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: use personal data, weather forecast data obtained through internet search tools, recipe data, and shopping product data to train a generative artificial intelligence model; use the generative artificial intelligence model to generate prompts based on the aforementioned data to instruct the user to have the optimal food delivered instantly; and identify the user's emotions and generate prompts to recommend corresponding dietary menus to the user based on the emotions.

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 speech in response to the user's speech.

[0003] The main problem this invention aims to solve is that existing dietary recommendation and food shopping support technologies are mostly rule-based or simple retrieval systems, which cannot comprehensively consider the user's multi-dimensional personalized information and external environmental information, thus making it difficult to provide users with truly "most suitable" food and menu suggestions in a timely and accurate manner. Specifically, the existing technology has at least the following problems: 1. Inability to integrate multi-source data for personalized decision-making: Traditional systems typically make recommendations based only on limited user preferences or historical records, rarely considering users' personal data (including allergy information, health status, taste preferences, etc.), real-time weather forecast data, in-system recipe data, and online shopping product data. This results in a lack of close correlation between the recommendation results and the user's current physical condition, emotional state, and environmental conditions.

[0004] 2. Difficulty in meeting users' needs for "instant delivery" and "readily available" food solutions in a timely manner: Many recommendation systems only provide recipe references and are not fully integrated with instant delivery or online shopping services. Users still need to search for ingredients, compare prices and place orders themselves, which cannot realize a fast and integrated food acquisition process based on recommendation results.

[0005] 3. Inability to effectively identify and utilize user emotional information: Most existing dietary recommendation programs do not pay sufficient attention to the user's current emotional state (such as high stress, low mood, excitement, etc.), and cannot provide dietary menus that match the emotion and have a calming or motivating effect. Therefore, they are significantly lacking in emotional care and psychological relief.

[0006] 4. Inability to generate optimal recipes and ingredient combinations under complex constraints: In real life, users are often constrained by multiple factors, such as budget, health goals, nutritional needs, availability of seasonal ingredients, personal cooking skills, and available preparation time. When faced with such multidimensional constraints, traditional systems often have overly simplistic recommendation logic, making it difficult to balance various factors and output a comprehensive optimal recipe and ingredient solution.

[0007] 5. Insufficient support for food-related open-ended questions: Existing systems mostly use preset options or fixed query interfaces. They often fail to provide flexible, semantically rich, and user-integrated answers to diverse food questions posed by users in natural language, such as "What should I eat today?" or "How can I make a low-fat, high-protein dinner in 20 minutes?"

[0008] Therefore, it is necessary to provide a system that can utilize generative artificial intelligence models to integrate personal data, weather forecasts, recipe data, and shopping product data, and combine them with user budget, health and nutrition needs, skills, and time constraints, while also recognizing user emotions and supporting natural language question answering, in order to solve the above-mentioned technical problems and improve the personalization, feasibility, and user experience of diet recommendations. Summary of the Invention

[0009] To address the aforementioned issues, this invention provides an information processing system comprising a processor that utilizes generative artificial intelligence models and multi-source data to provide users with high-precision, actionable dietary recommendations and food-related Q&A services.

[0010] On the one hand, to address the problem of insufficient integration of multi-source data for personalized decision-making in existing technologies, the processor in this invention is configured to: use personal data, weather forecast data obtained through internet search tools, recipe data, and shopping product data to train or fine-tune a generative artificial intelligence model. This enables the generative artificial intelligence model to learn the correlation between dietary preference patterns, health constraints, and environmental factors among different user groups. During actual operation, the generative artificial intelligence model generates prompts based on the aforementioned multi-source data to indicate the immediate delivery of optimal food to the user. In this way, the system can not only derive recipes or food solutions that meet the user's health and taste requirements, but also directly drive the integration with online shopping or instant delivery services in the form of prompts, achieving a rapid conversion of recommendation results into actual food delivery.

[0011] On the other hand, to address the problem of existing technologies failing to identify and utilize user emotional information, the processor in this invention is further configured to: perform emotion recognition on the user's voice, text input, or other interactive behaviors, determine the user's current emotional state, and generate prompts to recommend appropriate dietary menus to the user based on the stated emotion. By incorporating emotional information as one of the input conditions for a generative artificial intelligence model, the system can reflect emotional care in its recommendations. For example, it can recommend warm, easily digestible, and calming foods when the user is under stress or feeling down, and recommend menus more suitable for energy replenishment and physical recovery when the user is emotionally excited or after exercise, thereby improving the psychological fit of the recommendations.

[0012] Furthermore, to address the challenge of generating optimal recipes and ingredient combinations under complex constraints, the processor in this invention is further configured to: generate prompts for providing users with the optimal recipes and ingredients, taking into account the user's budget, health status, nutritional needs, seasonal ingredients, cooking skills, and preparation time, using the generative artificial intelligence model. Specifically, the processor can input the aforementioned constraint parameters, along with the recipe data and shopping product data stored in the system, into the generative artificial intelligence model. The model then comprehensively weighs factors such as cost, nutritional structure, seasonality, operational difficulty, and time constraints, automatically outputting a recipe recommendation that meets the constraints and is optimal overall, along with a corresponding ingredient purchase list. By encapsulating this output as prompts, the system can further drive the terminal interface or external shopping services to display or allow users to place orders for the specific ingredients and quantities required.

[0013] Furthermore, to address the insufficient support for open-ended questions related to food in existing technologies, the processor in this invention is further configured to: generate prompts using the generative artificial intelligence model to instruct users to provide answers to their food-related questions, enabling it to handle various food-related inquiries. In other words, when a user asks questions about food, dietary habits, recipe preparation, and nutritional balance in natural language on the terminal, the processor can input the user's question along with contextual information such as the user's personal data, weather data, existing recipes, and product data into the generative artificial intelligence model. The model then generates an answer that is semantically natural and relevant to the individual user. The processor then transmits this answer as a prompt to the corresponding display or interaction module, thereby achieving real-time intelligent responses to diverse and open-ended food questions.

[0014] Through the above technical means, the system of the present invention can: 1. Using generative artificial intelligence models as the core, and comprehensively utilizing personal data, weather forecasts, recipe data, and shopping product data within a unified framework, to achieve highly accurate personalized food recommendations; 2. Directly generate prompts to drive instant delivery or shopping processes, making the recommendation results highly actionable and implementable; 3. Introduce user emotion recognition to incorporate emotional factors into the dietary decision-making process, providing users with more emotionally considerate dietary menus; 4. Generate optimal recipes and ingredient combinations under multiple constraints such as budget, health, nutrition, seasonal ingredients, cooking skills, and time. 5. By using generative artificial intelligence models to intelligently answer food-related natural language questions, we can provide users with flexible, comprehensive, and personalized dietary consultation services.

[0015] Therefore, the present invention can effectively overcome the shortcomings of the prior art and significantly improve the overall performance of the diet recommendation and food service system in terms of personalization, intelligence and user experience.

[0016] "System" refers to an overall device or set of devices consisting of at least one processor and optional hardware and / or software modules such as memory, communication interface, input / output devices, etc., used to perform the data processing, model calling and information interaction functions described in this invention.

[0017] A “processor” refers to hardware or a combination of hardware that can execute computer program instructions, process and operate input data, and control the operation of various functional modules of the system, including but not limited to a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a programmable logic device (FPGA), or any combination thereof.

[0018] "Personal data" refers to various types of information data related to a specific user, including but not limited to the user's basic information (age, gender, height, weight, etc.), health status information (medical history, health goals, etc.), dietary preference information (taste preferences, foods to avoid, etc.), allergy information, and other user characteristic data related to dietary recommendations.

[0019] "Weather forecast data" refers to meteorological forecast information related to a specific geographical location and time, obtained through internet search tools or external weather service interfaces, including but not limited to temperature, humidity, weather conditions (sunny, rainy, snowy, etc.), wind force, air quality, and weather trend data for a period of time in the future.

[0020] "Recipe data" refers to a set of structured or unstructured data related to food preparation, including but not limited to the dish name, required ingredients and quantities, cooking steps, cooking time, difficulty level, nutritional information, target audience and applicable scenarios.

[0021] "Shopping product data" refers to product information data related to food and its raw materials obtained from online shopping platforms, e-commerce websites, or other product databases, including but not limited to product name, specifications, price, inventory status, sales link, place of origin, and user reviews.

[0022] "Generative artificial intelligence model" refers to an artificial intelligence model trained through machine learning or deep learning techniques that can automatically generate text, images, or other forms of output based on input data. In this invention, it is mainly used to generate prompts, recipe suggestions, and question-and-answer content, including but not limited to generative models based on large-scale language models.

[0023] "Prompt information" refers to instructional or descriptive information generated by the processor and provided to generative artificial intelligence models or other system modules to trigger or guide subsequent processing steps. In this invention, it is usually text information used to instruct the generation of optimal food delivery solutions, recommend diet menus, provide optimal recipes and ingredients, or generate question-and-answer content.

[0024] "Instant delivery" refers to a delivery method where, after a user confirms the food or ingredient plan, the corresponding food or ingredient is delivered to the user's designated location within a preset short time frame through an external delivery service platform or logistics system. Its time is usually significantly shorter than traditional scheduled delivery or regular mailing methods.

[0025] "User's emotions" refers to a user's psychological and emotional state at a specific point in time, including but not limited to feelings of joy, sadness, anxiety, fatigue, excitement, and relaxation. These emotions can be identified and determined by analyzing the user's voice, text input, interactive behavior, or other available signals.

[0026] A “meal menu” refers to a set of dishes or food combinations designed for a user for a specific period of time (such as a meal, a day, or a week) based on the user’s personal data, weather forecast data, and emotional state. The menu may include specific dish names, ingredient lists, portion size suggestions, and order of consumption.

[0027] "Budget" refers to the upper limit of money that a user is willing or able to spend within a certain time frame or a single meal, used to constrain the cost range of the system when generating recipes and ingredient recommendations.

[0028] "Health status" refers to various information related to a user's physical condition, including but not limited to medical conditions (such as high blood pressure, diabetes, etc.), weight management goals (fat loss, muscle gain, maintaining weight, etc.), exercise level, and advice from doctors or nutritionists.

[0029] "Nutritional needs" refers to the nutritional intake requirements that users need to meet or avoid in order to achieve specific health or body shape goals within a certain period of time. These requirements include, but are not limited to, the types and approximate intake ratios of nutrients such as protein, fat, carbohydrates, dietary fiber, vitamins, and minerals.

[0030] "Seasonal ingredients" refer to seasonal ingredients that are naturally abundant, of good quality, and relatively reasonably priced during a specific season or period of time, including but not limited to seasonal vegetables, fruits, grains, and other agricultural by-products.

[0031] "Cooking skills" refers to a user's proficiency and experience in cooking operations. It can be divided into different levels such as beginner, intermediate, and advanced, and is used to constrain and match the complexity of recipes and operational requirements.

[0032] "Meal preparation time" refers to the amount of time a user has available for all operations, including food preparation, cooking, and plating, during a meal preparation process. The system uses this as one of the time constraints when generating recommended solutions.

[0033] "Food-related questions" refer to inquiries made by users in natural language or other forms, which involve food, cooking, nutrition, eating habits, ingredient selection and pairing, etc., including but not limited to questions such as "What to eat", "How to cook it", and "Is it suitable for me". Attached Figure Description

[0034] Figure 1 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the first embodiment.

[0035] 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.

[0036] Figure 3 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the second embodiment.

[0037] 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.

[0038] Figure 5 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the third embodiment.

[0039] 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.

[0040] Figure 7 This is a conceptual diagram illustrating an example of the configuration of the data processing system according to the fourth embodiment.

[0041] 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.

[0042] Figure 9 This represents an emotion map that maps multiple emotions.

[0043] Figure 10 This represents an emotion map that maps multiple emotions.

[0044] Figure 11 This is a sequence diagram illustrating the processing flow of the data processing system of the first embodiment.

[0045] Figure 12This is a sequence diagram illustrating the processing flow of the data processing system in Application Example 1.

[0046] Figure 13 This is a sequence diagram illustrating the processing flow of the data processing system of the second embodiment.

[0047] Figure 14 This is a sequence diagram illustrating the processing flow of the data processing system in Application Example 2. Detailed Implementation

[0048] Hereinafter, an example of an implementation of the system according to the present disclosure will be described with reference to the accompanying drawings.

[0049] First, let me explain the terminology used in the following instructions.

[0050] 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.

[0051] 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.

[0052] 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.

[0053] 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.

[0054] 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.

[0055] First Implementation Method Figure 1 An example of the configuration of the data processing system 10 according to the first embodiment is shown.

[0056] 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.

[0057] 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).

[0058] 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.

[0059] 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.

[0060] 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.

[0061] 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.

[0062] Figure 2 The diagram shows an example of the main functions of the data processing device 12 and the smart device 14.

[0063] 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.

[0064] 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).

[0065] 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.

[0066] 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.

[0067] 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."

[0068] In existing technologies, computer-based diet recommendation systems typically only perform simple matching of users' basic preference information with static recipe data, or only use generative artificial intelligence models to output natural language suggestions, but they have significant shortcomings in the following aspects: (1) Within the computer system, multi-source heterogeneous data such as user attribute data, environmental condition data (e.g., meteorological data), cooking step data and purchase object data lack unified feature and standardization processing, which makes it impossible to form a structured feature set suitable for efficient reasoning of generative artificial intelligence models, thus limiting the model reasoning efficiency and recommendation quality.

[0069] (2) Existing systems often use generative artificial intelligence models directly for dialogue or simple question and answer, without constructing special prompt templates for personalized meal plans, and without explicitly encoding constraints such as budget, health status, nutritional needs, supply period, cooking skills, and preparation time in the prompt statements. This results in unstable model output results, poor controllability, and difficulty in achieving reliable automated decision-making processes at the computer level.

[0070] (3) Existing dietary recommendations mostly remain at the level of dish names or recipes, lacking a mechanism to computerize and optimize the generated meal plan with specific purchase objects. They cannot automatically match available goods, calculate costs, and form a structured purchase list data within the system, thus failing to form an efficient and programmable linkage with external purchasing services.

[0071] (4) Existing systems typically do not feed back user evaluation information and actual purchase history information in a closed loop to the feature set and prompt statement generation process, which makes it impossible for the computer to adaptively adjust the internal data representation and prompt strategy based on long-term interaction data. The system has technical bottlenecks in terms of personalization accuracy and long-term optimization capabilities.

[0072] (5) In response to the diverse inquiries raised by users in the food field (such as nutritional components, applicable population, alternative ingredients, etc.), traditional solutions often use rule bases or static knowledge bases for retrieval-based answers. It is difficult to dynamically construct high-quality prompts in generative artificial intelligence models by combining current user characteristics and environmental conditions, thus failing to fully leverage the model's reasoning ability and improve the human-computer interaction experience.

[0073] Therefore, how to design a data feature generation and prompt generation mechanism for generative artificial intelligence models at the computer system level, effectively integrating user multidimensional attributes, environmental conditions, recipe and product information, and interactive feedback to form a system that can automatically generate meal plans and purchase lists and continuously optimize itself, has become a technical issue that urgently needs to be solved in this field.

[0074] 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.

[0075] In this invention, the server includes a processing unit for acquiring user attribute information, environmental condition information, cooking step information, and purchase object information, and standardizing the information to generate a feature set; a processing unit for constructing prompt statements for inputting into a generative artificial intelligence model based on the feature set, and generating prompt statements instructing the generative artificial intelligence model to generate meal plan information based on user attribute information and environmental condition information; a processing unit for associating the components included in the meal plan information with the purchase object information, calculating a purchase candidate set and cost information for each required component, and generating a purchase list information containing the purchase candidate set; a display unit for presenting the meal plan information and the purchase list information to the user, and sending instruction information to a purchase processing device or external purchase service to instruct the user to acquire the purchase candidate set based on the user's selection information; and a processing unit for acquiring evaluation information and actual purchase history information from the user, reflecting the information in the feature set for updating, thereby correcting the subsequently generated prompt statements and meal plan information. This enables the implementation of a unified data feature generation and prompt generation framework for generative artificial intelligence models within a computer. The server can automatically construct high-quality prompts that incorporate multi-dimensional conditions such as budget constraints, health status, nutritional needs, availability, cooking skills, and preparation time. This drives the generative artificial intelligence model to generate structured, personalized meal plan information, which is then linked with product data to form an executable purchase list. Furthermore, based on user reviews and purchase history, the feature set and prompt strategies are continuously optimized, thereby improving the relevance, controllability, and overall computational efficiency of the generated results. This represents a substantial improvement in data processing workflows and human-computer interaction methods within computer technology.

[0076] A "system" refers to an information processing unit consisting of at least one processing device, a storage device, and optional display and communication devices, used to perform data acquisition, data processing, generative artificial intelligence model invocation, and result output.

[0077] "Processing device" refers to an electronic device capable of executing program instructions, performing calculations, logical judgments, and flow control on input data, including but not limited to a central processing unit, a graphics processing unit, or other programmable computing units.

[0078] "User" refers to the entity that uses the system to obtain meal plan information and purchase list information. It can be an individual or an organization, and is usually distinguished by user account or identifier in the system.

[0079] "User attribute information" refers to individual characteristic data related to the user, including but not limited to age, gender, physical condition, dietary preferences, allergy information, health goals, and historical selection behavior.

[0080] "Environmental conditions information" refers to data related to the objective environment in which the user is located, including but not limited to meteorological data, temperature, humidity, weather conditions, time information, geographical information, and other external conditions that affect dietary choices.

[0081] "Cooking Step Information" refers to data describing the food preparation process, including but not limited to the steps in the recipe, the time required, the order of operations, the heating method, the required kitchen utensils, and related precautions.

