system

Abstract

The system according to the present embodiment includes a receiving unit, a generating unit, and a providing unit. The receiving unit receives an instruction from a user. The generating unit analyzes the instruction received by the receiving unit and generates an SQL query. The providing unit extracts data according to the SQL query generated by the generating unit and returns a result thereof.

Description

Technical Field

[0001] The technology disclosed herein relates to a system. Background Technology

[0002] Patent Document 1 discloses a personalized chatbot control method executed by at least one processor, the method comprising: receiving user speech; adding the user speech to a prompt containing instructions related to a chatbot role; encoding the prompt; and inputting the encoded prompt into a language model to generate chatbot speech in response to the user speech.

[0003] Patent document 1: Japanese Patent Application Publication No. 2022-180282.

[0004] In existing technologies, users need SQL knowledge to interact directly with databases, and without professional knowledge, it is difficult to extract data, which presents a challenge. Summary of the Invention

[0005] The system described in this embodiment includes a receiving unit, a generating unit, and a providing unit. The receiving unit receives instructions from users. The generating unit parses the instructions received by the receiving unit and generates an SQL query. The providing unit extracts data based on the SQL query generated by the generating unit and returns the results. Attached Figure Description

[0006] Figure 1 This is a conceptual diagram illustrating an example of the configuration of a data processing system according to the first embodiment.

[0007] Figure 2 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and smart device according to the first embodiment.

[0008] Figure 3 This is a conceptual diagram illustrating an example of the data processing system configuration in the second embodiment.

[0009] Figure 4 This is a conceptual diagram illustrating an example of the main functions of the data processing device and smart glasses according to the second embodiment.

[0010] Figure 5 This is a conceptual diagram illustrating an example of the data processing system configuration in the third embodiment.

[0011] Figure 6 This is a conceptual diagram illustrating an example of the main functions of the data processing device and head-mounted terminal according to the third embodiment.

[0012] Figure 7 This is a conceptual diagram illustrating an example of the data processing system configuration in the fourth embodiment.

[0013] Figure 8 This is a conceptual diagram illustrating an example of the functions of the main parts of the data processing device and robot according to the fourth embodiment.

[0014] Figure 9 It represents an emotion graph that maps multiple emotions.

[0015] Figure 10 It represents an emotion graph that maps multiple emotions.

[0016] Explanation of reference numerals in the attached figures Data processing systems 10, 210, 310, and 410 12 Data processing device 14 Smart devices 214 Smart Glasses 314 Head-mounted terminal 414 Robot. Detailed Implementation

[0017] Hereinafter, an example of an implementation of the system involved in this disclosure will be described with reference to the accompanying drawings.

[0018] First, let's explain the terms used in the following description.

[0019] In the following embodiments, the processor (hereinafter referred to as "processor") can be a single computing device or a combination of multiple computing devices. Furthermore, a processor can be a single computing device or a combination of multiple computing devices. Examples of computing devices include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), or TPU (Tensor Processing Unit), etc.

[0020] In the following implementation, the labeled RAM (Random Access Memory) is a memory that temporarily stores information and is used by the processor as working memory.

[0021] In the following embodiments, the labeled memory is one or more non-volatile storage devices used to store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), disk (e.g., hard disk) or magnetic tape, etc.

[0022] In the following implementation, the labeled Communication I / F (Interface) is an interface that includes a communication processor and an antenna, etc. The Communication I / F is responsible for communication between multiple computers. Examples of communication standards applicable to the Communication I / F include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] In the following implementation, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it can be only A, only B, or a combination of A and B. Furthermore, in this specification, when "and / or" connects more than three items, the same approach as "A and / or B" applies.

[0024] [First Implementation] Figure 1 An example of the configuration of the data processing system 10 according to the first embodiment is shown.

[0025] like Figure 1 As shown, the data processing system 10 includes a data processing device 12 and an intelligent device 14. An example of the data processing device 12 is a server.

[0026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0027] The smart device 14 includes a computer 36, a receiver 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. In addition, the receiver 38, output device 40, and camera 42 are also connected to the bus 52.

[0028] The receiving device 38 includes a touchscreen 38A and a microphone 38B, etc., for receiving user input. The touchscreen 38A receives user input generated by contact with an indicator (e.g., a pen or finger). The microphone 38B receives user input generated by sound by detecting the user's voice. The control unit 46A sends data representing user input received via the touchscreen 38A and microphone 38B to the data processing unit 12. In the data processing unit 12, a specific processing unit 290 (see...) Figure 2 Get the data that represents the user input.

[0029] The output device 40 includes a display 40A and a speaker 40B, etc., and presents data to the user by outputting data in a user-perceptible form (e.g., sound and / or text). The display 40A displays visual information such as text and images according to instructions from the processor 46. The speaker 40B outputs sound according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0030] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for sending and receiving various information between processor 46 and processor 28 via network 54.

[0031] Figure 2 An example of the main functions of the data processing device 12 and the smart device 14 is shown.

[0032] like Figure 2 As shown, in the data processing apparatus 12, specific processing is performed by the processor 28. A specific processing program 56 is stored in the memory 32. The specific processing program 56 is an example of a "program" as understood in this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0033] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0034] In the smart device 14, specific processing is performed by the processor 46. A specific processing program 60 is stored in the memory 50. The specific processing program 60 is used in conjunction with the data processing system 10. The processor 46 reads the specific processing program 60 from the memory 50 and executes the read specific processing program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific processing program 60 executed on the RAM 48. Furthermore, the smart device 14 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0035] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of the processing performed by the data processing system 10 of the first embodiment will be described.

[0036] (Example) The database conversation system described in this invention allows users to interact with a database without using SQL. This system allows users to instruct an AI (AI platform) on the required data definition via chat. The AI ​​parses the instruction, converts it into SQL, executes it in a data warehouse (DWH), and returns the result to the user. For example, when a user instructs "I want to retrieve sales data for 2023," the AI ​​parses the instruction and generates an SQL query such as "SELECT * FROM sales WHERE year = 2023." This SQL query is executed in the DWH, extracting the sales data for 2023. The extracted data is then returned to the user, allowing the user to obtain the desired data as if directly requesting data extraction from the DWH. Through this mechanism, users can interact with the database without using SQL; they can easily obtain the required data simply by inputting instructions via chat. This simplifies database operations and enables efficient data utilization. Therefore, the database conversation system allows users to interact with the database and easily obtain the required data without using SQL. Specifically, this database conversation system consists of multiple modules, including a user interface unit, a natural language parsing unit, an SQL generation unit, a database access unit, and a response generation unit. This system accepts users' natural language instructions (e.g., "I want to get sales data for 2023") in the form of text data (UTF-8 encoded strings, approximately 4096 tokens at most). Input examples include "Show the customer list for 2022" and "Tell me the inventory quantity for this month." The natural language parsing unit uses a large-scale language model based on Transformer to perform word segmentation, syntax parsing (generating a syntax tree), semantic parsing (entity extraction, intent inference), and mapping to SQL structures on the input sentences. For example, it extracts entities such as "2023" and "sales data" from the input sentence and assigns them as components of the WHERE or SELECT clauses. The AI ​​model outputs an SQL query string (e.g., "SELECT * FROM sales WHERE year = 2023"), with output examples including "SELECT name FROM customers WHERE year = 2022" or "SELECT stock FROM inventory WHERE month = '2023-06'". The SQL generation unit constructs queries according to SQL syntax rules based on the extracted entities and conditions. In addition, the SQL generation department pre-obtains database schema information (table names, column names, data types) and performs consistency checks during query generation to reduce syntax and runtime errors. The database access department sends the generated SQL query to the data warehouse or relational database to obtain the result set (data in matrix form, such as equivalent to a Pandas DataFrame).The response generation department organizes the data into appropriate formats such as text, tables, and images (bar charts, line charts, etc.) based on the acquired data, user instructions, and past dialogue history, and then provides feedback to the user through the user interface. The AI ​​model integrates advanced processing techniques such as fine-tuning based on pre-trained weights, user-customized prompts, and error correction rules (e.g., preventing SQL injection and automatically escaping reserved words). Through these technologies, this system differs from traditional manual SQL writing or GUI operations, achieving improvements unique to computer technology such as semantic matching in high-dimensional vector spaces, rule-based syntax transformation, and high-speed response through parallel distributed processing. Technical benefits include: even users without SQL knowledge can extract data with high accuracy and speed, significantly reducing input errors and syntax mistakes, decreasing database maintenance costs, and shortening the overall system response time. Specific application areas include business analysis, automatic sales report generation, inventory management, medical data extraction, and learning history analysis in the education sector, and it is expected to be applied in various business scenarios where frontline personnel without SQL knowledge utilize large amounts of data.

[0037] The database session system described in this embodiment includes a receiving unit, a generation unit, and a providing unit. The receiving unit receives instructions from users. User instructions may include, but are not limited to, text input, voice instructions, and specific commands. The receiving unit may receive instructions via text chat, for example. Alternatively, it may receive instructions via voice chat. Furthermore, it may receive instructions via real-time chat. For example, the receiving unit may receive an instruction from a user via text chat stating "I want to get sales data for 2023." The generation unit uses generative AI to parse the instructions received by the receiving unit and generate SQL queries. The generation unit may, for example, use natural language processing technology to parse user instructions and generate appropriate SQL queries. For example, the generation unit parses the user instruction "I want to get sales data for 2023" and generates an SQL query such as "SELECT * FROM sales WHERE year = 2023". The generation unit may also use generative AI to extract important information from user instructions using keyword extraction technology and generate SQL queries based on this. For example, the generation unit extracts keywords such as "2023" and "sales data" from user instructions and generates SQL queries based on this. The generation unit can also utilize generation AI to parse user instructions and generate SQL queries using syntax parsing technology. For example, the generation unit parses the syntactic structure of the user instruction and generates an SQL query based on it. The provision unit extracts data based on the SQL query generated by the generation unit and returns the results to the user. The provision unit extracts data in an appropriate manner, for example, based on the database type. For example, when extracting data from a relational database, the provision unit executes an SQL query to extract data. Furthermore, when extracting data from a NoSQL database, the provision unit can also execute an appropriate query to extract data. The provision unit can also filter data based on extraction conditions, extracting only the required data. For example, the provision unit extracts sales data for 2023 based on the SQL query "SELECT * FROM sales WHERE year = 2023". The provision unit returns the extracted data to the user. For example, the provision unit returns the extracted data to the user in text form. In addition, the provision unit can also display the extracted data in a visual form, such as charts. Therefore, the database session system according to this embodiment allows users to communicate with the database without using SQL and easily obtain the required data. Some or all of the above-mentioned processes in the provisioning department can be implemented using AI, or they can be implemented without AI. For example, the provisioning department can take the SQL query generated by the generation department as input and use an AI model for data extraction and feedback to perform data extraction and feedback. Specifically, this database session system consists of multiple modules, including a user interface department, a natural language parsing department, an SQL generation department, a database access department, and a response generation department.This processing department accepts users' natural language instructions (such as "I want to get the sales data for 2023") via text data (UTF-8 encoded strings, maximum 4096 tokens), voice data (16kHz PCM, WAV format, maximum 30 seconds), or command IDs (integer values ​​or JSON structures). Input examples include "Show the customer list for 2022" and "Tell me the inventory quantity for this month." This natural language parsing department uses a large-scale language model based on Transformer to perform word segmentation, syntax parsing (generating a syntax tree), semantic parsing (entity extraction, intent inference), and mapping to SQL structures on the input sentences. For example, entities such as "2023" and "sales data" are extracted from the input sentence and assigned as components of the WHERE or SELECT clauses. The AI ​​model outputs an SQL query string (such as "SELECT * FROM sales WHERE year = 2023"), with output examples including "SELECT name FROM customers WHERE year = 2022" or "SELECT stock FROM inventory WHERE month = '2023-06'". This SQL generation unit constructs queries according to SQL syntax rules based on extracted entities and conditions. Furthermore, the SQL generation unit pre-obtains database schema information (table names, column names, data types) and performs consistency checks during query generation to reduce syntax and runtime errors. The database access unit sends the generated SQL query to a data warehouse or relational database to obtain the result set (data in matrix form, equivalent to a Pandas DataFrame). The response generation unit, based on the obtained data, user instructions, and past dialogue history, organizes the data into appropriate formats such as text, tables, and images (bar charts, line charts, etc.) and provides feedback to the user through the user interface. The AI ​​model integrates advanced processing such as fine-tuning based on pre-trained weights, user-customized prompt design, and error correction rules (such as preventing SQL injection and automatically escaping reserved words). Through these technologies, this system differs from traditional manual SQL writing or GUI operation, achieving semantic matching in high-dimensional vector spaces, rule-based syntax transformation, and high-speed response through parallel distributed processing—improvements unique to computer technology. The technical benefits include enabling users without SQL knowledge to extract data with high accuracy and speed, significantly reducing typos and syntax errors, lowering database maintenance costs, and shortening overall system response time. Specific application areas include business analysis, automated sales report generation, inventory management, medical data extraction, and learning history analysis in the education sector. It is expected to be applied in various business scenarios where frontline personnel without SQL knowledge utilize large amounts of data.

[0038] The processing unit can process user instructions in a chat-like manner. For example, it can process instructions via text chat. For instance, the processing unit can process instructions via text chat, such as "I want to get the sales data for 2023". Furthermore, it can process instructions via voice chat, such as "I want to get the sales data for 2023". Additionally, it can process instructions via live chat, such as "I want to get the sales data for 2023". Thus, users can input instructions in a chat-like manner. Specifically, this processing unit can process user instructions using text data (UTF-8 encoded strings, maximum 4096 tokens), voice data (16kHz PCM, WAV format, maximum 30 seconds), or live streaming data. This processing unit can directly forward input text data to the natural language processing unit. If it is voice data, it is converted into text by a speech recognition module (such as an RNN-based speech recognition model or a Transformer-based encoder-decoder model) before being passed to the natural language processing unit. During real-time chat, the reception department uses bidirectional communication protocols such as WebSocket to receive users' input step-by-step in real time, and performs preprocessing such as buffering and timestamping. Input examples include "Show customer list for 2022" and "Tell me this month's inventory quantity," while voice input examples include "Show latest sales data" and "Want to confirm inventory status." The reception department handles multiple input formats through a unified interface, which not only improves user experience but also automates technical processing such as input data standardization, noise removal, and timestamp management. The technical effect is that by integrating multiple input channels, it can flexibly handle input based on the user's usage environment and situation, significantly improving the overall accessibility and responsiveness of the system. In addition, by introducing speech recognition and real-time communication, compared with traditional single text input, it enhances the ability to process unstructured data and multimodal input. Specific application areas include call center business automation, hands-free operation for field workers, voice instruction handling in medical settings, and real-time question answering in the education field, among other business scenarios. Through these components, this reception department not only automates manual operations but also achieves data flow optimization and multimodal input processing efficiency improvements unique to computer technology.

