Information processing system

Abstract

This invention provides an information processing system. The information processing system is characterized by comprising: a processor; wherein the processor is configured to: receive planned information and actual information from employees; perform environmental load assessment on the planned information and actual information and calculate a carbon neutrality score; identify the user's emotional state and provide input to a generative artificial intelligence model for generating prompts, so that the generative artificial intelligence model generates prompts based on the emotional state to instruct the user to adjust the carbon neutrality score according to the emotional state; provide input to the generative artificial intelligence model for generating prompts, so that the generative artificial intelligence model generates prompts to instruct the user to recommend a plan with a high carbon neutrality score; and calculate performance monthly using statistical methods.

Description

Technical Field

[0001] The technology disclosed herein relates to an information processing system. Background Technology

[0002] Japanese Patent Application Publication No. 2022-180282 discloses a method for controlling a role-based chatbot executed by at least one processor. The method includes the following steps: receiving a user's speech; adding the user's speech to a prompt word, the prompt word containing instruction statements associated with an explanation of the chatbot's role; encoding the prompt word; and inputting the encoded prompt word into a language model to generate a chatbot response to the user's speech.

[0003] Existing enterprise carbon emission management and employee low-carbon behavior management systems are mainly limited to simple recording and statistics of employee travel modes and energy consumption data. They typically calculate carbon emissions or carbon neutrality scores using fixed algorithms, lacking a comprehensive consideration of employees' subjective states and behavioral motivations. Specifically, existing technologies have at least the following problems: First, when traditional systems calculate carbon neutrality scores, they mostly rely solely on objective data to assess environmental load without considering users' behavioral tendencies at different times and in different emotional states. This makes it difficult to achieve personalized score adjustments and feedback that vary from person to person and from time to time, thus affecting the system's ability to guide employees towards low-carbon behavior.

[0004] Second, existing systems have limited support for "future plans." They typically cannot intelligently compare the environmental impact of different travel or work arrangements, nor can they generate persuasive and relatable high-scoring recommendations that take into account the user's current mood and acceptance level. As a result, the recommendations are difficult for users to actually adopt.

[0005] Third, in terms of performance presentation and feedback, existing technologies mostly rely on simple statistics on a daily or project-by-project basis, lacking monthly performance analysis based on statistical methods. This makes it difficult for employees and managers to grasp low-carbon behavior trends and areas for improvement from a medium- to long-term perspective.

[0006] Fourth, although some systems can collect carbon emission data by individual or department, they are often presented only in the form of reports or rankings, without being automatically pushed through multiple channels such as email and notification systems. They also lack systematic gamification design, making it difficult to continuously stimulate employees' enthusiasm for participation and their sense of competition, thus making it difficult to form a sustainable low-carbon culture at the organizational level.

[0007] Therefore, it is necessary to provide a system that can combine generative artificial intelligence models to not only assess the environmental impact and calculate carbon neutrality scores of employee plans and actual behaviors, but also identify user emotional states and dynamically adjust scores, automatically recommend high-scoring plans, conduct monthly statistical analysis, and improve employee engagement through multi-channel notifications and gamification mechanisms, in order to solve the above problems. Summary of the Invention

[0008] To address the aforementioned issues, this invention provides an information processing system. The system includes a processor configured to: receive planned and actual information from employees; perform environmental load assessment on the planned and actual information and calculate a carbon neutrality score; identify the user's emotional state and provide input to a generative artificial intelligence model for generating prompts, so that the generative artificial intelligence model generates prompts based on the emotional state to instruct adjustments to the carbon neutrality score according to the emotional state; provide input to the generative artificial intelligence model for generating prompts, so that the generative artificial intelligence model generates prompts to instruct recommendations of plans with high carbon neutrality scores to the user; and calculate performance monthly using statistical methods.

[0009] With the above configuration, the system can calculate an initial carbon neutrality score based on the planned and actual information provided by employees, using a preset environmental load assessment model. It also incorporates a generative artificial intelligence model to identify and analyze the user's emotional state, enabling the system to be context-aware in its feedback and suggestions. This allows it to adjust prompts with appropriate tone and content, making necessary weight adjustments or explanations to the score, thus enhancing the user's acceptance of the results. The prompts generated by the generative artificial intelligence model not only include recommendations for high-scoring plans but can also guide users to adopt more environmentally friendly travel or work options in a more acceptable way, based on their current emotional state.

[0010] Furthermore, to strengthen the quantitative management and information dissemination of low-carbon behaviors at both the organizational and individual levels, the processor is also configured to generate carbon neutrality scores separately for individuals and organizations, and to provide input to a generative artificial intelligence model for generating prompts. This allows the generative AI model to generate prompts instructing users to notify the system of the carbon neutrality scores via email or a notification system. In this way, the system can automatically send individual or departmental scores to relevant personnel via email, push notifications, etc., enabling regular feedback and timely notification of low-carbon performance, and reducing the costs of manual statistics and communication.

[0011] Furthermore, to enhance sustained engagement and motivation for behavioral improvement, the processor is configured to calculate individual and departmental scores and provide input to a generative AI model for generating prompts. This allows the model to generate gamified elements that incentivize achieving high scores by notifying employees of their individual and departmental scores. Through this design, the system can construct gamified elements around carbon neutrality scores, such as leaderboards, levels, badges, and challenges, and notify employees of their own and their department's position within the company. This significantly improves employee engagement and sustainability in low-carbon activities while maintaining data objectivity, achieving a synergistic effect between individual low-carbon behavior improvement and the organization's overall emissions reduction goals.

[0012] "System" refers to a hardware and software system comprising at least one processor and optional storage devices, communication interfaces and user interfaces, for performing the functions described in this invention, and may be implemented as a server system, cloud platform, locally deployed system or a combination thereof.

[0013] A processor is an electronic circuit or computing unit that can execute program instructions to process, operate, and control the flow of input data, including but not limited to a central processing unit (CPU), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), and a processing module consisting of multiple processing cores.

[0014] "Employee planned information and actual information" refers to records of work-related behaviors that employees will implement or have already implemented within a predetermined time frame. These may include, but are not limited to, commuting methods, travel routes, travel distances, business trip methods, remote work arrangements, and related energy consumption data. Planned information refers to behaviors that have not yet occurred but are expected to occur, while actual information refers to behaviors that have already occurred.

[0015] "Environmental load assessment" refers to the process of quantitatively or qualitatively calculating and analyzing the environmental impact of related behaviors in terms of energy consumption, carbon emissions, etc., based on employees' planned and actual information, through a pre-set or dynamically updated assessment model.

[0016] "Carbon neutrality score" refers to a score that quantifies the environmental friendliness of an employee's behavior, time period, or set of behaviors based on environmental load assessment results. It is usually given in numerical form, and the higher the score, the closer the employee is to carbon neutrality or the smaller the environmental impact.

[0017] "User's emotional state" refers to the psychological and emotional state or tendency of a user at a specific point in time or in a specific interaction scenario. Specifically, it can include, but is not limited to, positive, negative, neutral, anxious, tired, excited, indifferent and other emotional categories. It is usually identified through voice, text, facial expressions, behavioral patterns or multimodal data.

[0018] "Generative artificial intelligence model" refers to an artificial intelligence model that can automatically generate text, images, speech and other content based on input data, including but not limited to large language models, dialogue generation models, multimodal generation models, etc. In this invention, this model is used to generate prompts, suggestions and descriptions of gamification elements.

[0019] "Prompts" refer to text or other forms of information generated by generative artificial intelligence models to guide the system or user to perform specific operations, understand specific results, or make specific decisions. These include, but are not limited to, explanatory statements, suggestive statements, notification texts, guiding phrases, and instructional content used to invoke subsequent functional modules.

[0020] "High carbon neutrality score plan" refers to a plan that, among multiple alternative plans, has a relatively high carbon neutrality score and a relatively small environmental impact after environmental load assessment and carbon neutrality score calculation. This includes, but is not limited to, low-carbon travel plans, energy-saving office plans, or other behavioral plans with good emission reduction effects.

[0021] "Statistical methods for calculating monthly scores" refers to using predetermined statistical methods, with a natural month or any arbitrarily defined monthly period as the unit, to aggregate, sum, average, perform distribution analysis or other statistical processing on the carbon neutrality scores within that period to generate performance indicators that reflect monthly environmental performance.

[0022] "Carbon neutrality scores calculated separately by individual and organization" refers to the results obtained by independently summarizing and statistically analyzing the carbon neutrality scores of relevant behavioral records for individual employees and departments, business units, branches or other organizational entities, so that low-carbon performance can be evaluated separately at the individual and organizational levels.

[0023] "Email or notification system" refers to an electronic communication mechanism used to proactively push information to users or relevant parties, including but not limited to email systems based on email protocols, enterprise internal messaging systems, mobile application push systems, instant messaging tools, and notification modules integrated into enterprise platforms.

[0024] "Individual score and departmental score" refer to the carbon neutrality score statistics calculated for an individual employee and for a department, respectively. The individual score reflects the low-carbon behavior level of an individual employee, while the departmental score reflects the overall low-carbon behavior level of the department.

[0025] "Gamification elements" refer to game-like design elements introduced into a system to increase user engagement and incentivize users to continuously improve their behavior. These include, but are not limited to, points, levels, badges, leaderboards, tasks, challenges, achievement unlocks, and virtual rewards. These elements are linked to carbon neutrality scores or departmental scores to enhance the fun and competitiveness of low-carbon behavior management. Attached Figure Description

[0026] Figure 1 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the first embodiment.

[0027] 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.

[0028] Figure 3 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the second embodiment.

[0029] 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.

[0030] Figure 5 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the third embodiment.

[0031] Figure 6 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and head-mounted terminal according to the third embodiment.

[0032] Figure 7 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the fourth embodiment.

[0033] Figure 8 This is a conceptual diagram illustrating an example of the main functions of the data processing device and robot according to the fourth embodiment.

[0034] Figure 9 This represents an emotion map that maps multiple emotions.

[0035] Figure 10 This represents an emotion map that maps multiple emotions.

[0036] Figure 11 This is a sequence diagram illustrating the processing flow of the data processing system of the first embodiment.

[0037] Figure 12 This is a sequence diagram illustrating the processing flow of the data processing system in Application Example 1.

[0038] Figure 13This is a sequence diagram illustrating the processing flow of the data processing system of the second embodiment.

[0039] Figure 14 This is a sequence diagram illustrating the processing flow of the data processing system in Application Example 2. Detailed Implementation

[0040] Hereinafter, an example of an implementation of the system to which the technology of this disclosure relates will be described with reference to the accompanying drawings.

[0041] First, let me explain the terminology used in the following instructions.

[0042] In the following embodiments, the processor (hereinafter referred to as "processor") with reference numerals may be a single computing device or a combination of multiple computing devices. Furthermore, the processor may 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), etc.

[0043] In the following embodiments, RAM (Random Access Memory), as indicated in the figures, is a memory that temporarily stores information and is used as working memory by the processor.

[0044] In the following embodiments, the memory, as indicated by the reference numerals, is one or more non-volatile storage devices that store various programs and parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), disks (e.g., hard disks), or magnetic tapes.

[0045] In the following embodiments, the communication I / F (Interface) with reference numerals is an interface that includes a communication processor and an antenna, etc. The communication I / F is responsible for communication between multiple computers. As an example of a communication specification applicable to the communication I / F, wireless communication specifications such as 5G (5th Generation Mobile Communication System), Wi-Fi (wireless fidelity) (registered trademark), or Bluetooth (registered trademark) can be listed.

[0046] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it can be only A, only B, or a combination of A and B. Furthermore, in this specification, when "and / or" connects to express more than three items, the same interpretation as "A and / or B" applies.

[0047] First Implementation Method Figure 1 An example of the configuration of the data processing system 10 according to the first embodiment is shown.

[0048] like Figure 1 As shown, the data processing system 10 includes a data processing device 12 and an intelligent device 14. A server can be cited as an example of the data processing device 12.

[0049] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. 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. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0050] The smart device 14 includes a computer 36, a receiving device 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 receiving device 38, output device 40, camera 42, and communication I / F 44 are also connected to the bus 52.

[0051] The receiving device 38 includes a touchscreen 38A and a microphone 38B, and receives user input. The touchscreen 38A receives user input via touch by detecting contact with an indicator (e.g., a pen or finger). The microphone 38B receives user input via sound by detecting the user's voice. The control unit 46A in the processor 46 sends data representing the user input received by the touchscreen 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data representing the user input.

[0052] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting data in a form perceptible to the user 20 (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.

[0053] Communication I / F44 is connected to network 54. Communication I / F44 and 26 are responsible for sending and receiving various information between processor 46 and processor 28 via network 54.

[0054] Figure 2 The diagram shows an example of the main functions of the data processing device 12 and the smart device 14.

[0055] 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 according to the specific processing program 56 executed on the RAM 30.

[0056] 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. In the emotion inference function (emotion-specific function) using the emotion-specific model 59, various inferences and predictions related to the user's emotions are performed, including inferences and predictions of the user's emotions, but this is not limited to this example. Furthermore, emotion inference and prediction may also include, for example, emotion analysis (parsing).

[0057] In the smart device 14, the processor 46 performs the acceptance output processing. The memory 50 stores the acceptance output program 60. The acceptance output program 60 is used in conjunction with the data processing system 10 and the specific processing program 56. The processor 46 reads the acceptance output program 60 from the memory 50 and executes the read acceptance output program 60 on the RAM 48. The acceptance output processing is implemented by the processor 46 acting as the control unit 46A according to the acceptance output program 60 executed on the RAM 48. Furthermore, the smart device 14 has the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and these models can also be used to perform the same processing as the specific processing unit 290. The acceptance output processing is implemented by the processor 46 acting as the control unit 46A according to the acceptance output program 60 executed on the RAM 48.

[0058] Alternatively, 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. Furthermore, the data processing device 12 may be a server device or a user-held terminal device (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of the processing of the data processing system 10 of the first embodiment will be described.

[0059] Example 1 The flow of a specific process in Example 1 will be described. Each part of the system described below is implemented by the data processing device 12 and the smart device 14. Furthermore, the data processing device 12 is referred to as the "server," and the smart device 14 is referred to as the "terminal."

[0060] In existing technologies, most computer systems used to assess the environmental impact of daily human activities simply aggregate travel and energy consumption data based on simple emission coefficients, resulting in a single emission value or a rough score. Such solutions typically suffer from the following technical problems: First, the server-side data processing lacks standardized cleaning and structuring mechanisms, making it difficult to uniformly manage and efficiently calculate multi-source, multi-format activity data, leading to insufficient accuracy and scalability in environmental load assessments. Second, existing systems only output static numerical results on the server side, lacking the ability to dynamically interpret and generate interactive suggestions using generative artificial intelligence models. They cannot automatically generate personalized explanations and emission reduction suggestions based on different users' contexts and historical performance, thus limiting the system's ability to guide user behavior. Third, traditional systems often fail to incorporate user emotional states and other human-computer interaction-related information into the server-side calculation and feedback processes, hindering the service... The server cannot adjust the score presentation and prompts according to the user's emotional characteristics, resulting in a monotonous interactive experience and low long-term user engagement. Fourth, existing technologies generally only perform one-way calculations on actual historical data, lacking a mechanism to distinguish between "real-world calculations" and "hypothetical scenario simulations" within the same computational framework. The server cannot efficiently visualize and compare different activity plans, making it difficult to provide users with data-driven decision support for behavior optimization. Fifth, most systems simply display results on the terminal, without forming a statistical analysis and gamification-driven mechanism on the server side that is oriented towards time series and different granularities (individuals, organizations). It is impossible to generate ranking, achievement, and reward information that can be used for incentives through a unified algorithm.

[0061] Therefore, it is necessary to provide a computer implementation scheme that integrates data acquisition, standardized processing, precise calculation of environmental load, and time series statistical analysis on the server side, and combines generative artificial intelligence models to generate prompts, so as to achieve explanation generation, behavioral suggestion output, emotion adaptive feedback, and comparison and display of real and simulated scenarios. This will improve the accuracy of environmental load assessment, enhance user interaction experience and behavioral guidance, and achieve the improvement of computer technology itself.

[0062] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Embodiment 1 is achieved by the following means.

[0063] In this invention, the server includes: a processing unit for acquiring and storing planning and execution information related to personnel activities, wherein the activity information includes at least travel mode category, travel distance, and energy consumption, and converting raw data from different sources and formats into structured activity data through preprocessing operations such as data cleaning, unit standardization, and format specification; a calculation unit for performing vectorized numerical operations on the structured activity data based on the emission coefficients corresponding to the travel mode and the emission coefficients corresponding to the energy type, thereby calculating the environmental load on the server side, and performing normalization or scaling transformation on the environmental load to generate a carbon neutrality score; a statistical unit for performing statistical analysis on time series data according to individual and organizational dimensions based on the carbon neutrality score and environmental load decomposition results, generating periodic statistics and comparative indicators; and a prompt statement containing environmental load, carbon neutrality score, trend information, and statistical results, and displaying the... The system includes: a generation interface unit that sends prompts to a generative AI model to obtain natural language explanations and emission reduction behavior suggestions for the current user and the current cycle; an emotion adaptation unit that identifies emotional state information collected during user interaction and dynamically adjusts the content of prompts provided to the generative AI model based on emotion recognition results, thereby generating explanatory text that adapts to the user's emotions in terms of tone, emphasis, and feedback format; a simulation comparison unit that distinguishes between carbon neutrality score calculation processes based on real historical records and simulation score calculation processes based on hypothetical activity conditions within the same data processing framework, and outputs both in a comparable data structure for the terminal to display in parallel in numerical and graphical formats; and an output unit that packages and outputs the above calculation results, statistical information, explanatory text returned by the generative AI model, and behavior suggestions to the front-end device, enabling the front-end to complete complex result display by rendering data with a predetermined structure. This enables the implementation of an efficient and scalable data processing pipeline for calculating environmental load and generating carbon neutrality scores on the server side. Simultaneously, by linking with generative artificial intelligence models, it provides personalized explanations and behavioral suggestions without increasing the terminal's computing burden. Furthermore, it dynamically adjusts the feedback content based on the user's emotional state and simulation comparison results, thereby improving the overall technical performance of the computer system in terms of data processing efficiency, the level of intelligence in human-computer interaction, and the ability to guide environmental behavior.

[0064] "Information processing device" refers to an electronic computing device with a processor and memory, used to execute program instructions to store, calculate, analyze and output input data. It can be a server, a computer terminal or a computing platform composed of multiple computing nodes.

[0065] "Planning information" refers to structured or semi-structured data related to future activities, such as the planned mode of transportation, travel time, travel route, estimated travel distance, and expected energy consumption, which are collected before personnel actually carry out the activities.

[0066] "Execution information" refers to structured or semi-structured data reflecting the actual behavior after a person has actually completed an activity, such as the actual mode of travel, actual travel distance, actual energy consumption, and actual time of occurrence.

[0067] "Activity information" refers to the general term for planning information and execution information, which includes at least the type of travel mode, travel distance, and energy consumption. It is used to represent people's travel and energy use behavior within a certain period of time.

[0068] "Travel mode category" refers to the classification information used to indicate the mode of transportation used by people in a specific activity, including but not limited to public transportation, private transportation, non-motorized transportation, and remote interaction.

