[0925] The server trains the machine learning model based on past weather data and the current hazard map. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." After training is complete, the latest weather data is input to predict future disaster risks.

[0926] Visualization / verbalization

[0927] The server visualizes the progress of the hazard map based on the prediction results. For visualization, it uses the map display library "Leaflet.js," color-coding areas with high disaster risk and displaying them on the map. It also uses natural language processing tools to generate explanations of the prediction results. For example, it automatically generates explanations such as "Flood risk will increase in certain areas in 2025."

[0928] Display and save results

[0929] Users can view the visualized hazard map and generated explanations through their device's web browser. Operations such as zooming in and out on the map are also possible. The server also stores the prediction results and generated explanations in a database for future analysis and model improvement.

[0930] Optimization by Emotion Engine

[0931] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display more detailed information and adjust the display to reduce the user's stress.

[0932] Prompt example

[0933] "Please use weather data from the past 10 years and the current hazard map to predict future disaster risks. Please also explain how to present information flexibly according to the user's emotions."

[0934] The above is a specific embodiment of the present invention.

[0935] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0936] Step 1:

[0937] The user uploads weather data from 2020 to 2022 and the current hazard map through the web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The inputs are these files, and the output is the data stored on the server.

[0938] Step 2:

[0939] The server receives the data uploaded by the user and checks the file format. If the format is invalid, it returns an error message. If the format is correct, it saves it to the database and records the metadata. Specifically, it records the file name, upload date and time, file format, etc. The input is the uploaded file and related metadata, and the output is the data and metadata as stored in the database.

[0940] Step 3:

[0941] The server rechecks the format of the received data and identifies improper data or missing values. Specifically, it checks whether there are missing values ​​in the weather data. The input is the weather data stored in the database, and the output is the complete data after missing values ​​are imputed.

[0942] Step 4:

[0943] The server completes and corrects missing values ​​and improper data. Specifically, it uses "pandas" to complete missing values ​​by the mean and reshape the data. The input is data with missing values, and the output is cleansed data after missing value completion.

[0944] Step 5:

[0945] The server converts the cleansed data into a format that is easy to analyze. For example, it converts weather data into time series data "timeseries_weather_data.csv" and hazard maps into "preprocessed_hazard_map.json." The input is cleansed data, and the output is data in an easy-to-analyze format.

[0946] Step 6:

[0947] The server trains a machine learning model using past weather data and current hazard maps. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." The input is past weather data and hazard maps, and the output is a trained machine learning model.

[0948] Step 7:

[0949] The server inputs the latest weather data to predict future disaster risks. Specifically, it feeds new data into the trained model and obtains prediction results. The input is the latest weather data, and the output is a prediction of future disaster risks.

[0950] Step 8:

[0951] The server visualizes the prediction results using the map display library "Leaflet.js." The visualized map displays areas with high disaster risk in different colors. Specifically, high-risk areas are highlighted in red. The input is the disaster risk prediction results, and the output is a color-coded visualized map.

[0952] Step 9:

[0953] The server generates explanatory text for the prediction results using natural language processing tools. For example, it generates an explanatory text such as "Flood risk will increase in certain areas in 2025." The input is the disaster risk prediction results, and the output is an explanatory text generated in natural language.

[0954] Step 10:

[0955] The user can view the visualized hazard map and generated explanation through a web browser on their device. Operations such as zooming in and out on the map are also possible. The input is the visualized map and explanation, and the output is an interface that the user can view and operate.

[0956] Step 11:

[0957] The server stores the prediction results and generated explanations in a database. The stored data is used for future analysis and model improvement. The input is the prediction results and explanations, and the output is the stored data.

[0958] Step 12:

[0959] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display detailed information. The input is voice and facial expression data, and the output is the emotion analysis results and optimized information display.

[0960] (Application example 2)

[0961] 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."

[0962] In recent years, the frequent occurrence of natural disasters has made disaster risk prediction and countermeasures increasingly important. However, current systems have difficulty accurately predicting disaster risks and are unable to provide information that reflects the user's emotions. This can lead to feelings of anxiety among users, which can delay appropriate countermeasures. The present invention aims to solve these problems and enable users to receive information with peace of mind.

[0963] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0964] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing the hazard maps and weather data, means for checking the format of the received data and performing any necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for the user to display and confirm the visualized and generated explanations, emotion recognition means for recognizing the user's emotions in real time, and means for displaying and optimizing information according to the user's emotional state based on the emotion recognition means. This allows the user to intuitively understand the highly accurate disaster risk prediction results and receive appropriate information according to their emotional state.

[0965] A "hazard map" is a map that shows the regional distribution of disaster risk, visually displaying the predicted extent of the impact and degree of danger of a disaster.

[0966] "Weather data" means data relating to past and current weather conditions, including temperature, precipitation, wind speed, wind direction, and the like.

[0967] "Data cleansing" refers to the process of filling in missing data values ​​and correcting inconsistencies and errors, thereby improving data quality.

[0968] A "machine learning model" is a collection of algorithms that are trained to make specific predictions or classifications using past data.

[0969] A "predictive model" is a mathematical model for predicting future events using machine learning models.

[0970] "Visualization" is the process of visually representing data and prediction results and presenting them in a way that is easy for users to understand.

[0971] "Means for generating explanations in natural language" refers to technology that automatically generates explanations in natural language for users based on prediction results and analysis results.

[0972] "Emotion recognition means" is a technology that analyzes and recognizes emotions from a user's facial expressions, voice, and text data.

[0973] "Means for displaying and optimizing information" refers to technology that appropriately adjusts the display method and content of information based on the user's emotional state.

[0974] This invention relates to a system that predicts disaster risks and displays the resulting information optimally. This system receives and stores hazard maps and past weather data, performs data cleansing, and builds machine learning models to predict future disaster risks. Furthermore, it uses emotion recognition to optimize the display of information according to the user's emotional state.

[0975] First, the server receives current hazard maps and historical weather data. This involves uploading user-provided data to the server, which then stores the data in a database for future access. Specifically, users upload weather data in point or grid format through a web interface. The server verifies the data format and records and saves the metadata.

[0976] The server then performs cleansing on the received data. This is the process of filling in missing values ​​and standardizing data formats. During this process, the data is converted into a format that is easier to analyze. For example, time series data is standardized into a format that allows time series analysis, and data in different units is reconciled.

[0977] The server then builds a machine learning model based on the cleansed data. This uses a training dataset of historical weather data and hazard maps. Algorithms used include random forest, LSTM, and CNN. This results in a model that can accurately predict future disaster risks.