[0039] The generation department can parse user instructions and generate SQL queries. It utilizes generative AI to parse instructions received by the processing department and generate SQL queries. For example, the generation department uses natural language processing technology to parse user instructions and generate appropriate SQL queries. For instance, if the user instruction is "I want to get sales data for 2023," the generation department can generate an SQL query such as "SELECT * FROM sales WHERE year = 2023." The generation department can also use generative AI to extract key information from user instructions through keyword extraction technology and generate SQL queries based on this. For example, it can extract keywords such as "2023" and "sales data" from user instructions and generate SQL queries based on this. Furthermore, the generation department can use generative AI to parse user instructions through syntax parsing technology and generate SQL queries. For example, it can parse the grammatical structure of user instructions and generate SQL queries based on this. Thus, it is possible to generate appropriate SQL queries based on user instructions. Specifically, this generation department employs a large-scale language model based on Transformer (such as an encoder-decoder architecture) to perform word segmentation, syntax parsing (generating syntax trees), semantic parsing (entity extraction, intent inference), and mapping to SQL structures on the text data (natural language sentences with a maximum of 4096 tokens) received from the receiving department. AI input examples include natural language sentences such as "Show the customer list for 2022" and "Tell me the inventory quantity for this month," which are input into the model as token sequences or arrays with grammatical labels. The AI ​​model extracts entities such as "2022," "customer list," "this month," and "inventory quantity" from the input sentences and assigns them as components of the WHERE or SELECT clauses. The AI ​​model's output is an SQL query string (such as "SELECT name FROM customers WHERE year = 2022" or "SELECT stock FROM inventory WHERE month = '2023-06'"), with each output value being a syntactically correct SQL statement conforming to the database schema. When generating SQL, the generation department pre-obtains database schema information (table names, column names, data types) and performs consistency checks during query generation to reduce syntax and runtime errors. Furthermore, the generation department enhances security and reliability by designing user-customized hints and error correction rules (such as preventing SQL injection and automatically escaping reserved words). The AI ​​model is trained using supervised learning (a large dataset of paired natural language sentences and corresponding SQL queries), transfer learning, and fine-tuning. The loss function employs cross-entropy loss or a syntax error rate minimization metric. The technical results are that, compared to traditional manual SQL writing or rule-based conversion, it eliminates semantic ambiguity, achieves semantic matching in high-dimensional vector spaces, and enables high-speed response through parallel distributed processing.Specific application areas include business analysis, automated sales report generation, inventory management, medical data extraction, and learning history analysis in the education sector. It is expected to be applied in various business scenarios involving large amounts of data by frontline personnel who lack SQL knowledge. Therefore, this generation department not only automates manual tasks but also significantly improves the accuracy and efficiency of natural language understanding and structured data generation, which are unique to computer technology.

[0040] The providing department can extract data based on the generated SQL queries and return the results to the user. The providing department extracts data based on the SQL queries generated by the generating department and returns the results to the user. For example, the providing department extracts data in an appropriate manner based on the database type. For instance, when extracting data from a relational database, the providing department executes an SQL query to extract data. Furthermore, when extracting data from a NoSQL database, the providing department can also execute an appropriate query to extract data. The providing department can also filter data based on extraction criteria, extracting only the required data. For example, the providing department extracts sales data for 2023 based on the SQL query "SELECT * FROM salesWHERE year = 2023". The providing department returns the extracted data to the user. For example, the providing department returns the extracted data to the user in text format. In addition, the providing department can also display the extracted data in a visual format such as charts. Thus, the extracted data can be returned to the user. Specifically, the Provision Department forwards the SQL query string from the Generation Department to the Database Access Department, which then initiates appropriate queries for relational databases (such as PostgreSQL and MySQL) or NoSQL databases (such as document-oriented databases and key-value databases). The Database Access Department, as the query execution result, retrieves data in matrix form (e.g., equivalent to a Pandas DataFrame, with each row representing a record and each column representing an attribute value), and performs filtering and sorting based on extraction conditions (WHERE clause, LIMIT clause, etc.) as needed. Extraction examples include retrieving the 2022 customer list using "SELECT name FROM customers WHERE year = 2022" and retrieving the June 2023 inventory quantity using "SELECT stock FROM inventory WHERE month = '2023-06'". The Provision Department then passes the retrieved data to the Response Generation Department, which, based on user instructions and past dialogue history, organizes the data into appropriate formats such as text (e.g., "Sales in 2023 were 10 million yuan"), tables (HTML tables or CSV), and images (bar charts, line charts, etc.). Chart generation can utilize data visualization libraries (such as matplotlib and Plotly) and output as image data (PNG, SVG, etc.). Furthermore, the response generation unit can automatically select the optimal display format based on the user's terminal environment and past usage history. The technical benefits include achieving a unified data extraction and response generation process by accommodating differences in database types and output formats, improving the overall flexibility and scalability of the system, and enabling rapid and accurate responses to diverse user needs. Specific application areas include business dashboards, automatic sales report generation, inventory management systems, medical data analysis, and learning history visualization in the education sector.As a result, this supply department has not only automated manual operations, but also improved the efficiency and reliability of data extraction, visualization, and response generation, which are unique to computer technology.

[0041] The processing department can infer a user's emotions and adjust the timing of instruction processing based on these inferred emotions. For example, when a user is stressed, the processing department temporarily delays instruction processing, waiting for the user to relax. For instance, the processing department can capture the user's facial expressions through a camera and use emotion inference algorithms to infer emotions. For example, the processing department can calculate an emotion score based on facial expression changes and temporarily delay instruction processing when it determines the user is stressed. Conversely, when a user is relaxed, the processing department immediately processes instructions. For example, the processing department can record the user's voice and use speech analysis technology to infer emotions. For example, the processing department can analyze the tone and speed of the voice to determine if the user is relaxed and immediately process the instruction. Furthermore, when a user is anxious, the processing department prioritizes processing instructions and processes them quickly. For example, the processing department can collect the user's biometric data (heart rate and skin conductance) through sensors and use emotion inference algorithms to infer emotions. For example, the processing department can calculate an emotion score based on heart rate changes and prioritize instruction processing when it determines the user is anxious. Thus, the timing of instruction processing can be adjusted according to the user's emotions. Emotion inference can be achieved through emotion engines or generative AI and other emotion inference functions. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Specifically, in order to infer the user's emotional state, this processing department simultaneously acquires multiple input data (image tensors, speech waveform data, biosensor values, etc.) and inputs them into the multimodal AI model. This AI model can, for example, accept image inputs such as a 128×128 pixel face image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), speech inputs such as a 16kHz sampled WAV waveform (maximum 30 seconds), and biosensor data such as heart rate (1Hz sampling, 30 samples) and electrodermal activity (1Hz sampling, 30 samples). An example input is a user changing facial expressions in front of a camera and saying "I need the data now," while wearable sensors detect an increase in heart rate. This AI model employs a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction from biometric data. Finally, it integrates feature vectors from all modalities to output emotion labels (e.g., stress, relaxation, tension, anxiety) and emotion scores (continuous values ​​from 0.0 to 1.0). Output examples include "Stress: 0.85, Relaxation: 0.10, Anxiety: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. The processing department inputs these emotion scores into threshold judgments (e.g., temporary delay when stress > 0.8, immediate processing when anxiety > 0.6) or priority assignment logic to dynamically control the timing of processing. For example, a 5-second delay timer is activated when stress is high, while immediate forwarding to the natural language processing department occurs when relaxation is high. The AI ​​model is trained using a large-scale paired dataset of facial expressions, speech, biometric data, and emotion labels, with cross-entropy loss or regression error as the loss function.Furthermore, the processing department can accumulate each user's emotional response patterns and automatically adjust personalized thresholds and delay parameters. The technical benefits include significantly reducing users' psychological burden and the risk of misoperation, substantially improving user experience quality, and enhancing the overall responsiveness and reliability of the system. Unlike traditional simple timing control or manual operator judgment, this system achieves unique processing optimization through combining multimodal emotion inference in a high-dimensional feature space with rule-based dynamic control. Specific application areas include automated call center response systems, stress management-enabled processing terminals in medical settings, question-and-answer systems with student emotion monitoring in the education sector, and hands-free processing terminals for field workers, suitable for various business scenarios requiring processing timing control based on emotional states.

[0042] The processing department can analyze a user's past instruction history and select an appropriate processing method. For example, it can prioritize recommending instruction methods frequently used by the user (voice, text, etc.). For instance, analyzing a user's past instruction history, the processing department prioritizes voice instructions for users who frequently use them. Furthermore, the processing department can predict and recommend instruction methods for specific time periods based on a user's past instruction history. For example, if the processing department finds that a user uses text instructions during a specific time period, it will prioritize text instructions during that time period. Additionally, the processing department analyzes a user's past instruction history to select the most efficient processing method. For instance, analyzing a user's past instruction history and determining that voice instructions are the most efficient, it will prioritize voice instructions. Thus, the optimal processing method can be selected based on a user's past instruction history. Some or all of the above processing by the processing department can be implemented using AI, or it can be done without AI. For example, the processing department can input the user's past instruction history into AI, which will then select the optimal processing method. Specifically, this processing department maintains a time-accumulated instruction history database for each user (e.g., a structured table containing the processing time, instruction method (text, voice, command ID, etc.), processing terminal type, processing result, and processing time for each instruction), and inputs this data into the AI ​​model. The AI ​​input example includes user ID "U123", instruction history for the past 30 days (e.g., 2024-06-01 09:00 text, 2024-06-01 12:00 voice, 2024-06-02 10:00 text, etc.), processing results for each instruction (success, failure), processing terminal (PC, smartphone, etc.), and environmental information at the time of processing (time period, day of the week, location, etc.). This AI model uses a historical parsing network based on RNN or Transformer, specifically designed for time-series data parsing, to extract features from the input history (frequency distribution of instruction methods, usage tendencies in different time periods, success rates of different terminals, etc.). The AI ​​model output is structured data such as recommendation scores for each processing method (e.g., text 0.7, voice 0.2, command 0.1) or recommendation methods for different time periods (e.g., recommending voice in the morning and text in the afternoon). Output examples include "09:00-12:00 recommended voice instructions, 12:00-18:00 recommended text instructions" or "Overall recommendation: voice 0.65, text 0.30, command 0.05," etc. Based on these recommendation scores, the service department prioritizes displaying recommended buttons on the user interface or automatically selects the default service method. The AI ​​model is trained using supervised learning with each user's historical data and actual service success rate and processing efficiency as teacher signals, or through clustering for user type classification. Furthermore, the service department can continuously learn from changes in user usage preferences and update the recommendation logic online. The technical effect is that by automatically recommending the optimal service method for each user, the tediousness and error rate of service operations can be significantly reduced, significantly improving the overall service efficiency and user satisfaction of the system.Unlike traditional methods that rely on uniform prompts or manual operators reviewing historical data, this system achieves optimized processing unique to computer technology through automatic parsing and dynamic recommendation logic of high-dimensional historical data. Specific applications include automating processing within enterprise business systems, providing optimal processing UI prompts for doctors and nurses in medical settings, customizing processing methods for students in the education sector, and reducing the workload of call center operators. It is suitable for various business scenarios requiring historical processing optimization.

[0043] The processing department can filter instructions based on the user's current project or area of ​​interest when processing them. For example, the processing department prioritizes instructions related to the user's current ongoing projects. For instance, it can prioritize instructions based on the user's project name or progress status. Furthermore, the processing department can prioritize highly relevant instructions based on the user's area of ​​interest. For instance, it can determine the area of ​​interest based on the user's past activity history or questionnaire results and prioritize related instructions. Additionally, the processing department can filter unnecessary instructions based on the user's current project or area of ​​interest. For instance, it can filter out instructions unrelated to the user's project or area of ​​interest. Thus, instructions can be filtered based on the user's current project or area of ​​interest. Some or all of the above processing by the processing department can be implemented using AI, or it can be done without AI. For example, the processing department can input the user's project or area of ​​interest information into AI, which will then perform the filtering. Specifically, this processing department maintains a structured database for each user, containing information on their current ongoing projects (e.g., project ID, name, progress status, start date / expected end date, related tags, etc.) and area of ​​interest information (e.g., keyword list, questionnaire answers, topic distribution extracted from past instructions, etc.). The AI ​​input example includes user ID "U456", current project "Business Analysis 2024" (80% progress, tags: sales, inventory), areas of interest "Sales Report" and "Inventory Management", and the content of the past 30 instructions (natural language text or command IDs). This AI model vectorizes the input project and areas of interest information with the new acceptance instruction (e.g., "Please display the sales trend for 2024 in a chart") and calculates semantic relevance. The AI ​​model outputs structured data such as relevance scores for each instruction (e.g., 0.92, 0.15, etc.) or acceptance / rejection labels (accept, retain, reject). Output examples include "Instruction A: Relevance 0.95 (accepted)" and "Instruction B: Relevance 0.10 (rejected)". The acceptance department inputs these relevance scores into threshold judgments (e.g., above 0.7 accepted, below 0.3 rejected) or priority assignment logic to dynamically control the acceptance order and filtering. The AI ​​model is trained using the correspondence between past acceptance performance and projects / areas of interest as supervised learning teacher signals, or by clustering in the semantic vector space. Furthermore, the processing department can continuously learn about each user's project progress and changes in their areas of interest, updating the filtering logic online. The technical effect is the efficient processing of instructions that are only relevant to the user's business situation or interests, significantly reducing unnecessary processing load and mishandling, and improving the overall processing efficiency and user satisfaction of the system. Unlike traditional simple keyword matching or manual operator judgment, this system achieves processing optimization unique to computer technology through relevance inference in a high-dimensional semantic space and dynamic filtering.Specific application areas include automated acceptance of project management systems, topic-based acceptance in R&D sites, optimization of student project acceptance in the education field, and case-based acceptance control in sales sites. It is applicable to various business scenarios that require acceptance optimization based on projects and areas of interest.