[0069] "Travel distance" refers to the spatial displacement of a person from the starting point to the end point in a certain activity. It is usually expressed in units of length and can be either a one-way distance or a round-trip distance. It can also be converted into a uniform unit as needed.

[0070] "Energy consumption" refers to the amount of energy consumed during a specific activity to complete office work, travel, or other activities, usually expressed in standard units of measurement for electricity or other forms of energy.

[0071] "Structured activity data" refers to activity information that has been preprocessed, including format standardization, field unification, and unit conversion, and is organized with predefined fields and data types. It has clear column and field meanings in data tables or data structures.

[0072] A "data processing program" refers to a set of software instructions that run on an information processing device and are used to perform operations such as reading, converting, calculating, statistically analyzing, and outputting active information.

[0073] A "general data processing library" refers to a collection of software functions that provide general functional interfaces for data cleaning, statistical analysis, matrix operations, or numerical calculations. These functions are used in data processing programs to reduce redundant development work.

[0074] "Preprocessing" refers to the operations performed on the original activity information before environmental load calculation, such as handling missing values, handling outliers, standardizing units, and formatting, in order to improve the accuracy and efficiency of subsequent calculations.

[0075] The "emission factor" is a parameter used to represent the amount of greenhouse gas emissions per unit of travel distance or unit of energy consumption, usually expressed as the mass of emissions per unit distance or unit of energy consumption.

[0076] "Environmental load" refers to a numerical indicator used to measure the degree of environmental impact of human activities, calculated based on activity information and emission coefficients. It is usually expressed as greenhouse gas emissions or other quantitative values ​​of environmental impact.

[0077] "Normalization processing" refers to the process of mapping the original environmental load to a numerical transformation within a preset range, so as to make comparisons and scores between different time periods or different individuals.

[0078] "Scale transformation" refers to the process of linearly or nonlinearly transforming environmental load based on a predefined mathematical function to adapt it to a specific scoring system or display requirement.

[0079] The "carbon neutrality score" is a score calculated based on environmental load through normalization or scaling transformation. It is used to represent the degree to which human activities are achieved relative to the carbon neutrality target on a uniform scale.

[0080] "Time-period statistics" refer to the statistical results obtained by aggregating data such as environmental load and carbon neutrality score within a predetermined time window (such as day, week, month or year), including sum, average, maximum, minimum and trend indicators.

[0081] "Prompt statements" refer to natural language or semi-structured text content used as input to generative artificial intelligence models, which contains key information and instructions related to environmental load, carbon neutrality score, and user behavior.

[0082] "Generative AI models" refer to AI models built on deep learning or other machine learning techniques that are used to automatically generate natural language text or other content based on input prompts.

[0083] "Explanation" refers to the natural language text output by generative artificial intelligence models based on prompts, which explains the composition of carbon neutrality scores, the reasons for changes, and the impact of different behaviors on environmental load.

[0084] "Behavioral suggestions" refer to specific action plans output by generative artificial intelligence models or information processing devices, which guide users to reduce environmental impact and improve carbon neutrality scores by changing travel methods, optimizing energy consumption, and other means.

[0085] "Emotional state" refers to the classification or numerical information inferred from sensor data, interactive behavior, or other identification methods, used to characterize a user's current psychological state or emotional tendency.

[0086] "Emotion Adaptive Unit" refers to a functional module in an information processing device that adjusts the content and output style of prompts based on the user's emotional state to generate feedback information that matches the user's emotions in terms of tone, emphasis, and presentation.

[0087] "Simulation score" refers to the carbon neutrality score obtained by using the same calculation process as the real score, based on the user-defined hypothetical activity conditions without changing the historical records. It is used to predict the environmental impact of different scenarios.

[0088] The “simulation comparison unit” refers to a functional module in an information processing device that distinguishes between the actual recorded calculation process and the hypothetical scenario calculation process, and outputs the actual score and simulation score in a comparable data structure for comparative analysis.

[0089] "Output device" refers to a display terminal, computing terminal, or electronic device with an interactive interface used to receive calculation results from an information processing device and present the corresponding information to the user.

[0090] "Gamification elements" refer to a set of incentive information generated based on carbon neutrality scores and statistics, including ranking information, achievement information, and reward information, which are used to promote continuous user participation and optimize behavior through game-like mechanisms.

[0091] In one embodiment of the invention, the server is configured as a computing device having a processor, memory, a network interface, and persistent storage. The server runs application service programs, database management programs, and generative artificial intelligence model interface programs on an operating system (e.g., a general-purpose server operating system). The server loads data processing programs for environmental load calculation and carbon neutrality score generation into memory and stores activity data tables, emission factor tables, and model call logs in the persistent storage unit.

[0092] In one embodiment of the present invention, the terminal is configured as an electronic device having a display device, an input device, and a network communication module. The terminal can be a mobile terminal or a desktop terminal, and runs a browser application or a native graphical user interface application locally. The terminal executes an interface rendering program locally to display a data input interface and a visualization result interface, and exchanges data with a server through the network communication module.

[0093] In this embodiment of the invention, the user uses a terminal to input raw data related to activity information on an interface. The user selects or inputs fields such as travel mode type, one-way distance, round trip status, office energy consumption, and home office energy consumption in a form provided by the terminal. After completing the input on the terminal, the user allows the terminal to send these fields as structured data to the server, thereby triggering the server-side data processing flow.

[0094] After receiving activity information from the terminal, the server writes the raw data into an activity data table defined in the database management system. This activity data table includes fields such as personnel identifier, date and time, travel mode, trip type, distance, energy type, energy consumption, and a flag field to indicate whether a record is real or simulated. During writing, the server appends a unique record identifier and a timestamp to each record, ensuring that subsequent calculations can be indexed and aggregated based on time and user dimensions.

[0095] When performing data cleaning and structuring, the server uses a two-dimensional tabular data structure in memory to store multiple active records read from the database. Within this data structure, the server defines column names and data types, and its data processing program performs missing value detection, outlier filtering, unit conversion, and format standardization. During unit conversion, the server standardizes all distances to kilometers and all power consumption to kilowatt-hours, and internally records the conversion factor. When filtering outliers, the server automatically discards or marks records exceeding a preset threshold range (e.g., distances that are positive and less than a certain upper limit) to avoid extreme values ​​unreasonably impacting subsequent environmental load calculations.

[0096] The server uses vectorized numerical operations instead of iterative processing record by record when calculating environmental load. It maintains an emissions coefficient table in memory, mapping travel mode type to emissions coefficient per unit distance and energy type to emissions coefficient per unit energy consumption. The server merges the travel mode column from the activity data table with the emissions coefficient table using a key join operation, adding corresponding travel emissions coefficient and energy consumption emissions coefficient columns to each record in the resulting structure. The server generates travel emissions (distance multiplied by the emissions coefficient per unit distance) and energy consumption emissions (energy consumption multiplied by the emissions coefficient per unit energy consumption) columns through linear operations, and then groups and sums the records for the same user within a certain time period to obtain the total environmental load.

[0097] When generating carbon neutrality scores, the server performs normalization or scaling transformation on the total environmental load using a defined scoring function. In one embodiment, the server uses a linear scoring function, parameterizing the "score reduction rate per unit emission" and the "maximum score for zero emissions" to map emissions to scores using a continuous function. In another embodiment, the server uses logarithmic or piecewise functions to impose stronger score penalties on areas with higher emissions. When calculating scores, the server utilizes a numerical computation library to simultaneously perform function operations on environmental load vectors from multiple users and multiple time periods, reducing redundant computation overhead and thus improving overall computation speed.

[0098] When performing time-period statistics, the server groups activity data in two ways: by user identifier and by time dimension. Within each group, the server calculates statistics such as total emissions, travel emissions, energy consumption emissions, average daily emissions, maximum daily emissions, and minimum daily emissions within the period. It also calculates the corresponding average carbon neutrality score, highest score, and lowest score within the same group. Furthermore, the server calculates the rate of change between adjacent periods, such as the percentage change in total emissions and the percentage change in average score between this month and last month, thus providing usable features for constructing subsequent prompts.

[0099] When interacting with the generative artificial intelligence model, the server uses a generation interface program to generate prompts based on the current user's statistical results and environmental load. In a specific example, the server generates the following Chinese prompt: "Based on the following data, please generate a concise Chinese description for employee A, explaining their carbon neutrality score and providing three specific emission reduction recommendations."

[0100] Total CO2 emissions: 22 kg; of which commuting emissions: 2 kg, office energy consumption emissions: 20 kg; Carbon neutrality score: 85 points; an improvement from last week's score of 80 points.

[0101] Please describe in an encouraging and specific tone, and keep it within 300 words. In another example, the server generates the following prompt for a comparison of commuting methods: "Employee A currently commutes by private car, with a total round-trip distance of 40km. The emission coefficient of the private car is 0.15 kgCO2 / km."

[0102] If commuting is changed to tram, the tram emission coefficient is 0.05 kgCO2 / km.

[0103] Please compare the CO2 emissions of the two commuting methods and explain to users in Chinese approximately how much emissions can be reduced each week and month. When generating the prompt, the server inserts data such as total environmental load, component emissions, score, time comparison statistics, and user identifiers into a predefined text template, thus forming a well-structured natural language input. The server then sends the prompt to a generative artificial intelligence model service deployed remotely or locally via a network interface.

[0104] The server uses a transformer-based neural network model in its generative AI model service. During model training, the server inputs a large amount of environmental behavior description text and corresponding data features as training data. During training, the server encodes numerical features such as environmental load vectors, carbon neutrality scores, and time trends into text or additional feature labels, which are then used as part of the model input. Internally, the server employs a multi-layer self-attention encoder and decoder structure, learning the mapping from prompts to natural language description text through sequence-to-sequence transformations. During training, the server uses a cross-entropy loss function to evaluate the difference between the model's output text and manually annotated reference text, updates network weights through backpropagation, and performs iterative training using an adaptive learning rate optimization algorithm. During the inference phase, the server keeps the model parameters fixed, performs only forward propagation, and generates corresponding natural language descriptions based on the input prompts.

[0105] The server can use a separate neural network module or traditional machine learning algorithms for emotion state recognition. In one implementation, the server receives user interaction trajectory data from the terminal, such as click frequency, page dwell time, and emotional features of text input. During the training phase, the server uses these interaction features along with pre-labeled emotion state tags as samples to input into an emotion classification model. The server processes the feature vectors using a multilayer perceptron or convolutional network in this model, outputting a probability distribution representing the emotion category. During inference, the server determines the current user's emotion state based on the highest probability category, such as "positive," "neutral," or "negative." When generating prompts, the server selects different templates or adjusts the tone according to the emotion state; for example, it reduces the proportion of critical language and increases encouraging expressions when the emotion is negative, thereby improving user acceptance.

[0106] To differentiate between real and simulated scenarios, the server adds a simulation flag field to each record. When receiving a hypothetical scenario from the terminal, the server sets this flag to simulation mode and performs calculations only in memory, without persisting it as historical data. The server performs the same environmental load calculation and score generation process for both real and simulated data within the same calculation module, ensuring comparability between the two results. When outputting results, the server places the real and simulated scores in separate fields for the terminal to view in a side-by-side graphical format. This structured output format reduces the complexity of the terminal's combination and inference logic, thereby reducing the terminal's processing burden and improving the overall system's response consistency.

[0107] After receiving the results returned by the server, the terminal maps the data to the input structure of the local visualization component. In one embodiment of the invention, the terminal displays the total score, historical score trend chart, pie chart of travel and energy consumption composition, and natural language descriptions returned by the generative artificial intelligence model in a graphical interface. When a simulation comparison needs to be displayed, the terminal represents the actual score and the simulation score with adjacent bars and adds explanatory text below, such as "By changing the commuting mode from private car to electric vehicle, it is estimated that emissions can be reduced by X kgCO2 per week, and the score can be improved by Y points." The terminal also provides interactive controls on the interface, allowing users to input new assumptions and re-trigger simulation calculations.

[0108] Users can continuously use both real-world recording mode and simulation mode through the terminal in this system. In real-world recording mode, users input daily commuting and energy consumption data, and the server generates official carbon neutrality scores and statistical reports based on this data. In simulation mode, users modify their travel methods, reduce energy consumption, or change their work patterns. The server quickly calculates new environmental loads and scores in memory based on the modified conditions. Users can intuitively compare the differences between the two, thus making more data-driven travel and energy consumption choices in the real world.

[0109] In this invention, the server improves the processing efficiency and result generation speed of large-scale user activity data by combining a unified data structure, vectorized numerical operations, time-series grouped statistics, and a generative artificial intelligence model. In environmental load calculation, the server utilizes pre-loaded emission coefficient tables and batch processing mechanisms to reduce multiple database accesses and redundant calculations, thereby supporting more users simultaneously with the same hardware resources. When generating natural language descriptions and suggestions, the server separates the interface with the generative artificial intelligence model through prompts, allowing the model to evolve independently without affecting the data pipeline, forming a scalable modular internal computer structure.

[0110] After adopting the above structure and algorithm, the server achieves technical improvements in environmental load assessment accuracy, calculation speed, result interpretation capability, and data management compared to traditional solutions that rely solely on manual aggregation or simple script calculations. Internally, the server appropriately utilizes generative artificial intelligence models to semantically expand structured data and generate natural language, combined with emotion recognition and simulation comparison mechanisms. This ensures that the system's behavior is not merely a simple automation of manual operations, but rather reflects an optimization of computer technology itself in terms of computational path design, algorithm structure, and data flow organization.

[0111] use Figure 11 The processing flow is explained.

[0112] Step 1: After launching the application, the terminal reads the user identifier from local storage or session information as input and renders the data input interface on the display device. The terminal generates components within the interface for inputting fields such as travel mode type, travel distance, round trip status, energy type, energy consumption, and date. The terminal performs local validation on each user input (e.g., checking if values ​​are non-negative and if required fields are empty) and temporarily stores valid input as a structured form data structure. When the user clicks the "Submit" button, the terminal uses this form structure as input, serializes it into a request message via the network communication module, and prepares to send it to the server.

[0113] Step 2: On the terminal interface, users are prompted to select their mode of transportation, enter the one-way or round-trip distance, input their office or home energy consumption, and choose the corresponding date. After confirming the input is correct, the user clicks the submit button on the terminal, triggering the generation of a structured request containing user identification, time information, and activity fields. In this step, the user does not perform calculations directly but instead converts real-world behavioral information into digital input through the terminal's interface components, providing raw data for subsequent server processing.

[0114] Step 3: After receiving a user's submission, the terminal combines the local form data with the user and terminal identifiers to form an activity record as input. The terminal encodes this activity record using a network protocol and sends it to the application interface address specified by the server through a secure channel. During transmission, the terminal appends authentication information and content type description to the request header. After completing data transmission, the terminal sends the transmission status as output to the local interface module, displaying a "submission successful" or "submission failed" message.

[0115] Step 4: After receiving a request message from the terminal, the server passes the request content as input to the application service program. The server parses the message within the application, extracting fields such as user identifier, date, mode of transport, distance, and energy consumption. The server performs server-side validation on the parsed fields (e.g., checking field type, range, and missing data), and writes the validated records to the active data table in the database. During the write process, the server generates a unique record identifier and timestamp for each record and returns a successful write message to the terminal. In this step, the server performs data reception, validation, and storage operations, outputting the persisted active records.

[0116] Step 5: When the server needs to perform calculations, it reads activity records from the database that meet specific conditions (such as a user or a time period) as input. The server loads these records into a two-dimensional tabular data structure in memory, where each row corresponds to one activity record and each column represents a field. The server performs preprocessing operations on this data structure, including deleting rows with missing key fields, filling optional fields with default values, converting distances to kilometers, and converting energy consumption to kilowatt-hours. The server outputs the preprocessed, standardized activity data for subsequent environmental load calculations.

[0117] Step 6: When calculating emissions, the server takes the pre-processed activity data table and the pre-stored emission coefficient table as input. In memory, the server matches the travel mode column of the activity data table with the travel mode key in the emission coefficient table, adding a travel emission coefficient column to each activity record. Similarly, the server matches the energy type column with the corresponding energy emission coefficient table, adding an energy consumption emission coefficient column to each record. The server then performs numerical calculations on each record: using "travel distance × travel emission coefficient" to obtain travel emissions, and using "energy consumption × energy consumption emission coefficient" to obtain energy consumption emissions, adding both as new numerical columns to the data table. The server outputs an expanded data table containing travel emissions and energy consumption emissions for subsequent summarization and scoring steps.

[0118] Step 7: When calculating the total environmental load and carbon neutrality score, the server uses the extended data table output in step 6 as input. The server groups the records according to user ID and time period (e.g., day, week, month), and sums the travel emissions and energy consumption emissions within each group to obtain the total environmental load for each group. The server then passes the total environmental load to a predefined scoring function, performing numerical transformations such as "score = max(0, 100 - α × total_emission)" to calculate the carbon neutrality score for different groups in batches. The server records the total environmental load and carbon neutrality score for each time period as output in a new results data table, providing basic data for time series statistics and explanatory data generation.

[0119] Step 8: When performing time-period statistical analysis, the server uses the data table output in step 7 as input. On this data table, the server continues to perform statistical calculations by user and time period, including the average daily emissions, maximum daily emissions, minimum daily emissions, average score, highest score, and lowest score within the calculation period. The server also calculates the difference and ratio between the total emissions and average scores of adjacent periods (e.g., this month and last month) to obtain the rate of change index. The server organizes these statistics into a structured statistical result set as output, containing various statistical characteristics for each user and each period, providing input for subsequent prompt statement construction and visualization.

[0120] Step 9: When preparing to interact with the generative AI model, the server takes total environmental load, carbon neutrality score, and time-period statistics as input. The server fills these numerical and categorical information into a predefined text template to construct prompts for input into the generative AI model. For example, the server can generate the following text prompts as output: "Based on the following data, please generate a concise Chinese description for employee A, explaining their carbon neutrality score and providing three specific emission reduction recommendations."

[0121] Total CO2 emissions: 22 kg; of which commuting emissions: 2 kg, office energy consumption emissions: 20 kg; Carbon neutrality score: 85 points; an improvement from last week's score of 80 points.

[0122] Please describe in an encouraging and specific tone, and keep it within 300 words. The server uses the generated prompts as input data for subsequent generative artificial intelligence model calls.

[0123] Step 10: The server uses user interaction characteristics and identified emotional states as input for emotion recognition and adaptive prompt processing. Based on the emotional state, the server selects different language style templates or adjusts the vocabulary in the prompts; for example, it reduces the proportion of negative wording and increases positive guidance and encouraging expressions when the user's emotion is negative. During this process, the server uses different word combinations on the same set of environmental load and score data to generate multiple candidate prompts. The server selects the prompt that matches the current emotional state as the final output and sends it to the generative artificial intelligence model service, thereby achieving adaptive adjustment of tone and content.

[0124] Step 11: When the server invokes the generative AI model service, it takes the prompt from step 9 or 10 as input and sends it to the inference server endpoint where the generative AI model is deployed via a network communication interface. Upon receiving the natural language text returned by the model, the server outputs this text as an explanation of the carbon neutrality score and behavioral suggestions. In this step, the server does not perform further deep semantic processing on the text; instead, it stores the entire text as a descriptive field in the result data structure, while simultaneously recording the model invocation time, execution time, and invocation identifier for subsequent monitoring and optimization.