[0978] The predictive model uses the latest weather data as input to predict future disaster risks. The prediction results are visualized using a map display library. For example, Leaflet or Mapbox can be used to color-code high-risk areas. In addition, natural language generation technology is used to automatically generate user-friendly explanations. This includes implementations using TensorFlow and Scikit-learn.

[0979] Users can check the prediction results through a smartphone app or web browser. The app uses GPS to provide detailed prediction results based on the user's current location in real time. Furthermore, the server uses emotion recognition to analyze the user's voice and facial expressions. This analysis uses the camera, microphone, and libraries such as Mediapipe and DeepFace. The content and format of the information displayed are optimized according to the user's emotional state. For example, if the user is feeling anxious, Saver will present information in a gradual and intuitive manner, providing a sense of security.

[0980] For example, when a user launches the app on a rainy day, it predicts that a particular area will be at increased risk of flooding over the next week based on past weather data and current hazard maps. When the emotion engine recognizes that the user is feeling anxious, the app will gradually display messages such as, "Caution is required for the next week. Would you like to see routes to the nearest evacuation shelter?" and provide evacuation routes and shelter information.

[0981] An example of a prompt sentence to input to the generative AI model is as follows:

[0982] "Use the past two years of rainfall data and current hazard maps to predict flood risk based on your location."

[0983] "Recognize the user's emotions from their voice and facial expressions, and optimize the information displayed to provide a sense of security."

[0984] The above is an embodiment of the present invention. This system allows users to intuitively understand the highly accurate disaster risk prediction results and to receive appropriate information according to their own emotional state.

[0985] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0986] Step 1:

[0987] Receiving and storing data

[0988] Users provide current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, when a user uploads data, the server checks its format, records the metadata, and saves it in the appropriate folder. The input is the hazard map and weather data, and the output is the uncleansed data stored in the database.

[0989] Step 2:

[0990] Data cleansing and formatting

[0991] The server checks the format of the received data and performs data cleansing. For example, it fills in missing values ​​and standardizes the format. The cleansed data is converted to make it easier to analyze. The input is the data before cleansing, and the output is the cleansed data that can be analyzed. Specific operations include unifying time series data and standardizing image data.

[0992] Step 3:

[0993] Building a machine learning model

[0994] The server builds a machine learning model based on the cleansed data. It trains a training dataset using past weather data and hazard maps to create a model that can predict future disaster risks. The input is the cleansed data, and the output is the constructed predictive model. Algorithms used include random forest, LSTM, and CNN.

[0995] Step 4:

[0996] Disaster risk prediction

[0997] The server inputs the latest weather data and uses the constructed model to predict future disaster risk. The inputs are the latest weather data and a prediction model, and the output is the predicted disaster risk value. Specifically, it executes the prediction algorithm and outputs the future risk as a numerical value.

[0998] Step 5:

[0999] Visualization of prediction results and natural language generation

[1000] To visually display the prediction results, the server uses a map display library to display the predicted disaster risk on a map in a color-coded manner. It also uses natural language generation technology to generate explanatory text that is easy for users to understand. The input is the prediction results, and the output is a visualized map and explanatory text. Specifically, it generates a map using Leaflet or Mapbox, and automatically generates explanatory text using TensorFlow or Scikit-learn.

[1001] Step 6:

[1002] Viewing and verifying results

[1003] Users can check the forecast results and their explanations via a smartphone app or web browser. The app uses GPS functionality to provide real-time disaster risk information based on the user's current location. The input is a visualized map and explanatory text, and the output is information displayed on the user's device. Specifically, the app displays the results on a user interface, allowing users to zoom in and out on the map and check detailed information.

[1004] Step 7:

[1005] Emotion recognition and information optimization

[1006] The device captures the user's voice and facial expressions and uses emotion recognition means to analyze the user's emotional state. Based on the analysis results, the server adjusts the way information is displayed and provides optimal information according to the user's emotional state. The input is the user's voice and facial expression data, and the output is optimized information display. Specifically, data is collected in real time using a camera and microphone, and analyzed using Mediapipe or DeepFace.

[1007] By executing the above processing steps, the present invention provides the user with highly accurate disaster risk prediction information and realizes optimization of appropriate information display according to the user's emotional state.

[1008] 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.

[1009] 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.

[1010] 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.

[1011] [Fourth embodiment]

[1012] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1013] 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.

[1014] 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).

[1015] 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.

[1016] 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.

[1017] 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).

[1018] 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. 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.

[1019] 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.

[1020] 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.

[1021] 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.

[1022] 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.

[1023] 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.

[1024] 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."

[1025] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[1026] Data acquisition and storage

[1027] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves them in the appropriate folder, and records metadata (such as the file name, upload date and time, and data type) in the database.

[1028] Data Preprocessing

[1029] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (for example, changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and converting image data into a format that allows image analysis.

[1030] Data analysis and model building

[1031] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[1032] Visualization / verbalization

[1033] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[1034] Display and save results

[1035] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[1036] Through the above process, the present invention realizes a system that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides this information to users.

[1037] The processing flow will be explained below.

[1038] Step 1:

[1039] The user uploads current hazard maps and historical weather data through a web interface. During this process, the user clicks the file selection button and selects the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, the user presses the upload button to send the data to the server.

[1040] Step 2:

[1041] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[1042] Step 3:

[1043] The server cleanses the incoming data by identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[1044] Step 4:

[1045] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into JSON format and converts weather chart image files into an analyzable format (PNG or JPEG). This ensures that subsequent analysis processes proceed smoothly.

[1046] Step 5:

[1047] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[1048] Step 6:

[1049] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[1050] Step 7:

[1051] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[1052] Step 8:

[1053] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[1054] Step 9:

[1055] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[1056] Step 10:

[1057] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[1058] Step 11:

[1059] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[1060] Example 1

[1061] 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."

[1062] Systems that use current hazard maps and past weather data to predict future disaster risks must address issues such as data format, inconsistencies, missing values, etc. They must also provide users with easy-to-understand forecast results and have a means of storing them for future data improvement and verification.

[1063] 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.

[1064] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing hazard maps and weather data, means for checking the format of the received data and performing necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for users to display and confirm the visualized and generated explanations, and means for saving the disaster risk prediction results in a database and retaining them for future model improvement and verification. This makes it possible to ensure data consistency and integrity, build highly accurate prediction models, provide users with easy-to-understand information, and retain data for future improvements.