[0082] "Purchase Target Information" refers to data related to purchasable entities such as ingredients or food, including but not limited to product name, specifications, unit price, inventory, supply location, delivery conditions, and seller information.

[0083] "Standardization processing" refers to the process of unifying heterogeneous raw data, including but not limited to unit conversion, format normalization, field mapping, missing value handling, encoding conversion, and numerical normalization, in order to generate standardized data suitable for subsequent calculations.

[0084] A “feature set” refers to a collection of multiple processed data features. Each feature represents one aspect of the user’s attributes, environmental conditions, cooking steps, or the object of purchase, and is used as input to a generative artificial intelligence model.

[0085] "Generative artificial intelligence models" refer to artificial intelligence models that are trained on large-scale data based on machine learning or deep learning techniques and can automatically generate text, structured data, or other output results based on input information.

[0086] "Prompt statements" refer to text or equivalent representations used as input to generative artificial intelligence models. They contain model roles, task objectives, constraints, contextual information, and output requirements to guide the model in generating expected output results.

[0087] "Dietary planning information" refers to the dietary arrangement results generated by generative artificial intelligence models for the user, including but not limited to recommended dishes, combination schemes, meal times, ingredient quantities, nutritional structure and related instructions.

[0088] "Components" refers to the basic entities required to constitute each dish or dietary plan in the meal plan information, including but not limited to ingredients, seasonings, beverages and other available food-related items.

[0089] "Purchase candidate set" refers to a set of one or more alternative products selected from the purchase object information for a certain component, used to provide purchase options for that component with different brands, specifications or prices.

[0090] "Cost information" refers to cost data related to the purchase of the candidate set, including but not limited to unit price, quantity, total amount, estimated total cost, and comparison results related to the budget.

[0091] "Purchase List Information" refers to structured list data generated based on meal plan information and purchase candidate set, listing the products, quantities, costs, and purchase channels corresponding to each component element, used to guide subsequent purchase operations.

[0092] "Display device" means a device used to present information to a user in a visual form, including but not limited to a display screen, a touch screen, or a terminal display module that can provide a graphical interface.

[0093] "Selection information" refers to the instruction data generated by the user through terminal interaction after viewing meal plan information or purchase list information, including but not limited to the operation information of checking, confirming, replacing, deleting or modifying purchase candidates.

[0094] "Purchase processing device" refers to a device or system used to execute a purchase process according to instructions, including but not limited to an order generation module, a settlement module, and a processing module that interacts with a payment system and an inventory system.

[0095] "External purchasing services" refers to a platform for purchasing goods or services provided by an external provider, including but not limited to online shopping platforms, online interfaces of offline retail systems, and other transaction systems accessible via the internet.

[0096] "Instruction information" refers to control data generated by the system and sent to the purchase processing device or external purchase service to instruct operations such as placing an order, adding to the shopping cart, or reserving for a specified set of purchase candidates.

[0097] "Evaluation information" refers to the subjective feedback data that users give to the system output results after using the meal plan information and purchase list information, including but not limited to ratings, text comments, preference tags, and satisfaction indicators.

[0098] "Actual purchase history information" refers to data related to the actual purchase behavior completed by the user, including but not limited to purchase time, purchased goods, quantity, amount, supplier, and order status.

[0099] "Update" refers to the process of adjusting, supplementing, or recalculating the existing set of features based on newly acquired evaluation information and actual purchase history information, in order to reflect the latest preferences and behavioral patterns of the users.

[0100] "Correction" refers to using the updated set of features to make directional adjustments to the subsequently generated prompts and meal plan information, so that they better meet the actual needs of the users and the optimization goals of the system.

[0101] "Funding constraint information" refers to data that indicates the acceptable range of expenses for a user in a certain period or a single dining scenario, including but not limited to budget limits, expected price range, and cost preferences.

[0102] "Body condition information" refers to data related to the user's physiological and health status, including but not limited to weight, body fat, blood pressure, blood sugar, medical history, and current health goals.

[0103] "Nutritional requirements information" refers to data related to the user's goals or restrictions in energy and nutrient intake, including but not limited to calorie targets, protein, fat, carbohydrate ratios, and requirements for specific micronutrients.

[0104] "Supply period information" refers to data related to the seasons in which ingredients or food are supplied, including but not limited to seasonal attributes, market cycle, regular supply period, and changes in availability.

[0105] "Cooking skill information" refers to data used to represent the user's level of ability in cooking, including but not limited to the classification of proficiency as beginner, intermediate, and advanced, as well as the degree of tolerance for complexity.

[0106] "Available preparation time information" refers to the amount of time a user can use to prepare and cook during a particular dining scenario, including absolute time values ​​and relative time constraints.

[0107] "Inquiry information" refers to natural language or structured question data that users send to the system regarding the ingested food or diet-related issues, including but not limited to questions about ingredients, nutrition, suitable populations, and alternative solutions.

[0108] "Response information" refers to the answer content generated by the generative artificial intelligence model in response to the query information, including explanatory text, suggested solutions, precautions, and personalized responses based on the current set of features and environmental conditions.

[0109] The embodiments of this invention will be described in detail, in conjunction with the information processing system structure described in the appendix, to illustrate the roles of the server, terminal, and user within the system, as well as the data structure, algorithm flow, generative artificial intelligence model structure, and learning method within the computer, so that those skilled in the art can implement this invention accordingly. This invention is not limited to the specific embodiments described below; related modules and steps can be replaced, deleted, or combined without departing from the spirit of this invention.

[0110] I. System Overall Structure Servers are deployed on computing devices with network communication capabilities, which can be physical servers, virtual servers, or cloud computing nodes. Servers typically include a central processing unit (CPU), optional graphics processing unit (GPU), main memory, non-volatile memory, a network interface, and an operating system and middleware. Servers can run backend programs based on general-purpose operating systems, such as backend service programs running on Unix-like operating systems. Servers can also use relational database management systems (such as General Database Service) or non-relational database systems to store user data, ingredient and product data, log data, etc.

[0111] The terminal can be a smartphone, tablet computer, or desktop terminal, running a mobile operating system or a desktop operating system. The terminal is equipped with a display device, input device (such as a touch screen or keyboard), communication module, etc., and interacts with the server through a communication network.

[0112] Users interact with the server through a terminal, providing basic configurations, query requests, and feedback. Users do not directly participate in the server's internal algorithm and data structure control.

[0113] II. Server-side module composition and data structure In this embodiment, the server includes multiple functional modules, which can be implemented by one or more processing units by executing program instructions stored in a storage medium. Typically, the server includes, but is not limited to, the following modules: 1. Data Acquisition and Standardization Module The server defines structured data tables or equivalent data sets in the database for different types of information. For example, the server can define: (1) User Basic Information Table: Fields include user ID, age, gender, height, weight, etc.

[0114] (2) User Health Information Form: Fields include user identifier, health status tags (such as hypertension, diabetes, etc.), health goals (fat loss, muscle gain, etc.), and allergen list.

[0115] (3) User Preferences and Behavior Information Table: Fields include dietary preference tags (light, low-salt, no spicy food, etc.), historical dish selections, rating records, etc.

[0116] (4) Environmental Conditions Information Table: Fields include user ID, geographic location, acquisition time, temperature, humidity, weather type, and air quality.

[0117] (5) Recipe Information Sheet: Fields include recipe identifier, dish name, main ingredient list, side ingredient list, standard dosage of each ingredient, step sequence, estimated time, and nutritional components.

[0118] (6) Product Information Sheet: Fields include product identifier, name, specifications, unit price, inventory, supplier identifier, delivery range, and estimated delivery time.

[0119] (7) Feedback and Purchase History Information Form: Fields include user ID, meal plan ID, rating, text review, purchased item ID and quantity, and actual transaction amount.

[0120] The server receives raw data from forms uploaded by terminals and external interfaces (such as weather services, recipe services, and product services) through a data acquisition and standardization module. The server uses a parsing library in a general-purpose programming language to transform the structured data (such as JSON) returned by the external interfaces into an internally unified data structure, converting units, cleaning strings, and mapping categorical data to integers or vector labels. Through these processes, the server transforms the raw, multi-source, heterogeneous data into feature fields that are easily computed in memory.

[0121] Based on the above table, the server constructs a "feature set". In one implementation, the server constructs a feature vector for each user, where each dimension of the vector corresponds to a predefined feature, for example: - Basic user characteristics: normalized values ​​such as age, gender, height, and weight.

[0122] - Health and allergy characteristics: represented by multi-label binary vectors, such as the presence of a certain disease or allergy to a specific food.

[0123] - Dietary preference features: Map users' preferred tastes to sparse vectors.

[0124] - Environmental characteristics: temperature, humidity, weather type coding, etc.

[0125] - Budget and time characteristics: acceptable budget limit, scaled values ​​of available cooking time.

[0126] The server can organize these features into structured objects (such as key-value pairs) and, when necessary, convert them into embedded vectors or text fragments for generative AI models to read.

[0127] 2. Feature Fusion and Candidate Set Generation Module The server filters candidate items from the recipe information table and the product information table based on a set of features. The server can perform the following operations: The server queries the recipe information table and removes recipes that do not meet health or allergy restrictions. For example, the server checks whether an ingredient in a recipe is included in the user's allergy list; if so, the recipe is excluded. The server can sort the remaining recipes by indicators such as cooking time, calories, and protein content. This sorting can be achieved using a simple weighted scoring function, such as calculating a score for each recipe as: w1·health fit + w2·time fit + w3·preference similarity.

[0128] After identifying several candidate recipes, the server searches for corresponding candidate products for each ingredient in the product information table. The server can perform mapping using string matching, thesaurus, and simple embedding similarity calculations. The server stores the results in a candidate mapping table, recording the list of candidate products and their prices for each ingredient. The server further calculates an estimated cost range based on combinations of candidate products, which is used to express budget-related constraints in subsequent prompts.

[0129] Through this preliminary candidate screening and mapping, the server can narrow the search space when providing contextual information to the generative artificial intelligence model, reduce the burden on the model when outputting, thereby reducing the amount of computation and improving the response speed.

[0130] 3. Prompt Statement Generation Module The server automatically constructs prompts for generative artificial intelligence models based on the feature set, candidate recipes, and product information. The prompts are in the form of natural language text.

[0131] The server can use string templates to write structured features into a fixed-format prompt statement. For example, in one implementation, the server generates the following prompt statement: "You are a smart assistant that combines the functions of a nutritionist and a chef."

[0132] The given conditions are as follows: 1. User basic information: 30 years old, female, height 165cm, weight 60kg, currently controlling body fat.

[0133] 2. User's health status: Slightly high blood pressure, requiring control of sodium intake; no diabetes; allergic to peanuts and shrimp.

[0134] 3. User dietary preferences: They prefer light flavors, like chicken, fish and various vegetables, and dislike offal.

[0135] 4. Today's weather at the user's location: High of 33°C, Low of 26°C, High humidity, No rain.

[0136] 5. User budget: The average budget for dinner today is no more than 40 yuan per person.

[0137] 6. User time limit: The time available for preparing dinner is no more than 30 minutes.

[0138] 7. The recipes available in this system from the recipe database include Chinese home-style dishes, Western-style light meals, and simple salads.

[0139] Based on the above conditions, please generate a meal plan suitable for today's dinner for the user, recommending only 1-2 main courses and 1 simple side dish. Requirements: - Overall energy content is moderate, suitable for controlling body fat; - Low sodium content, avoid making it too salty; - Does not contain peanuts and shrimp; - Keep the method as simple as possible, and it should be completed within 30 minutes; - Explain the general method and main ingredients of each dish.

[0140] Please output in a structured format, including: 'Reasons for Recommendation', 'List of Dishes', 'Main Ingredients and Quantities for Each Dishe', and 'Simple Steps'. The server can also generate simplified or task-oriented prompts, such as: "Based on the following information, generate dinner suggestions for today for the user:" - Health status: Needs a low-sodium diet, currently losing fat; - Food preferences: Prefers refreshing, cold dishes; does not eat beef. - Weather: Current temperature 30℃, feels hot; - Budget: Total price not exceeding 60 yuan; Please recommend two dishes, and provide the name of each dish, the ingredients needed, and a brief description of how to prepare it. The server can also include information on available ingredient inventory in the prompt message, such as: "The following ingredients are currently on sale at supermarkets near the user: chicken breast, tomatoes, lettuce, cucumbers, eggs, tofu, and canned corn."

[0141] User's health status: No serious illness, desires to increase protein intake and reduce fat intake.

[0142] Weather: Hot and humid.

[0143] Based on these available ingredients, please design a simple dinner meal plan for one person. Requirements: - The total intake should be kept around 700 kcal; - Prioritize using the listed discounted ingredients; - It's easy to make and perfect for quick home cooking; It will also output the dish name, ingredient list, and cooking steps. The server constructs prompts in the manner described above, aggregating discrete features into a coherent semantic context. This allows generative AI models to consider multiple constraints simultaneously in a single inference iteration. Compared to directly inputting scattered parameters, the structured and semantic prompts significantly reduce multi-turn interactions and the number of network requests, thereby reducing communication load and improving overall response efficiency.

[0144] 4. Generative Artificial Intelligence Model Module The server inputs the prompts into the generative artificial intelligence model. This model can be a sequence-to-sequence language model based on the Transformer architecture, deployed on an inference node with a GPU or accessed through an external inference service.

[0145] In one implementation, the server uses a pre-trained language model fine-tuned for a dietary scenario. The model's structure may include a multi-layered self-attention encoder and decoder, each layer containing a multi-head attention sublayer and a feedforward network sublayer. During the fine-tuning phase, the server performs supervised learning using large-scale training samples in the form of "user features + weather + recipe + reasonable output." During training, the server employs a cross-entropy loss function to measure the difference between the model's output sequence and the target sequence, and updates the model parameters using a backpropagation algorithm. Weight updates can be based on adaptive learning rate optimization methods, and gradient pruning, regularization, and data augmentation (e.g., perturbing some input values ​​or adding synonym variants) can be used during training to improve model robustness.

[0146] During the inference phase, the server processes the prompts through word segmentation and vector embedding, inputting them into the model's encoding section. The decoding section then generates the output sequence word by word. The server can control sampling parameters (temperature, top-k, top-p, etc.) to constrain the diversity and stability of the generated sequence. In certain implementations, the server requires the model to output structured text with specific markers (e.g., "[dish name]", "[ingredients]", "[steps]", etc.) so that the post-processing module can accurately parse it.

[0147] Compared to traditional systems that generate responses based on rules or templates, the aforementioned model utilizes the nonlinear representation capabilities of deep neural networks to map high-dimensional features to the natural language output space, enabling the generation of more detailed and personalized dietary plans under multiple constraints. Through pre-training and fine-tuning processes, the server equips the model with rich semantic knowledge and performs domain adaptation based on the system's specific data, thereby improving inference accuracy and generalization ability at the computer technology level.

[0148] 5. Results Analysis and Diet Plan Construction Module After receiving the output from the generative artificial intelligence model, the server parses the text. Using predefined tagging patterns or regular expressions, the server extracts information such as dish names, ingredient names and quantities, cooking steps, and reasons for recommendation from the output text. The server then maps this information back into a structured data structure to create a meal plan record.

[0149] For each ingredient, the server searches for matching products in the product information table and calculates the minimum cost, average cost, or cost range for different combinations. The server can mark high-priced products as alternatives and prioritize lower-priced products that meet the criteria. The server then generates a purchase candidate set and corresponding cost information, forming a purchase list.

[0150] In this way, the server transforms natural language output into structured data that can be used for subsequent machine-readable processing, realizing automatic conversion from language space to data space, reducing manual work, and improving the automation and accuracy of data processing.

[0151] 6. Feedback Processing and Model Input Optimization Module The server receives user reviews and actual purchase data uploaded by users and writes them into a feedback and purchase history information table. The server periodically analyzes this data, such as calculating the adoption rate of each recipe, the rating distribution, and the consistency between actual user purchases and suggested purchases.

[0152] Based on these statistics, the server can adjust the construction method of the feature set and the prompt template. For example, for food categories that users have repeatedly rejected, the server can reduce their weight, decreasing their probability of appearance in the candidate set. The server can also add content such as "Please avoid ingredients that users have frequently rejected before" to the prompt, thereby guiding the model to output suggestions that better align with users' long-term preferences.

[0153] This feedback-based iterative update enables the server to adaptively improve feature representation and suggestion strategies. Compared to a static system, it can continuously optimize the data processing and reasoning process inside the computer over time, improve long-term prediction accuracy and reduce the number of invalid recommendations, thereby reducing the overall consumption of computing and communication resources.

[0154] III. Composition and Function of Terminals In this embodiment, the terminal primarily serves as the user interface and handles local interactive control. The terminal can run client programs based on mobile or web applications and exchange data with the server via secure communication protocols.

[0155] After obtaining meal plan and purchase list information from the server, the terminal parses them into interface components. In one embodiment, the terminal displays the recommended menu in card format, with each card showing the dish name, image, main ingredients, estimated cooking time, and nutrition label. The terminal also displays a shopping list on the interface, grouped by ingredient type, showing candidate products and prices for each ingredient.

[0156] The terminal accepts user input, including confirmation, menu replacement, number of participants, and budget adjustment. The terminal converts these operations into structured instructions and sends them to the server, which then reconstructs the feature set and prompts to generate a new meal plan. This human-computer interaction method decouples the terminal's graphical interface layer from the server's intelligent inference layer, achieving low-latency and highly flexible collaboration.