[0125] Step 12: When integrating the final results, the server takes as input the environmental load decomposition results, carbon neutrality score, time-period statistics, simulation score (if applicable), and explanatory text output by the generative artificial intelligence model. The server packages this content into a unified response data structure, which includes total emissions, emissions from each source, current score, historical score sequences, statistical indicators, comparison data between real and simulation data, and natural language descriptions. The server outputs this response data structure as a response message to the terminal via the network interface.

[0126] Step 13: After receiving the response message from the server, the terminal uses it as input for the local interface rendering module. The terminal parses the numerical and text fields in the response, displays the carbon neutrality score in the numerical area, and inputs the travel and energy consumption composition into the visualization library as input to a chart component for drawing bar charts, pie charts, or line charts. Simultaneously, the terminal displays the explanatory text returned by the generative AI model in the description area, and when both the actual and simulated scores exist, it inputs their values ​​into a comparison chart component to display the differences using side-by-side bars or curves. The terminal then presents the rendered graphical interface to the user as output, helping the user understand the impact of their behavior on environmental impact and further adjust their behavior or re-enter new assumptions.

[0127] Application Example 1 The process flow corresponding to the specific processing in Use Case 1 will be described below. The various parts of the system described below are implemented by the data processing device 12 and the intelligent device 14. Furthermore, the data processing device 12 is referred to as the "server" and the intelligent device 14 is referred to as the "terminal".

[0128] In industrial production and operations, there is an increasing need to utilize information processing technologies to assess the environmental impact of employee behavior and production facility operations, in order to support energy conservation, emission reduction, and carbon neutrality management. While existing technologies can perform simple statistical analyses on data from single sources (such as employee travel information or energy consumption data from a single unit), they still have significant shortcomings in the following aspects: (1) At the computer system level, employee-related data and production equipment sensor data are often scattered in different systems. The lack of unified data modeling and efficient computing mechanism leads to a fragmented environmental load assessment process. It is impossible to achieve integrated quantitative evaluation of "employees-equipment-organization" in the same data processing pipeline, which increases the complexity of system integration and operation and maintenance costs.

[0129] (2) Traditional systems mostly use fixed rules or preset reports to display data. The computer is only used as a passive statistical tool. It lacks a mechanism to automatically generate targeted optimization suggestions using generative artificial intelligence models. It cannot convert large-scale time series data and historical behavior data into directly executable optimization schemes, resulting in low data utilization.

[0130] (3) In terms of user interaction, existing systems usually only return numerical indicators or simple alarms. They do not define the automatic construction logic of "prompt statements" and the interface protocol with generative artificial intelligence models at the system level. This results in the need to manually write prompt texts, making it impossible to form a stable and scalable prompt statement generation process. It is also difficult to realize repeatable and traceable optimization suggestion generation and iteration within the computer.

[0131] (4) In terms of closed-loop control, existing systems generally do not feed back feedback information such as whether users adopt suggestions or not, and users' operation history to the calculation logic of environmental load score. The computer system lacks a mechanism to adaptively adjust model parameters and scoring conditions based on actual adoption results, and cannot build an automatic cyclic processing flow of "plan-execution-evaluation-improvement", thus limiting the system's self-optimization capability.

[0132] (5) For scenarios involving multiple employees, multiple organizational units, and multiple devices, existing technologies rely heavily on temporary implementations of front-end applications for comparing scores, generating rankings, and constructing gamified elements within the computer. They lack a framework for unified data aggregation, evaluation information generation, and game element definition on the server side, which is not conducive to expansion and maintenance in large-scale environments.

[0133] Therefore, a technical solution that improves the computer system level is needed to enable the server to uniformly access employee data and device sensor data, automatically generate prompts and optimization solutions using data analysis components and generative artificial intelligence models, and integrate user feedback into the scoring algorithm, thereby achieving overall optimization of environmental load management in terms of software architecture and data processing flow.

[0134] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Application Example 1 is achieved by the following means.

[0135] In this invention, the server includes: a data acquisition and calculation unit for acquiring employees' planned and actual information, and quantifying environmental load based on travel mode, travel distance, device usage time, and energy consumption contained in the planned and actual information, and calculating an environmental load score for an individual or organization; a time-series data processing unit for acquiring time-series data on operating status and energy consumption from detectors installed on production devices, and calculating the environmental load score of each device in real time based on the time-series data; a scoring management unit for storing the environmental load scores of employees and each device, and summarizing the environmental load scores using statistical methods within a predetermined summarization period to calculate a periodic score; and a scoring management unit for calculating the environmental load score based on the environmental load scores of employees and each device. The system uses environmental load scores and periodic performance data to generate high-scoring plans and candidate device operation modes that can reduce environmental load. It provides prompts explaining these candidates to a generative artificial intelligence model (AI) and includes a prompt generation and model interface unit. An optimization scheme generation and distribution unit generates improved plans to optimize employee behavior and device operation conditions based on the natural language response of the AI ​​model to the prompts, and sends these improved plans to user terminals. A feedback learning and loop control unit automatically and continuously executes a plan-execution-evaluation-improvement cycle, incorporating the adoption of the improved plans into the calculation of the environmental load score based on user selection results and user operation history obtained from the user terminals. This creates a complete data processing chain on the server side, from multi-source data acquisition, unified quantitative calculation, automatic prompt construction, generative AI model invocation, optimization scheme generation and distribution, to user feedback feedback. This automates and closes the environmental load evaluation and optimization process, thereby improving the integration, scalability, and adaptive optimization capabilities of environmental management-related data processing at the computer technology level.

[0136] "Employee planned and actual information" refers to a set of data submitted by users through terminals or other input devices and received by the server, representing the employee's planned activities and actual activities within a specific time period, including but not limited to travel time, origin and destination, mode of travel, travel distance, planned device usage time, actual device usage time, and corresponding energy consumption information.

[0137] "Mode of travel" refers to the category of transportation or form of movement used by employees during commuting, business trips or other travel, including but not limited to walking, cycling, public transportation, rail transportation, motor vehicle transportation and combinations thereof.

[0138] "Travel distance" refers to the spatial displacement length of an employee during a predetermined time period due to commuting, business trips, or other mobile activities. It is usually expressed in length units (such as kilometers) and can be input by the user or calculated by positioning devices and sensor systems.

[0139] "Device usage time" refers to the cumulative time that an employee or system uses a device in a working or available state within a predetermined observation period. It is usually expressed in time units (such as minutes or hours) and can be calculated from log records, timing modules, or sensor data.

[0140] "Energy consumption" refers to the amount of various types of energy consumed by employee activities or equipment operation within a predetermined time period, including but not limited to electrical energy consumption, fuel consumption and other forms of energy consumption, usually measured in units such as energy, volume, or mass.

[0141] "Environmental load" refers to the quantitative representation of the impact of employee behavior and equipment operation on the environment within a predetermined period of time. It is usually calculated based on energy consumption, emission factors and related environmental indicators, and can be reflected as greenhouse gas emissions, pollutant emissions or their comprehensive indicators.

[0142] "Environmental load score" refers to a standardized evaluation value obtained by the server based on the environmental load calculation results and converting the environmental load according to the preset scoring rules. It is used to represent the degree of environmental impact of employees or equipment within a specific period, and is usually expressed within a predetermined score range (e.g., 0 to 100).

[0143] "Production equipment" refers to equipment or systems that perform processing, assembly, handling, testing, and other operations in the production or service process, including but not limited to industrial robots, machine tools, conveying equipment, production line units, and their auxiliary equipment.

[0144] "Detector" refers to a sensing or measuring unit configured on a production device or its power supply line to acquire data related to the device's operating status and energy consumption, including but not limited to electricity meters, power sensors, current sensors, position sensors, status sensors, and interface modules for acquiring the data from the aforementioned sensors.

[0145] "Time series data" refers to a sequence of multiple data records recorded in chronological order that reflect the operating status and energy consumption of production facilities. Each record includes at least a timestamp and one or more measurements or status values ​​associated with that point in time.

[0146] "Period score" refers to the summary score obtained by the server within a predetermined summary period (such as daily, weekly, monthly or custom period) by statistically processing the environmental load score (including summation, averaging, weighting or other statistical operations) to evaluate the environmental performance within that period.

[0147] "High-Score Plan" refers to a behavioral plan or business arrangement scheme that is estimated by the server based on the employee's historical environmental load score, equipment operation score and periodic performance, and can be expected to achieve a higher environmental load score under predetermined conditions. This includes, but is not limited to, adjustments to travel plans, adjustments to work methods and optimization of time arrangements.

[0148] "Device operation mode candidates" refers to multiple optional combinations of operating parameters or control strategies generated by the server based on the operating data and environmental load score of the production device, in order to reduce the environmental load. These include, but are not limited to, speed mode, load level, power on / off time period, standby strategy, and path planning.

[0149] "Prompt statements" refer to texts automatically generated by the server and provided to generative artificial intelligence models. These texts describe the input data, task requirements, and constraints in natural language or structured text, and are used to guide the generative artificial intelligence models to output corresponding analysis results or optimization suggestions.

[0150] "Generative artificial intelligence models" refer to artificial intelligence models that run on servers or external computing resources, are trained based on machine learning, and are capable of automatically generating natural language text or structured output based on input prompts, including but not limited to text generation models based on deep learning.

[0151] "Natural Language Response" refers to the answers, suggestions, analysis results, or other text content expressed in human natural language that a generative artificial intelligence model outputs after receiving a prompt, which is then parsed by the server and converted into an improvement plan.

[0152] "Improvement plan" refers to a set of specific implementation suggestions or operational steps generated by the server based on the natural language response output by the generative artificial intelligence model and information such as environmental load score, used to optimize employee behavior plans and device operating conditions.

[0153] "User terminal" refers to an electronic device that allows users to view environmental load-related information, receive improvement plans, and perform interactive operations, including but not limited to mobile terminals, fixed terminals, wearable terminals, and the applications or web interfaces running on them.

[0154] "User selection results" refers to the results of user actions on the user terminal regarding the high-scoring plans, device operation mode candidates, or improvement schemes provided by the server, such as adoption, rejection, partial adoption, or parameter adjustment. These results are collected by the user terminal and sent to the server.

[0155] "User operation history" refers to the collection of operation records performed by users on the user terminal for the environmental load management function, including but not limited to scheme browsing records, click records, confirmation or cancellation operation records, and parameter change records, which are used by the server for behavior analysis and model parameter updates.

[0156] "Calculation conditions for environmental load score" refers to the set of various algorithm parameters, weight coefficients, threshold settings, benchmark values, and configurations related to data preprocessing and scoring rules used by the server in the process of calculating environmental load score, which can be adjusted according to feedback information.

[0157] "Plan-Execute-Evaluate-Improve Cyclic Processing" refers to a closed-loop data processing flow that the server executes periodically over time. This includes generating a plan, acquiring or guiding actual execution data according to the plan, evaluating the environmental load on the execution results, and repetitively improving subsequent plans and calculation conditions based on the evaluation results and user feedback.

[0158] "Evaluation information" refers to a set of data generated by the server after comparing and analyzing the environmental load scores and periodic performance of employees or organizational units. It is used to represent the performance status, including but not limited to ranking information, goal achievement, trend analysis, and related explanatory text.

[0159] "Information providing device" refers to an electronic device or application system that receives evaluation information sent by a server through communication and presents the information to a target object, including but not limited to message push systems, notification service systems, email systems and their clients.

[0160] "Game elements" refer to the structured elements defined by the server to introduce gamification mechanisms to promote behavioral change, including but not limited to obtaining scores, levels, rankings, badges, virtual rewards, mission objectives, and related rules, which are used to construct game-like feedback based on environmental load scores and evaluation information.

[0161] "Game-style feedback information" refers to natural language text or structured content output by generative artificial intelligence models based on prompts provided by the server that define game elements. This content presents environmental load performance and behavioral suggestions in a gamified manner and is displayed to users in the form of leaderboards, tasks, reward prompts, etc. on the user terminal interface.

[0162] In an embodiment of this invention, the server is deployed in a data center or factory field server room. The server includes a multi-core central processing unit, main memory, non-volatile storage, and network interface devices. The server runs a general-purpose operating system, such as a Unix-like operating system, and installs data processing software components, including an interpreted language runtime environment (e.g., Python runtime environment), data analysis libraries (e.g., Pandas, NumPy), a database management system (e.g., a relational database system), a web service framework (e.g., an application framework based on the HTTP protocol), and an interface module for calling generative artificial intelligence models. The terminal includes a mobile terminal, a fixed terminal, or an industrial terminal. The terminal runs an operating system (e.g., a mobile operating system or a desktop operating system) and a human-machine interface program or browser for interacting with the server. Users input data related to employee behavior and equipment operation to the server through the terminal and receive environmental load scores and improvement plans output by the server.

[0163] In this implementation, the server manages various data table structures through a database management system. When storing employee-related data, the server uses a relational schema, including tables for basic employee information, planned employee information, actual employee information, and environmental load scores. The planned employee information table includes at least the following fields: employee ID, date, mode of transportation, planned travel distance, planned device usage time, and estimated energy consumption. The actual employee information table includes at least the following fields: employee ID, date, actual mode of transportation, actual travel distance, and actual energy consumption. The server employs transaction mechanisms and index structures when writing data to ensure low latency data insertion and retrieval even with large datasets, thereby improving the overall processing speed of environmental load calculations.

[0164] When processing sensor data from production units, the server employs a time-series data structure for management. The server creates a time-series table for each unit in the database, including fields such as timestamp, instantaneous power, periodic energy consumption, operating status, and operating mode. The server uses a timestamp-sorted index, enabling efficient execution of time-window-based queries and aggregation operations. After batch reading time-series data from the database using the Pandas library, the server loads it into in-memory DataFrames and performs vectorized operations using NumPy to perform data processing operations such as power integration, energy consumption accumulation, and idle time ratio calculation. Because vectorized processing avoids record-by-record interpretive loop operations, the server can significantly improve the throughput of energy consumption calculations and environmental load score calculations, while reducing CPU usage and response time.

[0165] In calculating environmental load scores, the server employs a multi-level scoring model. First, the server calculates a baseline environmental load value based on energy consumption and emission factors. Emission factors can be stored in a parameter table, with different coefficients set according to energy type and region. Then, the server normalizes the baseline environmental load value into a standardized score using a pre-defined function, such as logarithmic transformation or piecewise linear mapping, to ensure a more even distribution of scores within the 0-100 range, facilitating comparison and visualization. The server maintains scoring results at the employee and device levels separately, and performs statistical analysis based on daily, weekly, or monthly aggregation windows. During aggregation, the server uses a sliding window algorithm and incremental update strategy to avoid repeatedly scanning complete historical data. This allows the server to process only newly added data records during score updates, significantly reducing computational complexity and storage access load.

[0166] When generating prompts, the server structures and organizes the raw data and intermediate calculation results, then automatically constructs natural language text. In one implementation, the server generates prompts using a template-plus-rules approach: the server maintains multiple templates in storage, each containing placeholders for variables such as employee identification, time range, energy consumption data, and environmental load score. The server selects the appropriate template based on the specific scenario, such as a "Employee Commuting Optimization Scenario" template or a "Production Unit Energy Consumption Optimization Scenario" template. After determining key parameters through data analysis, the server fills in the templates to generate complete natural language prompts.

[0167] For example, in an employee commuting optimization scenario, the server can generate the following prompt: "You are an environmental management consultant. Based on the following employee commuting data, please design a more environmentally friendly commuting solution."

[0168] Current data: Employee ID: E12345 - This week's commute log: - Monday: 20 km drive - Tuesday: 20 km drive - Wednesday: 20 km drive - Thursday: 20 km drive - Friday: 20 km drive Public transportation at the company's location: There are rail transit and buses. The commute time is approximately 40 minutes, and the one-way distance is 20 kilometers.

[0169] Please complete the following tasks: 1. Estimate the environmental load score (0-100 points) of the employee's current commuting method.

[0170] 2. Based on low-carbon modes such as public transportation and cycling, three alternative commuting schemes are proposed, and the environmental load score of each scheme is estimated.

[0171] 3. Provide concise steps to explain how employees can implement the highest-scoring solution. In the scenario of energy consumption optimization in production facilities, the server can generate the following prompt statement: "You are an industrial energy efficiency optimization expert. Below is the operating data for robot A for the most recent day:" - Total energy consumption: 150 kWh - Running time: 10 hours, including 6 hours in high-speed mode, 3 hours in standard mode, and 1 hour in standby mode. - Idle running time: 2 hours (mainly occurring during round trips between workstation 1 and workstation 3) - Production volume: 500 units completed Please complete the following tasks: 1. Calculate the environmental load score (0-100 points) for robot A based on the above data, and briefly explain the calculation basis.

[0172] 2. Analyze the main reasons for the high energy consumption.

[0173] 3. Propose three action-level optimization suggestions that can be directly implemented (e.g., adjusting the path, changing energy-saving mode parameters, merging processes, etc.), and estimate the potential score improvement after implementation. In one specific implementation, the server interacts with the generative artificial intelligence model service via a network interface. Preferably, the server employs a sequence-to-sequence generation model based on a transformer structure, and the generative artificial intelligence model is built upon a multi-layer attention network. This generative artificial intelligence model includes a word embedding layer, a multi-layer self-attention encoding layer, a multi-head attention decoding layer, and an output probability layer. During the training phase, the model is pre-trained using a large amount of environment-management-related corpus and general language corpus, and then fine-tuned using small-scale environmental load optimization task data to enhance the generation quality on specific tasks. When the server invokes the model, it converts the prompt statements into a labeled sequence as the model's input sequence and sets generation parameters such as decoding length, temperature parameters, and sampling strategies to control the diversity and stability of the output text.

[0174] During model training, the server uses cross-entropy loss as the error function and employs gradient descent-based optimization algorithms (such as adaptive moment estimation) to update the model parameters. The server uses batch training and mitigates overfitting through gradient clipping, gradient explosion prevention, and regularization. In the data preprocessing stage, the server can also use data augmentation techniques, such as paraphrasing and word order transformation, to generate equivalent suggestions and improve model robustness. Through these methods, the generative AI model can generate semantically coherent and actionable improvement suggestions in a relatively short time when faced with data from different employees, devices, and timeframes.

[0175] When generating improvement proposals, the server further performs structured parsing of the natural language responses from the generative AI model. Based on predefined key fields and trigger words, the server extracts the suggested items, estimated scores, and implementation steps from the response text into structured records, which are stored in a suggestion data table. When necessary, the server can also perform constraint checks on the suggested content using a rules engine, such as filtering out operational instructions that do not comply with safety rules or production constraints. When presenting the improvement proposals to users through a terminal, the terminal converts this structured information into graphical interface elements, such as card lists and table views, allowing users to view the estimated environmental load score, expected energy savings, and specific implementation steps for each proposal.

[0176] After receiving the improvement proposals, users can interact with them through the terminal. The terminal provides options for each improvement proposal, such as adoption, partial adoption, or rejection, and allows users to adjust certain parameters, such as the number of days per week using low-carbon commuting methods and the activation time periods of energy-saving modes for production equipment. The terminal encodes the user's selections as operation records and sends them to the server via the network interface. After receiving the user's selections, the server writes them to the user's operation history table and adjusts the scoring criteria in the environmental load score calculation module. For example, it may increase the weight of behaviors corresponding to adopted energy-saving measures, or differentiate between baseline data for "optimized proposals" and "non-optimized proposals" during trend analysis.