[1065] A "hazard map" is a map that visually displays information about risks such as natural disasters.

[1066] "Weather data" refers to data that includes past weather observation results such as rainfall, temperature, and wind speed.

[1067] "Means for receiving" refers to the function for receiving data from the outside and incorporating it into the system.

[1068] "Means for storing" refers to a function for saving received data.

[1069] "Data cleansing" refers to the process of filling in missing data values ​​and correcting format inconsistencies and inconsistencies.

[1070] "Easy-to-analyze format" refers to converting data into a format suitable for data analysis and model building.

[1071] A "machine learning model" refers to an algorithm or mathematical model that makes predictions and classifications based on data analysis.

[1072] "Disaster risk" is an indicator that shows the possibility of a natural disaster occurring and the extent of its impact.

[1073] "Visualization" refers to the process of visually displaying data and prediction results to make information easier to understand intuitively.

[1074] "Natural language generation" refers to a technology that outputs the prediction results of a machine learning model in a text format that humans can understand.

[1075] "Database" refers to a system that manages the storage of structured data.

[1076] MODE FOR CARRYING OUT THE INVENTION

[1077] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data, and provides the results in an easy-to-understand format to users. Specific embodiments of this system are described below.

[1078] Data acquisition and storage

[1079] A user uploads current hazard maps and historical weather data (rainfall, weather charts, etc.) to the system through a web interface. The server receives this data and stores it in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in a file format (e.g., CSV or JSON). The server checks the file format, saves it in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[1080] Data Preprocessing

[1081] The server checks the format of the uploaded data and performs data cleansing. Specific examples include identifying and filling in missing values ​​in weather data and standardizing the data format (for example, changing all time notation to UTC). The server then converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and image data into a format that allows image analysis.

[1082] Data analysis and model building

[1083] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data from the past 10 years and hazard maps are used as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[1084] Visualization / verbalization

[1085] The server creates a predicted transition of the hazard map based on the prediction results. As a specific example, the predicted transition is visualized using a map display library (e.g., Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." The results are output in a format that is easy for humans to understand by utilizing a natural language generation (NLG) library.

[1086] Display and save results

[1087] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database, where they can be used for future model improvements and verification.

[1088] Specific examples

[1089] 1. Example of generating a prompt:

[1090] Prompt: "Please upload weather data and hazard maps for 2020-2022."

[1091] Example: A user uploads rainfall data in CSV format, and the server checks and saves the data.

[1092] 2. Data Preprocessing Example:

[1093] Prompt: "Please cleanse the data and standardize the format."

[1094] Example: The server will impute missing data with the average value and convert all times to UTC.

[1095] 3. Model Building Example:

[1096] Prompt: "Construct a flood risk prediction model using weather data from the past 10 years."

[1097] Example: The server trains a model using the LSTM algorithm and makes future predictions.

[1098] 4. Example of result visualization:

[1099] Prompt: "Visualize your prediction results and explain them in natural language."

[1100] Execution example: The server uses the Google Maps API to display the prediction results on a map and generates the sentence, "The risk of flooding will increase in certain areas in 2025."

[1101] In this way, a system is realized that uses past weather data and current hazard maps to predict future disaster risks with high accuracy and provides the results to users.

[1102] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1103] Step 1: Acquire and store data

[1104] Users upload current hazard maps and historical weather data through a web interface.

[1105] Input: Hazard map (e.g., GeoJSON, KML), historical weather data (e.g., CSV, JSON)

[1106] Specific operation: A user accesses the system's upload page in a web browser, selects a file, and clicks the upload button.

[1107] Output: The server saves the received data in a specific folder and records the metadata in a database.

[1108] Step 2: Format and cleanse the data

[1109] The server inspects the format of the uploaded data and performs any necessary data cleansing.

[1110] Input: Uploaded hazard map and weather data

[1111] What happens: The server checks the file format as CSV or JSON and performs cleansing, such as imputing missing values. For example, it converts all timestamps to UTC and fixes inconsistencies.

[1112] Output: Cleansed data

[1113] Step 3: Transform the data

[1114] The server converts the cleansed data into a format that is easy to parse.

[1115] Input: Cleansed hazard map and meteorological data

[1116] Specific operation: The server converts time series data into a format that can be used for time series analysis (e.g., Pandas DataFrame), and converts image data into a format that can be used for image analysis (e.g., NumPy array).

[1117] Output: Data in an easy-to-parse format

[1118] Step 4: Building a machine learning model

[1119] The server builds a machine learning model using past weather data and hazard maps.

[1120] Input: Rainfall data for the past 10 years and current hazard map

[1121] Specific operation: The server trains models using algorithms such as random forest, LSTM, and CNN, optimizes hyperparameters, and selects the best model.

[1122] Output: A trained machine learning model

[1123] Step 5: Predict disaster risk

[1124] The server inputs the latest weather data and uses a trained machine learning model to predict future disaster risks.

[1125] Input: Latest weather data and trained model

[1126] How it works: The server inputs the latest weather data into the model and retrieves the forecast results, scaling and transforming them as needed.

[1127] Output: Predicted disaster risk

[1128] Step 6: Visualize and verbalize the prediction results

[1129] The server visualizes the prediction results and generates explanations in natural language.

[1130] Input: Predicted disaster risk

[1131] Specific operation: The server uses a map display library (e.g., Leaflet or Google Maps API) to display the prediction results on a map, and generates explanatory text using a natural language generation (NLG) library.

[1132] Output: Visualized hazard map and natural language description

[1133] Step 7: View and save the results

[1134] The user displays and checks the prediction results on their device, and the server stores the prediction results in a database.

[1135] Input: Visualized and natural language prediction results

[1136] How it works: The user checks the prediction results in a web browser and zooms in and out on the map to see the details. The server stores the prediction results in a database and saves them for future model improvement and verification.

[1137] Output: Prediction information provided to the user and prediction results stored in a database

[1138] Through each processing step, the system uses past weather data and current hazard maps to accurately predict future disaster risks and provides the results to users.

[1139] (Application example 1)

[1140] 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."

[1141] In modern logistics centers, risk management due to natural disasters is extremely important, but systems that utilize past weather data and hazard maps are insufficient, and a system that can effectively predict future disaster risks and present specific countermeasures is needed.In addition, there is a lack of means to visualize these predictions and provide them to users in an easy-to-understand manner, which creates challenges in planning and executing logistics.

[1142] 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.