[0044] The processing department can infer a user's emotions and prioritize processing instructions based on these inferred emotions. For example, when a user is stressed, the processing department will postpone less important instructions and prioritize more important ones. For instance, the processing department can capture the user's facial expressions using a camera and use emotion inference algorithms to infer emotions. For example, the processing department can calculate an emotion score based on facial expression changes and, if it determines the user is stressed, postpone less important instructions. Furthermore, when the user is relaxed, the processing department will process all instructions equally. For example, the processing department can record the user's voice and use speech analysis technology to infer emotions. For example, the processing department can analyze the tone and speed of the voice and, if it determines the user is relaxed, process all instructions equally. Additionally, when a user is anxious, the processing department will prioritize instructions with high urgency. For example, the processing department can collect the user's biometric data (heart rate and skin conductance) using sensors and use emotion inference algorithms to infer emotions. For example, the processing department can calculate an emotion score based on heart rate changes and, if it determines the user is anxious, prioritize instructions with high urgency. Thus, the priority of instructions can be determined based on the user's emotions. Emotion inference can be achieved through emotion engines or emotion inference functions such as generative AI. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Specifically, to infer the user's emotional state, this processing department simultaneously acquires multiple input data (such as: a 128×128 pixel face image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), a 16kHz sampled WAV speech waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples), etc.), and inputs them into a multimodal AI model. This AI model uses a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction from biological data. It integrates feature vectors from all modalities and outputs emotional labels (such as: stress, relaxation, anxiety, etc.) and emotional scores (continuous values ​​from 0.0 to 1.0). An input example is a user changing facial expressions in front of a camera and saying "I need the data now," while wearable sensors detect an increase in heart rate, etc. Examples of AI model outputs include "Stress: 0.85, Relaxation: 0.10, Anxiety: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. This service department inputs these sentiment scores into threshold judgments (e.g., delaying low-importance indications when stress > 0.8, prioritizing urgent indications when anxiety > 0.6) or priority assignment logic to dynamically control the priority of indications. For example, when stress is high, low-importance indications are retained for 5 minutes; when urgency is high, they are immediately forwarded to the natural language processing department. The AI ​​model is trained using a large-scale paired dataset of facial expressions, voice, biometric data, and sentiment labels, with cross-entropy loss or regression error as the loss function. Furthermore, the service department can accumulate each user's sentiment response patterns and automatically adjust personalized threshold and priority parameters.The technical benefits include reducing users' psychological burden and the risk of misoperation, enabling rapid processing of important instructions and delaying unnecessary ones, thereby improving the overall responsiveness and reliability of the system. Unlike traditional simple priority control or manual operator judgment, this system achieves optimized processing unique to computer technology by combining multimodal sentiment inference from a high-dimensional feature space with rule-based dynamic priority control. Specific applications include automated call center response systems, emergency instruction priority processing terminals in medical settings, subject processing systems with student emotion monitoring in the education sector, and hands-free processing terminals for on-site workers, suitable for various business scenarios requiring instruction priority control based on emotional state.

[0045] The processing department can consider the user's geographic location information when processing instructions, prioritizing instructions with high relevance. For example, when the user is located in a specific area, the processing department prioritizes instructions related to that area. For instance, the processing department obtains the user's current location through GPS data or IP address and prioritizes instructions related to that area. Furthermore, the processing department can prioritize instructions with high relevance based on the user's current location. For example, the processing department prioritizes relevant instructions based on information about the user's current location. Additionally, when the user is moving, the processing department prioritizes the most suitable instructions based on the current location. For example, the processing department considers the user's movement path and prioritizes instructions with high relevance. Thus, instructions with high relevance can be prioritized based on the user's geographic location information. Some or all of the above processing by the processing department can be implemented using AI, or it can be done without AI. For example, the processing department can input the user's geographic location information into AI, which will then perform the priority determination for instructions with high relevance. Specifically, this service department obtains structured data from user terminals, including GPS coordinates (latitude and longitude, e.g., 35.6895, 139.6917), regional information inferred from IP addresses (prefecture, city / town / village level), and Wi-Fi access point information (BSSID list). An example of AI input is user ID "U789", current location "Chiyoda Ward, Tokyo", movement path (GPS trajectory over the past 30 minutes), current time, and past service instructions (e.g., "Check nearby store inventory" or "Get local event information"). This AI model vectorizes the geographic location information and service instructions, calculating a regional relevance score (0.0–1.0). The AI ​​model outputs the regional relevance score and priority label (e.g., high, medium, low) for each instruction, with examples such as "Instruction A: Relevance 0.95 (Priority)" and "Instruction B: Relevance 0.20 (Delay)". The service department inputs these scores into threshold judgment (e.g., priority is given to those with scores above 0.7) or priority assignment logic to dynamically control the service order. While the user is on the move, the relevance of regional information along the movement path to the instruction content is evaluated in a time-series manner, and the instruction is processed at the optimal time. The AI ​​model is trained using the correspondence between past processing performance and regional information as supervised learning teacher signals, or geographic clustering. In addition, the processing department can continuously learn each user's movement patterns and usage preferences in different regions, updating the priority logic online. The technical effect is that it can efficiently process instructions that are only related to the user's current location or movement status, significantly reducing unnecessary processing load and misprocessing, and improving the overall processing efficiency and user satisfaction of the system. Unlike traditional simple keyword matching or manual operator region determination, this system achieves processing optimization unique to computer technology through high-dimensional geographic spatial relevance inference and dynamic priority control.Specific application areas include on-site event navigation systems, location-based acceptance control at logistics sites, regional case acceptance at sales sites, and tourist navigation terminals, which are suitable for various business scenarios that require optimized acceptance based on geographical location information.

[0046] The processing department can analyze a user's social media activity when processing instructions. For example, it can prioritize instructions based on the user's mentions on social media. For instance, it can analyze the user's social media posts to prioritize relevant instructions. Furthermore, the processing department can prioritize instructions related to high-interest areas from the user's social media activity. For instance, it can analyze the user's social media activity history to prioritize instructions related to high-interest areas. Additionally, the processing department can analyze the user's social media activity to prioritize the most relevant instructions. For instance, it can prioritize highly relevant instructions based on the user's social media trends and number of followers. Thus, relevant instructions can be processed based on the user's social media activity. Some or all of the above processing by the processing department can be implemented using AI, or it can be done without AI. For example, the processing department can input the user's social media activity data into AI, which will then process the relevant instructions. Specifically, this processing department obtains the user's social media activity data (such as: the most recent 100 posts, posting time, number of likes, number of reposts, number of followers, frequency of trending keywords, etc.) as a structured database and inputs it into the AI ​​model. The AI ​​input example includes user ID "U234", content posted in the past 30 days (natural language text), interaction metrics for each post (e.g., 50 likes, 10 shares), and a list of trending keywords (e.g., "AI", "inventory management", "sales analysis"). This AI model vectorizes the posted content using a natural language processing model (e.g., a Transformer-based text classification network) and performs domain-specific clustering or topic extraction. It further calculates the semantic relevance between the indicated content and the posted content, outputting a relevance score (0.0–1.0) and priority labels (high, medium, low). Output examples include "Instruction A: Relevance 0.90 (Priority)," "Instruction B: Relevance 0.15 (Delayed)," or "Sales analysis related instruction: High priority," etc. The processing department inputs these scores into threshold judgments (e.g., priority processing for scores above 0.7) or priority assignment logic to dynamically control the processing order. The AI ​​model is trained using supervised learning with the correspondence between past processing performance and social media activities as teacher signals, or by using topic distribution clustering. Furthermore, the processing department can continuously learn each user's areas of interest and trend changes, updating priority logic online. The technical effect is the efficient processing of instructions only related to the user's latest interests or social trends, significantly reducing unnecessary processing load and erroneous processing, improving overall system processing efficiency and user satisfaction. Unlike traditional simple keyword matching or manual operator trend judgment, this system achieves processing optimization unique to computer technology through high-dimensional semantic space relevance inference and dynamic priority control. Specific application areas include market analysis support systems, trend-linked processing in sales settings, processing of student-related topics in the education sector, and topic-linked processing in customer support, suitable for various business scenarios requiring processing optimization based on social media activities.

[0047] The generation department can infer a user's emotions and adjust the generation method of SQL queries based on the inferred emotions. For example, when the user is relaxed, the generation department generates detailed SQL queries. For instance, it captures the user's facial expressions using a camera and uses emotion inference algorithms to infer emotions. For example, it calculates an emotion score based on facial expression changes and generates detailed SQL queries when the user is relaxed. Conversely, when the user is anxious, the generation department generates concise SQL queries. For instance, it records the user's voice and uses voice analysis technology to infer emotions. For example, it analyzes the tone and speed of the voice to determine if the user is anxious and generates concise SQL queries. Furthermore, when the user is stressed, the generation department generates SQL queries with fewer errors. For instance, it collects the user's biological data (heart rate and skin conductance) using sensors and uses emotion inference algorithms to infer emotions. For example, it calculates an emotion score based on heart rate changes and generates SQL queries with fewer errors when the user is stressed. Thus, the generation method of SQL queries can be adjusted according to the user's emotions. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Specifically, this generation unit acquires multiple input data simultaneously (such as: a 128×128 pixel face image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), a 16kHz sampled WAV speech waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples)) to infer the user's emotional state, and inputs them into the multimodal AI model. This AI model uses a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction from biological data. It integrates feature vectors from all modalities and outputs emotional labels (such as: stress, relaxation, anxiety, etc.) and emotional scores (continuous values ​​from 0.0 to 1.0). An example of AI input is a user changing facial expressions in front of a camera and saying "I need the data now," while wearable sensors detect an increase in heart rate, etc. Examples of AI model outputs include "Stress: 0.85, Relaxation: 0.10, Anxiety: 0.70" or "Relaxation: 0.95, Stress: 0.05". This generation department inputs these emotion scores into threshold judgments (e.g., stress > 0.8 reduces errors, relaxation > 0.8 increases detail, anxiety > 0.6 simplifies) or generation parameter control logic to dynamically adjust the SQL query generation method (number of items in the SELECT clause, number of conditions in the WHERE clause, presence or absence of the JOIN clause, setting of the LIMIT clause, etc.).For example, in a relaxed state, SQL queries with detailed summaries and multi-condition filtering are generated; in a rushed state, SQL queries extracting only the minimum necessary columns are generated; and in a stressed state, template-based safe SQL queries are generated to minimize the risk of syntax and execution errors. The AI ​​model is trained using a large-scale paired dataset of facial expressions, voice, biometric data, and sentiment labels, with cross-entropy loss or regression error as the loss function. Furthermore, the generation department can accumulate each user's emotional response patterns and past query generation history, automatically adjusting personalized generation parameters. The technical effect is the ability to generate SQL queries based on the user's psychological state and usage, reducing the rate of misoperation and errors, improving user experience quality, and enhancing the overall responsiveness and reliability of the system. Unlike traditional one-size-fits-all query generation or manual adjustments by operators, this method achieves query generation optimization unique to computer technology by combining multimodal sentiment inference in a high-dimensional feature space with rule-based dynamic generation control. Specific application areas include automated call center response systems, data extraction with stress management in medical settings, learning history analysis with student emotional monitoring in education, and hands-free terminals for on-site workers, suitable for various business scenarios requiring query generation control based on emotional state.

[0048] The generation unit can adjust the level of detail generated based on the importance of the instruction when generating SQL queries. For example, for instructions with high importance, the generation unit generates detailed SQL queries. For instance, the generation unit evaluates the importance of user instructions and generates detailed SQL queries for those with high importance. Conversely, for instructions with low importance, the generation unit generates concise SQL queries. For instance, the generation unit evaluates the importance of user instructions and generates concise SQL queries for those with low importance. Furthermore, the generation unit can dynamically adjust the level of detail of the SQL queries based on importance. For instance, the generation unit evaluates the importance of user instructions and dynamically adjusts the level of detail of the SQL queries accordingly. Thus, the level of detail of the SQL queries can be adjusted according to the importance of the instruction. Some or all of the above processing by the generation unit can be implemented using AI, or it can be implemented without AI. For example, the generation unit can input the importance of the user instruction into AI, and AI can then adjust the level of detail of the SQL queries. Specifically, this generation department calculates an importance score (e.g., a continuous value from 0.0 to 1.0 or low, medium, and high labels) for each user instruction and dynamically controls the SQL query generation parameters based on this score (number of columns in the SELECT clause, number of conditions in the WHERE clause, presence or absence of GROUP BY or ORDER BY, complexity of the JOIN clause, setting of the LIMIT clause, etc.). AI input examples include "Want to get 2023 sales data (importance: high)," "Please tell me this month's inventory (importance: medium)," and "Confirm yesterday's login history (importance: low)." This AI model takes the instruction text content, past usage history, business rules, user attributes, etc., as feature inputs and outputs an importance score through an importance inference network (e.g., a Transformer-based classification or regression model). AI model output examples include "Instruction A: Importance 0.95," "Instruction B: Importance 0.40," etc. When the importance is high, the generation department generates SQL containing detailed summaries and multi-condition filtering; when the importance is low, it generates concise SQL that extracts only the necessary minimum columns. Furthermore, query execution priority and resource allocation can be dynamically adjusted based on importance. The AI ​​model is trained using supervised learning with a dataset corresponding to past instructions and business deliverables, or through reinforcement learning utilizing user feedback. The technical benefits include generating SQL queries based on business importance and urgency, achieving overall system resource optimization, reduced response time, lower risk of misoperation, and improved user satisfaction. Unlike traditional one-size-fits-all query generation or manual adjustments by operators, this approach combines importance inference from a high-dimensional feature space with dynamic generation control, achieving query generation optimization unique to computer technology. Specific application areas include extracting key indicators from business analysis dashboards, extracting urgent data from medical settings, generating case priority reports from sales sites, and extracting topic importance data in the education field, making it suitable for various business scenarios requiring importance-based query generation control.

[0049] The generation department can apply different generation algorithms based on the category of the instruction when generating SQL queries. For example, for instructions related to sales data, the generation department applies a sales data-specific generation algorithm. Similarly, for instructions related to customer data, the generation department applies a customer data-specific generation algorithm. Likewise, for instructions related to inventory data, the generation department applies an inventory data-specific generation algorithm. Thus, an appropriate generation algorithm can be applied based on the category of the instruction. Some or all of the above processing by the generation department can be implemented using AI, or it can be done without AI. For example, the generation department can input the category of the instruction into the AI, which will then execute the application of the appropriate generation algorithm. Specifically, this generation department performs word segmentation and syntactic analysis on the instruction text received by the user through a natural language parsing unit, and automatically determines the instruction category (e.g., sales, customer, inventory, procurement, expense, etc.) through a classification network (such as a Transformer-based text classification model). Examples of AI inputs include "I want to get sales data for 2023," "Please tell me the inventory for this month," and "Display the customer list for 2022." The AI ​​model outputs category labels (e.g., sales, inventory, customers) and confidence scores for each category (e.g., sales 0.95, inventory 0.03, customers 0.02). Output examples include "Instruction A: Sales category," "Instruction B: Inventory category," etc. Based on the category determination results, the generation department selects a SQL template generation algorithm specific to sales data (e.g., sales table JOIN, period summary, by department GROUP BY), a algorithm specific to customer data (e.g., customer attribute filtering, historical JOIN), and an algorithm specific to inventory data (e.g., inventory table JOIN, latest inventory extraction), to generate SQL queries. Furthermore, different pattern information and business rules can be automatically applied to each category (e.g., monthly summary for sales, daily update for inventory) to reduce the risk of syntax errors and inconsistencies in business logic. The AI ​​model is trained using supervised learning with large-scale paired datasets of instruction text with category labels and corresponding SQL queries, or automatically generated using templates for each category. The technical effect is that by optimizing the algorithm selection based on the indicated content, the accuracy, efficiency, and maintainability of SQL query generation are significantly improved, reducing the risk of erroneous generation and deviations from business logic. Unlike traditional one-size-fits-all query generation or manual category selection by operators, this method combines category classification in a high-dimensional semantic space with dynamic algorithm selection, achieving query generation optimization unique to computer technology.Specific application areas include data extraction by indicator category in business analysis dashboards, patient / examination / medication category data extraction in medical settings, case / customer / sales category report generation in sales settings, and grade / attendance / topic category data extraction in the education field. It is suitable for various business scenarios that require query generation based on category control.