[0157] IV. The Role of Users In this implementation, users provide personal information and preferences through the terminal, initiate queries, browse generated results, and provide feedback. Users can also input free-form query statements on the terminal, such as: "I'm very tired today and don't want to cook anything too complicated. Can you recommend a low-calorie dinner that can be made in under 20 minutes?" The terminal sends the query information along with the server's existing set of features. Based on this, the server constructs a prompt containing the query information and instructs the generative AI model to generate not only a direct response text but also corresponding supplementary meal plan information. In this way, users can trigger personalized computational processes through natural language, rather than strictly relying on fixed menu selections, thereby enhancing the system's human-machine collaboration capabilities.

[0158] V. Explanation of Technical Effects and Causal Relationship Through the above modular design and specific implementation, the present invention achieves the following technical effects at the computer technology level: 1. The server uses a unified set of features to represent multi-source heterogeneous data, which enables generative artificial intelligence models to process more structured data during the input stage, thereby reducing redundant calculations in feature engineering and improving inference efficiency.

[0159] 2. By pre-screening recipes and product candidates, the server reduces the effective search range of the model in the output space, making the prompts more focused, significantly reducing the probability of the model generating irrelevant content, improving overall inference accuracy and shortening inference time.

[0160] 3. By explicitly embedding constraints such as budget, health status, and time limits in the prompt statements, the server moves complex constraints from post-rule checking to the inference stage inside the model, reducing the number of post-processing filters and lowering the computational and communication overhead in the overall processing chain.

[0161] 4. The server maps the natural language results into a unified data structure through structured parsing of the model output, and generates a purchase list in conjunction with product data. This makes the system output not only readable, but also able to directly drive subsequent processing units (such as the order generation module), realizing an integrated closed loop from intelligent reasoning to actual device and service control.

[0162] 5. By collecting user reviews and purchase history, the server dynamically adjusts feature values ​​and prompt strategies, continuously optimizing input data and model usage, rather than simply accumulating logs. This adaptive adjustment mechanism for data representation and prompt generation gradually improves the accuracy of the computer system and reduces useless output over long-term operation, thereby saving computing resources and improving response quality.

[0163] 6. Compared with traditional rule-based systems, the generative artificial intelligence model in this invention uses a deep neural network structure for nonlinear mapping, which has stronger expressive power. However, without constraints, it is prone to producing uncontrollable results. Through specially designed feature construction and prompt statements, the server technically applies soft constraints to the model's output space, achieving a trade-off between high expressive power and high controllability. This represents a technical improvement to the way generative models are invoked.

[0164] In summary, this implementation method, through the specific division of server, terminal, and user roles, and the detailed design of the collaborative operation of feature set, prompt statements, and generative artificial intelligence model, enables the system to not only automatically generate personalized meal plans and purchase lists, but also substantially improve upon traditional computer systems in terms of processing speed, accuracy, data management, and utilization of computing resources.

[0165] use Figure 11 The processing flow is explained.

[0166] Step 1: The server receives and standardizes user-related data.

[0167] Input: User's basic information (age, gender, height, weight, etc.), dietary preference information (taste preferences, dietary restrictions), health status information (disease tags, fat loss / muscle gain goals, etc.) uploaded by the terminal, allergen information, and user-authorized synchronized health records (weight history, steps, heart rate, etc.).

[0168] The server parses the request message from the terminal, performs integrity checks and type validation on the fields, unifies the units of numerical values ​​such as height and weight, maps text-based preferences to internal tags (such as mapping "light" to taste_light=1), and converts allergenic ingredients into a list structure.

[0169] The server processes heterogeneous input data into structured records through data standardization, writes them into user information tables and health information tables, and generates a draft of user features in memory for subsequent construction of feature sets.

[0170] Output: Standardized user records stored in the database, and a preliminary user feature data structure in memory.

[0171] Step 2: The server obtains and processes environmental condition information.

[0172] Input: User's geographical location information (coordinates reported by the terminal or preset city), current time information.

[0173] The server calls an external meteorological service interface, sends a request containing location and time parameters, and receives meteorological data such as temperature, humidity, and weather type.

[0174] The server performs format parsing, unit unification, and missing value handling on meteorological data, unifying temperature to degrees Celsius, converting humidity to percentage form, and generating weather labels (such as "hot and humid" or "cool and cloudy") based on the numerical range.

[0175] The server writes the processed meteorological data into the environmental conditions information table and updates the environmental feature vector corresponding to the user.

[0176] Output: Environmental condition records stored in the database, and environmental feature data attached to user characteristics.

[0177] Step 3: The server filters a candidate set from recipe and product data sources.

[0178] Inputs: Draft user characteristics (including health tags, allergens, and preferences), environmental characteristics, recipe information table, and product information table.

[0179] The server performs filtering operations on the recipe information table, removing recipes that contain ingredients that the user is allergic to or that conflict with major health contraindications. Then, it calculates a rating for each recipe based on cooking time, calories, protein content, user preference tags, and other factors.

[0180] The server selects several dishes as a candidate recipe set based on the scores from high to low, and performs a matching query on the product information table for the ingredient names in each candidate recipe. It then uses string similarity or a thesaurus to map the ingredients to several product entries.

[0181] The server sorts the matched product entries by price, inventory, and delivery time, and calculates the minimum price, average price, and number of available products for each ingredient, generating a candidate mapping table from ingredients to products.

[0182] Output: A set of candidate recipes, along with corresponding ingredient-product candidate mappings and preliminary cost statistics.

[0183] Step 4: The server constructs a set of features and generates prompt statements.

[0184] Inputs: Standardized user information, environmental condition records, candidate recipe set, ingredient-product candidate mapping, budget information, and time constraint information.

[0185] The server encodes the above types of information into a unified set of features, including numerical features (such as budget and time), discrete features (disease labels and weather labels), and multi-label features (preferences and allergens), and converts some of the information into readable text fragments.

[0186] The server uses a predefined template to concatenate user information, health status, taste preferences, weather description, budget and time constraints, and range of available recipes into natural language prompts in a fixed order. The prompts explicitly request the generative AI model to output structured information such as "reason for recommendation, list of dish names, main ingredients and quantities, and simple steps".

[0187] Output: One or more complete prompt texts for the current user scenario, and a corresponding set of feature objects.

[0188] Step 5: The server calls a generative artificial intelligence model to generate dietary plan information.

[0189] Input: Prompt text constructed by the server, and a set of internal features (which can be used for additional context of the external model or for the local model embedding layer).

[0190] The server sends the prompt to the inference interface of the generative artificial intelligence model. The model segments and vectorizes the text, and after multi-layer attention and feedforward network calculations, it generates the output text word by word.

[0191] The server sets sampling parameters (such as temperature, top-k, top-p, etc.) when invoked and may generate multiple times to select the higher-quality results. The server receives the complete text of the model's output as candidate outputs for meal plans.

[0192] Output: A natural language meal plan text containing recommended menus, reasons for recommendation, ingredients and quantities, cooking steps, etc.

[0193] Step 6: The server parses the generated results and structures the meal plan information.

[0194] Input: Meal plan text output by a generative artificial intelligence model, and predefined structured patterns (tag format or regular expression rules).

[0195] The server uses key tags, regular expressions, or simple parsing rules to identify dish names, corresponding ingredients, dosage descriptions, and steps from text, maps them into structured records, and stores them in a meal plan table.

[0196] For each parsed ingredient, the server extracts a list of products from the aforementioned ingredient-product candidate mapping, estimates the required quantity of each product based on usage, calculates the individual and total costs, and verifies whether it is within the budget. If it exceeds the budget, the server can record the plan as suboptimal and mark it as needing to be regenerated.

[0197] Output: Structured meal plan record, including a list of dishes, ingredient details, estimated prices, and association information with product IDs.

[0198] Step 7: The server generates the final purchase list information and links it to the product data.

[0199] Inputs: Structured meal plan records, ingredient-product candidate mapping, and user budget constraints.

[0200] The server aggregates the demand for the same ingredient in different dishes to obtain the total demand. Based on the aggregated demand, the server selects the appropriate product combination (e.g., multiple small packages or one large package) and calculates the total price for each combination.

[0201] The server sorts all product combinations from lowest to highest cost, prioritizes combinations that meet the budget and have sufficient inventory, and generates a purchase list containing product ID, name, specifications, quantity, unit price, subtotal, and total price.

[0202] Output: One or more purchase lists corresponding to the current meal plan, serving as an actionable shopping plan.

[0203] Step 8: The terminal displays meal plans and purchase lists, and receives user selection information.

[0204] Input: Structured meal plan data and purchase list information returned by the server.

[0205] The terminal parses the received JSON or equivalent data structure and maps information such as menu, ingredients, and prices to interface components, displaying them on the screen as a list or card, while also showing the total price and nutritional tips.

[0206] The terminal provides interactive controls (buttons, checkboxes, drop-down lists, etc.) that allow users to replace dishes, switch brands of goods, or delete certain items, and converts these operations into selection information (such as selection status and replacement instructions) and caches them locally.

[0207] After user confirmation, the terminal will upload the selection information to the server via the network for updating subsequent processing.

[0208] Output: The user's confirmed selection information and possible modification instructions on the terminal interface are sent to the server.

[0209] Step 9: The server updates the purchase instruction based on the user's selection information and triggers the external purchase service.

[0210] Input: Selection information from the terminal (including confirmed menu and product combinations, adjusted quantities, etc.) and original purchase list information.

[0211] The server compares the user's selection with the original plan, performs difference calculations, and generates the final purchase candidate set. The server packages this data into instruction information (such as product identifier, quantity, delivery location, etc.) adapted to external purchasing services.

[0212] The server sends the instruction information to the purchase processing device or external purchase service through the interface and receives the returned order confirmation or error information; if an error is returned (such as insufficient stock), the server can go back to the purchase list generation stage to reselect candidate combinations.

[0213] Output: Purchase instruction information successfully sent to the external purchasing service, along with corresponding order confirmation information or error status data.

[0214] Step 10: The terminal displays the order status to the user and assists in completing subsequent operations.

[0215] Input: Order confirmation information or error message returned by the server.

[0216] The terminal displays status information such as order number, estimated delivery time, and total amount on the interface; if an error occurs (such as some products being out of stock), the terminal displays the reason for the error and provides operation entry points such as "reselect" and "change products".

[0217] When the terminal requires additional confirmation or payment from the user, it guides the user to an external payment page or an integrated payment module and displays the corresponding status update (such as payment successful or payment failed).

[0218] Output: The user's confirmation of the order status, and possible secondary operations (such as payment completion information, cancellation instructions) are fed back to the server or external services.

[0219] Step 11: Users provide feedback after their meal.

[0220] Input: The user's subjective feelings about the implemented meal plan, including ratings, written reviews, and whether they would like to choose it again, are entered through the terminal.

[0221] Users select a star rating on the terminal, enter a short review (such as "tastes good but there are too many steps", "the portion is just right", "too oily", etc.), and click the submit button.

[0222] The terminal sends the feedback information to the server in a structured format (score value, text content, and corresponding meal plan ID).

[0223] Output: User review information and associated meal plan identifiers received by the server.

[0224] Step 12: The server uses feedback and purchased historical updated feature sets to optimize subsequent prompt statements.

[0225] Input: User review information, actual purchase history, original feature set, and prompt statement template.

[0226] The server writes the evaluation information and purchase history into the feedback and purchase history information table, and calculates statistical indicators, such as the average rating of a certain type of dish, the probability of a certain ingredient being selected again after use, and the consistency rate between suggestions and actual purchases.

[0227] The server adjusts feature weights based on statistical results (e.g., reducing the weight of dish categories with low ratings) and adds new dimensions to the feature set (e.g., "users dislike recipes with more than N steps"). Simultaneously, the server updates the prompt template, adding constraints such as "Please avoid overly complex dishes" and "Prioritize flavor types that have received high user ratings."

[0228] The server will then use the updated set of features and templates when generating prompts and meal plans for the user the next time, enabling the generative AI model to automatically favor results that better align with the user's long-term preferences during the inference phase.

[0229] Output: The updated set of features, the prompt statement generation template, and the long-term preferences and weight configurations recorded in the database, which will be used as input in subsequent loops.

[0230] 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".

[0231] With the development of information processing and network technologies, systems that utilize computing devices to provide users with dietary advice and ingredient procurement support are gradually increasing. However, existing technologies generally suffer from the following problems. First, most existing systems rely solely on static recipe data or simple rule engines for recommendations, lacking the ability to comprehensively process individual user information (including preferences, allergies, health status, etc.) and dynamic environmental information (such as current temperature and weather conditions), resulting in insufficient personalization of the generated dietary plans. Second, existing systems often treat "menu recommendation" and "ingredient procurement" as independent processes. Computing devices struggle to directly convert the structured results generated from the menu into data structures that can be used by product retrieval and ordering interfaces within the same data stream, forcing users to repeatedly operate across multiple applications or interfaces, increasing the interaction burden. Third, although generative artificial intelligence models can already generate complex text content based on natural language, existing technologies lack a unified mechanism at the system level for constructing, managing, and utilizing prompts for generative artificial intelligence models. This makes it difficult to effectively convert multi-source data scattered across databases, external service interfaces, and user terminals into high-quality model input, thus limiting the controllability and computability of the model output. Fourth, in the existing system, the server's calls to weather services, recipe data, and shopping platform interfaces are mostly discrete and functionally segmented. There is a lack of an integrated computing process centered on a menu structure that automatically connects "individual data acquisition → environmental data acquisition → prompt statement construction → model-generated menu → product retrieval → order generation," thus failing to fully leverage the advantages of computing devices in data integration and automated decision-making.

[0232] Therefore, how to improve the computing and data processing architecture on the server side, enabling the server to automatically acquire and parse individual user information and environmental information, generate high-quality prompts based on preset templates, parse the output of generative artificial intelligence models into executable structured menu data, and further automatically drive product retrieval and order generation processing, thereby reducing user operations while improving recommendation accuracy and overall system computing efficiency, has become an urgent computer technology issue to be solved in this field.

[0233] 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.

[0234] In this invention, the server includes a computing and storage device for receiving individual information, including user preferences, allergy information, health status information, and location information, from a user terminal and storing it in an information storage unit; a data acquisition and parsing device for acquiring weather forecast information from an external information providing device via a communication network and parsing it to generate an environmental information data structure; a prompting statement generation device for reading individual information from the information storage unit and combining it with environmental information, cooking recipe information, and product sales information; and a prompting statement generation device for constructing prompting statements for input into a generative artificial intelligence model according to a pre-set text template; and a device for calling the generative artificial intelligence model to generate multiple diet plans, a list of required raw materials, and other data based on the prompting statements. The system includes: a model calling and result parsing device for parsering recommended menu information, including production steps, into structured menu data; an interactive control device for sending the structured menu data to a user terminal and receiving user selection information for the target menu; a data matching and filtering device for constructing product search conditions based on the raw material list in the target menu and calling the search interface of the product providing device to obtain multiple sales product information, filtering and sorting the sales product information to generate procurement candidate information; and an order generation and execution device for receiving the final procurement content determined by the user on the user terminal, calling the order placement interface of the order processing device to generate a delivery order, and automatically executing the order generation and execution of the food procurement and delivery processing corresponding to the recommended menu. This allows for the formation of a unified data and computation chain within the server, encompassing "individual information collection, environmental information acquisition, automatic construction of prompts, invocation of generative AI models, generation of structured menus, product retrieval and filtering, and automatic order generation." This transforms the output of the generative AI model from natural language content into structured data that can be directly consumed by the program, reducing manual operations between different systems, lowering interaction complexity, and improving the accuracy of menu recommendations and the efficiency of resource allocation through centralized server computing and data processing. Ultimately, this enhances the integrated processing capabilities of dietary recommendations and ingredient procurement based on generative AI models at the computer technology level.

[0235] "Information processing device" refers to an electronic device that includes a computing unit, a storage unit, and a communication interface, used to process and store input data and send and receive data through a network.

[0236] The "computation unit" refers to the processing unit in an information processing device that executes program instructions, performs logical operations on input data, controls the flow of data, and performs data conversion.

[0237] The "information storage unit" refers to a data storage unit used to store data such as individual information, environmental information, cooking recipe information, product sales information, and output results from generative artificial intelligence models in a read-write manner.

[0238] "User terminal" refers to a human-computer interaction device operated by a user for inputting personal information, displaying recommended menu information, and confirming products and orders.

[0239] "Individual information" refers to a set of data related to a specific user, including information on preferences, allergies, health status, location, budget, cooking skills, and available preparation time.

[0240] "Preference information" refers to information about a user's dietary preferences, including preferred cuisines, taste preferences, and favorite and least favorite ingredients.

[0241] "Allergy information" refers to restrictive information indicating that a user has an allergic reaction to a specific food or ingredient, which is used for exclusion control during menu generation and product recommendation.

[0242] "Health status information" refers to information related to a user's current or long-term health status, including health management-related data such as weight management goals, chronic disease status, and dietary restrictions.

[0243] "Location information" refers to data used to indicate the geographical area where a user is located, including geographic coordinates, city names, or area identifiers, in order to obtain environmental information and information on deliverable goods in the corresponding area.

[0244] "Budget information" refers to data that indicates the range of total expenses a user can accept in a single transaction or over a certain period, and is used to limit the total price of recommended menus and product combinations.

[0245] "Nutritional needs information" refers to information indicating a user's needs or limitations in terms of nutrients such as energy, protein, fat, carbohydrates, vitamins, and minerals.

[0246] "Cooking skill level information" refers to data indicating a user's proficiency in cooking operations, which is used to control the complexity of the preparation steps in the recommended menu.

[0247] "Available preparation time information" refers to data indicating the length of time a user can use for ingredient preparation and cooking operations, which is used to limit the preparation time range of recommended menus.