[1143] In this invention, the server includes a means for receiving the current hazard map and past weather data, a means for storing the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to accurately predict future disaster risks and propose specific countermeasures in the operation of a logistics center.

[1144] A "hazard map" is a map that shows the risk of natural disasters in a specific area.

[1145] "Weather data" refers to data on past weather conditions such as rainfall, temperature, and wind speed.

[1146] A "server" is a computer system for receiving, storing, processing, and analyzing data.

[1147] "Data cleansing" is the process of preparing data to fill in missing values ​​and standardize the format.

[1148] A "machine learning model" is an algorithm built using past data to make future predictions.

[1149] A "predictive model" is a tool for predicting future events using a constructed machine learning model.

[1150] "Visualization" is the process of visually displaying the prediction results.

[1151] "Natural language generation" is a technology that outputs prediction results as text in a form that is easy for humans to understand.

[1152] A "logistics center" is a facility that stores, ships, and delivers goods.

[1153] "Specific measures" are detailed action plans that should be implemented in response to predicted disaster risks.

[1154] The present invention is directed to a system that predicts future disaster risks based on current hazard maps and past weather data, and provides specific countermeasures appropriate for the operation of a logistics center. Specific embodiments of this system are described below.

[1155] Data acquisition and storage

[1156] Users upload current hazard maps and historical weather data (e.g., rainfall, weather charts) to the system through a web interface. The server receives these data and stores them in a database. For example, a user uploads weather data from 2020 to 2022 and the current hazard map in file format (e.g., CSV or JSON). The server checks the format of these files, saves them in the corresponding folder, and records metadata (e.g., file name, upload date and time, data type) in the database.

[1157] Data Preprocessing

[1158] The server verifies the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in weather data and standardizes the data format (e.g., changing all times to UTC). The server then converts the cleansed data into a format that is easy to parse. This includes converting time series data into a format that allows time series analysis, and image data into a format that allows image analysis.

[1159] Data analysis and model building

[1160] The server uses past weather data and hazard maps to build a machine learning model. For example, it uses rainfall data and hazard maps from the past 10 years as training data to build a flood risk prediction model. Specific algorithms used include random forests, LSTM (long short-term memory networks), and CNN (convolutional neural networks). Once the model training is complete, the server provides the latest weather data as input to the model to predict future disaster risks.

[1161] Visualization / verbalization

[1162] The server creates a predicted transition of the hazard map based on the prediction results. For example, it visualizes the predicted transition using a map display library (Leaflet or Google Maps API). The server then converts the prediction results into natural language and provides it to the user. For example, it automatically generates a sentence such as "Flood risk will increase in certain areas in 2025." It uses a natural language generation (NLG) library to output the results in a format that is easy for humans to understand.

[1163] Display and save results

[1164] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. Specifically, users can zoom in and out on the map using a web browser interface to view detailed prediction results. The server also stores these prediction results in a database for future model improvement and verification.

[1165] Presentation of specific measures

[1166] Based on the prediction results, the server will suggest specific measures to be taken in accordance with the operation of the logistics center. For example, it could suggest a specific action plan to the logistics center manager, such as, "Due to the increased risk of flooding next week, we recommend that you bring forward your delivery schedule." This will enable quick and effective measures to be taken against disaster risks.

[1167] Prompt Sentence Examples

[1168] "The risk levels for the next days in your area are predicted to be high. Suggest appropriate actions to mitigate this risk."

[1169] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1170] Step 1:

[1171] Users upload current hazard maps and historical weather data through a web interface. The input is a CSV or JSON file uploaded by the user, and the server receives the data. The received data is saved in the corresponding folder, and metadata (file name, upload date and time, data type) is recorded in the database.

[1172] Step 2:

[1173] The server checks the format of the received data and performs data cleansing, i.e., imputing missing values ​​and standardizing the data format. The input is the uploaded raw data, and the output is cleansed data with missing values ​​imputed and standardized format.

[1174] Step 3:

[1175] The server converts the cleansed data into a format that is easy to analyze. For example, it converts time series data into a format that allows time series analysis, and image data into a format that allows image analysis. The input is the cleansed data, and the output is the data converted into an analyzable format.

[1176] Step 4:

[1177] The server uses historical weather data and hazard maps to build a machine learning model. Specifically, it trains a risk prediction model using algorithms such as random forest, LSTM, and CNN. The input is weather data and hazard map data from the past 10 years, and the output is the trained predictive model.

[1178] Step 5:

[1179] The server inputs the latest weather data into the model to predict future disaster risks. The input is the latest weather data, and the output is predicted disaster risk data.

[1180] Step 6:

[1181] The server creates a hazard map based on the prediction results. It uses a map display library (such as Leaflet or Google Maps API) to visually display areas of increased risk. The input is the predicted disaster risk data, and the output is an interactive map showing the predicted changes.

[1182] Step 7:

[1183] The server converts the prediction results into natural language and provides them to the user. Specifically, it uses a natural language generation (NLG) library to generate sentences such as "Flood risk will increase in certain areas in 2025." The input is the prediction results, and the output is a natural language explanation presented to the user.

[1184] Step 8:

[1185] Users can view the predicted hazard map trends and explanations of the prediction results on their devices. They can zoom in and out of the map through a web browser interface to view details. The input is an interactive map and natural language explanations provided by the server, and the output is disaster risk information visually confirmed by the user.

[1186] Step 9:

[1187] Based on the prediction results, the server proposes specific measures appropriate to the operation of the logistics center. For example, it proposes a specific action plan such as, "Due to the increased risk of flooding next week, we recommend bringing forward the delivery schedule." The input is the prediction results and past response data, and the output is a specific countermeasure proposal for the logistics center.

[1188] 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.

[1189] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. Furthermore, the present invention also relates to a system that combines an emotion engine that recognizes the user's emotions and displays appropriate information to the user. A specific embodiment of this system is described below.

[1190] Data acquisition and storage

[1191] Users upload current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, users upload weather data from 2020 to 2022 and current hazard maps in file format. The server checks the data format, saves it in the corresponding folder, and records metadata in the database.

[1192] Data Preprocessing

[1193] The server checks the format of the uploaded data and performs data cleansing. For example, it identifies and fills in missing values ​​in meteorological data and standardizes the data format. It also converts the cleansed data into a format that is easy to analyze. This includes converting time series data into a format that allows time series analysis and converting image data into an analyzable format.

[1194] Data analysis and model building

[1195] The server uses past weather data and hazard maps to build a machine learning model. For example, rainfall data and hazard maps from the past 10 years are used as training data to build a disaster risk prediction model. Algorithms applied include random forest, LSTM, and CNN. Once the model is trained, it inputs the latest weather data and predicts future disaster risks.