[0050] The generation unit can infer a user's emotions and adjust the length of SQL queries based on these inferred emotions. For example, when a user is anxious, the generation unit generates a short and concise SQL query. For instance, it captures the user's facial expressions using a camera and uses emotion inference algorithms to infer emotions. For example, it calculates an emotion score based on facial expression changes and generates a short and concise SQL query when it determines the user is anxious. Conversely, when a user is relaxed, the generation unit generates a longer SQL query with detailed explanations. For instance, it records the user's voice and uses speech analysis technology to infer emotions. For example, it analyzes the tone and speed of the voice and generates a longer SQL query with detailed explanations when it determines the user is relaxed. Furthermore, when a user is stressed, the generation unit generates a shorter SQL query with fewer errors. For instance, it collects the user's biometric data (heart rate and skin conductance) using sensors and uses emotion inference algorithms to infer emotions. For example, it calculates an emotion score based on heart rate changes and generates a shorter SQL query with fewer errors when it determines the user is stressed. Thus, the length of the SQL query can be adjusted according to the user's emotions. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Specifically, this generation unit acquires multiple input data simultaneously (such as: a 128×128 pixel face image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), a 16kHz sampled WAV speech waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples)) to infer the user's emotional state, and inputs them into the multimodal AI model. This AI model uses a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction from biological data. It integrates feature vectors from all modalities and outputs emotional labels (such as: stress, relaxation, anxiety, etc.) and emotional scores (continuous values ​​from 0.0 to 1.0). An example of AI input is a user changing facial expressions in front of a camera and saying "I need the data now," while wearable sensors detect an increase in heart rate, etc. Examples of AI model outputs include "Stress: 0.85, Relaxation: 0.10, Anxiety: 0.70" or "Relaxation: 0.95, Stress: 0.05". This generation department inputs these emotion scores into threshold judgments (e.g., shortening when anxiety > 0.6, increasing detail when relaxation > 0.8, and reducing errors when stress > 0.8) or generation parameter control logic to dynamically adjust the length of the SQL query (number of items in the SELECT clause, number of conditions in the WHERE clause, presence or absence of the JOIN clause, setting of the LIMIT clause, etc.).For example, under pressure, it generates minimal queries such as `SELECT * FROM sales WHERE year = 2023`; under relaxed conditions, it generates detailed queries such as `SELECT sales.id, sales.amount, sales.region, sales.date, customers.name FROM sales JOIN customers ON sales.customer_id = customers.id WHERE sales.year = 2023`; and under stress, it generates safe and concise queries based on templates. The AI ​​model is trained using a large-scale paired dataset of facial expressions, voice, biometric data, and sentiment tags, employing cross-entropy loss or regression error as the loss function. Furthermore, the generation department can accumulate each user's emotional response patterns and past query generation history, automatically adjusting personalized generation parameters. The technical effect is the ability to control SQL query length based on the user's psychological state and usage, reducing the incidence of misoperations and errors, improving user experience quality, and enhancing the overall responsiveness and reliability of the system. Unlike traditional methods that uniformly adjust query length or rely on manual adjustments by operators, this approach combines multimodal sentiment inference from a high-dimensional feature space with rule-based dynamic length control, achieving query generation optimization unique to computer technology. Specific application areas include call center automatic response systems, data extraction with stress management in medical settings, learning history analysis with student emotional monitoring in the education sector, and hands-free acceptance terminals for on-site workers, which are suitable for various business scenarios that require controlling the query length based on emotional state.

[0051] The generation department can determine the priority of SQL query generation based on the submission time of the instruction. For example, the generation department prioritizes generating SQL queries for instructions submitted earlier. For instance, the generation department evaluates the submission time of user instructions and prioritizes generating SQL queries for instructions submitted earlier. Conversely, the generation department delays generating SQL queries for instructions submitted later. For instance, the generation department evaluates the submission time of user instructions and delays generating SQL queries for instructions submitted later. Furthermore, the generation department can dynamically adjust the SQL query generation priority based on the submission time. For instance, the generation department evaluates the submission time of user instructions and dynamically adjusts the SQL query generation priority based on the submission time. Thus, the generation priority of SQL queries can be determined based on the submission time of the instruction. Some or all of the above processing by the generation department can be implemented using AI, or it can be implemented without AI. For example, the generation department can input the submission time of the user instruction into the AI, and the AI ​​will execute the determination of the SQL query generation priority. Specifically, this generation department records the submission time (timestamp, e.g., 2024-06-01 09:00:00) of each instruction received from the receiving department and manages the instruction queue in chronological order. Examples of AI inputs include Instruction A (2024-06-01 09:00), Instruction B (2024-06-01 09:05), and Instruction C (2024-06-01 09:10). This AI model takes the submission time, content, user attributes, and past processing history of the instructions as features as input, and outputs a generation priority score (0.0–1.0) for each instruction through a priority inference network (such as a temporal RNN or a priority inference model based on Transformer). Examples of AI model outputs include "Instruction A: Priority 0.95", "Instruction B: Priority 0.80", and "Instruction C: Priority 0.60". The generation department dynamically adjusts the order of SQL query generation and resource allocation based on the priority scores. Furthermore, composite priority control can be implemented by comprehensively considering submission time, importance, relevance, and user urgency indicators. The AI ​​model is trained using supervised learning with a dataset corresponding to past instruction submission times, actual processing priorities, and business deliverables, or through reinforcement learning using user feedback. The technical effect is to generate SQL queries based on the submission order of instructions and the urgency of business operations, thereby improving the overall responsiveness, fairness, and user satisfaction of the system, and reducing the risk of misoperation and delays. Unlike traditional simple FIFO processing or manual priority adjustment by operators, this technology achieves query generation optimization unique to computer technology by combining temporal priority prediction in a high-dimensional feature space with dynamic generation control. Specific application areas include call center automated response system handling order control, priority processing of urgent instructions in medical settings, case submission order control in sales settings, and project submission order control in the education field, making it suitable for various business scenarios that require query generation control based on submission timing.

[0052] The generation unit can adjust the generation order of SQL queries based on the relevance of the instructions. For example, the generation unit prioritizes generating SQL queries for instructions with high relevance. For instance, the generation unit evaluates the relevance of user instructions and prioritizes generating SQL queries for instructions with high relevance. Conversely, the generation unit delays generating SQL queries for instructions with low relevance. For instance, the generation unit evaluates the relevance of user instructions and delays generating SQL queries for instructions with low relevance. Furthermore, the generation unit can dynamically adjust the generation order of SQL queries based on the relevance of the instructions. For instance, the generation unit evaluates the relevance of user instructions and dynamically adjusts the generation order of SQL queries based on the relevance. Thus, the generation order of SQL queries can be adjusted according to the relevance of the instructions. Some or all of the above processing by the generation unit can be implemented using AI, or it can be implemented without AI. For example, the generation unit can input the relevance of user instructions into AI, and AI can then adjust the SQL query generation order. Specifically, this generation unit performs word segmentation and syntactic analysis on multiple instruction texts received from the receiving unit through a natural language parsing unit, calculating the semantic relevance between each instruction (such as cosine similarity, semantic vector distance, topic consistency, etc.). Examples of AI inputs include "Request sales data for 2023", "Please provide inventory figures for 2023", and "Display customer list for 2022". This AI model takes the instruction text content, past usage history, business rules, and user attributes as feature inputs, and outputs a relevance score (0.0–1.0) for each instruction through a relevance inference network (such as a Transformer-based semantic vector generation model). Examples of AI model outputs include "Instruction AB: Relevance 0.95" and "Instruction AC: Relevance 0.40". The generation department groups instructions with high relevance and generates and executes SQL queries in the same batch to improve database access efficiency and cache hit rate. Instructions with low relevance are processed later to achieve optimal resource allocation. Furthermore, the execution priority and resource allocation of queries can be dynamically adjusted based on relevance. The AI ​​model is trained using supervised learning of the corresponding datasets of past instructions and business deliverables, or by clustering in the semantic vector space. The technical effect is the ability to generate SQL queries based on business relevance and data dependency, achieving overall system resource optimization, reduced response time, reduced risk of misoperation, and improved user satisfaction. Unlike traditional one-size-fits-all query generation or manual relevance adjustment by operators, this system achieves query generation optimization unique to computer technology by combining relevance inference from a high-dimensional semantic space with dynamic generation order control. Specific applications include utilizing the correlation between indicators in business analytics dashboards, extracting patient / examination / medication-related data in medical settings, generating case / customer / sales-related reports in sales settings, and extracting grade / attendance / topic-related data in education. It is suitable for various business scenarios requiring relevance-based query generation control.

[0053] The data delivery unit can infer a user's emotions and adjust the data delivery method based on the inferred emotions. For example, when a user is relaxed, the delivery unit provides detailed data. For instance, it captures the user's facial expressions using a camera and uses an emotion inference algorithm to infer emotions. For example, it calculates an emotion score based on facial expression changes and provides detailed data when the user is relaxed. Furthermore, when a user is anxious, the delivery unit provides concise data. For instance, it records the user's voice and uses speech analysis technology to infer emotions. For example, it analyzes the tone and speed of the voice to provide concise data when the user is anxious. Additionally, when a user is stressed, the delivery unit provides less erroneous data. For instance, it collects the user's biometric data (heart rate and skin conductance) using sensors and uses an emotion inference algorithm to infer emotions. For example, it calculates an emotion score based on heart rate changes and provides less erroneous data when the user is stressed. Thus, the data delivery method can be adjusted according to the user's emotions. Emotion inference can be achieved through emotion engines or generative AI, etc. Generative AI can be text-generated AI (such as LLM) or multimodal generative AI, but is not limited to these. Specifically, to infer the user's emotional state, this provision simultaneously acquires multiple input data (such as: a 128×128 pixel face image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), a 16kHz sampled WAV speech waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples), etc.), and inputs them into a multimodal AI model. This AI model uses a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction from biological data. It integrates feature vectors from all modalities and outputs emotional labels (such as: stress, relaxation, anxiety, etc.) and emotional scores (continuous values ​​from 0.0 to 1.0). An example of AI input is a user changing facial expressions in front of a camera and saying "I need the data now," while wearable sensors detect an increase in heart rate, etc. Examples of AI model outputs include "Stress: 0.85, Relaxation: 0.10, Anxiety: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. This provision uses these sentiment scores as input thresholds for judgment (e.g., stress > 0.8 for reduced errors, relaxation > 0.8 for more detailed data, and anxiety > 0.6 for more concise data) or provides parameter control logic to dynamically adjust the data delivery method (number of columns in tables, level of detail in charts, presence or absence of explanatory text, limits on the number of data entries, etc.). For example, in the relaxed state, data including detailed summaries and multiple charts is provided; in the anxious state, only the minimum necessary summary data is returned; and in the stress state, only template-based safe data is provided to minimize the risk of syntax and execution errors.The AI ​​model is trained using a large-scale paired dataset of facial expressions, voice, biometric data, and emotional tags, with cross-entropy loss or regression error as the loss function. Furthermore, the delivery department can accumulate each user's emotional response patterns and past delivery history, automatically adjusting personalized delivery parameters. The technical effect is the ability to provide data based on the user's psychological state and usage, reducing misoperation and error rates, improving user experience quality, and enhancing the overall responsiveness and reliability of the system. Unlike traditional one-size-fits-all data delivery or manual adjustments by operators, this approach combines multimodal sentiment inference in a high-dimensional feature space with rule-based dynamic delivery control, achieving data delivery optimization unique to computer technology. Specific application areas include automated call center response systems, data delivery with stress management in medical settings, learning history analysis with student emotional monitoring in education, and hands-free terminals for field workers, suitable for various business scenarios requiring data delivery control based on emotional states.

[0054] The data delivery department can analyze a user's past data usage history to select the optimal data delivery method. For example, it can prioritize data delivery methods frequently used by the user. Furthermore, the department can predict and recommend data delivery methods for specific time periods based on the user's past data usage history. For instance, if the department detects that a user uses a specific data delivery method during a particular time period, it will prioritize that method during that period. Additionally, the department can select the most efficient data delivery method based on the user's past data usage history. Thus, the optimal data delivery method can be selected based on the user's past data usage history. Some or all of the above processing by the delivery department can be implemented using AI, or it can be done without AI. For example, the department can input the user's past data usage history into AI, allowing AI to select the optimal data delivery method. Specifically, this provision department maintains a time-series database of user data usage history for each user (e.g., a structured table showing the acceptance time of each data provision request, provision method (text, table, chart, image, etc.), terminal type, usage result, processing time, etc.), and inputs this data into the AI ​​model. An example of AI input is user ID "U123", the data usage history for the past 30 days (e.g., 2024-06-01 09:00 text, 2024-06-01 12:00 chart, 2024-06-02 10:00 table, etc.), usage results for each provision method (satisfied, dissatisfied, re-request), terminal (PC, smartphone, etc.), and environmental information during usage (time period, day of the week, location, etc.). This AI model uses a historical parsing network based on RNN or Transformer, specifically designed for time-series data parsing, to extract features from the input history (frequency distribution of provision methods, usage tendencies in different time periods, satisfaction levels for different terminals, etc.). The AI ​​model outputs structured data such as recommendation scores for each delivery method (e.g., text 0.7, charts 0.2, tables 0.1) or recommendation methods for different time periods (e.g., charts recommended in the morning, text recommended in the afternoon). Output examples include "09:00-12:00 recommend charts, 12:00-18:00 recommend text" or "Overall recommendation: charts 0.65, text 0.30, tables 0.05," etc. Based on these recommendation scores, the delivery department prioritizes displaying recommendation buttons on the user interface or automatically selects the default delivery method. The AI ​​model is trained using supervised learning with each user's historical data and actual user satisfaction and processing efficiency as teacher signals, or through user type classification via clustering. Furthermore, the delivery department can continuously learn changes in user usage preferences and update the recommendation logic online.The technical benefits include significantly reducing the complexity and error rate of data delivery by automatically recommending the optimal data delivery method for each user, thereby substantially improving the overall data delivery efficiency and user satisfaction. Unlike traditional methods that rely on uniform prompts or manual historical data retrieval, this system achieves optimized data delivery through automatic parsing of high-dimensional historical data and dynamic recommendation logic, a feature unique to computer technology. Specific application areas include automating data delivery within enterprise business systems, providing optimal delivery UI prompts for doctors and nurses in medical settings, customizing data delivery methods for students in the education sector, and linking user history data delivery to customer support. It is suitable for various business scenarios requiring optimized data delivery based on historical data.