[0248] "Weather forecast information" refers to forecast data on weather conditions such as temperature, weather conditions, and humidity in a specific area at a predetermined time, provided by an external information providing device via a network.

[0249] "Environmental information data structure" refers to structured data generated based on meteorological forecast information and extracting elements related to dietary recommendations (including current temperature, weather conditions, etc.).

[0250] "Cooking recipe information" refers to recipe data related to food preparation, including dish name, list of required raw materials, amount of raw materials, seasonings, and preparation steps.

[0251] "Product sales information" refers to data related to salable products provided by the product providing device, including product name, specifications, price, inventory status, and delivery conditions.

[0252] "Information providing device" refers to a network service device that provides external data such as weather forecast information, cooking recipe information, or product sales information to a server.

[0253] "Product provision device" refers to a service device that provides product search and order placement interfaces through the network and performs product sales and delivery processing.

[0254] "Order processing device" refers to an information processing system used to receive order information, generate actual transaction orders, and manage order status and delivery process.

[0255] "Generative artificial intelligence models" refer to artificial intelligence models built based on machine learning methods that can automatically generate text or structured information based on input prompts.

[0256] "Prompt statements" refer to text data constructed by the server based on individual information, environmental information, cooking recipe information, and product sales information according to a preset text template, which is used to input into a generative artificial intelligence model to control the content it generates.

[0257] "Recommended menu information" refers to text or data output by the generative artificial intelligence model based on prompts, which includes multiple dietary plans, a list of raw materials required for each plan, and instructions on how to prepare them.

[0258] "Structured menu data" refers to menu data parsed from recommended menu information and represented by a predetermined data structure, which includes at least fields that can be processed by the program, such as menu identifier, dish name, ingredient name and quantity, and preparation steps.

[0259] "Target menu" refers to the specific menu selected by the user from multiple candidate menus in the recommended menu information, which serves as the basis for subsequent product retrieval and order generation.

[0260] The "raw material list" refers to the collection of various ingredients and their quantities required to create the target menu, extracted from structured menu data.

[0261] "Product retrieval interface" refers to the programming interface that the product providing device opens to the outside world, allowing users to obtain information on multiple corresponding products for sale by inputting search criteria.

[0262] "Sales product information" refers to the data set of salable products that correspond to the names or categories of raw materials, obtained through the product search interface.

[0263] "Purchase candidate information" refers to a set of candidate purchase products available for users to choose from, obtained by filtering and sorting based on indicators such as price, inventory, and delivery conditions, according to sales product information.

[0264] "Order information" refers to data generated based on the final purchase content determined by the user on the user terminal, used to request the order processing device to create a delivery order, including product details, quantity, receipt information and payment-related information.

[0265] "Delivery order" refers to business order data generated by the order processing device based on order information, used to execute the picking, delivery and settlement of goods.

[0266] "Interactive control" refers to the data interaction and interface linkage control between the server and the user terminal, which guides users to complete operations such as inputting personal information, selecting menus, and confirming products and orders.

[0267] The embodiments of this invention will be described primarily using a server, a terminal, and a user. Those skilled in the art can replace or extend the hardware platform and software components without departing from the spirit of this invention.

[0268] I. Overall Hardware and Software Composition of the System A server runs on one or more computing devices, including a central processing unit (CPU), main memory, non-volatile storage devices (such as solid-state drives), and a network interface. Servers can be deployed in data centers or cloud computing platforms. Servers run operating systems, such as general-purpose server operating systems, and application server software.

[0269] At the software level, servers include: The server uses an interpreted programming language and its runtime environment as the backend logic execution environment; The server uses an HTTP-based application framework to build network interfaces and implements REST-style application programming interfaces. The server uses a relational database management system (RDBMS) as the information storage unit to persistently store data such as individual information, environmental information, cooking recipe information, product sales information, and output results of generative artificial intelligence models. The server uses an HTTP client library to send network requests to external information providing devices and goods providing devices; The server uses the application programming interface provided by the generative artificial intelligence service to call the generative artificial intelligence model.

[0270] A terminal is a user-operated computing device, such as a smartphone, tablet, or other portable information terminal with a display and input device. The terminal runs a terminal operating system and installs and runs applications developed using a cross-platform user interface framework. The terminal sends and receives data with the server through secure communication protocols.

[0271] Users input information and confirm results through the terminal's touchscreen and graphical user interface. Instead of directly interacting with generative AI models or external information providers, users interact indirectly through centralized processing on a server.

[0272] II. Server-side data structure and module composition In order to achieve the functions described in the claims, the server divides its internal processing into multiple logical modules, and the modules exchange information with each other through a pre-designed data structure.

[0273] The server establishes tables or equivalent data structures in the information storage department for at least the following data: The server creates a user information table for individual information, with fields including user identifier, hobby information field (e.g., text type, enumeration type), allergy information field, health status information field (e.g., weight management goals, disease markers), budget information field, location information field, cooking skill level field, available preparation time field, and timestamp field.

[0274] The server creates an environmental information table with fields including region identifier, temperature, weather conditions, humidity, weather forecast time, data source identifier, and update time.

[0275] The server creates a recipe table for cooking recipe information. Fields include recipe identifier, dish name, recipe category, raw material list (which can be implemented through sub-tables), raw material quantity, seasonings, preparation steps, average preparation time, etc.

[0276] The server creates a product table for product sales information, with fields including product identifier, product name, category, corresponding raw material name, price, inventory status, delivery area, and estimated delivery time.

[0277] The server generates a menu result table for the generative artificial intelligence model, with fields including menu identifier, user identifier, generation time, title, structured raw material list, production step text, and model parameter information.

[0278] The server divides its functional logic into the following modules: The server is equipped with an individual information management module, which is used to receive, verify, and store the individual information uploaded by users.

[0279] The server is equipped with an environmental information acquisition and parsing module, which is used to send requests to the meteorological information providing device and parse the response into an environmental information data structure.

[0280] The server is configured with a prompt message generation module, which reads individual information, environmental information, cooking recipe information, and product sales information from the database, and constructs prompt messages based on a predefined text template.

[0281] The server is configured with a generative artificial intelligence model invocation module, which is used to send prompts to the generative artificial intelligence model and receive the generated results.

[0282] The server is equipped with a menu parsing and structuring module, which is used to parse the natural language recommendation menu information output by the generative artificial intelligence model into unified structured menu data.

[0283] The server is equipped with a product retrieval and filtering module, which generates search criteria based on the raw material list in the structured menu and selects the best candidate products obtained from the product supply device.

[0284] The server is configured with an order generation and execution module, which is used to construct order information based on the final purchase content confirmed by the user on the terminal, and call the order processing device's order placement interface to generate a delivery order.

[0285] The server is equipped with an interactive control module to manage communication with the terminal, including data format conversion, error handling, and logical control to minimize interactive steps.

[0286] III. Structure and Learning Methods of Generative Artificial Intelligence Models The server uses a deep learning-based language generation model as the generative AI model in the generative AI model invocation module. Logically, this model can be viewed as a neural network with the following structure: The server assumes the generative AI model is a multi-layered self-attention sequence-to-sequence model, containing a word embedding layer, a positional encoding layer, several self-attention encoding layers, and a decoding layer. Each self-attention encoding layer includes a multi-head self-attention sublayer and a feedforward network sublayer, using residual connections and normalization operations. The model is trained by maximizing the log-likelihood of the target sequence on the training corpus.

[0287] In describing this invention, the server treats the model input as a sequence of tokens encoded from prompts. After converting the prompts into token indices, the server inputs them into a word embedding layer, followed by forward propagation through a multi-layered self-attention structure. The model automatically learns the relevance of different parts of the sentence through internal attention weights, such as the correspondence between constraints like "allergy: dairy products" and "recommended dishes should not contain cheese, milk, or butter." The model's output is a natural language text, which the server parses to obtain various fields of the dietary plan.

[0288] The server description states that the model uses cross-entropy loss as the error function during the training phase, updates the neural network weights through backpropagation, and can employ optimization algorithms (such as gradient descent with momentum or adaptive learning rate). Based on a pre-trained general language model, the server can fine-tune the model by introducing specialized corpora related to diet, nutrition, and food procurement, thereby improving the generation quality of the model in this domain.

[0289] During the deployment phase, the server calls the model through a remote service interface. At this point, the model's weights are fixed and are not updated during system operation. The server controls the structure of the input prompt statements, thereby improving the matching degree between the output results and the required data structure without changing the internal structure of the model, achieving controllability and parsability of the model output.

[0290] IV. Construction Methods and Technical Effects of Prompt Statements The server employs a structured text template in its prompt generation module, rather than simply concatenating existing text. It not only directly lists user preferences, allergies, and weather conditions, but also includes explicit constraints and output format requirements in the prompts, ensuring that the generative AI model automatically adheres to these technical constraints during generation. For example, the server can generate the following Chinese prompt: "You are a professional nutritionist and family cooking consultant."

[0291] Based on the user and weather information provided below, generate 2-3 suitable dinner menus for today. Requirements: 1. It is essential to strictly avoid the user's allergens; 2. Meets the user's health goals; 3. Consider the current weather and make the dishes comfortable to eat (for example, refreshing cold dishes are suitable for hot weather). 4. Each menu item includes: dish name, brief description, detailed ingredient list (including approximate quantities), and simple preparation steps.

[0292] User Information - Food preferences: Japanese cuisine, light flavors - Allergy: Dairy products - Health status: Currently on a weight loss program, needs to control fat and total calorie intake. Weather information - Temperature: 30℃ - Weather conditions: Sunny Please generate a dinner menu that meets the above criteria. By encoding constraints and output structures into prompts in the aforementioned manner, the server facilitates the generative AI model in assigning higher weights to constraints within its internal attention mechanism. This reduces the generation of content that does not conform to the constraints, thereby indirectly improving the accuracy of menu recommendations and the success rate of subsequent parsing. This method of controlling model behavior by constructing specific prompts, compared to traditional rule engines, leverages the ability of neural networks to model complex constraint relationships in high-dimensional space, representing an improvement in the computer's internal control strategy.

[0293] V. Menu Parsing and Structured Data Generation In the menu parsing and structuring module, the server converts the natural language text output by the generative artificial intelligence model into structured menu data. To ensure efficient parsing, the server pre-defines the prompts to require the model to output paragraphs divided by specific headings or numbers, and requires the ingredient list and steps to be presented in a fixed format, such as one ingredient or step per line.

[0294] During parsing, the server uses string analysis algorithms to segment the output text, locate keywords, and perform pattern matching. For example, the server can detect identifiers such as "dish name:", "ingredients:", and "steps:" and use these as boundaries to divide the text into multiple fields. The server writes each field into structured data, forming a data object containing menu identifiers, dish names, ingredient names and quantities, and preparation steps.

[0295] By employing this top-down parsing structure, the server avoids complex word-by-word natural language understanding processing, reducing computational load and enabling large-scale concurrent parsing on the server side. This results in superior computational efficiency compared to algorithms that rely on manual parsing or highly complex semantic analysis.

[0296] VI. Algorithm Design for Product Retrieval and Filtering In the product retrieval and filtering module, the server maps the raw material list in the structured menu data to the search criteria of the product providing device. The server constructs search keywords for each raw material and can further add filtering conditions (such as fat content, whether it is organic, etc.) based on the raw material's category, user budget, and health restrictions.

[0297] The server uses a sorting algorithm to calculate product scores based on the sales product information returned by the product providing device, taking into account factors such as price, estimated delivery time, inventory status, and user's historical purchase preferences. The server can assign weight coefficients to each indicator, obtain the overall score through weighted summation, and sort the scores from high to low to generate procurement candidate information.

[0298] By centrally calculating and sorting product candidates, the server reduces the need for users to manually compare a large number of products on their terminals. Furthermore, the server performs optimized calculations internally, enabling data to be combined optimally in a single search. This reduces the number of communication round trips between the server and the product delivery device, thereby reducing network load and improving response speed.

[0299] VII. Explanation of Technical Effects and Causal Relationship Through the modular design and data flow structure described above, the server ensures that this system does not merely mechanically automate human menu selection and shopping processes, but rather improves the way data is represented and processed within the computer: By using a unified individual information data structure and an environmental information data structure, the server enables heterogeneous user preference data, health restriction data, and meteorological data to be represented in a unified format within the server, reducing redundant data conversion and improving the consistency of data management.

[0300] The server uses a prompt generation module to construct the text of the input generative AI model in a templated and constrained manner, making the model output easier to parse, reducing parsing errors, and thus improving the accuracy of end-to-end processing.

[0301] The server leverages the high-dimensional representation capabilities of generative artificial intelligence models to map complex user and environmental constraints into a reasonable food combination space. Compared to traditional rule-based or simple scoring algorithms, it can converge to a candidate set that satisfies multiple conditions more quickly in multi-constraint scenarios, resulting in improved recommendation accuracy and diversity under the same hardware resources.

[0302] The server uses a unified structured menu data as an intermediate representation internally, decoupling menu recommendation results from product retrieval and order generation, but still associating them through data structures, thereby improving the overall system scalability and module reusability.

[0303] The server uses a centralized product filtering and sorting algorithm to calculate a large number of candidate products at the server end at once. The terminal only needs to display a simplified candidate set, which improves network transmission volume and terminal computing burden and helps reduce latency.

[0304] Based on the above causal relationship, this invention achieves a comprehensive improvement in data management and computing efficiency at the computer technology level through the collaborative design of data structure, prompt statement construction, model calling and result parsing on the server side.

[0305] VIII. Multiple Implementation Methods and Alternative Solutions Servers can replace specific software and hardware components in different implementations. For example, a server can replace a relational database with a document-oriented database, as long as it can achieve structured storage and retrieval functions for individual information, environmental information, etc. A server can replace generative artificial intelligence models with other multi-layer neural network models with text generation capabilities, such as sequence models based on recurrent neural networks or hybrid models combining convolutional structures.

[0306] The terminal is not limited to portable terminals; it can also be a fixed information terminal, such as a kitchen terminal with a display screen in the home. In these implementations, the terminal also exchanges JSON or equivalent structured data with the server via a network interface.

[0307] In some implementations, servers can incorporate caching modules to cache frequently used recipe, product, and environmental information in high-speed storage, thereby shortening data access time and further improving system response speed. Servers can also employ load balancing technology to distribute generative AI model invocation tasks and product retrieval tasks across multiple servers, enhancing concurrent processing capabilities.

[0308] The user's operation path can be simplified or extended under different implementation methods. For example, in some forms, the server can automatically pop up a recommendation notification on the terminal according to the time period, and the user only needs to confirm once to complete the order. In other forms, the server can allow the user to add temporary conditions (such as "I want to eat cold dishes today") before constructing the prompt statement, and then the server will embed such conditions into the prompt statement to further improve the degree of personalization.

[0309] Through the above-mentioned various embodiments, this invention, while maintaining the core technology defined in the claims—namely, constructing prompt statements based on individual and environmental information, calling generative artificial intelligence models to generate recommended menus, and automatically driving product retrieval and order generation—provides a wealth of implementation methods for different application scenarios, making it convenient for those skilled in the art to deploy and optimize according to specific system environments.

[0310] use Figure 12 The processing flow is explained.

[0311] Step 1: Users enter their personal information on the terminal.

[0312] Users select or fill in fields such as dietary preferences, allergens, health status, budget range, available preparation time, and location information in sequence on the terminal interface, and then click the "Save" or "Update" button.

[0313] Input: Individual information (text, options, values, etc.) entered by the user on the interface.

[0314] The terminal combines these fields into a key-value pair set, performs basic string validation (non-empty check, format check), and then temporarily stores it locally in a structured data format (e.g., key-value mapping) to await transmission.

[0315] Output: Individual information data object to be sent to the server.

[0316] Step 2: The terminal sends individual information to the server.

[0317] The terminal encapsulates the individual information data object generated in step 1 into a request message through the network communication module, sets the target address to the individual information receiving interface of the server, and sends it through a secure communication protocol.

[0318] Input: Individual information data object (including user identifier and multiple attribute fields).

[0319] Before sending, the terminal serializes the data into a transmission format, adds authentication information or session identifier to the request header, and then initiates a network request.

[0320] Output: A network request message containing individual information is sent to the server.

[0321] Step 3: The server receives and stores individual information.

[0322] The server receives network requests sent by the terminal at the interface of the application framework, deserializes the individual information data object from the request body, validates the required fields and field types, maps the fields to database table fields after validation, and performs insert or update operations.

[0323] Input: Individual information data object from the terminal.

[0324] The server queries the database to see if a corresponding record already exists based on the user identifier. If it exists, the record is updated; otherwise, a new record is created. At the same time, an update timestamp is written.

[0325] Output: Standardized individual information records stored in the information storage unit, and the processing result status (success / failure) returned to the terminal.

[0326] Step 4: The server acquires weather forecast information and generates environmental information data structures.

[0327] When the server is scheduled to perform a task or is triggered on demand, it calls the interface of an external meteorological information provider to request data such as temperature and weather conditions at the user's location within a predetermined time period. After receiving the response, it extracts food-related fields, such as the current or daily maximum temperature and weather type, and combines them into an environmental information data structure.

[0328] Input: User location information (individual information records from the database) and call parameters (time range, language, etc.).

[0329] The server parses and filters the raw meteorological data returned from the outside, retaining only key fields such as temperature and weather conditions, and converts them into data objects in a unified internal format.

[0330] Output: An environmental information data structure containing elements such as temperature and weather conditions, which can be optionally stored in an environmental information table.

[0331] Step 5: The server constructs prompts for generative artificial intelligence models.

[0332] The server reads the user's individual information records from the information storage department, reads the current weather information from the environmental information data structure, and can extract representative information or constraints from the formula information table and the product information table. Then, according to the predefined text template, it fills these multi-source data into the specified positions in the template to construct a complete prompt text.