[1196] Visualization / verbalization

[1197] The server visualizes the predicted trends on the hazard map based on the prediction results. Using a map display library, the prediction results are displayed in color on the map. The server also converts the prediction results into natural language and provides it to the user. For example, it automatically generates explanatory text such as, "Flood risk will increase in certain areas in 2025."

[1198] Display and save results

[1199] Users can view the predicted hazard map trends and generated explanations on their devices. They can also zoom in and out of the map via a web browser to view detailed prediction results. The server also stores these prediction results in a database for future analysis and model improvement.

[1200] Optimization by Emotion Engine

[1201] The system includes an emotion engine that recognizes the user's emotions. The device analyzes the user's voice, text, and facial expressions to identify the type and intensity of the emotion. For example, while the user is operating the system, the device captures the user's facial expressions and voice using a webcam and a voice input device and analyzes them in real time. The device then evaluates the user's stress level and interest through the emotion engine.

[1202] Based on emotion recognition, the server adjusts the display and explanation of the prediction results. For example, if the user is feeling anxious, the system will gradually display detailed information so that the user can easily understand. The system will gradually display disaster risks, reducing the user's stress level. In this way, the system provides interactive information according to the user's emotions.

[1203] As described above, the present invention uses past weather data and current hazard maps to predict future disaster risks with high accuracy, provides the information in a format that is easy for users to understand, and optimizes the information display according to the user's emotions.

[1204] The processing flow will be explained below.

[1205] Step 1:

[1206] Users upload current hazard maps and historical weather data through a web interface by clicking the file selection button and selecting the required data files (e.g., precipitation data in CSV format, weather map image files) from their local computer. After selecting the files, they press the upload button to send the data to the server.

[1207] Step 2:

[1208] The server receives the uploaded data, checks the file format, saves it in the appropriate folder, and records the metadata of each file (file name, upload date and time, data type, etc.) in the database. For example, it checks that weather data is in CSV format and stores it in the appropriate folder.

[1209] Step 3:

[1210] The server cleanses the incoming data, identifying and filling in missing values ​​in the weather data, checking the accuracy of the data range, and standardizing the data units (for example, standardizing rainfall units to millimeters).

[1211] Step 4:

[1212] The server converts the cleansed data into a format that is easy to parse. For example, it converts time series data into JSON format and converts weather chart image files into parseable formats (e.g., PNG or JPEG). This ensures that subsequent analysis can proceed smoothly.

[1213] Step 5:

[1214] In order for the server to build a machine learning model, it prepares past weather data and hazard maps as a training dataset. Specifically, it divides the data into a training dataset and a test dataset. For example, 70% of the data is used for training and 30% is used for testing.

[1215] Step 6:

[1216] The server selects the machine learning algorithm to apply and trains the model, for example, using a random forest algorithm or a long short-term memory network (LSTM). This model is then used to predict future disaster risks based on past data.

[1217] Step 7:

[1218] The server inputs the latest meteorological data into a pre-built machine learning model to predict future disaster risks. For example, it inputs rainfall forecasts for the following year into the model to identify areas at high risk of flooding.

[1219] Step 8:

[1220] The server visualizes the predicted transition of the hazard map based on the prediction results. For this, a map display library (e.g., Leaflet or Google Maps API) is used. The applied prediction results are displayed in color on the map, allowing users to understand them intuitively.

[1221] Step 9:

[1222] The server converts the forecast results into natural language and documents them. For example, it automatically generates explanatory text such as, "The risk of flooding will increase in a specific area in 2025." This process utilizes a natural language generation (NLG) library.

[1223] Step 10:

[1224] The user checks the predicted hazard map trends and generated explanations of the prediction results on the device. The user can use the functions to zoom in and out of the map and view details through a web browser.

[1225] Step 11:

[1226] The server stores the prediction results and their metadata in a database for future validation and model improvement. The stored data can then be used for further analysis and comparative studies.

[1227] Step 12:

[1228] The device analyzes voice, text, and facial expressions to recognize the user's emotions. Specifically, it uses a webcam and a voice input device to capture and analyze the user's facial expressions and voice in real time.

[1229] Step 13:

[1230] The server analyzes the user's emotional data via an emotion engine to identify the type and intensity of the emotion. For example, if the user is feeling stressed, the timing and level of detail of risk information displayed can be adjusted based on that emotional data.

[1231] Step 14:

[1232] The server optimizes the way it displays the prediction results based on the user's emotions. Specifically, it provides detailed information in stages to help users understand the information without straining them. It also displays disaster risks gradually, taking measures to reduce the user's stress level.

[1233] As described above, the present invention can predict future disaster risks with high accuracy using past weather data and current hazard maps, and can optimize the information provided to users in a way that is adaptive to the users' emotions.

[1234] Example 2

[1235] 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."

[1236] Although it is extremely important to predict future disaster risks using current hazard maps and past weather data, there is a problem that it is difficult for users to intuitively understand using only such maps. In particular, the lack of flexible information provision based on the user's emotions leads to insufficient user understanding and stress management. Therefore, there is a need for optimal information display that not only uses past data but also reflects the user's emotions.

[1237] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1238] In this invention, the server includes a means for receiving a current hazard map and past weather data, a means for saving the hazard map and weather data, and a means for checking the format of the received data and performing necessary data cleansing. This makes it possible to predict future disaster risks with high accuracy using past weather data and a current hazard map and provide the predictions in a format that is intuitively easy for users to understand. Furthermore, by recognizing the user's emotions and optimizing the information display accordingly, the server can deepen the user's understanding and reduce stress.

[1239] A "current hazard map" is a map showing the risk of natural disasters created using the latest geographic information systems.

[1240] "Historical weather data" refers to data relating to the past climate and weather in a particular region.

[1241] "Data cleansing" is the process of filling in missing values ​​in data and removing or correcting inappropriate data.

[1242] A "machine learning model" is a mathematical model that uses algorithms to make predictions or classifications based on input data.

[1243] "Means for generating explanations in natural language" is a process that automatically creates explanations of complex data and prediction results in terms that are easy for users to understand.

[1244] "Visualization" refers to the display of data and prediction results in a visual format such as a graph or map.

[1245] "Means for recognizing emotions and optimizing display content" refers to the process of identifying emotions from the user's voice and facial expressions and adjusting the information displayed according to those emotions.