[0177] After adopting the structured feedback mechanism described above, the server can dynamically update parameters in subsequent calculation rounds based on user adoption. For example, the server can automatically estimate the effectiveness of a certain type of suggestion by comparing energy consumption changes before and after implementation, and store this as a weighting coefficient in the parameter table to influence the priority of such solutions in subsequent recommendations. In this way, the server builds a self-learning rule adjustment mechanism within the computer, eliminating the need for manual adjustments to evaluation parameters item by item, thereby achieving continuous optimization of the environmental load scoring logic for a large number of employees and devices. This mechanism not only improves the accuracy of the scoring model but also reduces maintenance workload.

[0178] In terms of data structure and module partitioning, the server in this implementation divides its functions into multiple logical modules, including a data acquisition and storage module, an environmental load calculation module, a statistics and ranking module, a prompt statement generation module, a generative artificial intelligence model interface module, an improvement scheme generation module, and a feedback learning and loop control module. The server uses message queues or asynchronous task scheduling mechanisms to pass intermediate results between modules, improving system throughput and avoiding blocking. Since environmental load calculation and generative artificial intelligence model invocation typically involve large computational loads, the server can achieve horizontal scaling through multi-threaded or multi-process parallel processing, or by distributing computational tasks to multiple computing nodes, thus maintaining low response latency even in large-scale deployment scenarios.

[0179] In this embodiment, the terminal is not merely a simple display device; it also undertakes data preprocessing and interaction optimization functions. For example, when collecting user input, the terminal can perform local format validation and basic encoding to reduce server-side error handling overhead, and can merge multiple small requests through batch submission strategies, thereby reducing the number of network requests and communication load. During interface rendering, the terminal can perform local interpolation and smoothing on the time-series score data provided by the server, improving the smoothness of the graphical display without altering the accuracy of the original data on the server. Through this division of labor, the overall communication efficiency and user experience of the system are improved.

[0180] In practical applications, users can use the system of this invention to technically control real-world devices. For example, users can select "Enable energy-saving mode and extend standby threshold time" on the terminal based on the energy-saving operation mode recommended by the server. After receiving this selection, the server can send the corresponding control parameters to the field control device via an industrial communication protocol, and the field control device will then execute the mode switch. Because the server comprehensively considers multi-dimensional data such as output, idle time ratio, and peak power when generating suggestions, and forms nonlinear, multi-condition optimization decisions with the support of a generative artificial intelligence model, this invention can significantly reduce energy consumption and environmental impact while ensuring production capacity, compared with traditional systems that rely solely on human experience or simple rules.

[0181] In this technical solution, the server improves the efficiency and accuracy of computers in large-scale, multi-source data processing by using a unified data structure, vectorized operations, sliding window statistics, and self-learning scoring rules. Through the tight coupling of templated prompt generation with a generative artificial intelligence model, it achieves the ability to automatically construct high-quality prompts and generate executable optimization solutions, avoiding the inefficient practice of manually writing prompt text. By incorporating user feedback into the scoring and recommendation logic, a computer system that can adaptively adjust to the evolving operating environment and user behavior is constructed. Therefore, this invention is not merely a simple automation of human workflows, but rather represents a substantial technical improvement at the levels of computer architecture, data processing flow, and intelligent decision-making mechanisms.

[0182] use Figure 12 The processing flow is explained.

[0183] Step 1: The terminal obtains employee planned information and actual information as input, and performs local validation and preprocessing on the input.

[0184] The terminal displays input controls on the user interface for entering dates, modes of transportation, travel distances, device usage time, and energy consumption. Users enter or select relevant fields sequentially on the terminal.

[0185] The terminal performs format and range checks on the input data, such as checking whether the date format is correct, whether the travel distance is non-negative, and whether the travel mode is in the preset list.

[0186] After successful verification, the terminal encapsulates each field into a structured data object and generates request data containing employee identifier, timestamp, plan information, and actual information, which is then sent to the server as output.

[0187] Step 2: The server receives the planned and actual information sent by the terminal as input and writes it into the database.

[0188] The server receives request data from the terminal through the network interface and parses the fields in the request body, including employee ID, date, mode of travel, travel distance, device usage time, and energy consumption.

[0189] The server constructs insert statements or data records based on the parsing results, writes employee plan data into the employee plan information table, and writes actual employee data into the employee actual information table.

[0190] After a successful write operation, the server returns an acknowledgment response to the terminal, outputting the newly added record ID and status information in the database.

[0191] Step 3: The server reads employee planned and actual information from the database as input and calculates the employee's baseline environmental load value and environmental load score.

[0192] The server uses query statements to extract fields such as travel mode, travel distance, device usage time and energy consumption from the employee planned information table and the employee actual information table based on employee identifier and time range, forming a dataset.

[0193] The server performs data preprocessing on the dataset, including unit standardization, missing value imputation, and outlier filtering. Then, it maps different modes of travel and energy types to their corresponding emission coefficients according to a pre-stored emission factor table.

[0194] The server calculates commuting emissions by multiplying travel distance by emission coefficients through numerical calculations, calculates device energy consumption emissions by multiplying device usage time by power parameters and emission factors, and adds up the emissions from each part to obtain the basic value of environmental load.

[0195] The server further applies normalization and scoring functions to map the baseline environmental load value to a preset score range, obtaining the employee's environmental load score as the output, and writing the score into the employee environmental load score table.

[0196] Step 4: The server obtains sensor data from the production unit from the time series table as input and calculates the unit's environmental load score.

[0197] The server reads data such as instantaneous power, periodic energy consumption, operating status, operating mode, and timestamps from the device time series table based on the device identifier and time window.

[0198] The server sorts the time series data by time and calculates the total energy consumption, effective running time, idle time ratio, and the proportion of each operating mode within the time window.

[0199] The server multiplies the total energy consumption by the grid emission factor to obtain the device emissions, and then weights the emissions according to the proportion of idle time and operating mode to reflect the additional environmental burden of inefficient operation.

[0200] The server applies a scoring function to the adjusted emissions, generates an environmental load score for the device as output, and records the score in the device environmental load score table.

[0201] Step 5: The server uses employee environmental load scores and device environmental load scores as inputs to calculate periodic scores and statistical information.

[0202] The server groups and aggregates records from the employee environmental load score table and the device environmental load score table according to a predetermined aggregation period (e.g., daily, weekly, or monthly), and calculates the average score, highest score, and improvement trend for each employee and each device within that period.

[0203] The server uses a sliding window or incremental update method to calculate only the newly added time period data, avoiding duplicate processing of historical intervals and thus improving statistical efficiency.

[0204] The server writes the calculated periodic scores into the periodic score table and generates statistical results including the periodic average score, ranking, and historical trends, which serve as input and output for subsequent recommendation and suggestion statements.

[0205] Step 6: The server takes the environmental load scores and periodic performance of employees and equipment as inputs to generate high-scoring plans and candidate equipment operation modes.

[0206] The server analyzes environmental load scores for different time periods, travel modes, and operating modes to identify behavioral patterns and combinations of operating parameters that can significantly reduce emissions.

[0207] Based on the analysis results, the server constructs several feasible high-scoring plan candidates, such as replacing self-driving travel with public transportation, concentrating high-load operation of the device during specific periods, and enabling energy-saving mode.

[0208] The server estimates the potential environmental load score for each candidate solution after implementation and saves the candidate solution and its estimated score as a candidate plan record, which serves as the input and output for generating prompts and optimization suggestions later.

[0209] Step 7: The server takes candidate plans and related data as input and generates prompts for generative artificial intelligence models.

[0210] The server reads employee IDs, time ranges, current environmental load scores, candidate travel mode combinations, device operating parameters, and relevant constraints from the candidate plan records.

[0211] The server selects a prompt template that corresponds to the scenario and fills the placeholders in the template with the above data to form a complete natural language text.

[0212] Through this data filling and combination process, the server generates one or more prompt statements as output, which describe the current situation, task requirements, and expected output, and are prepared to be provided to the generative artificial intelligence model.

[0213] Step 8: The server takes the generated prompt as input, calls the generative artificial intelligence model, and obtains a natural language response.

[0214] The server encodes the prompt statement into text input and sends it to the generative AI model service deployed locally or remotely through the model interface module, setting parameters such as decoding length, generation temperature, and sampling strategy.

[0215] The server waits for the model to complete inference and then receives the natural language response text returned by the model. The response includes an environmental load score estimate, cause analysis, and specific improvement suggestions.

[0216] The server outputs the received natural language response and stores it in a response record table, providing a data foundation for subsequent structured parsing and the generation of improved solutions.

[0217] Step 9: The server takes the natural language response of the generative artificial intelligence model as input, parses it, and generates a structured improvement plan.

[0218] The server performs text analysis on the natural language response, identifies key information such as suggested items, execution steps, estimated scores, and implementation conditions, and maps them to structured fields according to predefined rules.

[0219] The server organizes the structured information obtained from the analysis into improvement plan records, including plan number, applicable objects, step list, expected environmental load score, and implementation priority.

[0220] The server records the improvement solutions in an improvement solution table and outputs them for use in pushing to the terminal and subsequent feedback learning processing.

[0221] Step 10: The server takes the improvement plan record and the target employee or device identifier as input and sends the improvement plan to the terminal.

[0222] The server selects solutions from the improvement solution table that are relevant to the user or device based on the organization to which the user or device belongs and the user's permissions.

[0223] The server converts the selected solutions into a data format suitable for terminal display, including a title, description, list of steps, and estimated score, and sends it to the terminal via a network interface.

[0224] After sending the message, the server records the push time and the receiving terminal identifier as output in the log so that it can track the propagation path of the solution and the user response status later.

[0225] Step 11: The terminal takes the improved solution data sent by the server as input, presents the solution content in the user interface, and receives the user's selection results.

[0226] Based on the received data, the terminal generates solution cards or lists on the interface, displaying the name of each improvement solution, the expected environmental load score, and key implementation steps.

[0227] The terminal provides interactive controls for each solution, allowing users to choose "accept", "partially accept" or "reject", and allows users to modify some parameters, such as execution frequency and start time.

[0228] After the user completes the selection, the terminal combines the user's selection result and the modified parameters into feedback data, which is then sent to the server as output.

[0229] Step 12: The server takes the user's selection results and user operation history as input, updates the environmental load score calculation conditions, and records the feedback.

[0230] After receiving the feedback data from the terminal, the server writes the user's adoption mark and modified parameters into the user operation history table and establishes a link with the corresponding improvement plan record.

[0231] The server analyzes the adoption rate and actual energy consumption changes under different users and scenarios, maps the adoption status to the validity weight of a certain type of suggestion, and updates the relevant parameters in the environmental load score calculation module, such as adjusting the weight or baseline value of certain behaviors.

[0232] The server outputs updated scoring parameters and recorded feedback data, enabling subsequent environmental load score calculations and recommendation generation to be adaptively adjusted based on the latest feedback results.

[0233] Step 13: The server takes the updated scoring parameters and new behavioral data as input and performs a new round of environmental load score calculation and periodic performance statistics.

[0234] Based on the updated calculation conditions, the server extracts data for the new time period from the employee actual information table and the device time series table, and recalculates the environmental load baseline value and environmental load score.

[0235] The server uses an incremental update strategy during the calculation process, applying the latest scoring parameters to the data of newly added time periods for data processing and numerical calculation, thereby avoiding repeated calculations of complete historical data.

[0236] The server writes the new scores and periodic statistics into the corresponding data table, which serves as the output to provide the latest data for subsequent prompt generation, improvement plan generation, and user display, thus realizing a cyclical process of planning-execution-evaluation-improvement.

[0237] Alternatively, an emotion engine for inferring user emotions can be combined. That is, the specific processing unit 290 can also use the emotion-specific model 59 to infer user emotions and perform specific processing using user emotions.

[0238] Example 2 The flow of a specific process in Example 2 will be described. Each part of the system described below is implemented by the data processing device 12 and the smart device 14. The data processing device 12 will be referred to as the "server," and the smart device 14 as the "terminal."

[0239] While various information processing systems exist for assessing environmental impact or carbon emissions, the following technical problems remain: First, traditional systems often only perform simple summation or threshold judgments on environmental data at the server side, lacking a unified and scalable environmental load indicator modeling and scoring mechanism. This makes it difficult for the system to make refined comparisons and rankings among multiple behavioral options, and the calculation results are difficult for users to understand intuitively. Second, traditional systems typically present results to users in fixed-format reports or static charts, lacking deep integration with generative artificial intelligence models. They cannot dynamically generate personalized, multi-layered explanatory texts based on users' natural language questions, making it difficult for users to make quick behavioral decisions based on complex environmental data. Third, existing systems often only perform statistical analysis for single users or single scenarios, lacking unified management and comparative display of multi-granularity environmental load indicators for individual units, group units, and regional units. They cannot fully utilize historical data for time series analysis and incentive mechanism design, making it difficult to form a feedback loop for continuous improvement at the organizational or societal level. Fourth, existing systems typically fail to organically integrate environmental load indicators with gamification elements (such as rankings, achievement status, and reward information), and they also fail to transform these gamification elements into natural language content with motivating effects through generative artificial intelligence models, resulting in insufficient user engagement and long-term sustainability.

[0240] Therefore, it is necessary to provide a new system and its implementation method to improve from the perspective of computer technology: to achieve efficient data processing of environmental load indicators and scalable scoring and comparison mechanisms on the server side; to build a collaborative interface between the system and generative artificial intelligence models based on structured data and prompts; and to conduct statistical analysis and gamified presentation of historical data at the individual and group levels, thereby improving the technical effectiveness of the entire environmental behavior recommendation and incentive process in terms of computing performance, data expression ability, interaction flexibility, and user participation.

[0241] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Embodiment 2 is achieved by the following means.

[0242] In this invention, the server includes means for acquiring information related to behavioral options and calculating environmental load index values ​​for each behavioral option based on the degree of environmental impact; means for comparing the calculated environmental load indices and determining the optimal behavioral option; means for generating structured data containing the optimal behavioral option and its corresponding environmental load index, and combining the structured data with natural language prompts from the user to form generation instruction information as prompts for a generative artificial intelligence model; means for providing the prompts to the generative artificial intelligence model to instruct the generation of a natural language response containing recommended content of the optimal behavioral option and explanatory text based on the environmental load index; and means for... The system includes a device for sending natural language responses and optimal behavior options to a terminal device for display, and a device for storing historical information related to behavior options and performing time-series aggregation of environmental load indicators using statistical methods. It may also include a device for calculating environmental load indicators by individual and group units and generating prompts that instruct the generative artificial intelligence model to output comparative explanatory text and notification text. Furthermore, it includes a device for calculating individual and regional unit evaluation values ​​based on environmental load indicators, generating prompts that instruct the generative artificial intelligence model to output game element information containing ranking information, achievement status information, and reward information, and sending the game element information to the terminal device for display. This allows for efficient calculation and detailed comparison of the environmental load of various behavior options on the server side using a unified data structure and scoring algorithm. By combining structured calculation results with natural language prompts, the system drives the generative artificial intelligence model to generate personalized and interpretable natural language feedback, significantly improving the computational efficiency and expressive power of environmental behavior recommendations. Simultaneously, it forms a closed-loop mechanism based on time-series statistics and gamified incentives at multiple levels—individual, group, and regional—improving the technical performance of the computer system in environmental indicator evaluation, result presentation, and user interaction.

[0243] "System" refers to a collection of devices consisting of at least one server, at least one terminal device, and a network for data communication between the server and the terminal device. This collection is a technical solution for processing and outputting environmental information by collaboratively performing functions such as data acquisition, computation, model invocation, and result presentation.

[0244] A "server" is a computing device equipped with a processor, memory, and communication interface, used to execute program code in a network environment, process behavior option data, calculate environmental load indicators, construct prompt statements and call generative artificial intelligence models, and provide processing results to terminal devices.

[0245] "Terminal device" refers to an information processing device operated by a user to send requests to a server and receive and display the results returned by the server, including but not limited to mobile phones, tablet computers, desktop computers, or other electronic devices with network communication and interface display functions.

[0246] "Behavioral options" refer to multiple executable action candidates that users or groups can choose in a specific scenario. Each behavioral option will generate corresponding resource consumption and environmental impact when executed.

[0247] "Environmental impact level" refers to a set of quantitative or semi-quantitative indicators used to characterize the extent of environmental impact caused by a certain behavioral option during its implementation. This set can be determined based on greenhouse gas emissions, energy consumption, pollutant emissions, and other environmentally relevant parameters.

[0248] The "environmental load index" is a numerical value or score calculated by the server based on the degree of environmental impact and multidimensional data related to the behavioral options. It is used to quantitatively represent the magnitude of the environmental load for each behavioral option.

[0249] "Environmental load index numerical value" refers to the specific numerical representation of the environmental load index, including scalar value, vector value, or score after normalization and weighting, which is used to compare and rank different behavioral options.

[0250] The “optimal behavior option” refers to the behavior option that best meets the predetermined optimization objective based on the comparison of environmental load index values ​​under at least one evaluation criterion. It is usually the behavior option with the lower environmental load or the highest comprehensive score.

[0251] "Structured data" refers to a collection of data organized in a predefined format, with clearly defined field names and data types, which can be directly parsed and processed by program logic to represent optimal behavior options and their related environmental load indicators.

[0252] "Natural language prompts" refer to textual information written in human-understandable natural language to instruct generative artificial intelligence models to perform specific generative tasks. This includes statements directly entered by the user and internal prompts automatically constructed by the server.

[0253] "Generation instruction information" refers to the comprehensive input information that the server uses to constrain and guide generative artificial intelligence models to generate specific content, based on a combination of structured data and natural language prompts.

[0254] "Generative AI models" refer to AI models that are trained on large-scale datasets and can automatically generate natural language text, structured text, or other forms of output based on input prompts, including but not limited to generative models used for dialogue, summarization, explanation generation, or content creation.

[0255] "Natural language response" refers to the text content that a generative artificial intelligence model outputs after receiving prompts and related data. This text content conforms to the grammatical and semantic rules of natural language and is used to explain or recommend relevant environmental information to the user.

[0256] "Historical information" refers to past data related to behavioral options, environmental load indicators, and user or group selection records within a certain time range, which is used for statistical analysis and time series processing.

[0257] "Statistical methods" refer to mathematical or algorithmic means used to aggregate, analyze, and model environmental load indicators and related data, including but not limited to summation, averaging, analysis of variance, time series analysis, and trend analysis.

[0258] "Time series aggregation" refers to the process of organizing, grouping, and summarizing environmental load index data collected or calculated at different times according to the time dimension, so as to obtain statistical results by day, week, month, or other time granularity.

[0259] "Individual unit" refers to the smallest entity that can be identified and evaluated individually in the system, including but not limited to a single user, a single account, or a single device.

[0260] "Group unit" refers to a subject unit composed of multiple individual units that can be used as a whole for the statistical analysis of environmental load indicators, including but not limited to organizations, teams, or departments.

[0261] "Regional unit" refers to a spatial or logical area defined by geography or business scope, used to statistically compare environmental load indicators at a higher level, including but not limited to cities, industrial parks, or business domains.

[0262] "Evaluation value" refers to a quantitative indicator or score calculated based on environmental load indicators and their statistical results, used to reflect the degree of excellence or inferiority of an individual unit, group unit, or regional unit in terms of environmental performance.

[0263] "Comparative explanatory text" refers to the natural language description content generated by generative artificial intelligence models based on the differences in environmental load indicators of different individual units or groups, used to explain and compare these differences.