[0333] Inputs include: individual information records (habits, allergies, health status, budget, etc.), environmental information data structures (temperature, weather conditions), and, if necessary, formula and product information.

[0334] During the construction process, the server performs string concatenation, conditional branching (such as adding constraint sentences based on the presence of allergy information), and format control (such as adding paragraphs and numbering) to ensure that the prompt statements contain constraints and output format requirements that can be recognized by the generative artificial intelligence model.

[0335] Output: Natural language prompts containing user constraints, environment constraints, and output requirements.

[0336] Step 6: The server uses a generative artificial intelligence model to generate recommended menu information.

[0337] The server takes the prompt statement generated in step 5 as input, sends it to the generative artificial intelligence model through the model call interface, specifies the model name, maximum generation length, randomness parameters and other configurations, and receives the natural language text generated by the model as a response.

[0338] Input: Prompt text and model call parameters (model identifier, maximum output length, temperature, etc.).

[0339] The server internally encodes the prompt statement via network request and sends it to the model server. The model performs encoding and decoding operations on the prompt statement in its internal neural network structure to generate recommended menu text that meets the constraints; the server then obtains this text as the result.

[0340] Output: A recommended menu in natural language text containing one or more dietary plans, a list of ingredients required for each plan, and preparation steps.

[0341] Step 7: The server parses the recommended menu text and generates structured menu data.

[0342] The server parses the recommended menu text obtained in step 6, identifies the dish name, ingredient list, quantity, and steps, and converts each menu item into a unified data structure list for easy subsequent product retrieval.

[0343] Input: Recommended menu Natural Language Text.

[0344] The server uses string search, delimiter splitting, and pattern matching to segment the text according to predetermined identifiers (such as "dish name:", "ingredients:", "steps:"). It further splits the ingredient rows to extract the raw material names and quantities, and constructs a structured menu data object array containing the fields.

[0345] Output: A structured menu data set, where each element includes a menu identifier, dish name, raw material list, and preparation steps.

[0346] Step 8: The server sends the structured menu data to the terminal, and the terminal displays the recommended menu.

[0347] The server encapsulates the structured menu data generated in step 7 into a response message and returns it to the terminal. After receiving the message, the terminal parses the data object and presents the menu name, brief description, and operable buttons in a list or card format in the graphical interface.

[0348] Input: A set of structured menu data from the server side.

[0349] The terminal reads the fields of the received data, maps the menu name, main ingredients and brief description to the interface components, and generates a scrollable or paginated menu list for users to browse.

[0350] Output: The recommended menu interface displayed on the terminal screen, and the status of interactive menu data in the terminal memory.

[0351] Step 9: The user selects the target menu on the terminal.

[0352] Users browse the recommended menu list on the terminal and click the "View Details" or "Select This Menu" button for a specific menu. The terminal then records the target menu identifier or complete menu object selected by the user.

[0353] Input: Structured menu data displayed in the terminal interface, and user click events.

[0354] The terminal internally updates the selected state, generating a selection data object containing the user identifier and the target menu identifier, which serves as input for subsequent procurement requests.

[0355] Output: The target menu selection data object representing the user's selection result.

[0356] Step 10: The terminal sends the target menu selection information to the server.

[0357] The terminal packages the target menu selection data object generated in step 9 into a request message and sends it to the menu selection interface of the server to notify the server that the current user has selected a specific menu.

[0358] Input: Target menu selection data object (including user identifier and menu identifier).

[0359] The terminal serializes the data object, adds necessary session or authentication information, and sends it to the server through the network layer.

[0360] Output: A network request message containing the target menu selection information.

[0361] Step 11: The server generates a list of raw materials and retrieves the corresponding products.

[0362] The server retrieves the data object corresponding to the target menu identifier from the structured menu data, extracts the raw material name and suggested dosage, and constructs a raw material list. Subsequently, the server generates search keywords and filtering conditions for each raw material, calls the product search interface of the product supply device, and obtains multiple candidate sales product information corresponding to that raw material.

[0363] Input: Target menu selection information (user ID and menu ID), and the corresponding structured menu data entries.

[0364] The server normalizes the raw material names according to an internal mapping table or rules (e.g., synonym unification), then assembles the retrieval parameters, queries the product database or external product services via the network, and receives and parses the returned product data set.

[0365] Output: A candidate set of raw materials and commodities indexed by raw materials, with each raw material corresponding to several sales commodity information.

[0366] Step 12: The server filters and sorts the candidate products to generate procurement candidate information.

[0367] The server scores and sorts the candidate product list obtained in step 11 based on price, inventory status, delivery time, and possible user preference history, filters out products that do not meet the basic constraints, retains only the top-scoring products as procurement candidates, and forms a procurement candidate information data structure with a unified format.

[0368] Input: Raw materials - candidate product set and filtering parameters (such as budget limit, delivery area restrictions, etc.).

[0369] The server calculates a comprehensive score for each product, such as weighting and summing indicators like price, delivery time, and score according to preset weights, and discards products with insufficient inventory or unable to be delivered to the user's address, ultimately forming an ordered list for each type of raw material.

[0370] Output: A data structure of procurement candidate information organized by raw materials, including recommended products and their attributes such as price and delivery time.

[0371] Step 13: The server sends the procurement candidate information to the terminal, and the terminal displays the candidate products.

[0372] The server sends the procurement candidate information generated in step 12 back to the terminal; the terminal parses the data structure, displays a list of recommended products for each raw material grouped by ingredient on the interface, and provides selection controls for the user to adjust.

[0373] Input: The procurement candidate information data structure generated on the server side.

[0374] The terminal maps the name, specifications, price, and estimated delivery time of each product to interface elements, providing a selection button and quantity input box for each product, thus forming an interactive shopping interface.

[0375] Output: The product candidate list displayed on the terminal, and the current shopping selection status maintained internally by the terminal.

[0376] Step 14: Users confirm the final purchase details on the terminal.

[0377] On the terminal's product candidate interface, users select one or more specific products for each raw material, adjust the quantity as needed, and finally click the "Confirm Order" button. The terminal then summarizes the user's selections to form the final procurement content data object.

[0378] Input: The procurement candidate list displayed on the terminal interface, as well as the user's selection and quantity input.

[0379] The terminal iterates through the current shopping status, records the product identifier, quantity, and associated raw material information for each selected product, calculates the estimated total price, and internally forms an order candidate structure.

[0380] Output: A final purchase data object containing complete shopping cart information.

[0381] Step 15: The terminal sends the final purchase details to the server, and the server generates a delivery order.

[0382] The terminal sends the final procurement content data object generated in step 14 as a request body to the server's order generation interface; after receiving it, the server parses the data structure, forms order information, calls the order processing device's order placement interface to generate the actual delivery order, and returns the order result to the terminal.

[0383] Input: Final procurement data object (including product identifier, quantity, user receipt information, etc.).

[0384] The server generates an order details list based on the data object, calculates the total order price, verifies it against inventory and delivery rules, creates the order by calling the order processing system over the network, obtains the order number and estimated delivery time, and then encapsulates these results and returns them to the terminal.

[0385] Output: Delivery order records created in the order processing device, and order confirmation information (order number, estimated delivery time, amount) returned to the terminal.

[0386] 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.

[0387] 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."

[0388] With the dramatic increase in the amount of diet-related information, computer-based dietary recommendation technologies are increasingly being used to help users make dietary choices under complex constraints. However, existing technologies have the following problems: (1) In many existing dietary recommendation systems, the server usually only performs simple searches in the database based on a few static parameters (such as a single budget or a single nutritional indicator). It lacks comprehensive modeling and constraint solving of multi-dimensional personalized conditions such as budget, health status, nutritional conditions, cooking skills, and time constraints. This results in insufficient data processing capabilities on the server side, making it impossible to efficiently generate candidate solutions that meet the real needs of users under complex constraints, and the utilization efficiency of computing resources is low.

[0389] (2) When calling generative artificial intelligence models, most existing systems use fixed or manually written prompts, failing to automatically construct highly relevant and highly constraining prompts based on the structured data processing results on the server side. This results in a low degree of matching between the output of the generative artificial intelligence model and the user's specific constraints, while also increasing unnecessary inference overhead and reducing the overall system's computational efficiency and response performance.

[0390] (3) Existing technologies typically process database retrieval results and generative artificial intelligence model outputs separately. There is no unified data processing flow established on the server side to perform cost calculations, nutritional value calculations, seasonal screening, and difficulty screening of candidate diets, and then integrate them with the model-generated results. This results in redundant calculations, data inconsistencies, and the need for users to filter results themselves, thereby reducing the overall processing capacity and scalability of the computer system in the dietary recommendation scenario.

[0391] (4) Existing systems often only present users’ emotional state and preferences on the terminal interface, failing to integrate such dynamic state information into the server-side prompt generation logic and data processing flow. This results in generative artificial intelligence models being unable to fully utilize this information during the inference stage, limiting the personalization of model output and the server’s ability to comprehensively process multimodal and multi-source information.

[0392] Therefore, a novel system architecture is needed that, on the server side, collaborates with generative artificial intelligence models through structured data processing, constraint numerical calculations, and automatic generation of prompts, in order to improve the server's data processing efficiency, computing resource utilization, and the accuracy and personalization of output results in dietary recommendation tasks, thereby achieving an overall performance improvement at the computer technology level.

[0393] 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.

[0394] In this invention, the server includes means for obtaining personal information containing lifestyle information from a user and storing the personal information in a storage device; means for performing query language-based retrieval processing on dietary information and food ingredient information stored in the storage device, and extracting a candidate set from the dietary information and food ingredient information based on budget information, health status information, nutritional condition information, cooking skill information, and cooking time information contained in the personal information; means for performing numerical calculations on each diet contained in the candidate set using food price information and nutritional value information to generate a dietary set that satisfies the constraints corresponding to the personal information; and means for automatically generating prompt statements for input into a generative artificial intelligence model based on the dietary set and the personal information, and instructing the generative artificial intelligence model to receive the prompt statements to generate dietary proposal information, and further integrating the dietary proposal information with the dietary set to generate display data for presentation to the user. This allows for the establishment of a unified processing chain on the server side, encompassing personal information collection, structured storage, efficient retrieval based on query languages, constraint-driven numerical computation, and adaptive prompt generation for generative artificial intelligence models. This enables the server to efficiently screen and calculate candidate diets while considering multidimensional constraints, providing highly relevant input to the generative artificial intelligence model. This reduces redundant reasoning, improves the matching degree between model output and user constraints, lowers the overall system's computational load and communication overhead, and enhances the technical indicators of the computer system in data processing efficiency, resource utilization, and response performance in dietary recommendation tasks, thus achieving a substantial improvement in computer technology.

[0395] "Personal information" refers to a collection of data provided by users and related to their living conditions, including but not limited to budget information, health status information, nutritional information, cooking skills information, cooking time information, and various attribute data reflecting users' living habits and preferences.

[0396] "Lifestyle information" refers to various types of information that reflect a user's daily lifestyle and environmental conditions, including time arrangements, eating habits, work and rest patterns, family composition, seasonal preferences, and other background information that can be used to set dietary recommendation constraints.

[0397] "Storage device" refers to hardware or software resources used to store and manage data, including database systems, file systems, cloud storage services, and other storage resources that can persistently store personal information, dietary information, and food ingredient information.

[0398] "Dietary information" refers to a structured set of data related to a diet plan, including the diet name, recipe steps, required food ingredients and quantities, estimated cooking time, difficulty level, cost estimate, and labeling information related to nutritional characteristics.

[0399] "Food ingredient information" refers to a set of data related to ingredients and their attributes, including ingredient name, unit price, proportion of edible portion, nutritional data, applicable season, storage method, and additional attributes related to cooking method.

[0400] "Query language" refers to a formal language used to retrieve and filter data in storage devices, including Structured Query Language and its variants, which enables efficient access to dietary and food ingredient information through conditional expressions, join operations, and aggregation operations.

[0401] The "candidate set" refers to an intermediate set of results obtained after retrieving dietary and food ingredient information based on query language. Each diet and food ingredient in this set meets some or all of the constraints in the user's personal information and is used for subsequent numerical calculations and further filtering.

[0402] "Food price information" refers to numerical data related to the cost of food ingredients, including unit price, specification price, historical price records, and various price parameters that can be used to calculate the cost per serving of food.

[0403] "Nutritional information" refers to quantitative data used to represent the nutritional components of food, including the content of nutrients such as calories, protein, fat, carbohydrates, dietary fiber, sodium, sugar, vitamins, and minerals.

[0404] "Constraints" refer to a set of restrictive rules composed of various parameters in personal information, including budget limits, calorie or nutrient ranges, health contraindications, cooking time limits, cooking skill level restrictions, and other conditions used to screen and optimize dietary plans.

[0405] A “dietary set” refers to the set of results obtained after performing cost calculations, nutritional value calculations, and constraint determinations on a candidate set. Each diet in this set satisfies all or a predetermined combination of constraints corresponding to the individual’s information.

[0406] "Generative artificial intelligence models" refer to models built on machine learning techniques, especially deep learning techniques, that can automatically generate text information based on input data. These include neural network models for natural language generation, which can generate dietary proposal information based on prompts.

[0407] "Prompt statements" refer to the instructional text input into a generative artificial intelligence model, which describes user constraints, existing dietary candidate information, and expected output format to guide the generative artificial intelligence model in generating dietary proposal information relevant to the target task.

[0408] "Dietary proposal information" refers to dietary-related suggestion data output by a generative artificial intelligence model after receiving prompts, including recommended meal plans, recipe combinations, food ingredient pairings, nutritional information, and meal arrangements.

[0409] "Display data" refers to organized and formatted output data used to present to users on terminal devices, including structured representations of diet sets, text or graphic content of dietary proposal information, and various layout and annotation information used for interface display.

[0410] "Emotional state information" refers to descriptive or quantitative data that reflects a user's current or recent emotional state, including emotional labels such as pleasure, tension, fatigue, and stress, or emotional characteristics inferred from physiological and behavioral signals.

[0411] "Preference information" refers to data that reflects a user's long-term or periodic dietary preferences, including taste preferences, cuisine preferences, degree of liking or disliking for specific ingredients, and preference orientation for health goals (such as fat loss, muscle gain, etc.).

[0412] In one embodiment of the invention, the server is constructed from general-purpose computer hardware, employing a hardware platform consisting of a multi-core central processing unit, an optional graphics processing unit, main memory, and a network interface. The server runs a general-purpose operating system, such as a Unix-like operating system, and deploys application server software, a database management system, and generative artificial intelligence model inference services on it. The server connects to a terminal via a communication network. The terminal can be a smartphone, tablet, or a computing device with browsing capabilities, running a mobile operating system or a browser environment. The user interacts with the server through the terminal.

[0413] The server maintains multiple data structures in the storage device. Within the database management system, the server establishes tables such as user information, dietary information, food ingredient information, nutritional information, and log information. Specifically, the user information table stores user budget information, health status information, nutritional conditions information, cooking skill information, cooking time information, and information reflecting user lifestyle habits and preferences. The dietary information table stores diet names, step descriptions, estimated cooking time, difficulty level, basic cost estimates, and label fields related to nutritional characteristics. The food ingredient information table stores unit price information, nutritional value information, applicable season, and ingredient identifiers that can be associated with dietary information. The nutritional information table maintains standard nutritional reference values ​​for subsequent calculations.

[0414] At the software level, the server divides its functions into independent modules, including a personal information management module, a data retrieval module, a cost and nutrition calculation module, a prompt statement generation module, a generative artificial intelligence model interface module, and a result integration and display data generation module. The server transmits structured data objects between these modules via an internal data bus or message queue. Each object contains clearly defined fields and types, thereby reducing redundant parsing and conversion operations.

[0415] In the personal information management module, the server uses the request parsing component provided by the application server framework to receive structured personal information sent by the terminal. The server maps this structured data into an internally unified user feature vector representation. Within this vector, the server assigns a fixed dimension to each type of feature; for example, budget ranges are mapped to numerical features, health status to multi-dimensional binary features, nutritional preferences to multi-label features, cooking skill levels to ordered numerical features, and time constraints to time limit features. This unified vector representation allows subsequent retrieval and model invocation to directly perform numerical calculations based on the vector, thereby reducing the computational overhead caused by multiple parsing and condition concatenation.

[0416] In the data retrieval module, the server uses Structured Query Language (SCL) to perform multi-condition combined searches on the dietary information table and the food ingredient information table. The server maps user feature vectors such as budget limits, cooking time limits, and difficulty level limits to query conditions, setting range filters for relevant fields. The server utilizes an index structure through the query optimizer, for example, creating composite indexes on the budget estimate, cooking time, and difficulty level fields to reduce the number of full table scans and improve query response speed. At this stage, the server selects only candidate diets that meet the basic constraints, forming a candidate set and reducing the load on subsequent computation modules.

[0417] The server performs numerical calculations on the candidate set in the cost and nutrition calculation module. Based on the ingredient list for each meal, the server batch-reads price and nutritional data from the food ingredient information table. Using a vectorized calculation method, the server multiplies the unit price of each ingredient by its usage or standardized serving size and sums the results to obtain a more accurate meal cost. For the nutritional information, the server weights and sums the nutrient vectors corresponding to each ingredient according to their usage to obtain the total nutritional vector for each meal. This total nutritional vector includes the content of calories, protein, fat, carbohydrates, sodium, sugar, and other specific nutrients. During the calculation process, the server can use a numerical calculation library to improve the efficiency of matrix and vector operations and avoid the overhead of item-by-item scalar operations.