[1246] "Means for users to display and confirm" refers to the means by which users can display and confirm data and prediction results through a web browser or application.

[1247] The present invention relates to a system that predicts future disaster risks using current hazard maps and past weather data. It also recognizes the user's emotions and optimizes the information display to make it easier for the user to intuitively understand. Specific embodiments are described below.

[1248] Hardware and Software Configuration

[1249] The server is the hardware that primarily performs processing, and it is desirable that it has high-performance computing capabilities. In addition, a relational database must be used as the database, and it must be capable of storing large amounts of data. The software used is "Python" for data management and analysis, "pandas" for data cleansing, and "scikit-learn" and "TensorFlow" for building machine learning models. "Leaflet.js" is used for map display, and a "natural language processing tool" is used to generate explanations in natural language.

[1250] The device used by the user is a PC or smartphone equipped with a camera and microphone for emotion recognition, which provides the interface. The software used is a web browser or dedicated application.

[1251] Data acquisition and storage

[1252] A user uploads weather data from 2020 to 2022 and the current hazard map in file format through a web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The server receives the data, checks the file format, and saves it to the database. When saving, metadata is also recorded. For example, metadata such as the file name, upload date and time, and file format are recorded.

[1253] Data Preprocessing

[1254] The server first checks the format of the data received and identifies inappropriate data or missing values. If missing values ​​are found, they are filled in with average values ​​or other data. The specific software used is "pandas." After this data cleansing, the data is converted into time series data "timeseries_weather_data.csv" for easier analysis. During this process, the data is converted into a format suitable for analysis.

[1255] Data analysis and model building

[1256] The server trains the machine learning model based on past weather data and the current hazard map. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." After training is complete, the latest weather data is input to predict future disaster risks.

[1257] Visualization / verbalization

[1258] The server visualizes the progress of the hazard map based on the prediction results. For visualization, it uses the map display library "Leaflet.js," color-coding areas with high disaster risk and displaying them on the map. It also uses natural language processing tools to generate explanations of the prediction results. For example, it automatically generates explanations such as "Flood risk will increase in certain areas in 2025."

[1259] Display and save results

[1260] Users can view the visualized hazard map and generated explanations through their device's web browser. Operations such as zooming in and out on the map are also possible. The server also stores the prediction results and generated explanations in a database for future analysis and model improvement.

[1261] Optimization by Emotion Engine

[1262] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display more detailed information and adjust the display to reduce the user's stress.

[1263] Prompt example

[1264] "Please use weather data from the past 10 years and the current hazard map to predict future disaster risks. Please also explain how to present information flexibly according to the user's emotions."

[1265] The above is a specific embodiment of the present invention.

[1266] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1267] Step 1:

[1268] The user uploads weather data from 2020 to 2022 and the current hazard map through the web interface. For example, "2020-2022_weather_data.csv" and "current_hazard_map.png" are uploaded. The inputs are these files, and the output is the data stored on the server.

[1269] Step 2:

[1270] The server receives the data uploaded by the user and checks the file format. If the format is invalid, it returns an error message. If the format is correct, it saves it to the database and records the metadata. Specifically, it records the file name, upload date and time, file format, etc. The input is the uploaded file and related metadata, and the output is the data and metadata as stored in the database.

[1271] Step 3:

[1272] The server rechecks the format of the received data and identifies improper data or missing values. Specifically, it checks whether there are missing values ​​in the weather data. The input is the weather data stored in the database, and the output is the complete data after missing values ​​are imputed.

[1273] Step 4:

[1274] The server completes and corrects missing values ​​and improper data. Specifically, it uses "pandas" to complete missing values ​​by the mean and reshape the data. The input is data with missing values, and the output is cleansed data after missing value completion.

[1275] Step 5:

[1276] The server converts the cleansed data into a format that is easy to analyze. For example, it converts weather data into time series data "timeseries_weather_data.csv" and hazard maps into "preprocessed_hazard_map.json." The input is cleansed data, and the output is data in an easy-to-analyze format.

[1277] Step 6:

[1278] The server trains a machine learning model using past weather data and current hazard maps. As a specific example, 10 years of rainfall data "rainfall_data_2010_2020.csv" and the current hazard map are used. Algorithms used include "random forest," "LSTM," and "CNN." The input is past weather data and hazard maps, and the output is a trained machine learning model.

[1279] Step 7:

[1280] The server inputs the latest weather data to predict future disaster risks. Specifically, it feeds new data into the trained model and obtains prediction results. The input is the latest weather data, and the output is a prediction of future disaster risks.

[1281] Step 8:

[1282] The server visualizes the prediction results using the map display library "Leaflet.js." The visualized map displays areas with high disaster risk in different colors. Specifically, high-risk areas are highlighted in red. The input is the disaster risk prediction results, and the output is a color-coded visualized map.

[1283] Step 9:

[1284] The server generates explanatory text for the prediction results using natural language processing tools. For example, it generates an explanatory text such as "Flood risk will increase in certain areas in 2025." The input is the disaster risk prediction results, and the output is an explanatory text generated in natural language.

[1285] Step 10:

[1286] The user can view the visualized hazard map and generated explanation through a web browser on their device. Operations such as zooming in and out on the map are also possible. The input is the visualized map and explanation, and the output is an interface that the user can view and operate.

[1287] Step 11:

[1288] The server stores the prediction results and generated explanations in a database. The stored data is used for future analysis and model improvement. The input is the prediction results and explanations, and the output is the stored data.

[1289] Step 12:

[1290] The device captures the user's voice and facial expressions in real time and analyzes their emotions using an emotion engine. Based on the analyzed emotional information, the server optimizes the display content and explanations. For example, if it determines that the user is feeling anxious, it will gradually display detailed information. The input is voice and facial expression data, and the output is the emotion analysis results and optimized information display.

[1291] (Application example 2)

[1292] 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."

[1293] In recent years, the frequent occurrence of natural disasters has made disaster risk prediction and countermeasures increasingly important. However, current systems have difficulty accurately predicting disaster risks and are unable to provide information that reflects the user's emotions. This can lead to feelings of anxiety among users, which can delay appropriate countermeasures. The present invention aims to solve these problems and enable users to receive information with peace of mind.

[1294] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1295] In this invention, the server includes means for receiving current hazard maps and past weather data, means for storing the hazard maps and weather data, means for checking the format of the received data and performing any necessary data cleansing, means for converting the cleansed data into an easily analyzable format, means for building a machine learning model and predicting future disaster risks using the prediction model, means for visualizing the prediction results and generating explanations in natural language, means for the user to display and confirm the visualized and generated explanations, emotion recognition means for recognizing the user's emotions in real time, and means for displaying and optimizing information according to the user's emotional state based on the emotion recognition means. This allows the user to intuitively understand the highly accurate disaster risk prediction results and receive appropriate information according to their emotional state.