[0264] "Notification text" refers to natural language content generated by generative artificial intelligence models based on environmental load indicators and their changes, used to inform users or groups of the current status, trends of change, or precautions.

[0265] "Game element information" refers to data and text information related to gamification mechanisms, constructed based on environmental load indicators and evaluation values, including but not limited to ranking information, achievement status information, and reward information, used to incentivize users to improve their environmental behavior.

[0266] "Ranking information" refers to the results of ranking individual units, group units, or regional units based on evaluation values, which is used to reflect the relative position of different subjects in environmental performance.

[0267] "Achievement Status Information" refers to a quantitative or textual description of the progress and achievement status of an individual or group of units in achieving preset environmental protection goals or tasks.

[0268] "Reward information" refers to incentive content that is linked to environmental performance or evaluation values ​​and is used to encourage individuals or groups to continuously improve environmental load indicators. This includes virtual badges, points, levels, or redeemable benefits.

[0269] "Electronic communication means" refers to communication methods used to transmit data and messages between servers and terminal devices or other systems, including but not limited to email, push notification services, short message services, or other network protocol-based transmission methods.

[0270] In the following description, the server, terminal, and user each play different functional roles. The server is mainly responsible for data processing, model inference, and result generation; the terminal is mainly responsible for interactive display and input collection; and the user mainly makes decisions by inputting natural language prompts through the terminal and referring to the server's output.

[0271] I. Example of System Hardware and Software Composition Servers are implemented using general-purpose computer hardware. A server may include a multi-core central processing unit (CPU), a graphics processing unit (GPU), main memory (RAM), non-volatile storage devices (solid-state drives or hard disks), and a network interface controller. At the operating system level, a server may run a Unix-like operating system, such as a distribution of Linux server operating system.

[0272] The following components can be deployed on the server at the software level: The server uses a scripting programming language runtime environment (such as the Python 3 runtime environment); The server uses relational database management software (such as a general-purpose relational database system) to store data related to behavioral options and environmental parameters; The server uses data processing libraries (such as NumPy and Pandas) to perform matrix operations and tabular data processing in memory; The server uses a web application framework (such as a general-purpose web framework based on the Python language) to implement the application programming interface; The server uses a generative artificial intelligence model inference service, which can run on a local GPU or a standalone inference server and communicate with the main server via HTTP or remote calls.

[0273] The terminal can take the form of a smartphone, tablet, laptop, or desktop computer. It includes a display screen, touch input device or keyboard, network communication module, and processor. At the software level, the terminal can run a mobile operating system or a desktop operating system and can install client applications or access a web interface provided by a server using a browser.

[0274] Users interact with the system through the terminal's user interface. Users input natural language prompts in the terminal interface, select behavior options, or confirm the behavior scenarios automatically extracted by the system, and then send this information to the server via the network.

[0275] II. Server-side functional modules and data structures When implementing the system described in the claims, the server can be logically divided into several modules, each of which is implemented by programs and data structures on the server.

[0276] 1. Behavior Option Data Management Module The server maintains a set of structured records for each behavior option. The server creates at least one table in the database for each behavior option, with each record representing an instance of a behavior option. The server can configure this table with fields including, but not limited to, the following: The server maintains an "option identifier" field in the data table to uniquely identify each action option; The server maintains a "behavior category" field in the data table to indicate whether the behavior belongs to a major category such as commuting, transportation, production activities, or equipment usage. The server maintains a field "emission factor per unit distance" in the data table to represent the greenhouse gas emissions per unit distance for this action; The server maintains a field called "Energy Consumption Factor per Unit Time" in the data table, which represents the energy consumption per unit time for that row. The server maintains a field "typical distance" or "typical duration" in the data table to represent the average distance or average time of this behavior option under normal use scenarios; The server maintains auxiliary attributes such as "cost parameters" and "reliability parameters" in the data table for later expansion into a comprehensive score.

[0277] The server uses the indexing and query optimization mechanisms provided by relational databases to efficiently retrieve the records corresponding to the behavior options after receiving terminal requests, thereby reducing data access latency.

[0278] 2. Environmental Load Calculation Module The server loads multiple action option records read from the database into memory using the Pandas data framework. The server uses NumPy as its underlying computation engine to perform batch operations on multidimensional arrays or vectors. For each action option, the server constructs a multidimensional feature vector, which may include: The server adds a total emissions component to the feature vector, which is obtained by multiplying the unit distance emission factor by the actual distance given by the user or inferred by the system. The server adds a total energy consumption component to the feature vector, which is obtained by multiplying the energy consumption factor per unit time by the actual duration. The server adds time components, cost components, and other numerical attributes related to environmental impact and user experience to the feature vector.

[0279] The server constructs a matrix from the feature vectors of all behavioral options, and then performs a standardization operation on this matrix. For example, the server can perform normalization by subtracting the mean and dividing by the standard deviation on each column (each feature dimension), or perform interval scaling to ensure that the features fall within a predetermined range. The server uses vectorized computation, allowing the CPU to utilize the SIMD instruction set to reduce loop overhead and thus improve computational speed.

[0280] The server can calculate environmental load indicators using either a weighted linear model or a weighted nonlinear model. For example, the server can assign weights to features such as emissions, energy consumption, time, and cost, which can be predetermined by environmental engineers or through machine learning methods. The server then multiplies the feature matrix with the weight vector using matrix multiplication to obtain the environmental load indicator value for each behavior option. Because the server uses NumPy's linear algebra interface, this calculation can leverage an efficient linear algebra library at the underlying level, thereby reducing overall computation time.

[0281] After obtaining the environmental load metric values, the server can further perform a sorting operation, using Pandas' sorting functions to arrange the behavior options from low to high environmental load. The server selects the behavior option with the lowest environmental load as the candidate optimal behavior option and extracts its structured information.

[0282] 3. Historical Information and Time Series Statistics Module The server creates a separate data table in the database for historical information. This table can record user identifiers, behavior option identifiers, timestamps, current environmental load index values, and related contextual information. After selecting a behavior, the server inserts the selected behavior and calculation results into the historical table.

[0283] The server periodically or on demand performs statistical summaries on historical tables. It can group historical records by time granularity (e.g., daily, weekly, monthly) and calculate average environmental load indicators, total emissions, and selection frequency for each time window. Using the database's aggregation query capabilities and Pandas' secondary processing functions, the server performs moving averages, trend fitting, and other processing on the time series data, generating aggregated time series results internally for use by generative artificial intelligence models and front-end displays.

[0284] The server maintains statistical views or materialized views for individual units, group units, and regional units respectively. This allows the server to generate comparison results without having to reconstruct all statistics from the original historical records for each request. Instead, it only needs to update incrementally, thereby reducing the amount of computation.

[0285] 4. Generative Artificial Intelligence Model Module The server deploys a generative artificial intelligence model. This model can employ a neural network architecture based on a multi-layer transformer structure. During model training, the server uses a large-scale corpus and descriptive text of the surrounding environment as training samples. During training, the server employs an autoregressive language modeling objective and uses a cross-entropy loss function to minimize prediction error. During backpropagation, the server uses gradient descent and its variants (such as adaptive learning rate algorithms) to update the model weights. The server can use data augmentation strategies during training, such as synonym substitution and sentence transformation, to improve the model's robustness to different expressions.

[0286] During the inference phase, the server loads the trained model weights into memory and deploys them to the inference engine. The server defines the input interface for the generative AI model, which includes natural language prompts and optional structured data encoding. The server can first perform text encoding on the structured data, such as converting fields like "behavior option name, total emissions, environmental load index, and time" into formatted text entries, and then concatenate them with the user prompts to form a complete prompt.

[0287] The server can design virtual tags or prefixes for different types of prompts within the generative AI model, enabling the model to distinguish between the "user-asked questions" and the "system-structured data." The server explicitly specifies output requirements in the prompts, such as requiring the model to answer in Simplified Chinese, limiting the number of sentences, or emphasizing differences in the interpretable environmental load. Because the server constrains the model beforehand using samples of a similar format during training or fine-tuning, the model can stably generate structured and interpretable natural language responses during inference.

[0288] This combination of structured data and natural language prompts enables the server to present complex numerical calculation results to the generative artificial intelligence model in a semantically clear form when calling the model. This allows the model to accurately reflect quantitative differences while maintaining linguistic fluency, improving the uncertainty brought about by traditional "black box" language generation and enhancing the accuracy and consistency of output interpretation.

[0289] 5. Prompt Statement Generation Module The server automatically generates prompts for generative artificial intelligence models based on the user's original input and calculation results. Users can enter the following text on the terminal: "Generative AI model, please tell me the most environmentally friendly option based on my commuting method choices." After completing the environmental load metric calculation, the server can construct the following internal prompt statement: The user's original prompt was: "Generative AI model, please tell me the most environmentally friendly option based on my commuting method choices." The system obtains the following commuting method information based on the database and calculations: 1. Option A: Total emissions of 500g CO2, commuting time of 25 minutes, environmental load index score of 90.

[0290] 2. Option B: Total emissions 0g CO2, commute time 20 minutes, environmental load index score 100.

[0291] 3. Option C: Total emissions of 2000g CO2, commuting time of 30 minutes, environmental load index score of 40.

[0292] As a generative artificial intelligence model, please use simplified Chinese to tell the user which behavioral option has the best environmental impact, and explain the reason in 1-3 sentences. By using this unconventional method of constructing prompts that combine structured results with user quotes, the server enables generative AI models to consider both user intent and precise numerical results when outputting. This avoids providing vague answers based solely on generalization experience from the corpus, thereby improving the controllability and determinism of the technical solution.

[0293] III. Terminal-User Interaction Patterns The terminal displays a text input box and a list of candidate behavior options related to the current scenario in a graphical user interface. The user manually selects several behavior options on the terminal, such as "tram," "bicycle," and "car" among modes of transportation, and enters a natural language request in the input box, for example: "Generating AI model: I commute 10 kilometers every day, and my options are tram, bicycle, and car. Please tell me which mode of transportation has the best environmental impact and briefly explain why." After the user confirms the input, the terminal transmits the natural language content and selected action options to the server via the network. Upon receiving the server's response, the terminal displays the server's recommended results and explanatory text on the interface, such as "The most environmentally friendly way to commute is by bicycle. Reason: Bicycles produce almost no carbon dioxide emissions during commuting, thus having the least impact on the environment." The terminal can further display statistical charts or gamified interfaces, showing the user's environmental load index trends over a certain period, their ranking within the group, and unlocked reward information. Because the server uniformly encapsulates this information as structured data and natural language text, the terminal can flexibly combine and display these elements at the presentation level without having to repeatedly implement complex computational logic.

[0294] IV. Technical Effects and Improvements in Computer Technology By employing a vectorized numerical computation library and a structured data framework, the server reduces the loop overhead of interpreted languages ​​in large-scale environmental data processing, significantly improving the efficiency of environmental load index calculation and statistical analysis. Through a unified data structure and feature vector construction method, the server allows for dimensional expansion of behavioral options simply by adding new components to the feature vectors and appending corresponding weights to the weight vectors, without requiring significant modifications to the program flow, thereby improving system maintainability and scalability.

[0295] By encoding quantitative results of environmental load indicators into prompts and combining them with user-generated natural language prompts, the server forces generative AI models to align their reasoning with specific numerical data, rather than relying solely on language patterns. This differs from traditional plain text question-answering systems, reducing instances where models "guess" answers based on experience, thereby improving the accuracy and stability of generated instructions.

[0296] The server employs a time-series aggregation strategy in historical data management, compressing previously scattered single-action records into statistical results aggregated by time windows. When subsequently calling generative AI models or transmitting statistical data to terminals, the server only needs to transmit the aggregated results instead of all detailed records, thereby reducing network bandwidth usage and the parsing burden on the terminal side, achieving a reduction in communication load.

[0297] The server establishes a unified evaluation value calculation and game element information generation mechanism at three levels: individual units, group units, and regional units. This enables the system to generate rankings, achievement status, rewards, and other information at different levels, and transforms it into highly interpretable and motivating natural language content through a generative artificial intelligence model. This multi-level, structured calculation and generation process is no longer limited to "digitizing human-made statistical reports," but rather organizes information within the computer using efficient data structures and algorithms, forming a new technological processing method.

[0298] The server employs specific loss functions and optimization algorithms during the training and inference processes of the generative AI model to reduce the deviation between the generated natural language and numerical facts. The server can introduce "numerical consistency" constraints during the fine-tuning phase to align environmental load indicators with the numerical descriptions in the model's output text, further reducing the probability of numerical errors in the model output and thus improving the system's reliability in environmental indicator description scenarios.

[0299] Through the specific hardware configuration, software components, data structures, feature vector design, prompt statement construction methods, and model training and inference strategies described above, the server, terminal, and user work collaboratively in the system of this invention. This not only enables quantitative evaluation and recommendation of environment-related behavioral options, but also demonstrates significant improvements in computational efficiency, data management, model controllability, and communication load. These improvements are improvements to computer technology itself, rather than simple automation of a business process.

[0300] use Figure 13 The processing flow is explained.

[0301] Step 1: The user enters a prompt statement and selects a behavior option on the terminal. Users can input natural language prompts via keyboard or touch on the terminal interface, such as "Generative AI model, I commute 10 kilometers every day, and my options are tram, bicycle, and car. Please tell me which mode of transportation has the best environmental impact and briefly explain why." Users can select or click one or more behavior options on the terminal interface, such as "tram", "bicycle" and "car".

[0302] The terminal encapsulates the user-input prompt text, the list of user-selected action options, and optional additional information (such as commuting distance, time period, and geographical location) into a request data structure.

[0303] Input: Natural language prompts entered by the user, action options selected by the user in the interface, and user-related contextual parameters.

[0304] Output: A locally constructed request data structure (including prompts, a list of action options, and context parameters) ready to be sent to the server.

[0305] Step 2: The terminal sends a request to the server. The terminal establishes a secure connection to the server through the network communication module and sends the requested data structure to the server's application interface using a predefined application layer protocol (such as HTTPS).

[0306] When sending data, the terminal converts the request data into a predefined encoding format (such as UTF-8 text encoding) and packages it into a message body.

[0307] Input: The request data structure generated in step 1.

[0308] Output: A request message sent to the server over the network, which the server can receive from the network interface.

[0309] Step 3: The server parses the request and extracts key parameters. After receiving a request message from the terminal at the network interface, the server uses a web framework at the application layer to parse the HTTP message.

[0310] The server reads the encoded request data from the message body and parses it into an internal data object according to a predefined format.

[0311] The server extracts parameters such as user natural language prompts, behavior option identifiers, commuting distance, time information, and user identifiers from internal data objects.

[0312] The server performs basic text normalization processing on the extracted natural language prompts, such as removing leading and trailing spaces and standardizing the encoding format.

[0313] Input: The request message sent by the terminal.

[0314] Output: A parameter object internal to the server, containing user prompts, a list of action options, and context parameters related to the calculation.

[0315] Step 4: The server reads basic data on behavior options from the database. The server constructs database query conditions based on the extracted list of behavior option identifiers.

[0316] The server accesses a relational database through a database driver and performs query operations on data tables that store basic information about behavioral options.

[0317] The server reads the environmental parameters corresponding to each behavior option from the data table, including fields such as emission factor per unit distance, energy consumption factor per unit time, typical distance or time, and historical statistical parameters.

[0318] The server loads the retrieved row data into an in-memory data table structure (such as a data frame), standardizes field types, and checks for missing or outlier values.

[0319] Input: The list of behavior option identifiers obtained in step 3.

[0320] Output: A collection of underlying data for behavioral options in server memory, in the form of a structured data table with multiple fields.

[0321] Step 5: The server constructs feature vectors and calculates environmental load indicators. The server constructs a feature vector for each behavior option based on the underlying data of the behavior option and user context parameters (such as actual distance).

[0322] For each behavior option, the server performs the following data processing and calculations: The server calculates the total emissions by multiplying the unit distance emission factor by the actual distance to generate the total emissions component; The server calculates the total energy consumption by multiplying the energy consumption factor per unit time by the expected duration to generate the total energy consumption component. The server incorporates time, cost, and other available parameters into the feature vector.

[0323] The server combines the feature vectors of all behavioral options into a matrix and performs standardization or normalization on each dimension to eliminate dimensional differences.

[0324] The server uses a pre-defined weight vector to perform linear or non-linear combination operations on the normalized feature matrix to obtain the environmental load index value for each behavior option.

[0325] Input: Basic data of behavior options obtained in step 4 and context parameters from step 3.

[0326] Output: A structured data table containing the environmental load index values ​​for each behavior option. Each row in the data table corresponds to a behavior option and its calculated index value.

[0327] Step 6: The server compares environmental load metrics and determines the optimal behavior option. The server performs sorting or search operations on the environmental load index data table, arranging the environmental load index values ​​in ascending order.

[0328] The server selects the option with the optimal environmental load index according to predetermined rules, usually the one with the lowest environmental load index value, and marks it as the "optimal behavior option".

[0329] The server extracts the identifier, name, total emissions, total energy consumption, time, cost, and environmental load index value of the optimal behavior option to form an optimal option data object.

[0330] The server can also retain information about other non-optimal options for later comparison or sorting display.

[0331] Input: The environmental load index data table output in step 5.

[0332] Output: Optimal behavior options data object and a fully sorted list of behavior options.

[0333] Step 7: The server constructs prompts for generative artificial intelligence models. Based on the optimal behavior option data object and the sorting results, the server converts the key numerical parameters of each behavior option into natural language description fragments.

[0334] The server combines the user's original prompt statement with these descriptive fragments to form a complete internal prompt statement, which includes information such as the behavior option name, total emissions, and environmental load index score.

[0335] The server explicitly requires the output format of the generative artificial intelligence model in the prompt statement, such as requiring a response in Simplified Chinese, limiting the number of sentences, and specifying the basis for comparison.

[0336] Input: The optimal behavior option data object generated in step 6 and the sorted list, and the original user prompt statement in step 3.

[0337] Output: Internal natural language prompts used to invoke generative artificial intelligence models.

[0338] Step 8: The server invokes a generative artificial intelligence model to generate a natural language response. The server takes the internal prompts as input parameters and sends them to the inference engine where the generative AI model resides through the inference interface.

[0339] The server loads necessary generation control parameters into the inference request, such as maximum generation length, temperature coefficient, and decoding strategy.

[0340] Generative AI models perform forward inference computation on loaded neural network weights to generate natural language response text based on prompts and language patterns trained internally.

[0341] The server receives the generated text from the inference engine and performs necessary post-processing, such as removing extra spaces and checking whether it meets the specified format.

[0342] Input: The internal prompt statement and generation control parameters generated in step 7.

[0343] Output: A natural language response text containing recommendations for the optimal behavior options and explanations of the environmental load indicators.

[0344] Step 9: The server integrates numerical results with natural language responses and generates a response message. The server combines the optimal behavior option data object, the sorted list of behavior options (optional), and the natural language response text output by the generative artificial intelligence model into a unified response data structure.

[0345] The server labels numerical fields (such as total emissions and environmental load index values) and text fields (explanations) in the response data structure so that the terminal can display them in sections.

[0346] The server encodes the response data structure into a format suitable for network transmission and encapsulates it into a response message.

[0347] Input: The structured numerical results retained in step 6 and the natural language response text generated in step 8.