[0418] The server further filters the candidate set based on user constraints in the cost and nutrition calculation module. The server compares the total cost of each meal with the user's budget limit, excluding meals whose costs exceed a certain threshold to ensure the selected meals are cost-effective. The server compares the total nutrient vector with the nutritional condition range set by the user. For example, for users requiring high protein and low fat, the server calculates the ratio of protein to total calories and the ratio of fat to total calories, retaining only meals with a protein ratio higher than the set lower limit and a fat ratio lower than the set upper limit. The server can also compare sodium content with the recommended daily intake for low-sodium requirements, excluding meals that significantly exceed the threshold. The server thus generates a set of meals that meet the constraints.

[0419] In the prompt generation module, the server generates prompts for a generative artificial intelligence model based on a set of dietary data and personal information. The server reconstructs user feature vectors into natural language descriptions, including budget range, cooking skill level, health goals, time limits, and seasonal requirements. The server selects several representative dishes from the dietary set, incorporating their names, main ingredients, and brief nutritional descriptions into the prompts as reference information for model generation. Finally, the server converts structured data into natural language text using templates and rules to form the prompts.

[0420] In a specific instance, the server can generate the following prompt statement: The basic requirements for users are as follows: Daily budget: 50 yuan; Health goal: Control weight, high protein, low fat; Cooking level: Beginner; Preparation time for each meal: no more than 30 minutes; Prioritize the use of seasonal vegetables.

[0421] The following are examples of candidate recipes obtained through rule filtering: 1) Chicken breast and broccoli fried rice (high protein, low fat, about 20 minutes); 2) Tomato and egg noodles (easy to make, about 15 minutes).

[0422] Based on the above conditions, please design a detailed menu that includes dinners for 3 days.

[0423] For each day, please submit at least two dishes, including the dish name, required ingredients and quantities, simple preparation steps, estimated cooking time, and key nutritional characteristics (e.g., high protein, low fat, suitable for weight control). The server can also generate the following example of a prompt statement based on the original request: "The user has a budget of under 20,000 yuan and wants to eat healthily. The user is a cooking beginner and wants to complete a meal in under 30 minutes. Please recommend several specific recipes based on these conditions, and provide the nutritional characteristics and preparation steps for each dish." The server invokes a generative AI model within its generative AI model interface module. The server can deploy a neural network model based on a transformer architecture, comprising a multi-layered self-attention encoder and decoder. Model parameters are pre-trained on a large corpus and fine-tuned using diet-related data during the training phase. During inference, the server converts prompts into word sequences, which are input into the model's embedding layers. In each layer, the model performs matrix multiplication, attention weight calculation, normalization, and non-linear activation operations to generate output labeled sequences. The server can set temperature parameters and sampling strategies to control output diversity and determinism.

[0424] During model training, the server uses the cross-entropy loss function to calculate the error between the predicted and target sequences and updates the model weights through backpropagation. The server constructs question-and-answer resources, menu examples, and nutritional information text related to dietary data in the training database. Through supervised learning, the model learns how to generate reasonable dietary recommendations based on constraints. The server can introduce data augmentation techniques during training, such as using different descriptions for the same recipe or perturbing constraints, to improve the model's robustness in practical applications.

[0425] The server receives dietary proposals from the generative AI model in the result integration and display data generation module. The server performs sentence segmentation and structured parsing on the output text, mapping the dish names, ingredients, steps, and nutritional information into internal data structures. The server compares these generated results with previous dietary sets, identifies matching portions and adds structured fields. For new combinations generated by the model, the server can estimate their nutritional characteristics and costs by calling the nutritional value calculation module again. Finally, the server constructs a unified display data object containing rule-filtered diets and model-generated extended schemes, along with source identifiers and confidence score fields.

[0426] In this embodiment of the invention, the terminal operates a graphical user interface. After receiving display data from the server, the terminal displays the meal name, estimated cost, cooking time, nutrition label, and recommended source identifier on the interface in the form of a list or cards. When the user selects an item, the terminal displays detailed steps and an ingredient list. The terminal can highlight text or graphics according to the steps and can generate a shopping list on the interface based on the structured data provided by the server. The user can provide feedback on the recommendation results through the terminal, such as marking "like" or "dislike". The terminal sends this feedback to the server, which can store the feedback as part of the user's preference information in the user information table for subsequent personalized recommendations.

[0427] In practical use, users can input their budget, health goals, and time constraints through the terminal. The server, based on the aforementioned modules, completes data retrieval, numerical calculations, prompt generation, and model inference, thereby providing users with multiple dietary suggestions within a limited time. By uniformly processing user feature vectors and structured dietary data, the server not only automates tasks that previously required extensive manual searching and calculations, but also implements multi-level filtering and constraint-driven optimization processes within the computer. Because the server first narrows down the candidate space using structured retrieval and numerical calculations, and then generates highly constrained prompts for the generative AI model, the computational load required for model inference is reduced, while the matching degree between the output results and user constraints is improved, and the amount of data transmitted over the network is also reduced.

[0428] By employing an unconventional processing sequence of rule filtering followed by model generation and then structured integration, the server significantly reduces redundant computation and invalid candidates compared to directly inputting all user conditions as unstructured text into the model. In the cost and nutrient calculation module, the server performs vectorized processing and batch calculations on the data, improving its throughput in multi-user concurrent scenarios. During the query phase, the server utilizes a combination of indexes and restricted fields to reduce the number of database disk read / write operations, thus optimizing the data access path.

[0429] In another implementation, the server can replace the generative AI model with a language model of different parameter counts or architectures, such as a shallow transformer structure or a hybrid model integrating attention and convolutional structures, to adapt to data centers or edge server environments of different sizes. The server can also extend its personal information management module to receive physiological data from sensor devices, such as steps and heart rate, converting it into part of health status information to make dietary recommendations more precise. In yet another implementation, the server can establish multiple sets of dietary information tables and nutritional standards based on dietary habits in different regions. After receiving the user's location information, the server automatically selects the corresponding database partition for retrieval and calculation.

[0430] Through the collaborative work of structured data processing, numerical computation, prompt generation, and generative artificial intelligence model inference, the server enables the system to achieve technical effects such as improved computational speed, enhanced recommendation accuracy, optimized data management structure, and reduced communication load in the specific application of dietary recommendations. The server employs internal vectorized feature representation, index-driven query optimization, batch numerical computation, and a two-stage generation process, forming a technical solution different from traditional systems that rely on manual rules or single database retrieval. This results in a fundamental improvement in the human-machine collaborative dietary recommendation processing capabilities under multi-dimensional constraints at the computer technology level.

[0431] use Figure 13 The processing flow is explained.

[0432] Step 1: Users enter personal information via the terminal. Users enter their budget, health status, nutritional preferences, cooking skill level, acceptable cooking time, and seasonal preferences in the terminal's graphical interface.

[0433] The terminal uses the above inputs as raw input data, which specifically includes text, numerical values, and option markers.

[0434] The terminal processes the raw input, reads the values ​​from different controls, assembles them into a key-value pair structure, and converts them into a structured data object.

[0435] The terminal takes this structured data object as output and prepares to send it to the server through the communication module.

[0436] Step 2: Terminal sends structured personal information to server The terminal uses the structured data object generated in step 1 as input and calls the network communication library to construct a network request.

[0437] The terminal serializes the data object, encodes it into a network transmission format, and attaches a content type marker to the request header.

[0438] The terminal sends the serialized data to the interface address specified by the server through the communication interface.

[0439] The terminal outputs the result of the network request and the status of waiting for the server response.

[0440] Step 3: The server receives and parses personal information. The server takes network requests sent by the terminal as input and reads the request body through the request processing component of the application server framework.

[0441] The server performs a deserialization operation on the serialized data in the request body, parsing it to obtain a structured representation of personal information, including budget values, health status markers, nutritional preference lists, skill level enumerations, time limits, etc.

[0442] The server performs format validation and integrity checks on the parsed data, and fills in or marks missing or abnormal items with default values.

[0443] The server outputs verified and formatted personal information records to internal modules.

[0444] Step 4: The server stores personal information and generates feature representations. The server takes the personal information records output in step 3 as input and accesses the storage device through the database interface.

[0445] The server performs insert or update operations in the user information table, writing personal information into the corresponding record row.

[0446] The server also performs feature processing on personal information, mapping budget, health status, nutritional preferences, skill level, and time limit into numerical and categorical features, combining them into a fixed-dimensional user feature vector.

[0447] The server outputs the updated database records and the generated user feature vectors to the subsequent retrieval and calculation modules.

[0448] Step 5: The server retrieves candidate diets and ingredients based on query language. The server takes user feature vectors and dietary information tables and food ingredient information tables from the database as input.

[0449] The server extracts the budget limit, time limit, and difficulty limit from the feature vector, converts them into query conditions, and constructs a multi-condition query statement.

[0450] The server uses indexing and query optimization mechanisms to execute the query, filters the dietary information table, selects dietary records that meet basic constraints in terms of cost estimation, cooking time, and difficulty level, and then queries the corresponding food ingredient records.

[0451] The server outputs a set of dietary records that satisfy the basic constraints and a corresponding list of ingredients, forming a candidate set.

[0452] Step 6: The server calculates the cost of candidate diets. The server takes the candidate set output from step 5 and food price information as input.

[0453] For each candidate meal, the server iterates through its ingredient list, reads the unit price field from the food ingredient information, calculates the cost of each ingredient based on its quantity, and then sums up the costs of all ingredients in the meal to obtain a more accurate cost value.

[0454] The server compares the calculated costs with the user's budget characteristics and marks meals that significantly exceed the budget or are inefficient as unacceptable.

[0455] The server outputs a set of candidate diets with labeled results and updated precise cost fields for nutritional calculations and further filtering.

[0456] Step 7: The server performs nutritional value calculations on the candidate diets. The server takes the candidate diet set and nutritional information table output in the previous step as input.

[0457] For each candidate meal, the server maps its ingredients and quantities to a nutrition facts table, reads the nutrient content vector of each ingredient, scales it up proportionally according to the quantity, and then performs vector summation on the nutrient vectors of all ingredients to obtain the total nutrient vector of the meal.

[0458] The server compares the total nutrient vector with the user's nutrient preference characteristics (such as high protein, low fat, low salt, etc.), and determines whether the diet meets the nutrient constraints based on preset ratio thresholds and upper and lower limits of values. Diets that do not meet the constraints are marked and removed from the set.

[0459] The server outputs a set of diets that have passed nutritional screening and have total nutrient vector annotations, forming an intermediate result that meets nutritional standards.

[0460] Step 8: The server further filters the diet based on season and skill level. The server takes the set of nutritionally qualified diets output in step 7, current season information, and user cooking skill characteristics as input.

[0461] The server reads the recommended ingredient list for the current season from the configuration data, checks the overlap between the main ingredients and seasonal ingredients for each diet, and calculates the seasonal correlation index.

[0462] The server also compares the difficulty level of the meal, the number of steps, and the cooking techniques involved with the user's skill level, and eliminates meals that are too complex to operate and beyond the user's skill range.

[0463] The server outputs a set of diets that meet seasonal preferences and skill constraints, along with seasonal relevance and difficulty matching scores, as the final set of rule-based filtering results.

[0464] Step 9: The server generates prompts for generative artificial intelligence models. The server takes the set of meals output in step 8 and the user feature vector from step 4 as input.

[0465] The server converts user budget, health goals, nutritional preferences, cooking time limits, skill level, and seasonal preferences into natural language descriptions, and briefly synthesizes the names, main ingredients, and nutritional characteristics of representative meals selected from the dietary set into text.

[0466] The server uses predefined text templates and combination rules to concatenate the above information into structured natural language paragraphs, forming a complete prompt statement to constrain the generation process of generative artificial intelligence models.

[0467] The server outputs generated prompts, providing highly relevant input for the model's inference phase.

[0468] Step 10: The server invokes a generative artificial intelligence model to generate dietary proposal information. The server takes the prompt output from step 9 as input and encodes it into a text sequence suitable for processing by a generative artificial intelligence model.

[0469] The server sends the text sequence to a generative artificial intelligence model deployed in the inference environment through the model interface. The model is based on a multi-layer transformer structure and performs data operations such as embedding mapping, self-attention calculation, feedforward network operation and layer normalization on the input sequence, and gradually outputs a text tag sequence representing the recommended menu and recipe content.

[0470] The server receives the generated text results from the model interface, performs sentence splitting and content extraction, and parses the dish names, ingredient lists, step descriptions, and nutritional information contained in the text into structured fields.

[0471] The server outputs structured dietary proposal information, including recommendations for multiple days and multiple meals, which are then integrated with the results of rule-based filtering.

[0472] Step 11: The server integrates the filtered results from the rules and the generated results from the model, and then generates and displays the data. The server takes as input the rules output in step 8 for filtering the diet set and the dietary proposal information output in step 10.

[0473] The server compares the two sets of data, associates items in the model output that are consistent with or similar to existing dietary records, unifies their identifiers, and estimates the cost and nutritional characteristics of the new combination schemes generated by the model through the nutrition calculation module.

[0474] The server sorts all candidate results according to a predetermined sorting strategy (such as priority, matching degree with constraints, seasonal relevance) and constructs a collection of display data objects containing fields such as diet name, cost, time, nutrition label, and source mark.

[0475] The server outputs this collection of display data objects via a network interface to the terminal for user visualization.

[0476] Step 12: The terminal displays recommendation results and receives user feedback. The terminal takes the set of display data objects sent by the server in the previous step as input, and uses the interface rendering engine to display each recommended meal to the user in the form of a list or card, including name, cost, time, nutritional characteristics and source of recommendation.

[0477] When a user clicks on a specific food item, the terminal extracts the corresponding detailed fields from the input data, presents a detailed list of ingredients and step instructions on a new interface, and can generate auxiliary information such as shopping lists based on structured data.

[0478] The terminal also provides feedback controls on the interface, allowing users to select "like," "dislike," or rate the meal. The terminal combines user feedback with dietary information to create new structured feedback data.

[0479] The terminal sends this feedback data as output to the server, so that the server can update the user information record with the feedback during subsequent processing, further optimizing the subsequent filtering and prompt statement generation process.

[0480] 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".

[0481] With the widespread application of generative artificial intelligence models in natural language processing and recommendation, existing food recommendation systems based on general dialogue models still suffer from the following technical problems: First, most existing systems only take simple user preferences as input, treating generative AI models as black-box text generators. They lack unified modeling and structured fusion of multi-source heterogeneous data such as user attribute information, environmental prediction information, cooking step information, and purchased item information. This makes it difficult for the generated results to directly drive downstream product matching and instant delivery processes, resulting in a low-automation state of "human-intermediate processing" in the overall computation process. Second, the prompts in traditional systems are often static or manually written according to rules. They cannot dynamically generate high-quality prompts based on multi-dimensional constraints such as user budget, health status, nutritional goals, seasonal ingredients, cooking skills, preparation time, and emotional state. This leads to a mismatch between the input expression of the generative AI model and the user's actual needs, thereby reducing the accuracy and reliability of recommendations. Third, when processing food-related questions and answers, existing systems typically do not consider user emotions, real-time environmental information, and historical selection behavior as a whole. This makes it difficult to achieve adaptive optimization of question-and-answer strategies and menu content, and hinders continuous improvement of model calling logic and data processing processes at the system level. Furthermore, many systems only output recipes or suggestions in natural language, lacking an automatic mapping mechanism between the natural language generation results and structured menus, ingredient lists, and delivery instructions. This prevents computers from directly using the generated results for subsequent data processing and service orchestration.

[0482] Therefore, how to construct a computer-based solution capable of: unified collection and preprocessing of multi-source data on the server side; automatic generation of high-quality prompts for generative artificial intelligence models; conversion of unstructured proposal information output by the model into structured data that can be directly processed by the computer; and continuous optimization of the prompt generation process using user selection history and evaluation information has become an urgent technical challenge in this field. This invention attempts to improve the performance of the entire food recommendation and instant delivery system in terms of computational resource utilization, automation level, and personalized recommendation quality by improving the data processing flow, prompt generation logic, and interaction mode with generative artificial intelligence models on the server side, thereby achieving an improvement in computer technology itself.

[0483] 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.

[0484] In this invention, the server includes a unit that acquires user-related attribute information, environment-related prediction information, cooking-related step information, and purchase-related item information, and uses the information to construct a learning dataset for a generative artificial intelligence model to learn from the learning dataset; a unit that parses user condition information, emotional state information, and natural language request text acquired from the terminal, and combines the information with the prediction information and item information to automatically generate prompt statements for input to the generative artificial intelligence model; a unit that parses the food-related suggestion information output by the generative artificial intelligence model, structures the food menu information, ingredient information, and cooking step information, and generates instruction information for issuing immediate delivery instructions to the distribution system based on the ingredient information; a unit that uses the generative artificial intelligence model to perform natural language processing to generate response information containing answer information for food-related questions and explanatory text corresponding to emotional state information, and generates prompt statements for sending the response information to the terminal; and a unit that records user selection history and evaluation information, and reflects the selection history and evaluation information when generating prompt statements later. This enables unified modeling and automatic preprocessing of multi-source heterogeneous data on the server side. By dynamically generating high-quality prompts, it precisely drives generative artificial intelligence models, allowing model outputs to directly participate in subsequent data operations and service orchestration in a structured form. Furthermore, it utilizes user feedback loops to optimize the prompt generation and model invocation logic, thereby improving the processing efficiency, automation level, and recommendation quality of computers in personalized diet recommendations and instant delivery scenarios, achieving substantial improvements to the overall architecture and processing flow of computer technology.

[0485] "System" refers to an integrated computer system consisting of at least one data processing device, at least one terminal device, and a communication network for transmitting data between said devices, used to perform information acquisition, data processing, model reasoning, and result output.