[0055] The data delivery department can customize the data delivery method based on the user's current lifestyle. For example, when the user is busy, the delivery department provides concise data. For instance, the delivery department assesses the user's lifestyle and determines that they are busy, providing concise data. Conversely, when the user is relaxed, the delivery department provides detailed data. For instance, the delivery department assesses the user's lifestyle and determines that they are relaxed, providing detailed data. Furthermore, the delivery department can dynamically customize the data delivery method based on the user's lifestyle. For instance, the delivery department assesses the user's lifestyle and dynamically customizes the data delivery method based on that lifestyle. Thus, the data delivery method can be customized according to the user's lifestyle. Some or all of the above processing in the delivery department can be performed using AI, or it can be performed without AI. For example, the delivery department can input the user's lifestyle into AI, which will then customize the data delivery method. Specifically, this delivery department integrates multiple input data representing the user's lifestyle (e.g., calendar events, current time, terminal activity logs, wearable sensor activity data, location information, sleep / exercise / diet records, etc.) and inputs them into the AI ​​model. Examples of AI inputs include: user ID "U567", today's calendar schedule (3 meetings, 2 moves), current time "14:00", activity level (8000 steps, heart rate 80), device usage (using PC, not using smartphone), sleep time (6 hours), and dietary records (lunch eaten). This AI model utilizes a lifestyle inference network (such as a temporal RNN or a Transformer-based multivariate classification model) to output lifestyle status labels or scores such as "busy," "relaxed," "on the move," and "resting" from the input data. Output examples include "Busy: 0.85, Relaxed: 0.10" and "On the move: 0.70, Resting: 0.20." The system provides threshold values ​​for these lifestyle status scores (e.g., simplification for "Busy" > 0.7, detail for "Relaxed" > 0.7, and priority for voice output for "On the move" > 0.6) or provides parameter control logic to dynamically customize data delivery methods (text summarization, simplified charts, voice broadcasting, notification formats, etc.). For example, when busy, only a summary text or key metrics are returned; when relaxed, detailed analysis reports or multiple charts are provided; and when on the move, data is delivered via voice output or notifications. The AI ​​model can be trained using supervised learning with datasets corresponding to lifestyle data and user satisfaction and usage efficiency, or through reinforcement learning using user feedback. Furthermore, the delivery department can accumulate users' individual lifestyle patterns and past delivery history, automatically adjusting personalized customization logic. The technological advantage lies in that by automatically optimizing data delivery methods based on users' lifestyles and usage environments, information overload and the risk of misoperation can be significantly reduced, resulting in a substantial improvement in overall system delivery efficiency and user satisfaction.Unlike previous standardized delivery methods or manual customization by operators, this system achieves data delivery optimization unique to computer technology through automatic parsing and dynamic customization of high-dimensional lifestyle data. Specific application areas include: context-linked dashboard output for business professionals, context-specific data delivery for doctors / nurses in medical settings, learning support linked to students' lifestyles in the education sector, and hands-free notifications for field workers. It is suitable for various business scenarios requiring optimized data delivery based on lifestyle conditions.

[0056] The data delivery department can infer a user's emotions and prioritize data delivery based on these inferences. For example, when a user is relaxed, all data is provided equally. For instance, the department captures the user's facial expressions using a camera and uses emotion inference algorithms to infer emotions. For example, the department calculates an emotion score based on facial expression changes and, if the user is relaxed, provides all data equally. Furthermore, when a user is in a state of urgency, the department prioritizes providing data of higher importance. For instance, the department records the user's voice and uses speech analysis technology to infer emotions. For instance, the department analyzes the pitch and speed of the voice and, if the user is in a state of urgency, prioritizes providing data of higher importance. Additionally, when a user is under stress, the department prioritizes providing data with fewer errors. For instance, the department collects the user's physiological data (heart rate and skin conductance) using sensors and uses emotion inference algorithms to infer emotions. For instance, the department calculates an emotion score based on heart rate changes and, if the user is under stress, prioritizes providing data with fewer errors. Thus, the priority of data delivery can be determined based on the user's emotions. Emotion inference can be achieved through emotion engines or emotion inference functions such as generative AI. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Specifically, this provision acquires multiple input data (such as a 128×128 pixel facial image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), a 16kHz sampled WAV audio waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples)) to infer the user's emotional state, and inputs them into a multimodal AI model. This AI model uses a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction of physiological data. It integrates feature vectors from all modalities and outputs emotional labels (such as stress, relaxation, urgency, etc.) and emotional scores (continuous values ​​from 0.0 to 1.0). An example of AI input is a user changing facial expressions in front of a camera and saying "I want the data now," while wearable sensors detect an increase in heart rate, etc. Examples of AI model outputs include "Stress: 0.85, Relaxation: 0.10, Urgency: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. This provision department inputs these sentiment scores into threshold judgments (e.g., stress > 0.8 emphasizes reduced errors, relaxation > 0.8 provides equal data, urgency > 0.6 prioritizes importance) or priority control logic to dynamically adjust the priority of data provision (e.g., prioritizing important data, prioritizing low-error-risk data, providing all data equally, etc.). For example, in an urgent state, only highly important data is returned first; in a stressful state, template-based safe data is prioritized; and in a relaxed state, all data is provided equally.The AI ​​model is trained using supervised learning on a large-scale paired dataset of facial expressions, voice, physiological data, and emotional labels, with cross-entropy loss or regression error as the loss function. Furthermore, the system can accumulate users' individual emotional response patterns and past data delivery history, automatically adjusting personalized priority logic. The technical benefits lie in prioritizing data delivery based on user psychological state and usage patterns, enabling rapid delivery of important data, reducing misoperation and error rates, improving user experience quality, and enhancing the overall responsiveness and reliability of the system. Unlike previous uniform priority control or manual adjustments by operators, this system achieves data delivery optimization unique to computer technology through multimodal emotion inference in a high-dimensional feature space and rule-based dynamic priority control. Specific application areas include: automated call center response systems, priority delivery of emergency data in medical settings, learning support with student emotion monitoring in the education sector, and hands-free terminals for field workers, suitable for various business scenarios requiring priority control of data delivery based on emotional state.

[0057] The data delivery unit can consider the user's geographic location information and select an appropriate delivery method when providing data. For example, when the user is located in a specific region, the delivery unit prioritizes providing data relevant to that region. For instance, the delivery unit uses GPS data or IP address to obtain the user's current location and prioritizes providing data relevant to that region. Furthermore, the delivery unit can also prioritize providing highly relevant data based on the user's current location. For instance, the delivery unit prioritizes providing relevant data based on information about the user's current location. Moreover, when the user is moving, the delivery unit can select the optimal data delivery method based on the current location. For instance, the delivery unit considers the user's movement path and prioritizes providing highly relevant data. Thus, the optimal data delivery method can be selected based on the user's geographic location information. Some or all of the above processing in the delivery unit can be performed using AI, or it can be performed without AI. For example, the delivery unit can input the user's geographic location information into AI, and the AI ​​can then select the optimal data delivery method. Specifically, this provider acquires structured data from user terminals, including GPS coordinates (latitude and longitude, e.g., 35.6895, 139.6917), regional information inferred from IP addresses (prefecture, city / town / village level), and Wi-Fi access point information (BSSID list). AI input examples include: user ID "U789", current location "Chiyoda Ward, Tokyo", movement path (GPS trajectory over the past 30 minutes), current time, and past data usage (e.g., "checking nearby store inventory" or "getting local event information"). This AI model vectorizes the geographic location information and data provision request content, calculating a regional relevance score (0.0–1.0). The AI ​​model outputs the regional relevance score and priority label (e.g., high, medium, low) for each data provision request, with output examples such as "Request A: Relevance 0.95 (priority)" and "Request B: Relevance 0.20 (post-processing)". The provision department inputs these scores into threshold judgments (e.g., prioritizing data above 0.7) or priority assignment logic to dynamically control the provision order and method (map-linked charts, regional summary tables, on-site notifications, etc.). While the user is on the move, the relevance of regional information and data content along the movement path is evaluated time-series, providing data at the optimal time. The AI ​​model is trained using the correspondence between past data provision performance and regional information as teacher signals for supervised learning, or employs geographic clustering. Furthermore, the provision department can gradually learn users' individual movement patterns and usage preferences in different regions, updating the priority logic online. The technical effect is that by efficiently providing only data relevant to the user's current location and movement, unnecessary processing load and erroneous provision can be significantly reduced, improving the overall system's provision efficiency and user satisfaction. Unlike previous simple keyword matching or manual operator regional determination, this system achieves data provision optimization unique to computer technology through high-dimensional geographic spatial relevance inference and dynamic priority control.Specific application areas include: on-site event navigation systems, location-based data provision and control at logistics sites, regional case data provision at sales sites, and tourist navigation terminals, which are suitable for various business scenarios that require data provision optimization based on geographic location information.

[0058] The data delivery department can analyze users' social media activities and propose data delivery methods when providing data. For example, the department can prioritize providing relevant data based on users' mentions on social media. For instance, the department can analyze users' social media posts and prioritize relevant data. Furthermore, the department can prioritize providing data related to high-interest areas from users' social media activities. For instance, the department can analyze users' social media activity history and prioritize data related to high-interest areas. Additionally, the department can analyze users' social media activities and prioritize providing the most relevant data. For instance, the department can prioritize providing highly relevant data based on users' social media trends and follower counts. Thus, the optimal data delivery method can be proposed based on users' social media activities. Some or all of the above processing in the delivery department can be performed using AI, or it can be performed without AI. For example, the department can input users' social media activity data into AI, which will then propose the optimal data delivery method. Specifically, this delivery department acquires users' social media activity data (such as the most recent 100 posts, posting time, likes, reposts, follower count, frequency of trending keywords, etc.) as a structured database and inputs it into the AI ​​model. Examples of AI inputs include: user ID "U234", content posted in the past 30 days (natural language text), interaction metrics for each post (e.g., 50 likes, 10 shares), and a list of trending keywords (e.g., "AI", "inventory management", "sales analysis"). This AI model vectorizes the posted content using a natural language processing model (e.g., a Transformer-based text classification network) for domain-specific clustering and topic extraction. It further calculates the semantic relevance between the requested content and the posted content, outputting a relevance score (0.0–1.0) and priority labels (high, medium, low). Output examples include "Request A: Relevance 0.90 (priority)," "Request B: Relevance 0.15 (post-processing)," or "Sales analysis related data: high priority," etc. The delivery department inputs these scores into threshold judgments (e.g., prioritizing those above 0.7) or priority assignment logic to dynamically control the delivery order and delivery methods (trend-linked charts, domain-specific reports, notification formats, etc.). The AI ​​model is trained using supervised learning by correlating past performance with social media activity as teacher signals, or by employing topic-based clustering. Furthermore, the system can progressively learn users' areas of interest and trend changes, updating priority logic online. The technological advantage lies in efficiently providing only data relevant to users' latest interests and social trends, significantly reducing unnecessary processing load and erroneous provision, thus improving overall system efficiency and user satisfaction. Unlike simple keyword matching or manual trend judgment, this system achieves data provision optimization unique to computer technology through high-dimensional semantic space relevance inference and dynamic priority control.Specific application areas include: market analysis support systems, providing trend-linked data at the sales site, providing learning support for students' areas of interest in the education field, and providing topic-linked data for customer support, etc., which are suitable for various business scenarios that require data optimization based on social media activities.

[0059] The system described in this embodiment is not limited to the examples above. For instance, various modifications can be made. Specifically, this system allows for diverse design changes to various technical elements, such as the internal algorithms and data flows, AI model architecture, input / output data formats, user interface, database type, communication protocol, sensor device types, learning methods, parameter optimization techniques, and cloud-edge collaborative structure of the main components, including the receiving unit, generation unit, and providing unit. For example, the receiving unit can adopt multimodal input methods such as voice, text, image, and gesture; the generation unit can combine various generation algorithms, such as large-scale language models based on Transformers, time-series analysis models based on RNNs, and rule-based generation engines; the providing unit can dynamically select the output format based on user attributes and terminal environment; and the database access unit can initiate queries on heterogeneous databases such as relational databases, NoSQL databases, time-series databases, and graph databases, thus achieving diverse implementation options. Furthermore, this system can apply various AI model learning methods, such as supervised learning, semi-supervised learning, reinforcement learning, transfer learning, and self-supervised learning, to different scenarios, and the loss function and optimization algorithm (such as Adam, SGD, RMSprop, etc.) can also be selected according to the application. Furthermore, to enhance the overall scalability of the system, cutting-edge technologies such as microservice architecture, API gateway, event-driven design, distributed cluster structure, and high-speed inference platforms based on GPUs / TPUs can be introduced. The technical benefits lie in significantly improving the system's flexibility, scalability, and maintainability through the aforementioned diverse change tolerance, enabling it to quickly and accurately respond to changes in user needs and business environments. Specific application areas include: enterprise business analysis dashboards, data extraction and visualization systems in medical settings, learning history analysis in the education sector, real-time inventory management in logistics settings, and automated customer support, suitable for various business scenarios customized according to industry, business type, and scale of use. Thus, this system not only automates human operations but also achieves the flexible system design, operation, and optimization unique to computer technology.

[0060] The processing department can analyze a user's past instruction history and prioritize suggestions for frequently used instruction methods. For example, if a user has used voice instructions multiple times in the past, the processing department will prioritize voice instructions. Furthermore, the processing department can predict and propose instruction methods for specific time periods based on the user's past instruction history. For example, if a user uses text instructions during a specific time period, text instructions will be prioritized for that time period. Further, the processing department can analyze the user's past instruction history to select the most effective processing method. For example, if analyzing the user's past instruction history determines that voice instructions are the most effective, voice instructions will be prioritized. Thus, the optimal processing method can be selected based on the user's past instruction history. Specifically, this processing department holds a database of instruction history accumulated chronologically by user (such as a structured table containing the processing time, instruction method (text, voice, command ID, etc.), processing terminal type, processing result, processing time, etc. for each instruction), and inputs it into an AI model. This AI model uses a historical analysis network based on RNN or Transformer, specifically designed for time-series data analysis, to extract features from the input history (frequency distribution of instruction methods, usage tendencies in different time periods, success rates of different terminals, etc.). Examples of AI inputs include: user ID "U123", instruction history for the past 30 days (e.g., text at 09:00 on June 1, 2024, voice at 12:00 on June 1, 2024, text at 10:00 on June 2, 2024, etc.), processing results (success / failure) for each instruction, processing terminal (PC, smartphone, etc.), and environmental information at the time of processing (time period, day of the week, location), etc. The AI ​​model outputs structured data such as recommendation scores for each processing method (e.g., text 0.7, voice 0.2, command 0.1) or recommendation methods for each time period (e.g., voice recommended in the morning, text recommended in the afternoon). Output examples include "09:00-12:00 recommended voice instruction, 12:00-18:00 recommended text instruction" or "Overall recommendation: voice 0.65, text 0.30, command 0.05", etc. Based on these recommendation scores, the processing department prioritizes displaying the recommended button on the user interface or automatically selects the default processing method. The AI ​​model is trained using supervised learning, employing users' historical data and actual acceptance success rates and processing efficiency as teacher signals, or using clustering for user type classification. Furthermore, the service department can gradually learn changes in users' individual usage preferences and update the recommendation logic online. The technical effect is that by automatically recommending the optimal service method for each user, the workload and error rate of service processing can be significantly reduced, resulting in a substantial improvement in the overall system efficiency and user satisfaction. Unlike previous methods that relied on uniform service method prompts or manual operators reviewing history, this system achieves unique service optimization through the automatic parsing of high-dimensional historical data and dynamic recommendation logic.Specific application areas include: automation of internal business system processing, optimal processing UI prompts for doctors / nurses at medical sites, customization of processing methods for students in the education sector, and reduction of operator workload in call centers. It is applicable to various business scenarios that require processing optimization based on historical data.