[0348] Output: The response message to be sent to the terminal, containing the recommendation results, detailed numerical values, and explanatory text.

[0349] Step 10: The server sends a response message to the terminal. The server sends the encoded response message back to the requesting terminal via the network communication interface.

[0350] The server sets appropriate response header information before transmission to identify the data type and encoding method.

[0351] Input: The response message generated in step 9.

[0352] Output: The response data packet transmitted to the terminal over the network, which the terminal can parse and display.

[0353] Step 11: The terminal parses the server response and displays the results on the interface. After receiving the response data packet from the server, the terminal uses the local parsing module to decode the message into an internal data structure.

[0354] The terminal extracts numerical information such as the name of the optimal behavior option, the environmental load index value, the total emissions, and the time from the parsing results, as well as the natural language interpretation text returned by the server.

[0355] The terminal displays structured results on the interface, such as highlighting the optimal behavior options in a list and displaying natural language explanation text in the description area.

[0356] The terminal can simultaneously display the sorting of other behavior options and their corresponding metrics, allowing users to make horizontal comparisons.

[0357] Input: The response data packet sent by the server in step 10.

[0358] Output: The updated terminal interface content is displayed for users to read and use for decision-making.

[0359] Step 12: Users make behavioral decisions based on the results displayed on the terminal. Users can read the optimal behavior options and the explanatory text provided by the generative artificial intelligence model on the terminal interface to understand the differences in environmental load between different behavior options.

[0360] Users select a specific action plan based on their own needs and environmental load information, such as choosing a bicycle as their mode of commuting.

[0361] Users can confirm their selections on the terminal. After confirmation, the terminal can send the selection results back to the server so that they can be recorded in the historical information table for time series statistics.

[0362] Input: The recommended results and explanations displayed on the terminal interface, and the user's actual preferences and judgments.

[0363] Output: The user's final selected action plan and optional confirmation information, for use in subsequent historical records and statistical analysis.

[0364] Application Example 2 The process flow corresponding to the specific processing in Use Case 2 will be described below. The various parts of the system described below are implemented by the data processing device 12 and the intelligent device 14. In addition, the data processing device 12 is referred to as the "server" and the intelligent device 14 is referred to as the "terminal".

[0365] With the increasing demands for environmental protection and the pursuit of carbon neutrality, more and more organizations are looking to leverage information processing technologies to quantitatively assess employee travel patterns, business activities, and the operational status of production equipment, thereby guiding behavior towards lower environmental impact. However, existing technologies suffer from the following limitations at the computer technology level: (1) Traditional environmental load assessment systems mostly use fixed rules or simple weighted calculations, and only perform statistical analysis on structured data such as energy consumption and travel distance. They cannot process behavioral data, time series data and sentiment data in a unified calculation framework at the same time, resulting in a rigid calculation process for carbon neutrality scores and a lack of individualization and context adaptability in the evaluation results of environmental load.

[0366] (2) In the existing system, the environmental load assessment module, recommendation module and human-computer interaction module are often loosely coupled. Even if the generation model is introduced, it is only used as a general text generation tool. There is a lack of structured prompts on how to calculate and adjust scores, how to generate high-scoring behavior recommendations, and how to generate statistical analysis and feedback. As a result, the software architecture on the server side is difficult to form a scalable and programmable intelligent decision pipeline.

[0367] (3) For environmental load assessment and behavior recommendation of complex objects such as autonomous vehicles and operating equipment, most solutions only select strategies locally on the terminal and do not centralize multi-source time-series data such as routes, operating status, and equipment conditions to the server for unified modeling and calculation. This makes it difficult to calculate and compare fine-grained carbon neutrality scores for route combinations, operating modes, and equipment operating conditions, thus limiting the decision-making ability of computer systems in multi-entity and multi-dimensional environments.

[0368] (4) At the user interaction level, emotion recognition technology typically operates independently in existing systems, and its output is not systematically integrated into the server's score calculation and recommendation process. The server software lacks a standardized data structure and calling process, which prevents the computational chain of "emotional state - data processing - score adjustment - recommendation generation" from being executed automatically within the processor. This results in the system being unable to perform differentiated and dynamic computational processing for users with different emotional states while maintaining the fairness of the algorithm.

[0369] (5) For the statistics and feedback of monthly or periodic performance, existing solutions mostly output a single report, lacking a unified calculation model and an automated prompt generation mechanism. They cannot uniformly model time-series environmental data, emotional data, and comparative data (comparison between individuals and organizations, individuals and groups) and interpret them through intelligent text generation. It is also difficult to implement configurable gamified evaluation logic and automatic push logic within the server, thus limiting the scalability and interaction quality of the system in large-scale user scenarios.

[0370] Therefore, a new computer implementation is needed to enable servers to operate within a unified architecture: - Automatically acquire and integrate behavioral data, device data, and sentiment data using a data processing pipeline; - Using programmable prompts as an interface, the system systematically drives generative artificial intelligence models to perform environmental load assessment, score calculation, score adjustment, high-scoring behavior recommendation, and statistical analysis. - The calculation results are presented and controlled in real time through various terminal devices, thereby improving the ability to manage environmental loads and enhancing the technical performance of the computer system in terms of data processing efficiency, scalability, and intelligent interaction.

[0371] The specific processing performed by the specific processing unit 290 of the data processing apparatus 12 in Application Example 2 is achieved by the following means.

[0372] In this invention, the server includes a processing unit for acquiring data related to staff schedules and actual activities, assessing the environmental load generated by travel modes and business activities based on the data, and calculating a carbon neutrality score as an environmental load evaluation indicator for combinations of various behavioral options or routes and operating states; a processing unit for recognizing users' emotional states based on their facial expressions, voice, or text information and generating emotional information; and a processing unit for automatically generating multiple types of structured prompt statements and providing them to a generative artificial intelligence model to instruct on environmental load assessment, carbon neutrality score calculation or adjustment, recommendation of high-scoring behavioral options, and generation of periodic performance and analysis results, using staff schedules and actual activities, the emotional information, and the carbon neutrality score as input information. This allows for unified data modeling and pipelined processing of behavioral data, device operation data, and emotional data within the same server processing framework. Programmable prompts drive generative AI models to participate in score calculation, score adjustment, recommendation generation, and statistical analysis. This upgrades the carbon neutrality score calculation process from fixed-rule calculation to a configurable and scalable intelligent calculation process, thereby improving the data processing efficiency, decision-making accuracy, and interactive intelligence level of the computer system in environmental load assessment scenarios. Furthermore, it supports automatically applying high-scoring behavioral selection results to terminal devices via control signals, achieving end-to-end computer function improvement from data analysis to actual control.

[0373] A "system" refers to a computer implementation scheme consisting of at least one information processing device, at least one terminal device, and one or more storage media and communication devices, used to perform data acquisition, data processing, result output, and control instruction generation.

[0374] "Staff" refers to individual users who engage in business activities within an organization, including but not limited to employees, operators, or managers, whose schedule information and actual activity information are used by the system for environmental load assessment.

[0375] "Users" refers to individuals who interact with the system, including staff members themselves or other users who view and operate the system, whose emotional states and behavioral choices are collected and processed by the system.

[0376] "Schedule information" refers to structured information about time arrangements, including planned activities such as meetings, travel, assignments, and operational tasks, as well as their times and locations.

[0377] "Actual activity information" refers to the record of activities performed by staff or equipment in the real environment, including time-series data related to completed trips, business processing, and equipment operation.

[0378] "Mode of travel" refers to the form of transportation used to move people or objects, including walking, cycling, public transportation, motor vehicles, etc.

[0379] "Business activities" refer to all kinds of behaviors carried out in order to complete work tasks, including meetings, business trips, production operations, office operations, etc.

[0380] "Environmental load" refers to the degree of impact on the natural environment caused by modes of travel, business activities, and equipment operation, and is usually quantified through indicators such as energy consumption and emissions.

[0381] "Environmental load assessment" refers to the process of calculating and analyzing collected behavioral data, energy consumption data, and related parameters to quantitatively evaluate the environmental impact of corresponding activities.

[0382] "Behavioral options" refer to a variety of behavioral options available in a specific situation, including different travel routes, modes of transportation, business arrangements, or operational strategies.

[0383] "Operating route" refers to the path or trajectory of an autonomous vehicle or other mobile entity in space.

[0384] "Operating status" refers to the dynamic operating conditions such as working mode, speed, and load of an automated vehicle or work equipment during operation.

[0385] "Working equipment" refers to mechanical or electronic devices that perform specific tasks in a production or service setting, including robots, production equipment, production line units, etc.

[0386] "Carbon neutrality score" refers to a quantitative indicator calculated based on environmental load assessment results to reflect the degree of carbon emissions and environmental impact of a certain behavior, route, operating status or equipment condition. The higher the value, the closer it is to the carbon neutrality target.

[0387] "Emotional state" refers to a user's psychological and emotional state at a specific point in time, including pleasure, tension, stress, sadness, etc., which is identified through facial expressions, voice, or text content.

[0388] "Emotional information" refers to the structured data output by the emotion recognition module, which describes the user's emotional state and its intensity, and is used for subsequent score calculation or adjustment.

[0389] "Information processing device" refers to a general-purpose or special-purpose computing device with a processor, memory and communication interface, used to execute the software program related to this invention.

[0390] "Mobile device" refers to terminal equipment that can be carried or installed on a mobile body, including vehicle-mounted terminals, portable terminals, mobile communication terminals, etc.

[0391] "Display device" refers to an output device used to present text, graphics, or image information to a user, including displays, terminal interfaces, or large screens.

[0392] "Communication device" refers to hardware and protocol components used to transmit data between system components or between a system and an external network, including wired communication modules, wireless communication modules and related interfaces.

[0393] "Generative AI models" refer to AI models built on machine learning or deep learning techniques that can automatically generate text, strategies, or other outputs based on input data and prompts.

[0394] "Prompt statements" refer to natural language or structured text used to explicitly instruct generative artificial intelligence models on task objectives, input content, output format, or constraints.

[0395] "Statistical methods" refer to the computational methods used to summarize, average, sort, aggregate, or otherwise perform mathematical and statistical processing on time series score data and behavioral data.

[0396] "Performance" refers to the comprehensive evaluation result obtained after statistical processing of multiple carbon neutrality scores within a predetermined period (such as by month or by other time periods), which is used to reflect the environmental performance of an individual or organization within that period.

[0397] "Performance analysis results" refers to analytical information derived from performance and its historical changes and comparative relationships, including trend judgment, difference analysis, and improvement directions.

[0398] "Gamified evaluation information" refers to evaluation data generated based on scores and achievements, including elements such as ranking, level, and reward conditions, in order to enhance user engagement and motivation.

[0399] "Control signals" refer to instruction signals generated by the server and sent to the terminal or device to drive the terminal to automatically apply recommended behavior options or adjust operating conditions.

[0400] In the following description, the server acts as the central computing node, the terminal as the data acquisition and presentation node, and the user as the data provision and selection node. Each embodiment embodies the functions defined in the claims, focusing on the server's internal data structure, algorithm flow, the design of prompts for the generative artificial intelligence model, and its integration with real-world device control, thereby reflecting improvements in computer technology itself.

[0401] I. System Overall Structure A server includes at least one processor, memory, and communication interface, and is equipped with an operating system and runtime environment to execute various software components. Servers can be deployed in data centers or cloud computing platforms. The main software components used by a server include: data processing libraries (e.g., data analysis libraries for tabular data processing, numerical computation libraries for numerical operations), database management systems (e.g., relational databases), deep learning frameworks (e.g., general neural network frameworks), web service frameworks (e.g., general web frameworks), and generative artificial intelligence model invocation components.

[0402] The terminal can be a mobile terminal, a vehicle-mounted terminal, an industrial control terminal, or a general-purpose computing terminal, equipped with a display device, an input device, and optional camera and microphone. The terminal installs client software for communicating with the server, an image processing library for emotion recognition (e.g., a module based on a general-purpose image processing library), and a front-end program for interface display.

[0403] Users input schedule information, actual activity information, and subjective feedback text through the terminal, and authorize the terminal to use the camera and microphone to obtain facial expression and voice data.

[0404] II. Server-side data structure and module composition The server maintains corresponding data structures in memory for different functional modules. Typical structures include: 1. Behavioral Data Table The server defines multiple tables in the relational database for "Schedule Information" and "Actual Activity Information," for example: - Schedule: Fields include user ID, time period, location, planned travel method, and planned activity type.

[0405] - Actual Activity Table: Fields include user ID, timestamp, actual travel mode, device usage duration, production task ID, etc.

[0406] The server uses a data analysis library to read the above tables as data frame objects, associates them through keys (user identifier, timestamp), and forms a unified behavioral feature matrix for subsequent environmental load assessment.

[0407] 2. Environmental Load Factor Table The server maintains one or more "environmental load factor tables," with fields including: behavior type, travel mode, emission coefficient per unit mileage, and energy consumption coefficient per unit time. The server uses this table to map raw behavioral events into standardized energy consumption and emission characteristics, thereby reducing the complexity of online calculations.

[0408] 3. Emotion Data Table The server establishes a dedicated table for emotion information, with fields including: user ID, collection time, emotion type (happiness, stress, sadness, etc.), and confidence score. This table is populated with emotion recognition results uploaded by the terminal, and the server uses a time alignment algorithm to correlate emotion data with behavioral data.

[0409] 4. Carbon Neutrality Score Sheet and Grade Sheet The server maintains the database: - Score table: Records the carbon neutrality score for each event or time slice. Fields include user ID, time, score value, calculation version number, etc.

[0410] - Score Sheet: Records scores aggregated by month or other periods. Fields include user ID, period ID, total score, average score, ranking, and analysis tags.

[0411] The server loads the above table into memory in the form of a data frame and performs vectorization processing with columns as the feature dimension, so that batch calculations can be optimized using linear algebra to improve processing speed.

[0412] III. Environmental Load Assessment and Carbon Neutrality Score Calculation The server performs environmental load assessment based on behavioral data tables and environmental load factor tables. The server uses a numerical calculation library to calculate for each behavioral record: - Energy consumption = Behavioral parameters (e.g., mileage, time) × Corresponding coefficient; - Estimated emissions = Energy consumption × Emission factor

[0413] The server adds these calculation results as feature columns to the data frame. The server further generates a carbon neutrality score through normalization and weighted summation, for example: - The server assigns weights to different dimensions (emissions, energy consumption, proportion of renewable energy, etc.); - The server performs weighted calculations on thousands of records in a single operation within the vector space using matrix operations, thereby avoiding loop-by-loop calculations and significantly improving throughput.

[0414] By using this vectorized computing method, the server can reduce the number of memory accesses and CPU instruction calls compared to the traditional record-by-record processing mode, thereby improving computing efficiency and reducing computing time.

[0415] IV. Emotion Recognition and Emotion Information Generation The terminal uses a camera to capture facial image frames of the user and uses an image processing library to perform face detection, keypoint extraction, and image preprocessing (grayscale conversion, normalization, and size unification). The terminal then inputs the preprocessed image into a locally or remotely deployed convolutional neural network model. This convolutional neural network can employ a multi-layered structure of convolutional layers, pooling layers, and fully connected layers, and is trained through supervised learning on a large number of labeled emotion images. The model's loss function can be cross-entropy loss, and gradient descent and its variants are used for weight updates.

[0416] The terminal converts the neural network's output vector into an emotion label and a confidence score, for example, selecting the category with the highest probability as the emotion state. The terminal then uploads this emotion state and its confidence score to the server. The server filters out or smooths low-confidence emotion results, thereby reducing noise interference with subsequent score adjustments and improving the overall system stability.

[0417] The server aligns emotional information with behavioral records by time. For example, it uses a sliding time window strategy to assign the same emotional label to a group of behavioral records within a certain time range, thereby constructing a combined feature vector containing "behavioral features + emotional features".

[0418] V. Generative Artificial Intelligence Model and Prompt Statement Design The server invokes the generative artificial intelligence model through a unified interface. The server formats behavioral data, sentiment data, and score data into structured text fragments in memory, and then constructs prompt statements. These prompt statements are templated and parameterized; a typical example is as follows: - Example of a prompt statement: "You are an environmental consultant for a company. Below is a breakdown of employee A's daily carbon neutrality score for January, along with the total score and the total score for the previous month. Please write a feedback message of about 150 words in simplified Chinese, including: (1) an overall evaluation of the performance this month; (2) a comparison with the previous month; and (3) three specific suggestions for improvement in a friendly tone: {Employee A's daily score table and statistical summary}." - Example of a prompt statement 2: "You are a smart mobility consultant. Below are several candidate routes and their corresponding estimated energy consumption, CO2 emissions, and carbon neutrality scores. Based on this data, please answer in simplified Chinese which route and which autonomous driving mode is most conducive to carbon neutrality, and briefly explain your reasoning in 3-5 sentences: {Structured Route Data Table}." The server sends the prompt statement along with embedded data fragments as input to the generative AI model. This model employs a multi-layer encoder-decoder neural network structure based on a self-attention mechanism, pre-trained on a large-scale text corpus, and fine-tuned on historical data from environmental assessment scenarios. The server stores and displays the natural language text output by the model as part of the explanatory or recommendation results.

[0419] Compared to the traditional method of manually maintaining fixed templates, centralized management of prompt statement templates and data mapping rules on the server side allows the model to generate fine-grained and personalized feedback based on dynamic data, reducing the amount of hard-coded logic, making the system easier to expand and maintain, and effectively improving the diversity and adaptability of feedback content.

[0420] VI. Score Adjustment and Behavioral Recommendation Rules Based on Emotion The server adjusts carbon neutrality scores not through simple linear addition and subtraction, but by introducing a weight matrix of emotional characteristics. For example, the server can define: under high-stress conditions, the weight of scores related to high-intensity travel is reduced to encourage reducing high-load behaviors when in a bad mood; under positive emotional conditions, the scores for environmentally friendly behaviors are slightly amplified to strengthen positive incentives.

[0421] The server multiplies the emotional weights by the environmental load features using matrix operations to obtain an adjusted score vector. This weight matrix-based adjustment method, compared to simple threshold judgment, can perform continuous and smooth score transformations in a high-dimensional feature space, reducing boundary effects and improving the continuity and robustness of the overall decision-making.

[0422] When recommending high-scoring behavioral options, the server incorporates three types of information—behavioral score, user constraints (time, location, business needs), and emotional state—into a unified ranking algorithm. The server can employ a multi-objective ranking approach, using minimizing environmental load and maximizing user experience as a joint objective function. By setting weights and constraints, it performs Pareto ranking, outputting a set of candidate behavioral options and prompting a generative AI model to generate explanatory text.

[0423] VII. Integration of Automated Driving and Equipment Control Technologies In autonomous driving scenarios, the server calculates the carbon neutrality scores of multiple candidate routes based on sensor data such as vehicle location, speed, and battery level, as well as road condition data provided by map services. The server then provides structured data, including route node coordinates, speed limits, and estimated energy consumption, along with prompts, to the generative artificial intelligence model to obtain recommended routes and driving mode descriptions.

[0424] The server sends the recommendation results to the in-vehicle terminal via a communication interface. Upon receiving the results, the terminal not only displays them on the navigation interface but also converts the recommended driving route into a sequence of waypoints for the autonomous driving controller to parse. It also maps the recommended driving mode into a set of parameters (such as maximum acceleration limit, energy recovery intensity, etc.), thereby directly affecting the vehicle's longitudinal and lateral control behavior.