[0486] "User" refers to an individual or group of users who provide attribute information, condition information, emotional state information or natural language request text to the system through a terminal device and receive dietary suggestion information or response information.

[0487] "Attribute information" refers to basic characteristic information related to users, including but not limited to age, gender, allergy information, chronic disease information, dietary preferences, historical dietary records and historical purchase records, etc., which are used to characterize the long-term or relatively stable characteristics of users.

[0488] "Environmental prediction information" refers to predictive data related to the user's environment, obtained through external information acquisition methods, including but not limited to weather forecast information, temperature, humidity, weather conditions, and environmental conditions related to time and location.

[0489] "Cooking step information" refers to procedural information about the food or dish preparation process, including the order of raw material processing, heating methods, time control, and key operational points, which is a set of data used to guide the preparation of a specific dish.

[0490] "Purchase item information" refers to product-related information that can be obtained through retail or distribution platforms, including but not limited to product name, standardized category, specifications, unit price, inventory status, supplier, and estimated delivery time.

[0491] "Learning datasets" refer to a set of structured or semi-structured data that consists of attribute information, environmental prediction information, cooking step information, purchase item information, and associated labels or target values, and is used to train or fine-tune generative artificial intelligence models.

[0492] "Generative artificial intelligence models" refer to artificial intelligence models that are trained on a large amount of sample data and can automatically generate text, structured data or other forms of output based on input prompts, including but not limited to natural language generation models based on deep learning.

[0493] "Terminal" refers to an information processing device operated by a user and interacting with a server, including but not limited to smartphones, tablets, personal computers, or other electronic devices with a graphical user interface.

[0494] "User condition information" refers to the constraints or expected parameters that a user inputs regarding their dietary needs at a specific time, including but not limited to budget information, health status information, nutritional goal information, information related to seasonal ingredients, cooking skill level information, and food preparation time information.

[0495] "Emotional state information" refers to identifying information used to characterize a user's current psychological or emotional state, including but not limited to "happy," "sad," "fatigued," and "tense," which is obtained through text analysis, voice analysis, image analysis, or physiological signal analysis.

[0496] "Natural Language Request Text" refers to free text content input by the user in natural language, including but not limited to descriptions of dietary needs, physical condition, taste preferences, or food issues, which are used as the basis for the system to parse and generate prompt statements.

[0497] "Prompt statements" refer to the instructional text or message sequence generated by the server based on attribute information, user condition information, emotional state information, environmental prediction information, and purchased item information. These instructions are provided as input to the generative artificial intelligence model to guide the model to produce output that meets the expected task requirements.

[0498] "Dietary suggestion information" refers to the dietary-related suggestions output by the generative artificial intelligence model after receiving prompts, including but not limited to dish candidates, dietary structure suggestions, food pairing schemes, and nutritional information.

[0499] "Menu information" refers to structured information extracted from dietary suggestion information that indicates one or more combinations of dishes, including dish names, quantities, order of serving, or categories of dishes.

[0500] "Ingredient information" refers to structured data related to the ingredients required for each dish in the menu, including ingredient name, quantity, unit, alternative options, and association identifiers with the purchased items.

[0501] "Instant delivery instructions" refer to order-level or task-level instructions generated based on food information, which are used to prompt the distribution system to quickly deliver the required goods to the user's designated location within a predetermined time period.

[0502] "Instruction information" refers to data used to send operation requests to the distribution system, retail platform or other external service system, including order requests, product lists, delivery information and delivery time, etc., to drive the external system to perform delivery or supply actions.

[0503] "Response information" refers to comprehensive response data generated based on the output of a generative artificial intelligence model and used to return to the terminal. It includes at least answers to food-related questions and explanatory text or reassuring content corresponding to emotional state information.

[0504] "User selection history" refers to the record of choices made by users in one or more dietary suggestion information provided by the system, including behavioral trajectory data such as selecting menus, abandoning menus, clicking to view details, and triggering purchases or placing orders.

[0505] "Evaluation information" refers to explicit or implicit feedback information that users provide regarding the recommendation results or service experience during the use of the system, including ratings, text reviews, repeat purchase behavior, and dwell time, which can be used to measure satisfaction with the recommendation results.

[0506] "Cost constraint information" refers to the upper limit or price range of expenditure that users can accept for a single meal, a certain time period, or a specific order. It is used to limit the cost range when the system selects ingredients or dish combinations.

[0507] "Health status information" refers to data that reflects a user's current or long-term physical condition, including but not limited to medical history, body mass index, allergens, medical restrictions, and other information related to dietary safety or nutritional control.

[0508] "Nutritional goal information" refers to the nutritional intake goals set by users or systems for dietary purposes, including but not limited to target parameters such as fat loss, muscle gain, blood sugar control, salt control, increasing protein intake, or a balanced diet.

[0509] "Seasonal ingredient information" refers to information on the availability and suitability of ingredients related to a specific season or period, including seasonal vegetables, fruits, seafood, or raw materials that are of superior flavor and nutritional value during that season.

[0510] "Cooking skill level information" refers to the graded information that represents the user's cooking experience and proficiency, such as beginner, intermediate, and proficient, which is used to constrain the complexity and difficulty of the steps when the system recommends dishes.

[0511] "Food preparation time information" refers to the time constraints that users have available to prepare and complete this meal, including food pretreatment time and cooking time, which are used to limit the total time spent on recommended dishes.

[0512] "Food-related question text" refers to natural language questions posed by users that relate to food, nutrition, cooking, healthy eating, or ingredient selection. These questions are used to drive the system to generate question-and-answer responses.

[0513] In one embodiment of the present invention, the server comprises a computing device having a central processing unit, main memory, and a network interface, and the operating system can be a general-purpose server operating system. The server is connected to a terminal via a communication network, and the terminal can be a smartphone, tablet computer, or personal computer. The user interacts with the server through the terminal.

[0514] The server, at the hardware level, includes a multi-core processing unit, a graphics processing unit, non-volatile memory, and a high-speed network interface. At the software level, it includes a database management component, a network communication component, a data processing component, a generative artificial intelligence model inference component, and a prompt generation component. In a preferred embodiment, the server uses a scripting language environment and a deep learning framework, where the deep learning framework can be a tensor computation framework or a tensor computation library, and the generative artificial intelligence model can be a multi-layer encoder-decoder model based on a self-attention mechanism.

[0515] The server uses relational or key-value databases for database management, storing user attribute information, environmental prediction information, cooking step information, purchased item information, and historical interaction information in indexed table structures. In its data processing component, the server uses a data frame processing library to clean, normalize, and extract features from meteorological, product, and recipe data obtained from external interfaces. For example, the server encodes temperature, weather type, and time information as numerical features, maps ingredient names in recipes to standard category identifiers, and converts price and inventory information into comparable scalars.

[0516] The server uses a pre-trained language model and performs domain-specific fine-tuning in the generative AI model's inference component. The server constructs a sample set from the learning dataset, containing input text sequences and target text sequences. The input text sequences include encoded representations of attribute information, environmental prediction information, user conditional information, and emotional state information, while the target text sequences include dietary suggestions or response information. During training, the server uses cross-entropy loss as the error function and updates model parameters through backpropagation. The optimization objective is to minimize the difference between the generated text and the reference text. The server can perform weight updates using stochastic gradient descent or its improved versions. The server can increase the diversity of training samples through data augmentation techniques, such as synonym substitution, order perturbation, and noise injection into the input text, thereby improving the model's robustness to different request expressions.

[0517] In the prompt generation component, the server uses a combination of rule templates and learning strategies to generate prompts. First, the server maps user condition information (budget, health status, nutritional goals, seasonal ingredients, cooking skill level, and food preparation time) and emotional state information into structured key-value pairs. Then, a template engine inserts these key-value pairs into predefined natural language templates to construct basic prompts. The server further adjusts the wording and structure of the prompts based on historical selection history and evaluation information, such as emphasizing health or budget constraints, to guide the generative AI model in the solution space towards regions that better align with the user's long-term preferences. Because the server explicitly controls the content and structure of the prompts, the output range of the generative AI model is limited to a subspace highly relevant to the food recommendation task, reducing irrelevant generation and computational waste.

[0518] The server generates the following example prompt statement in a specific implementation: Example of a prompt statement 1: The user information is as follows: - Age: 32, Female - Allergy: Shrimp - Chronic disease: Mild hypertension - Health goal: Fat loss Current mood: Fatigue Current weather: 8 degrees Celsius, light rain, feels rather cold. - Budget: No more than 1000 yen - Cooking Skills: Beginner - Maximum cooking time: 20 minutes User's natural language description: 'Budget under 1000 yen, I'm a bit tired today, and I'd like a warm but not too oily dinner.' Based on the information above, please recommend 2-3 suitable family meal plans for tonight's dinner. Requirements: 1. Overall, it's relatively healthy, with a moderate amount of fat; 2. Considering the cold weather, the dishes should be served warm. 3. Do not use any shrimp ingredients, and be careful to control the salt content; 4. Each recipe should include: the name of the dish, a list of main ingredients (including approximate weights), brief steps (3-5 steps), and a 1-2 sentence explanation of why it is suitable for relieving fatigue. In another implementation, the server is used to process food-related questions and answers, generating prompts such as: Example of a prompt statement 2: User question: 'My stomach hasn't been feeling well lately, do you have any suggestions for a suitable dinner?' User health information: Has chronic gastritis and excessive stomach acid; no other major illnesses.

[0519] Please answer from the perspective of a nutrition consultant: 1. Explain the principles to keep in mind when having dinner if you have an upset stomach; 2. Recommend 2-3 suitable dinner examples, each explaining the main ingredients and simple cooking methods; 3. Remind about the types of food to avoid (such as overly spicy, overly sour, or overly oily foods).

[0520] Please answer in concise and easy-to-understand language. In another implementation, the server generates prompts for emotion-driven dietary recommendations, such as: Example of a prompt statement 3: "User's current mood: sad, feeling down."

[0521] Today's weather: Cloudy and rainy, with low temperatures.

[0522] User health status: No serious illness; goal is to maintain weight.

[0523] Users' favorite foods: chicken, potatoes, and cheese.

[0524] Please design 2-3 dinner dishes for users that can soothe emotions and bring warmth, suitable for home cooking, and with a cooking time of no more than 40 minutes.

[0525] Please explain why each dish is suitable for relieving this emotion, and provide the main ingredients and key steps. The server controls the input to the generative AI model through the aforementioned prompts, focusing the model's internal attention mechanism on key diet-related features such as nutritional constraints, temperature preferences, and emotional labels, thereby forming a more discriminative contextual representation during the encoding phase. During the decoding phase, the server can constrain the output length and structure, for example, requiring the model to output the solution as a numbered list, facilitating subsequent parsing into a structured record. The server utilizes this structured constraint to reduce the post-processing complexity of the generated results and improve overall computational efficiency.

[0526] When parsing the generated results, the server converts natural language text into structured menu and ingredient information. The server can use dictionary-based and pattern-matching parsing algorithms, or a lightweight sequence labeling model, to label the generated text with tags such as "dish name," "ingredient name," "quantity," "unit," and "step description." The server represents each dish as a record, containing the dish name, a list of ingredients, and a list of steps, mapping ingredient names to standardized names in the purchased item information. Through this data structure transformation, the server converts the unstructured output of the generative AI model into structured data that can be directly used for database queries and subsequent logical operations, thereby enabling automatic integration with the distribution system.

[0527] When interacting with the distribution system, the server generates instructions based on ingredient and purchase item information. Based on the required quantity and inventory status of each ingredient, the server selects items with lower unit prices and reliable supply, combines them into an order, and generates order parameters for immediate delivery. The server sends these order parameters to the distribution system's interface, which then uses them to access actual delivery equipment and logistics resources to deliver the goods to the user's designated location. This process allows the system to go beyond mere recommendation, technically implementing an automated control chain from result generation to actual physical delivery.

[0528] In this invention, the terminal serves as the user interface implementation device. The terminal executes the application and displays the input interface and recommendation results interface. The terminal guides the user to input budget, health status, taste preferences, and natural language descriptions through a graphical interface. The terminal serializes this information into structured data and sends it to the server. Upon receiving the menu and ingredient information from the server, the terminal displays it in a list or card format, allowing the user to view details, rate the food, or initiate a purchase. The terminal can also invoke local or cloud-based emotion recognition services, acquiring the user's facial expressions and voice features through a camera and microphone to generate emotional state information, further enhancing the personalization of recommendations.

[0529] In this invention, users control and provide feedback to the system via a terminal. After viewing the recommended menu returned by the server, users can select a menu item and trigger the "purchase ingredients" operation. After delivery is completed and the food is prepared or consumed, users can rate the menu on the terminal interface, such as by giving a star rating or a text review. The server records this rating information as part of the subsequent training data, used to update the learning dataset and adjust the prompt generation strategy.

[0530] In terms of technical effectiveness, the server improves computer technology through various mechanisms. By fusing structured multi-source data and generating dynamic prompts, the server reduces the probability of generative AI models outputting irrelevant content, thereby reducing ineffective inference overhead and improving inference speed and resource utilization. By automatically adjusting the content of prompts and template weights using user selection history and evaluation information, the server implements a feedback-driven prompt optimization algorithm. Compared to a fixed prompt strategy, this achieves higher recommendation accuracy and user satisfaction under the same hardware conditions. By automatically parsing the generated results into structured data, the server avoids manual interpretation and input, reducing error rates and providing a high-quality labeled data source for subsequent data statistics and model retraining.

[0531] In another implementation, the server can employ different generative AI model structures. For example, the server can use a multi-layer bidirectional attention network for the encoder and a unidirectional autoregressive network for the decoder. Through a multi-head attention mechanism, attribute information, environmental prediction information, and emotional state information are encoded into different subspace vectors. Then, during the decoding stage, attention weights automatically select the features that contribute most to the generated content. This structure enables the model to distinguish between long-term stable features (such as health status) and short-term contextual features (such as current weather and current mood), improving the relevance and stability of the generated content.

[0532] In another alternative approach, the server can break down the recommendation task into two stages: the first stage uses a structured recommendation model based on gradient boosting trees or matrix factorization to select several candidate menus from the database; the second stage inputs these candidate menus, along with the user's current conditions and sentiment information, as prompts into the generative AI model, allowing the model to provide natural language descriptions and fine-tuning for the candidate menus. This hierarchical architecture reduces the search space of the generative AI model, further reducing the computational load and improving the overall system response speed.

[0533] Through the various embodiments described above, the server, terminal, and user work collaboratively in the system constituted by this invention. Internally, the server implements a series of specific data structure designs, prompt generation algorithms, result parsing and structured processing flows, and automatic instruction generation and interaction flows with the circulation system. This makes the system not merely a simple automation of the traditional manual recommendation workflow, but rather an improvement in computer technology at the levels of generative artificial intelligence model invocation, data management, and overall system architecture. This results in improved processing speed, higher recommendation accuracy, lower error rates, and improved resource utilization efficiency.

[0534] use Figure 14 The processing flow is explained.

[0535] Step 1: Users input conditional information and natural language request text at the terminal. Users input information such as budget, health status, nutritional goals, seasonal food preferences, cooking skill level, and acceptable preparation time into the terminal's graphical interface, and also input natural language request text (e.g., "Budget under 1000 yen, I'm a bit tired today, I'd like to have a warm and not too oily dinner.").

[0536] Input: Structured conditional fields and natural language text entered by the user in the interface.

[0537] The terminal converts each field into key-value pairs according to a preset data structure, encodes natural language text into strings, and performs basic validation locally (e.g., checking if the budget is a number and if the preparation time is a positive integer). The terminal can also call local or remote sentiment analysis services to infer emotional state information (e.g., "fatigue") based on text or speech.

[0538] Output: A request data packet containing user condition information, natural language request text, and emotion state information.

[0539] Step 2: The terminal sends a request data packet to the server. The terminal sends the data packet generated in step 1 to the server using a security protocol via its communication module.

[0540] Input: A structured request data packet (including user identifier, condition information, request text, emotion state information, etc.).

[0541] The terminal serializes the request data packet into a predefined format (such as a JSON string), appends device identifier and application version information to the request header, and then sends it to the interface address specified by the server via the network interface. The terminal can compress the data before sending to reduce transmission load.

[0542] Output: The data stream transmitted to the server over the network.

[0543] Step 3: The server parses the request and supplements the user attribute information. After receiving the data stream sent by the terminal, the server restores it into a request data packet through the network communication component and performs syntax and integrity checks on the content.

[0544] Input: A request data packet from the terminal.

[0545] The server first parses the user identifier in the data packet, then queries the database for the user's attribute information (including age, gender, allergy information, chronic disease information, historical dietary records, historical purchase records, etc.), and merges this attribute information with the condition information in the request data packet. The server uses database query statements to extract records from multiple tables and uses a data processing library to transform the query results into a unified internal data structure.

[0546] Output: A merged context object containing user attribute information, condition information, emotion state information, and natural language request text.

[0547] Step 4: The server obtains environmental prediction information and purchase item information. The server obtains environmental prediction information and purchase item information from external information sources based on the location and time information in the context object.

[0548] Input: A context object containing fields such as user location and current time.

[0549] The server calls an external interface to obtain weather forecast data (such as current and future temperature, weather type, humidity, etc.) and calls a product service interface to obtain a list of purchasable products in the corresponding region (including product name, specifications, price, inventory, estimated delivery time, etc.). The server uses data processing tools to parse, extract fields, and unify units (e.g., convert temperature to degrees Celsius, unify price to a specific currency unit) the raw data returned from the external interface.

[0550] Output: The updated context object, which includes new environmental prediction information and standardized purchase item information.