[0061] The generation department can parse user instructions and adjust the level of detail generated based on the importance of the instruction when generating SQL queries. For example, for instructions with high importance, a detailed SQL query is generated. The importance of user instructions is evaluated, and detailed SQL queries are generated for those with high importance. Conversely, for instructions with low importance, a concise SQL query can be generated. Furthermore, the generation department can dynamically adjust the level of detail of the SQL query based on its importance. Specifically, this generation department calculates an importance score (such as a continuous value from 0.0 to 1.0 or a low / medium / high label) for each user's instruction and dynamically controls the SQL query generation parameters (number of columns in the SELECT clause, number of conditions in the WHERE clause, presence or absence of GROUP BY or ORDER BY, complexity of the JOIN clause, setting of the LIMIT clause, etc.) based on this score. Examples of AI input include "I want to get sales data for 2023 (importance: high)," "Please tell me the inventory for this month (importance: medium)," and "Confirm yesterday's login record (importance: low)," etc. This AI model takes the text content of the instruction, past usage history, business rules, and user attributes as feature inputs, and outputs an importance score through an importance inference network (such as a Transformer-based classification or regression model). Examples of the AI ​​model's output include "Instruction A: Importance 0.95" and "Instruction B: Importance 0.40". The generation department generates SQL queries with detailed summaries and multi-condition filtering when the importance is high, and generates concise SQL queries that extract only the minimum necessary columns when the importance is low. Furthermore, the execution priority and resource allocation of queries can be dynamically adjusted based on importance. The AI ​​model is trained using supervised learning with a dataset corresponding to past instructions and business deliverables, or through reinforcement learning using user feedback. The technical benefits include the ability to generate SQL queries based on business importance and urgency, achieving overall system resource optimization, reduced response time, lower risk of misoperation, and improved user satisfaction. Unlike previous uniform query generation or manual adjustments by operators, this model achieves query generation optimization unique to computer technology through importance inference and dynamic generation control in a high-dimensional feature space. Specific application areas include: extracting key indicators for business analysis dashboards, extracting emergency data from medical sites, generating case priority reports from sales sites, and extracting topic importance data in the education field. It is suitable for various business scenarios that require controlling query generation based on importance.

[0062] The data delivery department can extract data based on generated SQL queries and, when feeding the results back to the user, analyze the user's past data usage history to select the optimal data delivery method. For example, it prioritizes data delivery methods frequently used by the user in the past. Furthermore, it can predict and propose data delivery methods for specific time periods based on the user's past data usage history. Moreover, it can analyze the user's past data usage history to select the most effective data delivery method. Thus, the optimal data delivery method can be selected based on the user's past data usage history. Specifically, this delivery department maintains a data usage history database accumulated chronologically by user (such as a structured table containing the acceptance time of each data delivery request, delivery method (text, table, chart, image, etc.), terminal type, usage result, processing time, etc.), and inputs this data into an AI model. Examples of AI inputs include: user ID "U123", data usage history over the past 30 days (e.g., text at 09:00 on June 1, 2024, chart at 12:00 on June 1, 2024, table at 10:00 on June 2, 2024, etc.), usage results for each delivery method (satisfied / dissatisfied / request), terminal (PC, smartphone, etc.), and environmental information during usage (time period, day of the week, location, etc.). This AI model uses a historical analysis network based on RNN or Transformer, specifically designed for time-series data analysis, to extract features from the input history (frequency distribution of delivery methods, usage tendencies in different time periods, satisfaction levels for different terminals, etc.). The AI ​​model outputs structured data such as recommendation scores for each delivery method (e.g., text 0.7, chart 0.2, table 0.1) or recommendation methods for different time periods (e.g., chart recommended in the morning, text recommended in the afternoon). Output examples include "Recommended charts from 09:00-12:00, recommended text from 12:00-18:00" or "Overall recommendation: Charts 0.65, Text 0.30, Tables 0.05," etc. Based on these recommendation scores, the delivery department prioritizes displaying recommendation buttons on the user interface or automatically selects the default delivery method. The AI ​​model is trained using supervised learning with users' historical data and actual usage satisfaction and processing efficiency as teacher signals, or by using clustering for user type classification. Furthermore, the delivery department can gradually learn changes in users' usage preferences and update the recommendation logic online. The technical effect is that by automatically recommending the optimal data delivery method for each user, the operational burden and error rate can be significantly reduced, significantly improving the overall delivery efficiency and user satisfaction of the system. Unlike previous uniform delivery method prompts or manual operators reviewing history, this system achieves data delivery optimization unique to computer technology through automatic parsing of high-dimensional historical data and dynamic recommendation logic.Specific application areas include: automating data provision for internal business systems, providing optimal UI prompts for doctors / nurses in medical settings, customizing data provision methods for students in the education sector, and providing data linked to user history in customer support. These applications are suitable for various business scenarios that require data provision based on historical optimization.

[0063] The processing unit can infer a user's emotions and adjust the timing of instruction processing based on this inference. For example, when a user is feeling stressed, instruction processing can be temporarily delayed until the user relaxes. The unit captures the user's facial expressions using a camera and employs an emotion inference algorithm to predict emotions. An emotion score is calculated based on facial expression changes, and if the user is determined to be stressed, instruction processing is temporarily delayed. Conversely, when the user is relaxed, instructions can be processed immediately. The unit also records the user's voice and uses speech analysis technology to infer emotions. By analyzing the pitch and speed of the voice, if the user is determined to be relaxed, instructions can be processed immediately. Thus, the timing of instruction processing can be adjusted based on the user's emotions. Specifically, to infer the user's emotional state, this processing unit simultaneously acquires multiple input data (image tensors, audio waveform data, physiological sensor values, etc.) and inputs them into a multimodal AI model. This AI model accepts 128×128 pixel facial image tensors (RGB, 8-bit, 1 frame or 5 frames temporally) as image input, 16kHz sampled WAV waveforms (maximum 30 seconds) as audio input, and heart rate (1Hz sampling, 30 samples) and electrodermal activity (1Hz sampling, 30 samples) as physiological data input. An example input is a user changing facial expressions in front of a camera and saying "I want the data now," while wearable sensors detect an increase in heart rate. The AI ​​model uses a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for audio input, and a temporal attention model for feature extraction from physiological data. Finally, it integrates the feature vectors from all modalities to output emotion labels (such as stress, relaxation, tension, urgency, etc.) and emotion scores (continuous values ​​from 0.0 to 1.0). Example outputs are "Stress: 0.85, Relaxation: 0.10, Urgency: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. The processing department inputs these emotion scores into threshold judgments (e.g., temporary delay when stress > 0.8, immediate processing when urgency > 0.6) or priority assignment logic to dynamically control the timing of processing. For example, a 5-second delay timer is activated when stress is high, while immediate transfer to the natural language processing department occurs when relaxation is high. The AI ​​model is trained using supervised learning on a large-scale paired dataset of facial expressions, speech, physiological data, and emotion labels, with cross-entropy loss or regression error as the loss function. Furthermore, the processing department can accumulate individual user emotion response patterns and automatically adjust personalized thresholds and delay parameters. The technical effect is that it significantly reduces the psychological burden on users and the risk of misoperation, significantly improves the quality of user experience, and enhances the overall responsiveness and reliability of the system. Unlike previous simple timing control or manual operator judgment, this system achieves processing optimization unique to computer technology through multimodal emotion inference in a high-dimensional feature space and rule-based dynamic control.Specific application areas include: call center automatic response systems, medical field reception terminals with stress management, education field question and answer systems with student emotional monitoring, and hands-free reception terminals for field workers, which are suitable for various business scenarios that require controlling the timing of reception based on emotional state.

[0064] The generation unit can infer the user's emotions and adjust the generation method of SQL queries based on the inferred user emotions. For example, when the user is relaxed, a detailed SQL query is generated. The user's facial expressions are captured by a camera, and emotions are inferred using an emotion inference algorithm. An emotion score is calculated based on facial expression changes to determine when the user is relaxed, and a detailed SQL query is generated. Furthermore, when the user is anxious, a concise SQL query can also be generated. The user's voice is recorded, and emotions are inferred using speech analysis technology. The pitch and speed of the voice are analyzed to determine when the user is anxious, and a concise SQL query can be generated. Thus, the generation method of SQL queries can be adjusted according to the user's emotions. Specifically, to infer the user's emotional state, this generation unit simultaneously acquires multiple input data (such as a 128×128 pixel facial image tensor (RGB, 8-bit, 1 frame or 5 frames temporally), a 16kHz sampled WAV audio waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples), etc.) and inputs them into a multimodal AI model. This AI model employs a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for speech input, and a temporal attention model for feature extraction from physiological data. It integrates feature vectors from all modalities to output emotion labels (such as stress, relaxation, and urgency) and emotion scores (continuous values ​​from 0.0 to 1.0). An example of AI input is a user changing facial expressions in front of a camera and saying "I want the data now," while wearable sensors detect an increase in heart rate. Example outputs from the AI ​​model are "Stress: 0.85, Relaxation: 0.10, Urgency: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. The generation department uses these emotion scores as input thresholds (e.g., stress > 0.8 for reduced errors, relaxation > 0.8 for more detailed analysis, and urgency > 0.6 for more concise analysis) or generation parameter control logic to dynamically adjust the SQL query generation method (number of items in the SELECT clause, number of conditions in the WHERE clause, presence or absence of the JOIN clause, setting of the LIMIT clause, etc.). For example, under relaxed conditions, SQL queries with detailed summaries and multi-condition filtering are generated; under urgent conditions, SQL queries extracting only the minimum necessary columns are generated; and under stress conditions, template-based safe SQL queries are generated to minimize the risk of syntax and execution errors. The AI ​​model is trained using supervised learning on a large-scale paired dataset of facial expressions, speech, physiological data, and emotional labels, employing cross-entropy loss or regression error as the loss function. Furthermore, the generation department can accumulate users' individual emotional response patterns and past query generation history, automatically adjusting personalized generation parameters. The technical effect is that SQL queries can be generated based on the user's psychological state and usage patterns, resulting in reduced error rates, improved user experience, and enhanced overall system responsiveness and reliability.Unlike previous methods of uniform query generation or manual adjustments by operators, this system achieves unique query generation optimization through multimodal sentiment inference in a high-dimensional feature space and rule-based dynamic generation control. Specific applications include: automated call center response systems, data extraction with stress management in medical settings, learning history analysis with student emotional monitoring in education, and hands-free terminals for field workers. It is suitable for various business scenarios requiring query generation control based on emotional states.

[0065] The data delivery unit can infer a user's emotions and adjust the data delivery method based on the inferred emotions. For example, it can provide detailed data when the user is relaxed. It captures the user's facial expressions using a camera and uses an emotion inference algorithm to infer emotions. Based on the changes in facial expressions, it calculates an emotion score to determine when the user is relaxed and provides detailed data. Conversely, it can provide concise data when the user is anxious. It records the user's voice and uses speech analysis technology to infer emotions. By analyzing the pitch and speed of the voice, it determines when the user is anxious and provides concise data. Thus, the data delivery method can be adjusted according to the user's emotions. Specifically, to infer the user's emotional state, this data delivery unit simultaneously acquires multiple input data (such as a 128×128 pixel facial image tensor (RGB, 8-bit, 1 frame or 5 frames sequentially), a 16kHz sampled WAV audio waveform (maximum 30 seconds), heart rate (1Hz sampling, 30 samples), and electrodermal activity (1Hz sampling, 30 samples), etc.) and inputs them into a multimodal AI model. This AI model employs a CNN-based facial expression recognition network for image input, an RNN or Transformer-based speech emotion recognition network for voice input, and a temporal attention model for feature extraction from physiological data. It integrates feature vectors from all modalities to output emotion labels (such as stress, relaxation, and urgency) and emotion scores (continuous values ​​from 0.0 to 1.0). An example of AI input is a user changing facial expressions in front of a camera and saying "I want the data now," while wearable sensors detect an increase in heart rate. Example outputs from the AI ​​model are "Stress: 0.85, Relaxation: 0.10, Urgency: 0.70" or "Relaxation: 0.95, Stress: 0.05," etc. This provision uses these emotion scores as input thresholds (e.g., stress > 0.8 for reduced errors, relaxation > 0.8 for more detailed data, and urgency > 0.6 for more concise data) or provides parameter control logic to dynamically adjust the data delivery method (number of columns in tables, level of detail in charts, presence or absence of explanatory text, and limits on the number of data entries, etc.). For example, in a relaxed state, data with detailed summaries and multiple charts is provided; in an urgent state, only the minimum necessary summary data is returned; and in a stressful state, only template-based safe data is provided to minimize the risk of syntax and execution errors. The AI ​​model is trained using supervised learning on a large-scale paired dataset of facial expressions, speech, physiological data, and emotional labels, with cross-entropy loss or regression error as the loss function. Furthermore, the provisioning department can accumulate users' individual emotional response patterns and past data provisioning history, automatically adjusting personalized provisioning parameters. The technical effect is that data can be provided based on the user's psychological state and usage, reducing the incidence of misoperations and errors, improving user experience quality, and enhancing the overall responsiveness and reliability of the system. Unlike previous uniform data provisioning or manual adjustments by human operators, this approach achieves data provisioning optimization unique to computer technology through multimodal sentiment inference in a high-dimensional feature space and rule-based dynamic provisioning control.Specific application areas include: call center automatic response systems, data provision with stress management in medical settings, learning history analysis with student emotional monitoring in the education sector, and hands-free acceptance terminals for on-site workers, which are suitable for various business scenarios that require data provision based on emotional state control.

[0066] The processing department can filter instructions based on the user's current project or area of ​​interest when processing them. For example, instructions related to the user's current ongoing project are prioritized. Instructions related to the user's project name and progress are prioritized. Furthermore, instructions with high relevance can be prioritized based on the user's area of ​​interest. Areas of interest can be determined based on the user's past activity history or questionnaire results, prioritizing related instructions. Thus, instructions can be filtered based on the user's current project or area of ​​interest. Specifically, this processing department maintains a structured database for each user containing information on their current ongoing projects (such as project ID, name, progress status, start / expected end date, related tags, etc.) and areas of interest (such as keyword lists, questionnaire responses, topic distribution extracted from past instructions, etc.). Examples of AI input include: user ID "U456", current project "Business Analysis 2024" (80% progress, tags: sales, inventory), areas of interest "sales report" and "inventory management", and 30 past instructions (natural language text or command IDs). This AI model vectorizes the input project / area of ​​interest information with new processing instructions (such as "show sales trends in 2024 in a chart") and calculates semantic relevance. The AI ​​model outputs structured data such as relevance scores for each instruction (e.g., 0.92, 0.15, etc.) or acceptance / rejection labels (accept, retain, reject). Output examples include "Instruction A: Relevance 0.95 (accepted)" and "Instruction B: Relevance 0.10 (rejected)". The acceptance department inputs these relevance scores into threshold judgments (e.g., above 0.7 accepted, below 0.3 rejected) or priority assignment logic to dynamically control the acceptance order and filtering. The AI ​​model is trained using supervised learning with the correspondence between past acceptance performance and project / area of ​​interest as teacher signals, or through semantic vector space clustering. Furthermore, the acceptance department can gradually learn the changes in users' project progress and areas of interest, updating the filtering logic online. The technical effect is that it efficiently accepts only instructions relevant to users' business status and interests, significantly reducing unnecessary processing load and mis-acceptance, improving the overall acceptance efficiency and user satisfaction of the system. Unlike simple keyword matching or manual operator acceptance judgment in the past, this system achieves unique acceptance optimization through high-dimensional semantic space relevance inference and dynamic filtering. Specific application areas include: automated acceptance of project management systems, topic-based acceptance in R&D sites, optimization of student project acceptance in the education field, and case-based acceptance control in sales sites. It is applicable to various business scenarios that require optimized acceptance based on projects / areas of interest.