[0425] In industrial equipment scenarios, servers collect energy consumption and output data to calculate the equipment's carbon neutrality score and automatically adjust operating parameters based on threshold rules. For example, when the server detects that a robot's energy consumption per unit output is significantly higher than the historical average and it is performing a non-urgent production task, the server can instruct the terminal to reduce its operating speed or switch to energy-saving mode via control signals. Because the server processes energy consumption data in batches using data frames and compares it with statistical benchmarks, this control strategy can respond to changes in real time while suppressing misjudgments caused by random fluctuations.

[0426] In this way, the system does not remain at the abstract level of "displaying scores", but changes the behavior of autonomous driving systems and industrial equipment through control signals, so that the calculation results directly affect the physical world, realizing a closed loop from data analysis to equipment control.

[0427] VIII. Performance Statistics, Visualization, and Communication Optimization When calculating monthly scores, the server uses a data analysis library to group and aggregate a large number of score records in memory. The server uses columnar storage and batch queries to load data for the same period into memory at once, reducing multiple database round trips and thus lowering communication overhead and I / O latency.

[0428] When generating performance analysis, the server uses statistical methods to calculate metrics such as mean, median, variance, and trend slope, and detects and filters outliers. Because these statistical operations are based on vectorized operations and efficient mathematical libraries, the server can complete the analysis of large amounts of user data within an acceptable timeframe. Subsequently, the server uses standardized prompts to pass these statistical results to a generative artificial intelligence model, generating natural language analysis reports without requiring separate programming logic for each type of report, thus reducing application-layer code complexity and improving scalability.

[0429] The server synchronizes results and feedback to the terminal via a push notification interface. The terminal presents the results and text reports in the form of charts and paragraphs, allowing users to view their individual and team comparisons on the terminal.

[0430] IX. Technical Effects and Improvements in Computer Technology Through the above structure and process, the server achieves substantial improvements in computer technology in the following aspects: 1. Improved processing efficiency The server uses data frames and vectorized operations to batch process a large number of behavior and device records. Compared with the traditional record-by-record loop calculation method, it significantly reduces processing time and improves throughput, thereby supporting real-time evaluation and control of more users and devices with the same hardware resources.

[0431] 2. Improved assessment accuracy and stability The server introduces a multi-dimensional feature joint decision-making mechanism through data structures such as environmental load factor tables, statistical baselines, and sentiment weight matrices. This reduces the discrete jumps caused by simple rules in score calculation and adjustment, resulting in smoother and more realistic evaluation results.

[0432] 3. Improved Data Management The server stores behavioral data, emotion data, device data, and performance data in a structured manner and processes them in a unified model, avoiding duplicate storage and redundant calculations between subsystems, simplifying data flow management, and improving system maintainability.

[0433] 4. Communication load optimization The server reduces redundant data transmission by periodically batch fetching and pushing, columnar querying, and vectorized computation. It also reduces network bandwidth requirements by centralizing complex calculations on the server side and sending only results and necessary control signals to the terminal.

[0434] 5. Non-traditional automation methods The server utilizes generative AI models and prompts to not simply automate the manual analysis process mechanically, but rather dynamically adjusts the decision and interpretation logic through a configurable text interface. This allows the system to maintain flexibility and scalability in constantly changing environments and business needs. In this model, the complex logic of the decision and interpretation parts is supported by the high-dimensional representations learned by the model, while the server imposes constraints and guides the model's behavior through the structure of prompts, demonstrating an automation path different from traditional rule engines.

[0435] 10. Other Implementation Methods In other implementations, the server can employ different types of generative artificial intelligence models, such as a Chinese language model based on the Transformer architecture, or embed environmental data into the model's context, identifying data types by adding structured labels. The server can also adjust the emotion weight matrix, behavior score weights, or device control strategy parameters according to different organizations' strategies to adapt to different scenario needs.

[0436] For emotion recognition, the terminal can use a lightweight model that runs only locally to reduce its dependence on the network, or it can send the original image to a server for unified recognition by a centralized model to achieve higher accuracy.

[0437] Users can choose different privacy levels to control the frequency and scope of emotion data collection. The server can anonymize and hierarchically store sensitive data to meet privacy protection requirements.

[0438] Through the above-described various implementation methods, the system can realize the various functions defined in the claims in different hardware and software environments, thereby ensuring the feasibility and scalability of the invention in actual deployment.

[0439] use Figure 14 The processing flow is explained.

[0440] Step 1: Users input schedule and actual activity information through the terminal. The input is structured data including fields such as date, time period, location, planned mode of transportation, actual mode of transportation, activity type, and device usage time. The terminal performs format validation on the input (e.g., checking required fields and numerical ranges), encapsulates the validated data into a request message, and sends it as output to the server.

[0441] Step 2: The server receives schedule and activity information from the terminal. The input is the request message sent by the terminal. The server uses a data parsing module to parse the message into an internally unified data structure, writes each field into a behavior data table, and uses a data processing library to load these records into data frame objects, which serve as input for subsequent environmental load assessment. The output consists of behavior records stored in the database and behavior data frames in memory.

[0442] Step 3: The server calculates energy consumption and emissions based on the behavior data frame and the environmental load factor table. The input consists of the unit energy consumption coefficient and emission coefficient for each mode of travel and activity type in the behavior data frame and the environmental load factor table. The server performs vectorized operations on each behavior record, calculating energy consumption and estimated emissions, and adds the results as new columns to the data frame. The output is an expanded behavior data frame containing energy consumption and emission columns.

[0443] Step 4: The server calculates the carbon neutrality score based on extended behavioral data frames. The input consists of energy consumption, emissions, and other environmentally relevant features for each behavioral record. The server normalizes and weights each feature according to preset weights, calculates the score for the entire data column at the vector level using a numerical computation library, and writes the score to a carbon neutrality score table. The output consists of the carbon neutrality score for each record and the updated score data table.

[0444] Step 5: The terminal captures the user's facial expressions and voice signals via a camera and an optional microphone. Input consists of real-time captured image frames and audio clips. The terminal uses an image processing library for face detection and image preprocessing, and employs an emotion recognition model to extract emotion feature vectors and infer emotion categories and confidence levels. Output is a record of emotion information including user identifier, timestamp, emotion category, and confidence level, which is then sent to the server.

[0445] Step 6: The server receives emotion information uploaded by the terminal and performs time alignment. The input consists of emotion information records and a stored behavior data table. The server uses a time window matching algorithm to associate behavior records within a specified time range with corresponding emotion states, adding emotion category and confidence level columns to the behavior data frames. The output is a combined behavior data frame containing emotion features, providing input for subsequent score adjustments.

[0446] Step 7: The server adjusts the carbon neutrality score based on emotional characteristics. The input consists of a combined behavioral data frame and a pre-defined emotional weight matrix. The server applies the emotional weights to the environmental load characteristics through matrix multiplication, continuously and smoothly amplifying or reducing the original score, and updating the corresponding records in the carbon neutrality score table. The output consists of the adjusted carbon neutrality score and the updated score data table.

[0447] Step 8: The server constructs prompts for generative AI models. Inputs include the user's behavior records over a period, adjusted carbon neutrality score, and sentiment information. The server formats this data into structured text fragments and generates prompts according to a template, such as: "You are a corporate environmental consultant. Below is employee A's daily carbon neutrality score details for January, along with the total score and last month's total score. Please write a feedback message of approximately 150 characters in Simplified Chinese, including: (1) an overall evaluation of this month's performance; (2) a comparison with last month; and (3) three specific improvement suggestions in a friendly tone: {data fragment}." Outputs the complete prompt text and embedded data.

[0448] Step 9: The server sends the prompt text and embedded data to the generative AI model and retrieves the generated results. The input consists of the prompt text and structured data generated in step 8. The server calls the model interface, submitting the input to the generative AI model, which encodes the text and generates a response sequence based on an attention mechanism. The server receives the natural language text output by the model, such as feedback reports or recommendations, and stores it in a feedback results table. The output is either environmental feedback text or behavioral recommendation text.

[0449] Step 10: The server selects high-scoring behavioral options based on the adjusted carbon neutrality score and user constraints. Inputs include a score data table, time and location constraints, business requirement constraints, and emotional state information. The server uses a multi-objective ranking algorithm to sort candidate behavioral options or routes, prioritizing high-scoring options with low environmental impact and that meet the constraints. Output is a list of ranked behavioral options and their corresponding scores, used for recommendation.

[0450] Step 11: The server generates explanatory prompts for high-scoring behavior options and calls the generative AI model again. The input is the list of behavior options selected in step 10 and their scores. The server constructs prompts, such as: “You are a smart travel advisor. Below are several candidate routes and their corresponding estimated energy consumption, CO2 emissions, and carbon neutrality scores. Based on this data, please answer in Simplified Chinese which route and which autonomous driving mode is most conducive to carbon neutrality, and briefly explain your reasoning in 3–5 sentences: {Route Data Table}.” The server sends the prompts to the generative AI model and obtains a natural language explanation of the recommendation reasons. The output is a text description of the behavior recommendation.

[0451] Step 12: The server sends the carbon neutrality score, high-scoring behavior options, and generated text descriptions to the terminal. Input includes the score data, a list of recommended options, and feedback text from the generative AI model. The server packages this information via a communication interface and routes it to the corresponding terminal based on the user identifier. Output is a terminal-oriented response message.

[0452] Step 13: The terminal receives the server's response and presents the results to the user. The input is the response message sent by the server. The terminal parses the scoring data and text report, displaying the user's carbon neutrality score, recommended behavior options, and explanations on the display device in the form of charts, lists, and text. For autonomous driving or operational equipment scenarios, the terminal converts the recommended route into a waypoint sequence and the recommended mode into control parameters, preparing to send control commands to the lower-level control system. The output is a visual interface display and the control parameters to be sent.

[0453] Step 14: The user selects whether to accept recommended behaviors on the terminal. The input consists of a list of recommendations and explanatory text displayed on the terminal. The user confirms acceptance, partial acceptance, or rejection of certain recommendations via touch or button operations. The terminal records the user's selections as structured data and sends it to the server. The output is the user selection record, used for subsequent learning and strategy optimization.

[0454] Step 15: With user confirmation, the terminal sends control parameters to the autonomous driving system or operating equipment. The input consists of the control parameters prepared in step 13 and the user's acceptance. The terminal sends parameters such as waypoints, speed limits, and power limits to the vehicle controller or equipment controller via a local control bus or dedicated communication protocol to actually change the operating route or operating status. The output is the adjusted equipment operation control signal, realizing the execution of the recommended behavior in the physical world.

[0455] Step 16: The server calculates carbon neutrality scores and generates results within a predetermined period. Input consists of the score records of all users or devices within the period. The server uses a data analysis library to group the score data by user or organization, calculates statistical indicators such as total score, average score, variance, and ranking, and writes the results to a score table. The server then constructs new prompts and passes the statistical results to a generative artificial intelligence model to generate periodic feedback text. Output consists of periodic score data and corresponding feedback text for notification and visualization.

[0456] 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 sound representing user input regarding the result of the specific processing. The control unit 46A sends the sound data representing user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the sound data.

[0457] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 are prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers from the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms, such as sound data, text data, and image data. 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 induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, 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 can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0458] Furthermore, the processing of the aforementioned data processing system 10 is performed 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 performed by both 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.

[0459] For example, the collection unit is implemented by the control unit 46A of the smart device 14 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the smart device 14 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12, which uses a generation AI to generate a menu. For example, the serving unit is implemented by the output device 40 of the smart device 14 or the specific processing unit 290 of the data processing device 12, which provides the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0460] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the smart device 14.

[0461] Second Implementation Method Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0462] like Figure 3 As shown, the data processing system 210 includes a data processing device 12 and smart glasses 214. A server can be cited as an example of the data processing device 12.

[0463] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. 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. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0464] 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, camera 42, and communication I / F 44 are also connected to the bus 52.

[0465] Microphone 238 receives instructions from user 20 by receiving sounds emitted by user 20. Microphone 238 captures sounds emitted by user 20 and converts the captured sounds into sound data, which is then output to processor 46. Speaker 240 outputs sound according to instructions from processor 46.

[0466] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and imaging elements such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the user 20's surroundings (e.g., the field of view defined by an angle equivalent to the field of vision of an average healthy person).

[0467] Communication I / F44 is connected to network 54. Communication I / F44 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 / F44 and 26 is performed in a secure state.

[0468] Figure 4 This illustrates an example of the main functions of the data processing device 12 and the smart glasses 214. For example... Figure 4 As shown, in the data processing device 12, specific processing is performed by the processor 28. The specific processing program 56 is stored in the memory 32.

[0469] The specific processing program 56 is an example of a "program" involved in the technology of 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 according to the specific processing program 56 executed on the RAM 30.

[0470] 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. In the emotion inference function (emotion-specific function) using the emotion-specific model 59, various inferences and predictions related to the user's emotions are performed, including inferences and predictions of the user's emotions, but this is not limited to this example. Furthermore, emotion inference and prediction may also include, for example, emotion analysis (parsing).

[0471] In the smart glasses 214, the processor 46 performs reception and output processing. The memory 50 stores the reception and output program 60. The processor 46 reads the reception and output program 60 from the memory 50 and executes the read reception and output program 60 on the RAM 48. The reception and output processing is implemented by the processor 46 operating as a control unit 46A according to the reception and output program 60 executed on the RAM 48. Furthermore, the smart glasses 214 has the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and these models can also be used to perform the same processing as the specific processing unit 290.

[0472] Next, the specific processing of the specific processing unit 290 of the data processing device 12 will be described. Each part of the system described below is implemented by the data processing device 12 and the smart glasses 214. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0473] Example 1 The process is the same as that of the specific process described in Embodiment 1 in the first embodiment above, so the description is omitted.

[0474] Application Example 1 The process is the same as that in the specific processing described in Application Example 1 of the first embodiment above, so the description is omitted.

[0475] Example 2 The process is the same as that of the specific process in Embodiment 2 described in the first embodiment above, so the description is omitted.

[0476] Application Example 2 The process is the same as that in the specific processing described in Application Example 2 of the first embodiment above, so the description is omitted.

[0477] 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 outputs the result of the specific processing to the speaker 240. The microphone 238 acquires sound input representing the user's input regarding the result of the specific processing. The control unit 46A sends the sound 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 sound data.

[0478] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 are prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers from the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms, such as sound data, text data, and image data. 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 induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, 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 can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0479] Furthermore, the processing of the aforementioned data processing system 10 is performed 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 performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. Additionally, 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.

[0480] For example, the collection unit is implemented by the control unit 46A of the smart glasses 214 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the smart glasses 214 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12, which uses a generation AI to generate a menu. For example, the serving unit is implemented by the speaker 240 of the smart glasses 214 or the specific processing unit 290 of the data processing device 12, which provides the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0481] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the smart glasses 214.

[0482] Third Implementation Method Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0483] like Figure 5 As shown, the data processing system 310 includes a data processing device 12 and a head-mounted terminal 314. A server can be cited as an example of the data processing device 12.

[0484] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. 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. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0485] 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, display 343, and communication I / F 44 are also connected to the bus 52.

[0486] Microphone 238 receives instructions from user 20 by receiving sounds emitted by user 20. Microphone 238 captures sounds emitted by user 20 and converts the captured sounds into sound data, which is then output to processor 46. Speaker 240 outputs sound according to instructions from processor 46.

[0487] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and imaging elements such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, which captures images of the user 20's surroundings (e.g., the field of view defined by an angle equivalent to the field of vision of an average healthy person).

[0488] Communication I / F44 is connected to network 54. Communication I / F44 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 / F44 and 26 is performed in a secure state.

[0489] Figure 6 This illustrates an example of the main functions of the data processing device 12 and the head-mounted terminal 314. For example... Figure 6 As shown, in the data processing device 12, specific processing is performed by the processor 28. The specific processing program 56 is stored in the memory 32.

[0490] The specific processing program 56 is an example of a "program" involved in the technology of 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 according to the specific processing program 56 executed on the RAM 30.

[0491] The memory 32 stores the data generation model 58 and the emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290.

[0492] In the head-mounted terminal 314, the processor 46 performs the acceptance / output processing. The memory 50 stores the acceptance / output program 60. The processor 46 reads the acceptance / output program 60 from the memory 50 and executes the read acceptance / output program 60 on the RAM 48. The acceptance / output processing is implemented by the processor 46 operating as a control unit 46A according to the acceptance / output program 60 executed on the RAM 48.

[0493] Next, the specific processing of the specific processing unit 290 of the data processing device 12 will be described. Each part of the system described below is implemented by the data processing device 12 and the head-mounted terminal 314. In the following description, the data processing device 12 will be referred to as the "server" and the head-mounted terminal 314 will be referred to as the "terminal".

[0494] Example 1 The process is the same as that of the specific process described in Embodiment 1 in the first embodiment above, so the description is omitted.

[0495] Application Example 1 The process is the same as that in the specific processing described in Application Example 1 of the first embodiment above, so the description is omitted.

[0496] Example 2 The process is the same as that of the specific process in Embodiment 2 described in the first embodiment above, so the description is omitted.

[0497] Application Example 2 The process is the same as that in the specific processing described in Application Example 2 of the first embodiment above, so the description is omitted.

[0498] 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 outputs the result of the specific processing to the speaker 240 and the display 343. The microphone 238 acquires sound input representing the user's input regarding the result of the specific processing. The control unit 46A sends the sound 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 sound data.

[0499] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 includes prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms such as sound data, text data, and image data. 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 induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, 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 can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0500] Furthermore, the processing of the aforementioned data processing system 10 is performed 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 performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the head-mounted terminal 314. Additionally, 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.

[0501] For example, the collection unit is implemented by the control unit 46A of the head-mounted terminal 314 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the head-mounted terminal 314 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12 to analyze the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12 to generate a menu using a generation AI. For example, the serving unit is implemented by the speaker 240 and display 343 of the head-mounted terminal 314 or the specific processing unit 290 of the data processing device 12 to provide the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0502] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the head-mounted terminal 314.

[0503] Fourth Implementation Method Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0504] like Figure 7 As shown, the data processing system 410 includes a data processing device 12 and a robot 414. A server can be cited as an example of the data processing device 12.

[0505] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. 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. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0506] Robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a controlled 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, controlled object 443, and communication I / F 44 are also connected to the bus 52.

[0507] Microphone 238 receives instructions from user 20 by receiving sounds emitted by user 20. Microphone 238 captures sounds emitted by user 20 and converts the captured sounds into sound data, which is then output to processor 46. Speaker 240 outputs sound according to instructions from processor 46.

[0508] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and imaging elements such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, to photograph the area around robot 414 (e.g., the field of view defined by a perspective equivalent to the field of vision of an average healthy person).

[0509] Communication I / F44 is connected to network 54. Communication I / F44 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 / F44 and 26 is performed in a secure state.

[0510] The controlled object 443 includes a display device, LEDs (light-emitting diodes) for the eyes, and motors for driving the arms, hands, and feet. The posture or movement of the robot 414 is controlled by controlling the motors in the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. In addition, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0511] Figure 8 This illustrates an example of the main functions of the data processing device 12 and the robot 414. For example... Figure 8 As shown, in the data processing device 12, specific processing is performed by the processor 28. The specific processing program 56 is stored in the memory 32.