[0551] Step 5: The server needs to build the feature representation of generative artificial intelligence models. The server constructs input features for use by generative artificial intelligence models based on the merged context objects.

[0552] Input: A context object containing user attribute information, condition information, emotional state information, environmental prediction information, and purchase item information.

[0553] The server uses feature engineering algorithms to normalize numerical data (budget, temperature, price, etc.), performs one-hot encoding or embedding encoding on categorical data (weather type, sentiment label, etc.), and converts the natural language request text into a vector sequence through word segmentation and word embedding methods. The server then concatenates these features in a predefined order into a multi-channel input representation for subsequent use in prompt generation and model inference.

[0554] Output: A set of structured features that can be used by prompt generation components and generative AI models.

[0555] Step 6: The server generates prompts for generative artificial intelligence models. The server uses a prompt generation component to convert structured features into prompts in natural language.

[0556] Input: Key fields from a structured feature set and a context object.

[0557] The server fills in user attribute information, conditional information, emotional state information, environmental prediction information, and constraints into the template based on predefined templates and dynamic rules, forming a complete text prompt statement. For example, the server generates the following prompt statement: The user information is as follows: - Age: 32, Female - Allergy: Shrimp - Chronic disease: Mild hypertension - Health goal: Fat loss Current mood: Fatigue Current weather: 8 degrees Celsius, light rain, feels rather cold. - Budget: No more than 1000 yen - Cooking Skills: Beginner - Maximum cooking time: 20 minutes User's natural language description: 'Budget under 1000 yen, I'm a bit tired today, and I'd like a warm but not too oily dinner.' Based on the information above, please recommend 2-3 suitable family meal plans for tonight's dinner. Requirements: 1. Overall, it's relatively healthy, with a moderate amount of fat; 2. Considering the cold weather, the dishes should be served warm. 3. Do not use any shrimp ingredients, and be careful to control the salt content; 4. Each recipe should include: the name of the dish, a list of main ingredients (including approximate weights), brief steps (3-5 steps), and a 1-2 sentence explanation of why it is suitable for relieving fatigue. During the generation process, the server will adjust the weight of some content in the template based on historical selection and evaluation information, such as increasing the emphasis on salt content control.

[0558] Output: Prompt text for generative artificial intelligence models.

[0559] Step 7: The server invokes a generative artificial intelligence model and obtains dietary suggestion information. The server takes the prompt generated in step 6 as input and calls the generative artificial intelligence model to perform inference.

[0560] Input: Prompt statements in natural language.

[0561] The server encodes the prompts into a sequence of input statements for the model via a model inference interface. The generative AI model internally processes this input sequence using an encoder, calculating self-attention weights and generating a context vector. The decoder then progressively generates the output labeled sequence. During inference, the server sets the maximum output length, temperature parameters, and sampling strategies to control the diversity and determinism of the generated content.

[0562] Output: Natural language dietary suggestions output by the generative artificial intelligence model, which typically include several candidate menus, ingredient descriptions, and step instructions.

[0563] Step 8: The server parses the dietary suggestion information and structures the dietary menu and ingredient information. The server parses the natural language output of the generative artificial intelligence model and converts it into structured menu and ingredient information.

[0564] Input: Dietary suggestion information in natural language form.

[0565] The server uses rule matching, pattern recognition, or sequence labeling models to segment the output text and identify it into dish names, ingredient names, quantities, units, and step descriptions. For example, the server identifies "Chicken and Vegetable Warm Soup" as the dish name and "150g chicken breast, 80g carrot, 100g potato" as ingredient records. The server represents each dish as a menu record, containing fields such as menu identifier, dish name, ingredient list, and step list, and stores it in temporary structured storage.

[0566] Output: A structured set of food menu and ingredient information.

[0567] Step 9: The server maps ingredient information to purchase item information and generates instruction information. The server searches for corresponding products in the purchase category information based on the ingredient information and generates instructions for immediate delivery.

[0568] Input: Structured food information and standardized purchase item information.

[0569] The server uses string similarity matching, classification coding, or vector similarity calculation to align ingredient names with standard product names in the product list, and calculates the required quantity of goods based on specifications and minimum packaging unit. The server calculates the total price for each combination and selects suitable product combinations based on the user's budget and inventory status. Subsequently, the server constructs an instruction data structure containing product identifiers, quantities, delivery addresses, and time windows for later calls to the distribution system interface.

[0570] Output: A set of order candidate instruction information generated for each candidate menu.

[0571] Step 10: The server generates recommendation results data to return to the terminal. The server combines structured menu information, ingredient mapping information, and price calculation results into recommendation results data for the end user.

[0572] Input: Structured menu information, order candidate instruction information, and user constraints.

[0573] The server sorts candidate menus based on a comprehensive score, which includes health fit (how well they align with the user's health status and nutritional goals), cost fit (the gap with budget), time fit (consistency with preparation time constraints), and emotional fit (consistency between menu type and emotional tags). The server then packages the top-ranked menus into a recommendation list, with each menu including the dish name, estimated cost, cooking time, ingredient details, and a brief reason for recommendation.

[0574] Output: Structured recommendation results data, used to display to the terminal.

[0575] Step 11: The terminal receives and displays the recommendation results. After receiving the recommendation results data returned by the server, the terminal parses the data format and displays it visually in the user interface.

[0576] Input: Recommendation results data from the server.

[0577] The terminal displays each menu item as a card or list, including the dish name, estimated price, cooking time, and a brief description, and provides interactive buttons such as "View Details" and "Buy Ingredients" for each menu item. The terminal can optimize the text and icons layout according to the device resolution and user interface style to improve readability.

[0578] Output: A recommended menu interface displayed on the terminal screen, and a next step request generated based on the user's actions (such as a details request or an order request).

[0579] Step 12: The user selects a menu item and initiates an order request on the terminal. After viewing the recommendations, users can select a menu item via the terminal and click "Buy Ingredients" or a similar button as needed.

[0580] Input: The list of recommended menus displayed in the terminal interface.

[0581] When a user makes a selection on the interface, the terminal captures this event, reads the corresponding menu item and associated ingredient information from its local cache, and combines this information with the user's currently selected delivery address and payment method to generate an order request. The terminal then sends this order request to the server over the network.

[0582] Output: An order request data packet containing the selected menu identifier and delivery-related information.

[0583] Step 13: The server processes order requests and interacts with the distribution system. After receiving the order request from the terminal, the server uses the previously generated order candidate indication information to construct a formal order and call the circulation system interface.

[0584] Input: Order request data packet from the terminal and order candidate indication information.

[0585] The server selects the corresponding product combination and quantity based on the menu item chosen by the user, fills in parameters such as the user's shipping address and payment method, and generates an order request that conforms to the distribution system interface specifications. The server sends the order request to the distribution system through the interface, receives the order confirmation information and estimated delivery time from the distribution system, and records the order status.

[0586] Output: Order confirmation information including order number, estimated delivery time, and order details.

[0587] Step 14: The server returns order confirmation information to the terminal. The server encapsulates the order confirmation information returned by the distribution system into a simplified structure and sends it to the terminal.

[0588] Input: Order confirmation information returned by the distribution system.

[0589] The server extracts core fields (order number, delivery time window, total amount, etc.) from the order confirmation data, generates a confirmation message structure for user display, and sends it to the terminal over the network. Simultaneously, the server updates the order status in its internal order database for subsequent tracking and querying.

[0590] Output: Terminal-oriented order confirmation message data.

[0591] Step 15: The terminal displays the order results, and the user completes the interaction. After receiving the order confirmation message, the terminal displays the order number, delivery time, and total amount to the user.

[0592] Input: Order confirmation message data from the server.

[0593] The terminal displays a "Order Confirmed" message on the interface, along with information such as "Estimated Delivery Time" and "View Order Details" for user confirmation. After the user reads the confirmation message, the interaction ends, and the terminal writes the current status to local storage for displaying historical orders upon the next startup.

[0594] Output: Order confirmation screen display and updated terminal status data recorded locally.

[0595] 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.

[0596] 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.

[0597] 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.

[0598] 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.

[0599] 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.

[0600] Second Implementation Method Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0601] 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.

[0602] 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).

[0603] 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.

[0604] 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.

[0605] 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).

[0606] 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.

[0607] 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.

[0608] 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.

[0609] 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).

[0610] 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.

[0611] 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".

[0612] 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.

[0613] 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.

[0614] 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.

[0615] 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.

[0616] 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.

[0617] 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.

[0618] 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.

[0619] 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.

[0620] 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.

[0621] Third Implementation Method Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0622] 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.

[0623] 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).

[0624] 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.

[0625] 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.

[0626] 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).

[0627] 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.

[0628] 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.

[0629] 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.

[0630] 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.

[0631] 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.

[0632] 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".

[0633] 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.

[0634] 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.

[0635] 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.

[0636] 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.

[0637] 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.

[0638] 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.

[0639] 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.

[0640] 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.

[0641] 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.

[0642] Fourth Implementation Method Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0643] 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.

[0644] 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).

[0645] 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.

[0646] 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.

[0647] 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).

[0648] 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.

[0649] 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.

[0650] 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.

[0651] 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.

[0652] 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.

[0653] 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.

[0654] 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".

[0655] 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.

[0656] 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.

[0657] 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.

[0658] 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.

[0659] 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.

[0660] 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.

[0661] 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.

[0662] 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.

[0663] 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.

[0664] 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 [reference]). 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.

[0665] 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.

[0666] 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.

[0667] 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).

[0668] 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.

[0669] 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."

[0670] 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.

[0671] 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).

[0672] 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.

[0673] 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.

[0674] 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.

[0675] 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.

[0676] 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.

[0677] 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.

[0678] 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.

[0679] 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.

[0680] 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.

[0681] 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.

[0682] In addition, the following notes are provided in response to the above explanation.

[0683] Example 1 (Note 1) An information processing system, characterized in that it comprises: An apparatus that acquires user attribute information, environmental condition information, cooking step information, and purchase object information from a processing device, and standardizes the information to generate a set of feature quantities; A device that uses a processing unit to construct prompt statements for inputting into a generative artificial intelligence model based on the set of features, and generates prompt statements to instruct the generative artificial intelligence model to generate dietary plan information based on the user's attribute information and environmental condition information. A device that associates the components contained in the meal plan information with the purchase object information, calculates the purchase candidate set and cost information for each required component, and generates a purchase list information containing the purchase candidate set. A device that presents the meal plan information and the purchase list information to the user through a display device, and sends an instruction to a purchase processing device or an external purchase service to obtain the purchase candidate set based on the selection information from the user; A device that obtains evaluation information and actual purchase history information from the user through a processing device, reflects the information in the feature set for updating, and thereby corrects the subsequently generated prompts and meal plan information.

[0684] (Note 2) The information processing system according to Appendix 1 is characterized in that, The processing device incorporates the user's financial constraints, physical condition, nutritional needs, supply period, cooking skills, and available preparation time into the feature set, and generates prompt statements by explicitly recording the information in the prompt statements to indicate the use of the generative artificial intelligence model to provide information on cooking objects and components that meet the user's conditions.

[0685] (Note 3) The information processing system according to Appendix 1 is characterized in that, The processing device acquires inquiry information related to the ingestion target, combines the inquiry information with the feature set and the environmental condition information to generate prompt statements, and generates prompt statements to instruct the use of the generative artificial intelligence model to generate response information to the inquiry information and additional dietary plan information based on the response information.

[0686] Application Example 1 (Note 1) An information processing system, characterized in that it comprises: An apparatus for receiving individual information, including user preference information, allergy information, health status information and location information, from a user terminal in the computing unit of an information processing device, and storing the individual information in an information storage unit for subsequent reading and use; An apparatus for acquiring meteorological forecast information, including current temperature and weather conditions, from an information providing device via a communication network, parsing the meteorological forecast information, and generating an environmental information data structure related to dietary choices. A device for reading individual information from the information storage unit, combining the individual information with the environmental information data structure, cooking recipe information, and product sales information, and constructing prompt statements for input into a generative artificial intelligence model based on a pre-set text generation template; An apparatus for providing the prompt statements to the generative artificial intelligence model, enabling the generative artificial intelligence model to generate recommended menu information including multiple dietary plans, a list of required raw materials and preparation steps based on the individual information, the environmental information, the cooking recipe information and the product sales information, and for parsing the recommended menu information to generate structured menu data; A device for sending the structured menu data to the user terminal, enabling the user terminal to display the recommended menu information and receive the user's selection information for the target menu; An apparatus for extracting a list of required raw materials from the structured menu data based on the user's selection information of the target menu, generating query conditions for each raw material, calling the product retrieval interface of the product providing device to obtain multiple sales product information corresponding to each raw material, and filtering and sorting the sales product information to generate procurement candidate information. The device is used to send the procurement candidate information to the user terminal, so that the user terminal can receive the user's confirmation operation on the specific goods and quantities, and receive the order information containing the final procurement content. Based on the order information, the device calls the order placement interface of the order processing device to generate a delivery order, thereby automatically executing the food procurement and delivery processing device corresponding to the recommended menu. An interactive control device that enables the user terminal to complete a series of processes, from providing individual information and selecting menus to automatically ordering ingredients, with minimal user operations.

[0687] (Note 2) The information processing system according to Appendix 1 is characterized in that, When generating the prompt statement, the computing unit considers the budget information, health status information, nutritional needs information, seasonal raw material information, cooking skill level information, and available preparation time information contained in the individual information, and generates a prompt statement that instructs the generative artificial intelligence model to suggest a diet plan and corresponding raw materials based on the conditions.

[0688] (Note 3) The information processing system according to Appendix 1 is characterized in that, When the computing unit receives inquiry information related to food and diet management from the user terminal, it combines the inquiry information with the individual information and the environmental information to generate prompt statements for input into the generative artificial intelligence model, so that the generative artificial intelligence model generates answer information and additional dietary advice information in response to the inquiry information.

[0689] Example 2 (Note 1) An information processing system, characterized in that it comprises: Means for obtaining personal information, including information about a user's living conditions, from a user and storing the personal information in a storage device; A means for performing query language-based retrieval processing on dietary information and food ingredient information stored in the storage device, and for extracting a candidate set from the dietary information and food ingredient information based on budget information, health status information, nutritional condition information, cooking skill information and cooking time information contained in the personal information; A means for generating a set of diets that satisfy the constraints corresponding to the personal information by performing numerical calculations on each diet included in the candidate set using food price information and nutritional value information. Means for automatically generating prompts for input into a generative artificial intelligence model based on the dietary set and the personal information, and instructing the generative artificial intelligence model to receive the prompts to generate dietary proposal information; A means for integrating the dietary proposal information with the diet set to generate display data for presentation to the user.

[0690] (Note 2) The information processing system according to Appendix 1 is characterized in that it further includes: Means for receiving the personal information on the display interface based on user input, converting the personal information into structured data and sending it to the storage device through a communication channel, and receiving the diet proposal information and the display data and presenting them to the user visually.

[0691] (Note 3) The information processing system according to Appendix 1 is characterized in that, During the process of generating the dietary proposal information, by adding the user's emotional state information and preference information to the prompt statement, the generative artificial intelligence model is instructed to generate corresponding dietary plans and food ingredient plans based on the emotional state information and preference information.

[0692] Application Example 2 (Note 1) An information processing system, characterized in that it comprises: The means of acquiring user-related attribute information, environment-related prediction information, cooking-related step information, and purchase-related item information, and using the attribute information, prediction information, step information, and item information to construct a learning dataset, and enabling a generative artificial intelligence model to learn from the learning dataset; Based on user condition information, emotional state information, and natural language request text obtained from the terminal, the condition information, emotional state information, and request text are parsed and combined with the prediction information and the category information to generate prompt statements for input into the generative artificial intelligence model. The method involves parsing the food-related suggestion information output by the generative artificial intelligence model, structuring the food menu information, ingredient information and cooking step information, and generating instruction information based on the ingredient information to issue an instant delivery instruction to the distribution system. The generative artificial intelligence model is used to perform natural language processing to generate response information that includes answer information for food-related questions and explanatory text corresponding to the emotional state information, and means to generate prompt statements for sending the response information to the terminal. A means of recording user selection history and evaluation information for the learning dataset and the dietary suggestion information, and reflecting the selection history and evaluation information when generating the prompt statement later.

[0693] (Note 2) The information processing system according to Appendix 1 is characterized in that, The means of generating the prompt statement is configured to consider conditional information, including user cost constraints, health status, nutritional goals, information related to seasonal ingredients, cooking skill level, and food preparation time, when generating the prompt statement for the generative artificial intelligence model to output optimal cooking steps and ingredient information.

[0694] (Note 3) The information processing system according to Appendix 1 is characterized in that, The means of generating the prompt statement is configured to: generate prompt statements based on food-related question text, user-related attribute information, environment-related prediction information, and the emotional state information, so that the generative artificial intelligence model can output answer information for the question and dietary menu information corresponding to the emotional state information.

Claims

1. An information processing system, characterized in that, include: processor; The processor is configured to: use personal data, weather forecast data obtained through internet search tools, recipe data, and shopping product data to train a generative artificial intelligence model; use the generative artificial intelligence model to generate prompts based on the above data to indicate the immediate delivery of the user's optimal food; identify the user's emotions and generate prompts to recommend corresponding food menus to the user based on the emotions.

2. The information processing system according to claim 1, characterized in that, The processor is further configured to: generate prompts using the generative artificial intelligence model, taking into account the user's budget, health status, nutritional needs, seasonal ingredients, cooking skills, and meal preparation time, to indicate the optimal recipes and ingredients to the user.

3. The information processing system according to claim 1, characterized in that, The processor is further configured to: in order to address various food-related issues, utilize the generative artificial intelligence model to generate prompts indicating the generation of answers to users' questions about food.

Images