[0067] The generation department can apply different generation algorithms based on the category of the instruction when generating SQL queries. For example, for instructions related to sales data, a generation algorithm specifically for sales data is applied. When receiving instructions related to sales data, the SQL query is generated using a generation algorithm specifically for sales data. Similarly, for instructions related to customer data, a generation algorithm specifically for customer data can be applied. When receiving instructions related to customer data, the SQL query is generated using a generation algorithm specifically for customer data. Thus, an appropriate generation algorithm can be applied based on the category of the instruction. Specifically, this generation department performs word segmentation and syntactic analysis on the instruction text received by the user through the natural language processing department, and automatically determines the instruction category (e.g., sales, customer, inventory, purchasing, expense, etc.) through a category classification network (e.g., a Transformer-based text classification model). Examples of AI inputs include "I want to get the sales data for 2023," "Please tell me the inventory count for this month," and "Show the customer list for 2022." The AI ​​model outputs category labels (e.g., sales, inventory, customer) and confidence scores for each category (e.g., sales 0.95, inventory 0.03, customer 0.02). Output examples include "Instruction A: Sales Category" and "Instruction B: Inventory Category". Based on the category determination results, the generation department selects a SQL template generation algorithm specific to sales data (such as sales table JOIN, period summary, departmental GROUP BY, etc.), a customer data-specific algorithm (such as customer attribute filtering, historical JOIN, etc.), and an inventory data-specific algorithm (such as inventory table JOIN, latest inventory extraction, etc.) to generate SQL queries. Furthermore, different pattern information and business rules can be automatically applied to each category (such as monthly summaries for sales and daily updates for inventory) to reduce the risk of syntax errors and inconsistencies in business logic. The AI ​​model is trained using supervised learning with large-scale paired datasets of instruction text labeled with categories and corresponding SQL queries, or automatically generated using category templates. The technical effect is that by optimizing algorithm selection based on instruction content, the accuracy, efficiency, and maintainability of SQL query generation are significantly improved, reducing the risk of erroneous generation and deviations from business logic. Unlike previous unified query generation or manual category selection by operators, this system achieves query generation optimization unique to computer technology through high-dimensional semantic space category classification and dynamic algorithm selection. Specific application areas include: data extraction by indicator category in business analysis dashboards, data extraction by patient / examination / medication category in medical settings, generation of case / customer / sales category reports in sales settings, and data extraction by grade / attendance / topic category in the education field. It is suitable for various business scenarios that require query generation based on category control.

[0068] The data delivery department can consider the user's geographic location information and select the appropriate delivery method when providing data. For example, when the user is located in a specific area, data related to that area can be prioritized. Using GPS data or IP address to obtain the user's current location, data related to that area can be prioritized. Furthermore, based on the user's current location, highly relevant data can be prioritized. Based on information about the user's current location, relevant data can be prioritized. Thus, the optimal data delivery method can be selected based on the user's geographic location information. Specifically, this delivery department obtains geographic location information such as GPS coordinates (latitude, longitude, e.g., 35.6895, 139.6917) obtained from the user's terminal, regional information inferred from the IP address (prefecture, city / town / village level), and Wi-Fi access point information (BSSID list) as structured data. Examples of AI input include: user ID "U789", current location "Chiyoda Ward, Tokyo", movement path (GPS trajectory over the past 30 minutes), current time, and past data usage content (e.g., "checking nearby store inventory" or "obtaining local event information"). This AI model vectorizes geographic location information and data provision requests, calculating a regional relevance score (0.0–1.0). The AI ​​model outputs the regional relevance score and priority label (e.g., high, medium, low) for each data provision request, with examples such as "Request A: Relevance 0.95 (priority)" and "Request B: Relevance 0.20 (post-processing)." The provision department inputs these scores into threshold judgment (e.g., prioritizing requests with scores above 0.7) or priority assignment logic to dynamically control the provision order and method (map-linked charts, regional summary tables, on-site notifications, etc.). While the user is on the move, the relevance of regional information and data content along the movement path is evaluated in time series, providing data at the optimal time. The AI ​​model is trained using the correspondence between past data provision performance and regional information as teacher signals for supervised learning, or geographic clustering. Furthermore, the provision department can gradually learn users' individual movement patterns and usage preferences in different regions, updating the priority logic online. The technical effect is that by efficiently providing only data relevant to the user's current location and movement, unnecessary processing load and erroneous provision can be significantly reduced, improving the overall system provision efficiency and user satisfaction. Unlike traditional simple keyword matching or manual operator region determination, this system achieves data delivery optimization unique to computer technology through high-dimensional geospatial relevance inference and dynamic priority control. Specific application areas include: on-site activity navigation systems, location-based data delivery and control at logistics sites, regional case data delivery at sales sites, and tourist navigation terminals, suitable for various business scenarios requiring optimized data delivery based on geographic location information.

[0069] The data delivery department can analyze users' social media activities and propose data delivery methods when providing data. For example, the department can prioritize providing relevant data based on users' mentions on social media. It can also prioritize providing relevant data based on users' social media posts. Furthermore, it can prioritize providing data related to high-interest areas from users' social media activities. It can also prioritize providing data related to high-interest areas based on users' social media activity history. Thus, the optimal data delivery method can be proposed based on users' social media activities. Specifically, this delivery department acquires users' social media activity data (such as the most recent 100 posts, posting time, number of likes, number of reposts, number of followers, frequency of trending words, etc.) as a structured database and inputs it into the AI ​​model. Examples of AI input include: user ID "U234", posts (natural language text) from the past 30 days, interaction metrics for each post (such as 50 likes, 10 reposts), and a list of trending words (such as "AI", "inventory management", "sales analysis"), etc. This AI model vectorizes the posted content using a natural language processing model (such as a Transformer-based text classification network) to perform interest area clustering and topic extraction. The system further calculates the semantic relevance between the requested and published content, outputting a relevance score (0.0–1.0) and priority labels (high, medium, low). Output examples include "Request A: Relevance 0.90 (Priority)," "Request B: Relevance 0.15 (Post-processing)," or "Sales analysis related data: High priority," etc. The delivery department inputs these scores into threshold judgments (e.g., prioritizing data above 0.7) or priority assignment logic to dynamically control the delivery order and delivery methods (trend-linked charts, reports on areas of interest, notification formats, etc.). The AI ​​model is trained using supervised learning by correlating past data delivery performance with social media activity as teacher signals, or by employing topic distribution clustering. Furthermore, the delivery department can gradually learn users' individual areas of interest and trend changes, updating the priority logic online. The technical effect is that by efficiently providing only data related to users' latest interests and social trends, unnecessary processing load and erroneous delivery can be significantly reduced, improving the overall delivery efficiency and user satisfaction of the system. Unlike traditional simple keyword matching or manual trend judgment, this system achieves data delivery optimization unique to computer technology through relevance inference in a high-dimensional semantic space and dynamic priority control. Specific application areas include: market analysis support systems, trend-linked data provision in sales settings, student learning support in the education sector focusing on specific areas, and topic-linked data provision in customer support. It is suitable for various business scenarios requiring optimized data delivery based on social media activities.

[0070] The following is a brief description of the implementation process. Specifically, this system consists of three modules: the receiving department, the generation department, and the providing department. These modules work collaboratively to receive diverse user input (text, voice, commands, etc.) and perform a series of technical processing steps, including natural language parsing, AI inference, SQL query generation, database access, data shaping, response generation, and feedback to the user. The receiving department acquires various attribute data from the user, such as input content, past instruction history, emotional state, project information, geographical location information, and social media activities. It then performs AI model parsing, filtering, and priority assignment. The generation department automatically generates SQL queries based on the instruction content and attribute information received from the receiving department, using natural language processing technologies (such as large-scale language models based on Transformer, keyword extraction, and syntactic analysis). It dynamically controls the generation parameters based on the importance, category, emotional state, and relevance of the instruction. The providing department forwards the generated SQL queries to the database access department, which extracts data from heterogeneous databases such as relational databases or NoSQL databases. Based on the user's emotions, usage history, geographical location, and social media activities, the data is shaped in the optimal form (text, tables, charts, images, etc.) and fed back. The modules interact through diverse data structures such as structured databases, vector representations, time-series data, and image tensors. The input / output examples and subsequent processing flows of the AI ​​model are also clearly designed. The technical effect is that by flexibly controlling the acceptance, generation, and delivery of data based on user attributes and business conditions, the overall responsiveness, accuracy, reliability, and user satisfaction of the system are significantly improved. Specific application areas include: business analysis dashboards, data extraction and visualization in medical settings, learning history analysis in education, real-time inventory management in logistics, and automated customer support.

[0071] Step 1: The Reception Department receives instructions from users. User instructions include text input, voice commands, and specific commands. For example, the Reception Department can receive instructions via text chat, voice chat, or live chat. Step 2: The Generation Department parses the instructions received by the Reception Department and generates SQL queries. The Generation Department uses AI-powered generation, combined with natural language processing, keyword extraction, and syntax parsing techniques, to analyze user instructions and generate appropriate SQL queries. For example, parsing the user instruction "I want to get sales data for 2023" generates an SQL query such as "SELECT * FROM sales WHERE year = 2023". Step 3: The Provision Department extracts data based on the SQL queries generated by the Generation Department and returns the results to the user. The Provision Department extracts data using appropriate methods based on the database type and filters the data according to extraction criteria. For example, it extracts 2023 sales data based on the SQL query "SELECT * FROM sales WHERE year = 2023" and displays it visually in text, charts, and graphs. Specifically, this system's processing department receives user input instructions (such as text like "I want to get sales data for 2023" or voice like "Please tell me the inventory for this month") and forwards the input as structured data (command ID, attribute values, time, etc.) to the generation department. The generation department uses natural language processing AI (such as a large-scale language model based on Transformer) to perform word segmentation and syntactic analysis on the instruction text and inputs it into an SQL query generation network. Examples of AI input are "I want to get sales data for 2023" and "Please tell me the inventory for this month." The AI ​​model outputs SQL query strings (such as "SELECT * FROM sales WHERE year = 2023" and "SELECT stock FROM inventory WHERE month = '2023-06'"), with output examples like "SELECT * FROM sales WHERE year = 2023" and "SELECT stock FROM inventory WHERE month = '2023-06'." The provisioning department then forwards the SQL queries output by the generation department to the database access department, which initiates queries against relational or NoSQL databases. The database access department retrieves matrix-style data (e.g., each row represents a record and each column represents an attribute value) as the query execution results, and performs filtering and sorting processing based on extraction conditions (WHERE clause, LIMIT clause, etc.) as needed. The provision department then hands the retrieved data over to the response generation department, which, based on user instructions and past dialogue history, reshapes it into appropriate formats such as text (e.g., "Sales in 2023 were 10 million yuan"), tables (HTML tables or CSV), and charts (bar charts, line charts, etc.).Chart generation can utilize data visualization libraries (such as matplotlib and Plotly) to output image data (PNG, SVG, etc.). Furthermore, the response generation department automatically selects the optimal display format based on the user's terminal environment and past usage history. The technical advantage lies in achieving a unified data extraction and response generation process by accommodating differences in database types and output formats, thus enhancing the overall flexibility and scalability of the system and enabling rapid and accurate responses to diverse user needs. Specific application areas include: business dashboards, automatic sales report generation, inventory management systems, medical data analysis, and learning history visualization in the education sector. Therefore, this system not only automates human tasks but also achieves efficiency and reliability improvements in data extraction, visualization, and response generation—qualities unique to computer technology.

[0072] The specific processing unit 290 sends 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 voice representing the user's input to the result of the specific processing. The control unit 46A sends the voice data representing the user's 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 voice data.

[0073] Data generation model 58 is what is known as generative AI (Artificial Intelligence). An example of data generation model 58 includes ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Generative AI, such as data generation model 58, is obtained by deep learning through a neural network. The data generation model 58 is input with a prompt containing instructions, and with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the above-described specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts without instructions; in this case, the data generation model 58 can output inference results from prompts without instructions. The data processing device 12, etc., includes multiple data generation models 58, including AI other than generative AI. AI other than generative AI includes, but is not limited to, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes. Furthermore, AI can also act as an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to this example. Moreover, processing performed by AI, including generative AI, can be replaced by rule-based processing, and vice versa.

[0074] Furthermore, the processing performed by the aforementioned data processing system 10 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart device 14 or external devices, and the smart device 14 acquires or collects information required for processing from the data processing device 12 or external devices.

[0075] Each of the aforementioned elements, including the receiving unit, the generating unit, and the providing unit, can be implemented, for example, in at least one of the smart device 14 and the data processing device 12. For instance, the receiving unit can be implemented by the control unit 46A of the smart device 14, for receiving text input or voice instructions from the user. The generating unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, using a generation AI to parse the user's instructions and generate an SQL query. The providing unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, extracting data based on the generated SQL query and feeding the results back to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and can be varied.

[0076] [Second Implementation] Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0077] like Figure 3 As shown, the data processing system 210 includes a data processing device 12 and smart glasses 214. One example of the data processing device 12 is a server.

[0078] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN and / or LAN, etc.

[0079] 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, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0080] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0081] Camera 42 is a small digital camera equipped with an optical system such as 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, used to capture the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0082] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0083] Figure 4 An example of the main functions of the data processing device 12 and the smart glasses 214 is shown. Figure 4 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0084] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0085] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0086] In the smart glasses 214, specific processing is performed by the processor 46. A specific processing program 60 is stored in the memory 50. The processor 46 reads the specific processing program 60 from the memory 50 and executes the read specific processing program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific processing program 60 executed on the RAM 48. Furthermore, the smart glasses 214 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0087] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0088] The specific processing unit 290 sends 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 voice input representing the user's input to the specific processing result. The control unit 46A sends the voice data representing the user's 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.

[0089] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0090] The data processing system 210 of the second embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 210 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart glasses 214 or external devices, and the smart glasses 214 acquires or collects information required for processing from the data processing device 12 or external devices.

[0091] Each of the aforementioned elements, including the receiving unit, the generating unit, and the providing unit, can be implemented, for example, in at least one of the smart glasses 214 and the data processing device 12. For instance, the receiving unit can be implemented by the control unit 46A of the smart glasses 214, for receiving text input or voice instructions from the user. The generating unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, using a generation AI to parse the user's instructions and generate an SQL query. The providing unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, extracting data based on the generated SQL query and feeding the results back to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and can be varied.