[1296] A "hazard map" is a map that shows the regional distribution of disaster risk, visually displaying the predicted extent of the impact and degree of danger of a disaster.

[1297] "Weather data" means data relating to past and current weather conditions, including temperature, precipitation, wind speed, wind direction, and the like.

[1298] "Data cleansing" refers to the process of filling in missing data values ​​and correcting inconsistencies and errors, thereby improving data quality.

[1299] A "machine learning model" is a collection of algorithms that are trained to make specific predictions or classifications using past data.

[1300] A "predictive model" is a mathematical model for predicting future events using machine learning models.

[1301] "Visualization" is the process of visually representing data and prediction results and presenting them in a way that is easy for users to understand.

[1302] "Means for generating explanations in natural language" refers to technology that automatically generates explanations in natural language for users based on prediction results and analysis results.

[1303] "Emotion recognition means" is a technology that analyzes and recognizes emotions from a user's facial expressions, voice, and text data.

[1304] "Means for displaying and optimizing information" refers to technology that appropriately adjusts the display method and content of information based on the user's emotional state.

[1305] This invention relates to a system that predicts disaster risks and displays the resulting information optimally. This system receives and stores hazard maps and past weather data, performs data cleansing, and builds machine learning models to predict future disaster risks. Furthermore, it uses emotion recognition to optimize the display of information according to the user's emotional state.

[1306] First, the server receives current hazard maps and historical weather data. This involves uploading user-provided data to the server, which then stores the data in a database for future access. Specifically, users upload weather data in point or grid format through a web interface. The server verifies the data format and records and saves the metadata.

[1307] The server then performs cleansing on the received data. This is the process of filling in missing values ​​and standardizing data formats. During this process, the data is converted into a format that is easier to analyze. For example, time series data is standardized into a format that allows time series analysis, and data in different units is reconciled.

[1308] The server then builds a machine learning model based on the cleansed data. This uses a training dataset of historical weather data and hazard maps. Algorithms used include random forest, LSTM, and CNN. This results in a model that can accurately predict future disaster risks.

[1309] The predictive model uses the latest weather data as input to predict future disaster risks. The prediction results are visualized using a map display library. For example, Leaflet or Mapbox can be used to color-code high-risk areas. In addition, natural language generation technology is used to automatically generate user-friendly explanations. This includes implementations using TensorFlow and Scikit-learn.

[1310] Users can check the prediction results through a smartphone app or web browser. The app uses GPS to provide detailed prediction results based on the user's current location in real time. Furthermore, the server uses emotion recognition to analyze the user's voice and facial expressions. This analysis uses the camera, microphone, and libraries such as Mediapipe and DeepFace. The content and format of the information displayed are optimized according to the user's emotional state. For example, if the user is feeling anxious, Saver will present information in a gradual and intuitive manner, providing a sense of security.

[1311] For example, when a user launches the app on a rainy day, it predicts that a particular area will be at increased risk of flooding over the next week based on past weather data and current hazard maps. When the emotion engine recognizes that the user is feeling anxious, the app will gradually display messages such as, "Caution is required for the next week. Would you like to see routes to the nearest evacuation shelter?" and provide evacuation routes and shelter information.

[1312] An example of a prompt sentence to input to the generative AI model is as follows:

[1313] "Use the past two years of rainfall data and current hazard maps to predict flood risk based on your location."

[1314] "Recognize the user's emotions from their voice and facial expressions, and optimize the information displayed to provide a sense of security."

[1315] The above is an embodiment of the present invention. This system allows users to intuitively understand the highly accurate disaster risk prediction results and to receive appropriate information according to their own emotional state.

[1316] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1317] Step 1:

[1318] Receiving and storing data

[1319] Users provide current hazard maps and historical weather data through a web interface. The server receives this data and stores it in a database. Specifically, when a user uploads data, the server checks its format, records the metadata, and saves it in the appropriate folder. The input is the hazard map and weather data, and the output is the uncleansed data stored in the database.

[1320] Step 2:

[1321] Data cleansing and formatting

[1322] The server checks the format of the received data and performs data cleansing. For example, it fills in missing values ​​and standardizes the format. The cleansed data is converted to make it easier to analyze. The input is the data before cleansing, and the output is the cleansed data that can be analyzed. Specific operations include unifying time series data and standardizing image data.

[1323] Step 3:

[1324] Building a machine learning model

[1325] The server builds a machine learning model based on the cleansed data. It trains a training dataset using past weather data and hazard maps to create a model that can predict future disaster risks. The input is the cleansed data, and the output is the constructed predictive model. Algorithms used include random forest, LSTM, and CNN.

[1326] Step 4:

[1327] Disaster risk prediction

[1328] The server inputs the latest weather data and uses the constructed model to predict future disaster risk. The inputs are the latest weather data and a prediction model, and the output is the predicted disaster risk value. Specifically, it executes the prediction algorithm and outputs the future risk as a numerical value.

[1329] Step 5:

[1330] Visualization of prediction results and natural language generation

[1331] To visually display the prediction results, the server uses a map display library to display the predicted disaster risk on a map in a color-coded manner. It also uses natural language generation technology to generate explanatory text that is easy for users to understand. The input is the prediction results, and the output is a visualized map and explanatory text. Specifically, it generates a map using Leaflet or Mapbox, and automatically generates explanatory text using TensorFlow or Scikit-learn.

[1332] Step 6:

[1333] Viewing and verifying results

[1334] Users can check the forecast results and their explanations via a smartphone app or web browser. The app uses GPS functionality to provide real-time disaster risk information based on the user's current location. The input is a visualized map and explanatory text, and the output is information displayed on the user's device. Specifically, the app displays the results on a user interface, allowing users to zoom in and out on the map and check detailed information.

[1335] Step 7:

[1336] Emotion recognition and information optimization

[1337] The device captures the user's voice and facial expressions and uses emotion recognition means to analyze the user's emotional state. Based on the analysis results, the server adjusts the way information is displayed and provides optimal information according to the user's emotional state. The input is the user's voice and facial expression data, and the output is optimized information display. Specifically, data is collected in real time using a camera and microphone, and analyzed using Mediapipe or DeepFace.