[0512] The specific processing program 56 is an example of a "program" involved in the technology of 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 according to the specific processing program 56 executed on the RAM 30.

[0513] The memory 32 stores the data generation model 58 and the emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290.

[0514] In robot 414, the processor 46 performs the acceptance and output processing. The memory 50 stores the acceptance and output program 60. The processor 46 reads the acceptance and output program 60 from the memory 50 and executes the read acceptance and output program 60 on RAM 48. The acceptance and output processing is implemented by the processor 46 acting as the control unit 46A according to the acceptance and output program 60 executed on RAM 48.

[0515] Next, the specific processing of the specific processing unit 290 of the data processing device 12 will be described. Each part of the system described below is implemented by the data processing device 12 and the robot 414. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 will be referred to as the "terminal".

[0516] Example 1 The process is the same as that of the specific process described in Embodiment 1 in the first embodiment above, so the description is omitted.

[0517] Application Example 1 The process is the same as that in the specific processing described in Application Example 1 of the first embodiment above, so the description is omitted.

[0518] Example 2 The process is the same as that of the specific process in Embodiment 2 described in the first embodiment above, so the description is omitted.

[0519] Application Example 2 The process is the same as that in the specific processing described in Application Example 2 of the first embodiment above, so the description is omitted.

[0520] The specific processing unit 290 sends the result of the specific processing to the robot 414. In the robot 414, the control unit 46A outputs the result of the specific processing to the speaker 240 and the controlled object 443. The microphone 238 acquires sound input representing the result of the specific processing. The control unit 46A sends the sound 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 sound data.

[0521] Data generation model 58 is a so-called generative AI (Artificial Intelligence). Examples of data generation models 58 include ChatGPT (registered trademark) (accessible via the internet (URL: https: / / openai.com / blog / chatgpt)). Data generation model 58 is obtained through deep learning on a neural network. Input to data generation model 58 are prompt words containing instructions, and inference data such as sound data representing sound, text data representing text, and image data representing images (e.g., still image data or animation data). Data generation model 58 infers from the input inference data based on the instructions represented by the prompt words and outputs the inference result in one or more data forms, such as sound data, text data, and image data. 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 induction. The specific processing unit 290 performs the aforementioned specific processing while using data generation model 58. The data generation model 58 can also be a model finely tuned to output inference results from prompts that do not contain instructions. In this case, the data generation model 58 can output inference results based on prompts that do not contain instructions. The data processing apparatus 12, etc., includes various data generation models 58, including AI other than the generation AI. AI other than the generation AI can be, 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 can perform various processes, but is not limited to this example. Furthermore, the AI ​​can also be an AI agent. Furthermore, when the processing of the above-mentioned parts is performed by AI, the processing can be performed partially or entirely by AI, but is not limited to this example. Furthermore, the processing performed by the AI ​​including the generation AI can be replaced by processing in the rule base, and the processing in the rule base can also be replaced by processing performed by the AI ​​including the generation AI.

[0522] Furthermore, the processing of the aforementioned data processing system 10 is performed 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 performed by both the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. Additionally, 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.

[0523] For example, the collection unit is implemented by the control unit 46A of the robot 414 or the specific processing unit 290 of the data processing device 12. For example, the acquisition unit uses the camera 42 or communication I / F 44 of the robot 414 to acquire step data, which is then processed by the specific processing unit 290 of the data processing device 12. For example, the analysis unit is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the data from the collection unit and the acquisition unit. For example, the generation unit is implemented by the specific processing unit 290 of the data processing device 12, which uses a generation AI to generate a menu. For example, the serving unit is implemented by the speaker 240 of the robot 414 and the control object 443 or the specific processing unit 290 of the data processing device 12, which provides the generated menu to the user. The correspondence between each unit and the device or control unit is not limited to the above examples and various changes can be made.

[0524] In the above embodiments, examples of specific processing by the data processing device 12 are given, but the technology disclosed herein is not limited to this, and specific processing may also be performed by the robot 414.

[0525] Furthermore, the emotion-specific model 59, acting as an emotion engine, can determine a user's emotion based on a specific mapping. Specifically, the emotion-specific model 59 can determine a user's emotion based on an emotion graph that serves as a specific mapping (see...). Figure 9 The emotion-specific model 59 can also determine the robot's emotion, and the specific processing unit 290 performs specific processing based on the robot's emotions.

[0526] Figure 9 This is a diagram representing an emotion map 400 that maps multiple emotions. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotion is. On the outer side of the concentric circles, emotions representing states or behaviors arising from mood are arranged. Emotions are concepts that include feelings and mental states. Emotions generated by reactions occurring in the brain are arranged roughly to the left of the concentric circles. Emotions derived from situational judgments are arranged roughly to the right of the concentric circles. Emotions generated by reactions occurring in the brain and derived from situational judgments are arranged roughly above and below the concentric circles. Furthermore, "pleasant" emotions are arranged above the concentric circles, and "unpleasant" emotions are arranged below them. Thus, in the emotion map 400, multiple emotions are mapped based on the structure that generates emotions, and emotions that are likely to occur simultaneously are mapped close to each other.

[0527] These emotions are distributed at the three o'clock position of the emotion map 400, typically fluctuating between peace and anxiety. In the right half of the emotion map 400, situational awareness dominates over internal sensation, thus resulting in an impression of calm.

[0528] The inner side of the emotion map 400 represents the inner state, while the outer side represents behavior. Therefore, the further outward you are from the emotion map 400, the more visible the emotion becomes (manifested in behavior).

[0529] Here, human emotions are based on various balances such as posture and blood sugar levels. When these balances deviate from an ideal state, it indicates an unpleasant state; when they approach the ideal state, it indicates a pleasant state. Emotions in robots, cars, motorcycles, etc., can also be created in the following way: based on various balances such as posture and remaining battery power, when these balances deviate from an ideal state, it indicates an unpleasant state; when they approach the ideal state, it indicates a pleasant state. Emotion maps can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a Brain Physiological Signal Analysis System for Voice Emotion Recognition and Emotion, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). In the left half of the emotion map, emotions belonging to the sensory-dominated region, called "response," are arranged. Furthermore, in the right half of the emotion map, emotions belonging to the situational cognition-dominated region, called "situation," are arranged.

[0530] In the emotion map, two types of emotions that promote learning are defined. One is a negative emotion on the situational side, in the middle or peripheral region of "repentance" or "reflection." This occurs when the robot experiences negative emotions such as "I don't want to experience this feeling again" or "I don't want to be blamed again." The other is a positive emotion on the response side, near the "desire" region. This occurs when there are positive feelings such as "wanting more" or "wanting to know more."

[0531] The emotion-specific model 59 inputs user input into a pre-trained neural network to obtain emotion values ​​representing each emotion shown in the emotion map 400, thereby determining the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network... Figure 10 As shown in the sentiment graph 900, it was trained in a way that sentiments that are configured close to each other have similar values. Figure 10 The text shows examples of emotions such as "peace of mind", "stability", and "reassurance" that have similar emotion values.

[0532] The above description focuses on the functions of the data processing device 12, but the system of this disclosure is not necessarily installed on a server. The system of this disclosure can also be installed as a general information processing system. This disclosure can also be installed, for example, as a software program running on a personal computer, an application running on a smartphone, etc. The method of this disclosure can also be provided to users in the form of SaaS (Software as a Service).

[0533] In the above embodiments, an example of a specific process being performed by a single computer 22 is given. However, the technology disclosed herein is not limited to this, and the specific process can also be distributed among multiple computers, including computer 22. For example, the data generation model 58 can be located on an external device of the data processing apparatus 12, where data is generated based on the input data.

[0534] In the above embodiments, examples of storing a specific processing program 56 in the memory 32 have been described, 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 into the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0535] Alternatively, a specific processing program 56 may be pre-stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 according to the requirements of the data processing device 12.

[0536] In addition, it is not necessary to store all the specific processing program 56 in the storage device such as the server connected to the data processing device 12 via the network 54 or in the memory 32; a portion of the specific processing program 56 may be stored in advance.

[0537] As hardware resources for performing specific processes, various processors, as shown below, can be used. For example, a CPU can be listed as a processor, which functions as a general-purpose processor that performs specific processes by executing software, i.e., a program. Furthermore, processors can be listed as special-purpose circuits such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application-Specific Integrated Circuits), which are processors with circuitry specifically designed to perform specific processes. Each processor has built-in or connected memory, and each processor executes specific processes using that memory.

[0538] 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). Alternatively, the hardware resources for performing a specific process can be a single processor.

[0539] As an example of a single processor, there are two approaches: First, a processor is composed of a combination of one or more CPUs and software, which functions as a hardware resource to perform a specific process; second, as represented by a SoC (System-on-a-chip), a processor is used to implement the functionality of the entire system, which includes multiple hardware resources for performing a specific process, using a single IC (Integrated Circuit) chip. In this way, the specific process is implemented by using one or more of the aforementioned processors as hardware resources.

[0540] Furthermore, the hardware architecture of these various processors, more specifically, can utilize circuits that combine semiconductor elements and other circuit components. Moreover, the specific process described above is just one example. Therefore, without departing from the main point, unnecessary steps can certainly be deleted, new steps added, or the processing order changed.

[0541] The descriptions and illustrations above are detailed explanations of a portion of the technology disclosed herein, and are merely one example of the technology disclosed herein. For example, the above descriptions of the structure, function, effect, and results are just one example of the structure, function, effect, and results of a portion of the technology disclosed herein. Therefore, without departing from the spirit of the technology disclosed herein, unnecessary parts may be deleted, new elements added, or replacements may be made to the descriptions and illustrations above. Furthermore, to avoid confusion and facilitate understanding of a portion of the technology disclosed herein, explanations of common technical knowledge that do not require special explanation under the premise of being able to implement the technology disclosed herein have been omitted from the descriptions and illustrations above.

[0542] All documents, patent applications and technical specifications set forth in this specification are incorporated herein by reference to the same extent that each document, patent application and technical specification is specifically and individually described therein and referenced by reference.

[0543] In addition, the following notes are provided in response to the above explanation.

[0544] Example 1 (Note 1) An information processing system, characterized in that it comprises: A means for enabling an information processing device to acquire planning and execution information related to human activities, wherein the activity information includes at least structured input information such as travel mode type, travel distance, and energy consumption; An apparatus for enabling the information processing device to store the activity information into a storage device, and for preprocessing the activity information using a data processing program and a general data processing library, wherein the preprocessing includes at least missing value processing, unit unification processing, and data format standardization processing. A device for converting activity information into environmental load based on emission coefficients corresponding to travel modes and energy types, and calculating it in numerical form. A device for enabling the information processing apparatus to perform normalization or scaling transformation on the environmental load, thereby generating a carbon neutrality score; An apparatus for enabling the information processing device to generate prompts for inputting into a generative artificial intelligence model based on the carbon neutrality score and the activity information, and for using the prompts to obtain an explanation of the carbon neutrality score and behavioral suggestions for reducing environmental impact from the generative artificial intelligence model; An apparatus for enabling the information processing device to recognize the user's emotional state, and based on the recognition result to generate prompt statements for adjusting the carbon neutrality score or the content displayed by the carbon neutrality score, and providing them to the generative artificial intelligence model; A means for enabling the information processing device to calculate statistics divided by time period based on the carbon neutrality score and the environmental load, and to generate prompts for providing to the generative artificial intelligence model based on the statistics to obtain recommended information for high-scoring activity plans; A device for enabling the information processing apparatus to distinguish between carbon neutrality score calculation processing based on actual measurement data and simulation score calculation processing based on assumed activity conditions, and to output the actual score and simulation score in a comparable form. A means for enabling the information processing device to send the carbon neutrality score and the statistics to an output device and display them on the output device in numerical and graphical form, and a means for enabling the output device to present the explanations and behavioral suggestions obtained from the generative artificial intelligence model in natural language.

[0545] (Note 2) According to the information processing system described in Appendix 1, the information processing device is configured to aggregate the carbon neutrality score and the environmental load by individual and organizational units, and to provide a generative artificial intelligence model with prompt statements for distributing the carbon neutrality score and the environmental load to a notification device via a communication device.

[0546] (Note 3) According to the information processing system described in Appendix 1, the information processing device is configured to generate gamified elements containing ranking information, achievement information, and reward information based on the carbon neutrality scores of individual units and group units and the statistics, and to provide prompt statements to a generative artificial intelligence model for outputting the gamified elements.

[0547] Application Example 1 (Note 1) An information processing system, characterized in that it comprises: A device for acquiring employees’ planned and actual information, and for quantifying environmental load based on travel mode, travel distance, device usage time and energy consumption contained in the planned and actual information, and for calculating an individual’s or organization’s environmental load score. A device for acquiring time-series data on operating status and energy consumption from detectors installed on production units, and for calculating environmental load scores for each unit in real time based on the time-series data; A device for storing the environmental load scores of the employees and the environmental load scores of each device, and for summarizing the environmental load scores using statistical methods within a predetermined summarization period to calculate the periodic score. A device for generating high-scoring plans and device operation mode candidates that can reduce environmental load based on the environmental load scores of the employees, the environmental load scores of each device, and the periodic performance, and providing prompts explaining the candidates to a generative artificial intelligence model. An apparatus for generating an improved plan to optimize the employee's behavior plan and the operating conditions of the device based on the natural language response output by the generative artificial intelligence model to the prompt statement, and sending the improved plan to the user terminal; An apparatus for reflecting the adoption of the improvement scheme into the calculation conditions of the environmental load score based on the user selection results and user operation history obtained from the user terminal, and for automatically and continuously executing a cycle of planning-execution-evaluation-improvement.

[0548] (Note 2) The information processing system according to Appendix 1 is characterized in that, The device is configured to compare environmental load scores calculated by employee and by organizational unit with the periodic results, generate evaluation information including ranking information and target achievement, and send the evaluation information to an information providing device via communication to provide information to promote actions that reduce environmental load.

[0549] (Note 3) The information processing system according to Appendix 1 is characterized in that, The device is configured to generate prompt statements that define game elements including obtaining scores, rankings, reward indicators, and achieving goals, based on environmental load scores per employee and per organizational unit, and the evaluation information. The prompt statements are then provided to a generative artificial intelligence model, and the game-like feedback information output by the generative artificial intelligence model is presented to the user terminal, thereby promoting employees' behavior to reduce environmental load.

[0550] Example 2 (Note 1) An information processing system, characterized in that it comprises: A device for acquiring information related to behavioral options and calculating environmental load index values ​​for each behavioral option based on the degree of environmental impact; A device for comparing calculated environmental load indices and determining optimal behavioral options based on the environmental load indices; An apparatus for generating structured data containing the determined optimal behavior options and their corresponding environmental load indicators, and combining the structured data with natural language prompts from the user to form generation instruction information as prompts for generative artificial intelligence models. A device for providing the prompt statement to the generative artificial intelligence model to instruct the generation of a natural language response containing recommended optimal behavior options and explanatory text based on environmental load indicators; A means for sending the generated natural language response and the optimal behavior option to a terminal device for display; An apparatus for storing historical information related to behavioral options and for performing time-series aggregation of the environmental load indicators using statistical methods.

[0551] (Note 2) The information processing system according to Appendix 1 is characterized in that, It also includes a device for calculating environmental load indicators by individual and group units, generating prompts for generative artificial intelligence models to instruct the generative artificial intelligence models to generate explanatory and notification texts for comparing the environmental load indicators of the individual and group units, and transmitting the explanatory and notification texts by means of electronic communication.

[0552] (Note 3) The information processing system according to Appendix 1 is characterized in that, It also includes a device for calculating evaluation values ​​for individual units and regional units based on environmental load indicators, generating prompt statements for generative artificial intelligence models, instructing the generative artificial intelligence models to generate game element information containing ranking information, achievement status information and reward information corresponding to the evaluation values, and sending the game element information to a terminal device for display, so as to incentivize improvement of environmental load indicators.

[0553] Application Example 2 (Note 1) An information processing system, characterized in that it comprises: This is used to acquire data related to staff schedules and actual activities, and to assess the environmental load generated by travel modes and business activities based on the data. The combination of various behavioral options or routes and operating states is used to calculate the carbon neutrality score as a unit for environmental load evaluation indicators. A unit for recognizing a user's emotional state based on facial expressions, voice, or text information, and generating emotional information that reflects the emotional state in the calculation or correction of the carbon neutrality score. A unit for taking staff's schedule and actual activity information, the emotional information, and the carbon neutrality score as input information, generating prompt statements to instruct the generative artificial intelligence model on the calculation or adjustment methods of environmental load assessment and carbon neutrality score, and providing the prompt statements to the generative artificial intelligence model. A unit for determining high-scoring behavioral options with lower environmental impact based on the carbon neutrality score, including the worker's schedule, travel mode, operating route and status of autonomous vehicles, or operating conditions of work equipment, and generating prompt statements to indicate and recommend the high-scoring behavioral options to the generative artificial intelligence model, and providing the prompt statements to the generative artificial intelligence model. This unit is used to store the carbon neutrality scores and behavioral selection results in a time series manner according to individual staff or users and organizational units, calculate the scores on a monthly or predetermined periodic basis using statistical methods, generate prompt statements to instruct the generative artificial intelligence model to generate the scores and their analysis results, and provide the prompt statements to the generative artificial intelligence model. A unit for presenting the calculated carbon neutrality score, the high-scoring behavior option, and the score to the user via an information processing device, mobile device, or display device, and for generating control signals for automatically applying the behavior option when necessary.

[0554] (Note 2) The information processing system according to Appendix 1 is characterized in that the system further comprises: a unit for distributing, via a communication device, the carbon neutrality scores and the results generated by individual staff members or users and organizational units respectively in the form of electronic messages or notification messages, and for generating prompt statements for instructing a generative artificial intelligence model to generate feedback text explaining the distributed content using the generative artificial intelligence model, and providing the prompt statements to the generative artificial intelligence model.

[0555] (Note 3) According to the information processing system described in Appendix 1, the system further comprises: a unit for comparing the carbon neutrality scores and achievements of individual units and group units, generating gamified evaluation information including ranking information, achievement status and reward conditions, generating prompt statements for instructing a generative artificial intelligence model on the content and display method of generating the gamified evaluation information, and providing the prompt statements to the generative artificial intelligence model.

Claims

1. An information processing system, characterized in that, include: processor; The processor is configured to: receive planned and actual information from employees; perform environmental load assessment on the planned and actual information and calculate a carbon neutrality score; identify the user's emotional state and provide input to a generative artificial intelligence model for generating prompts, so that the generative artificial intelligence model generates prompts based on the emotional state to indicate adjustments to the carbon neutrality score according to the emotional state; provide input to the generative artificial intelligence model for generating prompts, so that the generative artificial intelligence model generates prompts to indicate recommendations for plans with high carbon neutrality scores to the user; and calculate scores monthly using statistical methods.

2. The information processing system according to claim 1, characterized in that, The processor is also configured to generate carbon neutrality scores, both by individual and by organization, and to provide input to a generative artificial intelligence model for generating prompts, such that the generative artificial intelligence model generates prompts to instruct notifications of the carbon neutrality scores via email or a notification system.

3. The information processing system according to claim 1, characterized in that, The processor is also configured to calculate individual scores and departmental scores, and to provide input to a generative artificial intelligence model for generating prompts, such that the generative artificial intelligence model generates prompts to instruct the provision of gamified elements that incentivize achieving high scores by notifying individual and departmental scores.

Images