[0092] [Third Implementation] Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0093] like Figure 5 As shown, the data processing system 310 includes a data processing device 12 and a head-mounted terminal 314. One example of the data processing device 12 is a server.

[0094] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN and / or LAN, etc.

[0095] The head-mounted 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, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0096] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0097] Camera 42 is a small digital camera equipped with an optical system such as 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, used to capture the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0098] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0099] Figure 6 An example of the main functions of the data processing device 12 and the head-mounted terminal 314 is shown. Figure 6 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0100] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0101] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0102] In the head-mounted terminal 314, specific processing is performed by the processor 46. A specific program 60 is stored in the memory 50. The processor 46 reads the specific program 60 from the memory 50 and executes the read specific program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific program 60 executed on the RAM 48. Furthermore, the head-mounted terminal 314 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0103] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0104] The specific processing unit 290 sends the result of the specific processing to the head-mounted terminal 314. In the head-mounted terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires voice representing the user's input to the specific processing result. The control unit 46A sends the voice data representing the user's 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.

[0105] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0106] The data processing system 310 of the third embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 310 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the head-mounted terminal 314, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the head-mounted terminal 314. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the head-mounted terminal 314 or external devices, and the head-mounted terminal 314 acquires or collects information required for processing from the data processing device 12 or external devices.

[0107] Each of the aforementioned elements, including the receiving unit, the generating unit, and the providing unit, can be implemented, for example, in at least one of the head-mounted terminal 314 and the data processing device 12. For instance, the receiving unit can be implemented by the control unit 46A of the head-mounted terminal 314, for receiving text input or voice instructions from the user. The generating unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, using a generation AI to parse the user's instructions and generate an SQL query. The providing unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, extracting data based on the generated SQL query and feeding the results back to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and can be varied.

[0108] [Fourth Implementation] Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0109] like Figure 7 As shown, the 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.

[0110] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN and / or LAN, etc.

[0111] Robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control object 443. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and control object 443 are also connected to the bus 52.

[0112] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0113] The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and an imaging element such as a CMOS image sensor or a CCD image sensor, used to photograph the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0114] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0115] The controlled object 443 includes a display device, LEDs for the eyes, and motors for driving the arms, hands, and feet. The posture and movements of the robot 414 are controlled by controlling the motors for the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. In addition, facial expressions of the robot 414 can also be expressed by controlling the illumination state of the LEDs for the robot 414's eyes.

[0116] Figure 8 An example of the main functions of the data processing device 12 and the robot 414 is shown. For example... Figure 8 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0117] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0118] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0119] In robot 414, specific processing is performed by processor 46. A specific program 60 is stored in memory 50. Processor 46 reads the specific program 60 from memory 50 and executes the read specific program 60 on RAM 48. Specific processing is achieved by processor 46 acting as control unit 46A based on the specific program 60 executed on RAM 48. Furthermore, robot 414 may also have the same data generation model and emotion-specific model as data generation model 58 and emotion-specific model 59, and use these models to perform the same processing as specific processing unit 290.

[0120] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0121] The specific processing unit 290 sends the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires voice representing the user's input regarding the result of the specific processing. The control unit 46A sends the voice data representing 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.

[0122] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0123] The data processing system 410 of the fourth embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 410 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the robot 414 or external devices, and the robot 414 acquires or collects information required for processing from the data processing device 12 or external devices.

[0124] Each of the aforementioned elements, including the receiving unit, the generating unit, and the providing unit, can be implemented, for example, in at least one of the robot 414 and the data processing device 12. For instance, the receiving unit can be implemented by the control unit 46A of the robot 414, for receiving text input or voice instructions from the user. The generating unit can be implemented, for example, by a specific processing unit 290 of the data processing device 12, which uses a generating AI to parse the user's instructions and generate an SQL query. The providing unit can be implemented, for example, by the specific processing unit 290 of the data processing device 12, which extracts data based on the generated SQL query and feeds the results back to the user. The correspondence between the various units and the device or control unit is not limited to the above examples and can be varied.

[0125] Furthermore, the emotion-specific model 59, serving as an emotion engine, can determine the user's emotion based on a specific mapping. Specifically, the emotion-specific model 59 can determine the user's emotion based on an emotion graph that serves as a specific mapping (see...). Figure 9 The robot's emotions can be determined by the emotion-specific model 59. In addition, the emotion-specific model 59 can also determine the robot's emotions in the same way, and the specific processing unit 290 can also perform specific processing using the robot's emotions.

[0126] Figure 9 This is a diagram representing an emotion map 400 that maps various emotions. In the emotion map 400, emotions are arranged radially from the center in concentric circles. The closer to the center of the concentric circles, the more primitive the emotion is. Further out on the concentric circles, emotions are arranged representing states or actions arising from mood. Emotion is a concept that includes both feelings and mental states. To the left of the concentric circles, emotions generated by reactions occurring in the brain are arranged roughly. To the right of the concentric circles, emotions guided by situational judgments are arranged roughly. Above and below the concentric circles, emotions generated by reactions occurring in the brain and guided by situational judgments are arranged roughly. Furthermore, the emotion of "pleasure" is arranged above the concentric circles, and the emotion of "unpleasantness" is arranged below. Thus, in the emotion map 400, various emotions are mapped according to the structure of emotion generation, while easily generated emotions are mapped nearby.

[0127] These emotions are distributed at the 3 o'clock position on the Emotion Chart 400, and usually fluctuate between peace and unease. In the right half of the Emotion Chart 400, because situational awareness is more dominant than internal feelings, it gives a sense of calm.

[0128] The inner side of the emotion diagram 400 represents the mind, and the outer side of the emotion diagram 400 represents actions. Therefore, the further you go to the outer side of the emotion diagram 400, the more the emotion can be seen (manifested in actions).

[0129] Here, human emotions are based on a balance of various factors such as posture and blood sugar levels. When these balances deviate from the ideal, it indicates unhappiness; when they approach the ideal, it indicates pleasure. In robots, cars, and motorcycles, emotions can also be created based on a balance of factors such as posture and remaining battery power. When these balances deviate from the ideal, it indicates unhappiness; when they approach the ideal, it indicates pleasure. Emotion maps can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Speech Emotion Recognition and Brain Physiological Signal Analysis Systems for Emotion, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). In the left half of the emotion map, emotions belonging to the "Reaction" domain, where sensation is dominant, are arranged. Furthermore, in the right half of the emotion map, emotions belonging to the "Situation" domain, where situational cognition is dominant, are arranged.

[0130] The emotion map defines two types of emotions that promote learning. One is a negative emotion located near the middle of "repentance" or "reflection" on the situation side. That is, when the robot experiences negative emotions such as "I never want to feel this way again" or "I never want to be scolded again." The other is a positive emotion located near "desire" on the response side. That is, when the robot experiences positive feelings such as "wanting more" or "wanting to know more."

[0131] The emotion-specific model 59 feeds user input into a pre-learned neural network to obtain emotion values ​​representing each emotion shown in the emotion graph 400, and determines the user's emotion. This neural network is pre-learned based on multiple learning data sets that combine user input with emotion values ​​representing each emotion shown in the emotion graph 400. Furthermore, this neural network is learned to... Figure 10 As shown in sentiment graph 900, sentiment values ​​in nearby configurations are similar to each other. Figure 10 Examples show how emotions such as "peace of mind," "stability," and "reassurance" can result in similar emotional values.

[0132] In the above embodiments, a specific processing is described by a single computer 22, but the technology disclosed herein is not limited to this, and distributed processing by multiple computers, including computer 22, is also possible.

[0133] In the above embodiments, an example of storing a specific processing program 56 in memory 32 is illustrated, but the technology disclosed herein is not limited thereto. For example, the specific processing program 56 may also 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 performs specific processing according to the specific processing program 56.

[0134] Alternatively, the specific processing program 56 can be stored in a storage device such as a server connected to the data processing device 12 via a network 54, and the specific processing program 56 can be downloaded and installed into the computer 22 upon request from the data processing device 12.

[0135] Furthermore, it is not necessary to store the entire specific process 56 in a storage device such as a server connected to the data processing device 12 via the network 54, nor is it necessary to store the entire specific process 56 in the memory 32; a portion of the specific process 56 may also be stored.

[0136] As a hardware resource for performing specific processing, various processors can be used. For example, a CPU is a general-purpose processor that functions as a hardware resource for performing specific processing by executing software, i.e., programs. Additionally, a dedicated circuit can be listed as a processor; it is a processor with a circuit structure specifically designed for performing specific processing, such as a FPGA (Field-Programmable Gate Array), PLD (Programmable Logic Device), or ASIC (Application Specific Integrated Circuit). Every processor has built-in or connected memory, and every processor executes specific processing by using that memory.

[0137] The hardware resources for performing a specific process can consist of one of these various processors, or a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Furthermore, the hardware resources for performing a specific process can also be a single processor.

[0138] As an example of a single processor, the first type consists of a combination of one or more CPUs and software, which functions as a hardware resource to perform specific processing. The second type uses a processor, such as a System-on-a-chip (SoC), which implements the entire system functionality, including multiple hardware resources performing specific processing, using a single IC chip. In this case, the specific processing is implemented using one or more of the aforementioned processors that serve as hardware resources.

[0139] Furthermore, as the hardware architecture of these various processors, more specifically, circuits combining semiconductor elements and other circuit components can be used. Moreover, the specific process described above is merely an example. Therefore, it goes without saying that, without departing from the main point, unnecessary steps can be removed, new steps can be added, or the processing order can be changed.

[0140] Furthermore, although the above examples have been described in terms of first to fourth embodiments, some or all of these embodiments can also be combined. Additionally, the smart device 14, smart glasses 214, head-mounted terminal 314, and robot 414 are just examples; they can be combined separately or are other devices.

[0141] The foregoing descriptions and illustrations are detailed explanations of the parts covered by this disclosure and are merely one example of this disclosure. For instance, the descriptions of the above-described structure, function, role, and effect are just one example of the structure, function, role, and effect of the parts covered by this disclosure. Therefore, it goes without saying that, without departing from the spirit of this disclosure, unnecessary parts can be deleted, new elements can be added, or replacements can be made to the foregoing descriptions and illustrations. Furthermore, to avoid confusion and facilitate understanding of the parts covered by this disclosure, explanations of technical common sense that does not require special explanation for implementing this disclosure have been omitted from the foregoing descriptions and illustrations.

[0142] All documents, patent applications and technical standards described in this specification are incorporated herein by reference as if they were specifically and individually described as incorporated by reference.

[0143] (Note 1) A system comprising: The receiving department is used to receive instructions from users. The generation unit parses the instruction received by the receiving unit and generates an SQL query; and The providing unit extracts data based on the SQL query generated by the generating unit and returns the result.

[0144] (Note 2) The system as described in Appendix 1 is characterized in that, The receiving department accepts instructions from users in a chat-like manner.

[0145] (Note 3) The system as described in Appendix 1 is characterized in that, The generation unit parses the user's instructions and generates an SQL query.

[0146] (Note 4) The system as described in Appendix 1 is characterized in that, The providing unit extracts data based on the generated SQL query and feeds back the results to the user.

[0147] (Note 5) The system as described in Appendix 1 is characterized in that, The receiving department infers the user's emotions and adjusts the timing of the instruction based on the inferred user emotions.

[0148] (Note 6) The system as described in Appendix 1 is characterized in that, The receiving department analyzes the user's past instruction history and selects an appropriate receiving method.

[0149] (Note 7) The system as described in Appendix 1 is characterized in that, When accepting instructions, the receiving department filters them based on the user's current project or area of ​​interest.

[0150] (Note 8) The system as described in Appendix 1 is characterized in that, The receiving department infers the user's emotions and determines the priority of the receiving instructions based on the inferred user emotions.

[0151] (Note 9) The system as described in Appendix 1 is characterized in that, When accepting instructions, the receiving department takes into account the user's geographical location information and prioritizes instructions that are highly relevant.

[0152] (Postscript 10) The system as described in Appendix 1 is characterized in that, When receiving instructions, the receiving department analyzes the user's social media activity and processes the relevant instructions.

[0153] (Postscript 11) The system as described in Appendix 1 is characterized in that, The generation unit infers the user's emotions and adjusts the generation method of SQL queries based on the inferred user emotions.

[0154] (Postscript 12) The system as described in Appendix 1 is characterized in that, When generating SQL queries, the generation department adjusts the level of detail based on the importance of the indication.

[0155] (Postscript 13) The system as described in Appendix 1 is characterized in that, When generating SQL queries, the generation unit applies different generation algorithms according to the indicated category.

[0156] (Postscript 14) The system as described in Appendix 1 is characterized in that, The generation unit infers the user's emotions and adjusts the length of the SQL query based on the inferred user emotions.

[0157] (Postscript 15) The system as described in Appendix 1 is characterized in that, When generating SQL queries, the generation department determines the priority of generation based on the indicated submission timing.

[0158] (Postscript 16) The system as described in Appendix 1 is characterized in that, When generating SQL queries, the generation unit adjusts the generation order according to the indicated relevance.

[0159] (Postscript 17) The system as described in Appendix 1 is characterized in that, The method of providing data by inferring user sentiment and adjusting the data provided based on the inferred user sentiment.

[0160] (Postscript 18) The system as described in Appendix 1 is characterized in that, When providing data, the providing department analyzes the user's past data usage history and selects an appropriate providing method.

[0161] (Postscript 19) The system as described in Appendix 1 is characterized in that, When providing data, the providing department customizes the delivery method based on the user's current living situation.

[0162] (Postscript 20) The system as described in Appendix 1 is characterized in that, The provider infers the user's emotions and determines the priority of data provision based on the inferred user emotions.

[0163] (Postscript 21) The system as described in Appendix 1 is characterized in that, When providing data, the providing unit considers the user's geographical location information and selects an appropriate providing method.

[0164] (Postscript 22) The system as described in Appendix 1 is characterized in that, When providing data, the providing department analyzes users' social media activities and proposes methods for providing the data.

Claims

1. A system, characterized in that, include: The receiving department is used to receive instructions from users; The generation department parses the instructions received by the receiving department and generates an SQL query; as well as The providing unit extracts data based on the SQL query generated by the generating unit and returns the result.

2. The system as described in claim 1, characterized in that, The receiving department accepts instructions from users in a chat-like manner.

3. The system as described in claim 1, characterized in that, The generation unit parses the user's instructions and generates an SQL query.

4. The system as described in claim 1, characterized in that, The providing unit extracts data based on the generated SQL query and feeds back the results to the user.

5. The system as described in claim 1, characterized in that, The receiving department infers the user's emotions and adjusts the timing of the instruction based on the inferred user emotions.

6. The system as described in claim 1, characterized in that, The receiving department analyzes the user's past instruction history and selects an appropriate receiving method.

7. The system as described in claim 1, characterized in that, When accepting instructions, the receiving department filters them based on the user's current project or area of ​​interest.

8. The system as described in claim 1, characterized in that, The receiving department infers the user's emotions and determines the priority of the receiving instructions based on the inferred user emotions.

Images