[1338] By executing the above processing steps, the present invention provides the user with highly accurate disaster risk prediction information and realizes optimization of appropriate information display according to the user's emotional state.

[1339] 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.

[1340] 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.

[1341] 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.

[1342] 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.

[1343] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left 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.

[1344] 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.

[1345] 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).

[1346] 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.

[1347] 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."

[1348] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1349] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1350] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1351] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1352] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1353] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1354] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1355] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1356] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1357] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1358] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1359] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1360] The following is further disclosed regarding the above embodiment.

[1361] Yes, here are the draft claims:

[1362] (Claim 1)

[1363] means for receiving current hazard maps and historical weather data;

[1364] a means for storing the hazard map and meteorological data;

[1365] means for verifying the format of the received data and performing any necessary data cleansing;

[1366] means for converting the cleansed data into a format that is easy to analyze;

[1367] A means for constructing a machine learning model and predicting future disaster risks using the predictive model;

[1368] means for visualizing the prediction results and generating explanations in natural language;

[1369] A means for a user to display and confirm the visualization and generated explanation.

[1370] Including system.

[1371] (Claim 2)

[1372] The data cleansing means includes means for completing missing values ​​in the data and checking the consistency of the data.

[1373] 10. The system of claim 1.

[1374] (Claim 3)

[1375] The machine learning model is constructed using historical weather data and hazard map data.

[1376] 10. The system of claim 1.

[1377] (Claim 4)

[1378] The prediction results are displayed on a map, and an interface is provided that allows users to check them in detail.

[1379] 10. The system of claim 1.

[1380] (Claim 5)

[1381] and means for storing said visualizations and generated explanations in a database and preserving them in a usable format for future analysis and model improvement.

[1382] 10. The system of claim 1.

[1383] "Example 1"

[1384] (Claim 1)

[1385] means for receiving current hazard maps and historical weather data;

[1386] a means for storing the hazard map and meteorological data;

[1387] means for verifying the format of the received data and performing any necessary data cleansing;

[1388] means for converting the cleansed data into a format that is easy to analyze;

[1389] A means for constructing a machine learning model and predicting future disaster risks using the predictive model;

[1390] means for visualizing the prediction results and generating explanations in natural language;

[1391] A means for a user to display and confirm the visualization and generated explanation;

[1392] A means for storing the disaster risk prediction results in a database and retaining them for future model improvement and verification.

[1393] Including system.

[1394] (Claim 2)

[1395] The data cleansing means includes means for completing missing values ​​in the data and checking the consistency of the data.

[1396] 10. The system of claim 1.

[1397] (Claim 3)

[1398] The machine learning model is constructed using historical weather data and hazard map data.

[1399] 10. The system of claim 1.

[1400] "Application Example 1"

[1401] (Claim 1)

[1402] means for receiving current hazard maps and historical weather data;

[1403] a means for storing the hazard map and meteorological data;

[1404] means for verifying the format of the received data and performing any necessary data cleansing;

[1405] means for converting the cleansed data into a format that is easy to analyze;

[1406] A means for constructing a machine learning model and predicting future disaster risks using the predictive model;

[1407] means for visualizing the prediction results and generating explanations in natural language;

[1408] A means for a user to display and confirm the visualization and generated explanation;

[1409] A means for proposing specific measures according to the operation of the logistics center based on the results of the prediction;

[1410] A system including:

[1411] (Claim 2)

[1412] 2. The system according to claim 1, wherein the data cleansing means includes means for filling in missing values ​​in the data and for checking the consistency of the data.

[1413] (Claim 3)

[1414] 2. The system of claim 1, wherein the machine learning model is constructed using historical weather data and hazard map data.

[1415] "Example 2: Combining Emotion Engines"

[1416] (Claim 1)

[1417] means for receiving current hazard maps and historical weather data;

[1418] a means for storing the hazard map and meteorological data;

[1419] means for verifying the format of the received data and performing any necessary data cleansing;

[1420] means for converting the cleansed data into a format that is easy to analyze;

[1421] A means for constructing a machine learning model and predicting future disaster risks using the predictive model;

[1422] means for visualizing the prediction results and generating explanations in natural language;

[1423] means for recognizing a user's emotions and optimizing the content displayed;

[1424] A means for a user to display and confirm the visualization and generated explanation.

[1425] Including system.

[1426] (Claim 2)

[1427] The data cleansing means includes means for completing missing values ​​in the data and checking the consistency of the data.

[1428] 10. The system of claim 1.

[1429] (Claim 3)

[1430] The machine learning model is constructed using historical weather data and hazard map data.

[1431] 10. The system of claim 1.

[1432] "Application example 2 when combining emotion engines"

[1433] (Claim 1)

[1434] means for receiving current hazard maps and historical weather data;

[1435] a means for storing the hazard map and meteorological data;

[1436] means for verifying the format of the received data and performing any necessary data cleansing;

[1437] means for converting the cleansed data into a format that is easy to analyze;

[1438] A means for constructing a machine learning model and predicting future disaster risks using the predictive model;

[1439] means for visualizing the prediction results and generating explanations in natural language;

[1440] A means for a user to display and confirm the visualization and generated explanation;

[1441] emotion recognition means for recognizing the user's emotions in real time;

[1442] and a means for displaying and optimizing information according to the emotional state of the user based on the emotion recognition means.

[1443] Including system.

[1444] (Claim 2)

[1445] The data cleansing means includes means for completing missing values ​​in the data and checking the consistency of the data.

[1446] 10. The system of claim 1.

[1447] (Claim 3)

[1448] The machine learning model is constructed using historical weather data and hazard map data.

[1449] 10. The system of claim 1. [Explanation of symbols]

[1450] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving current hazard maps and historical weather data; a means for storing the hazard map and meteorological data; means for verifying the format of the received data and performing any necessary data cleansing; means for converting the cleansed data into a format that is easy to analyze; A means for constructing a machine learning model and predicting future disaster risks using the predictive model; means for visualizing the prediction results and generating explanations in natural language; A means for a user to display and confirm the visualization and generated explanation. Including system.

2. The data cleansing means includes means for completing missing values ​​in the data and checking the consistency of the data. The system of claim 1 .

3. The machine learning model is constructed using historical weather data and hazard map data. The system of claim 1 .

4. The prediction results are displayed on a map, and an interface is provided that allows users to check them in detail. The system of claim 1 .

5. and means for storing said visualizations and generated explanations in a database and preserving them in a usable format for future analysis and model improvement. The system of claim 1 .

Images