Information processing system

Abstract

The application provides an information processing system. An information processing system, characterized by comprising: a processor; wherein the processor is configured to: receive a prompt text input by a user; input the prompt text to a generative artificial intelligence model; cause the generative artificial intelligence model to generate a concept or design according to the prompt text; upload the generated concept or design to an electronic market; and endow the generated concept or design with a non-fungible token.

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] While existing technologies utilize generative artificial intelligence models to automatically generate concepts or designs, the following problems still exist: First, after users input the prompt text, there is a lack of an integrated system that can automatically complete the process from parsing the prompt text, generating concepts or designs, to publishing and trading the generated results in the e-market. This results in a fragmented creation process, requiring users to switch between multiple platforms, which increases the barrier to entry and operating costs.

[0004] Second, when the generated concepts or designs circulate in the electronic market, their uniqueness and ownership are difficult to be reliably proven. The lack of effective integration with blockchain technologies such as non-fungible tokens (NFTs) can easily lead to problems such as unclear copyright ownership, misappropriation, and infringement.

[0005] Third, there is a lack of a unified mechanism for automatically matching transactions, calculating fees, and recording income for transactions between users based on generated or uploaded concepts or designs. Platform operators find it difficult to accurately and timely track and manage transaction revenue, and users also find it difficult to understand their own income status in a timely manner.

[0006] Therefore, it is necessary to provide an information processing system that can complete the following within the same platform: receiving prompt text, calling generative artificial intelligence models to generate concepts or designs, issuing and assigning non-fungible tokens in electronic markets, and simultaneously matching transactions between users and automatically calculating and recording transaction fee revenue, thereby solving the above problems. Summary of the Invention

[0007] To address the aforementioned issues, this invention provides an information processing system, comprising: a processor; wherein the processor is configured to: receive prompt text input by a user; input the prompt text into a generative artificial intelligence model; cause the generative artificial intelligence model to generate a concept or design based on the prompt text; upload the generated concept or design to an online marketplace; and assign a non-fungible token to the generated concept or design. Through this configuration, this invention achieves an integrated processing flow from user input to work generation, uploading, and digital ownership marking within a single system, lowering the user's barrier to entry and improving the efficiency of work management and publishing.

[0008] Furthermore, to improve the consistency between the generated results and user needs, in this invention, the processor is configured to: parse the prompt text using natural language processing technology and instruct the generation of the concept or design. By performing semantic analysis and structured processing on the prompt text, the system can more accurately understand the user's intent, thereby driving the generative artificial intelligence model to output a concept or design that better meets the needs, achieving a higher quality automatic generation effect.

[0009] Furthermore, to achieve efficient monetization of generated or uploaded works and platform revenue management in the electronic marketplace, the processor in this invention is configured to: act as an intermediary in transactions between users; and calculate transaction fees based on the transactions and record the fees as revenue. By integrating transaction matching and settlement logic into the system, this invention can automatically complete the transaction process between users, accurately calculate and record the transaction fees for each transaction, thereby providing a reliable revenue management tool for platform operation, while providing users with transparent transaction and revenue information.

[0010] In summary, by comprehensively utilizing the aforementioned technical means, this invention can complete generative AI creation, electronic market release, non-fungible token assignment, and transaction and fee management within the same system, thereby effectively solving problems such as fragmented creation processes, difficulty in proving ownership, and inconvenience in managing transaction revenue in existing technologies.

[0011] "System" refers to an overall collection of devices consisting of one or more hardware and / or software components for executing the processing flow described in this invention, and may include servers, processors, memory, network interfaces, and related software modules.

[0012] A processor is a hardware unit or its logical equivalent that is capable of executing computer program instructions to process input data and output results. It can be a single central processing unit (CPU), graphics processing unit (GPU), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or any combination thereof.

[0013] "User" refers to an entity that interacts with the system through a terminal device, inputs prompt text, browses or trades concepts or designs, etc., and can be a natural person, a legal person or other organization with legal qualifications.

[0014] "Prompt text" refers to text information input by the user and used to drive generative artificial intelligence models to generate and process data. It can include natural language descriptions of style, content, purpose, constraints, etc.

[0015] "Generative artificial intelligence models" refer to data-driven models that can automatically generate content semantically related to the input prompt text. These models can include deep learning-based text generation models, image generation models, audio generation models, or multimodal generation models.

[0016] "Concept or design" refers to creative content generated by a generative artificial intelligence model based on prompt text, or managed as a transaction object in this system, including but not limited to visual design schemes, product concept schemes, images, illustrations, icons, layout designs, interface designs, logo designs, copywriting schemes, or any combination thereof.

[0017] "Electronic marketplace" refers to a network-based trading platform provided and operated by a system for displaying, publishing, searching, and trading the concepts or designs described therein, including the front-end interface, back-end management modules, and server-side programs for processing orders and settlements.

[0018] "Upload" refers to the process by which a processor transfers data related to a generated or existing concept or design from a local or intermediate processing environment to a server or storage system used by an electronic marketplace so that the concept or design can be displayed and traded in the electronic marketplace.

[0019] "Non-Fungible tokens" refer to digital tokens issued based on blockchain technology that have unique identifiers and cannot be substituted for one another. They are used to represent the uniqueness and ownership information of a specific concept or design in the digital space, and usually correspond to NFTs (Non-Fungible Tokens).

[0020] "Assigning non-fungible tokens" refers to the process of associating a generated concept or design with a non-fungible token issued on the blockchain. This includes creating NFT metadata for the concept or design, minting the corresponding token on the blockchain, and recording the correspondence between the token and the concept or design.

[0021] Natural Language Processing (NLP) technology refers to a class of algorithms and models used to analyze, understand, and process human natural language text, including but not limited to word segmentation, part-of-speech tagging, syntactic analysis, semantic parsing, intent recognition, and text vectorization.

[0022] "Parsing the prompt text" refers to the process by which the processor uses natural language processing technology to perform syntactic and semantic analysis on the prompt text input by the user, extracting key elements, constraints, and generation goals, so that the generative artificial intelligence model can generate corresponding concepts or designs based on this information.

[0023] "Acting as an intermediary in transactions between users" means that the processor receives transaction requests from buyers and sellers regarding a concept or design through an electronic marketplace platform, generates and manages orders, and coordinates the payment and delivery process, but does not necessarily actually hold or use the concept or design as a party to the transaction.

[0024] "Transaction fee" refers to the service fee charged by the platform from the transaction amount according to a preset percentage or rule when a user transacts on a concept or design. This fee is used to compensate the platform for the costs incurred in matching, settlement, and maintaining the system.

[0025] "Recording transaction fees as revenue" means that after the processor completes a transaction between users, the transaction fee amount is recorded in the system's accounting or database records and stored and statistically analyzed as the platform's operating revenue or profit data. Attached Figure Description

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

[0027] Figure 2 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and smart device according to the first embodiment.

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

[0029] Figure 4 This is a conceptual diagram illustrating an example of the main functions of the data processing device and smart glasses according to the second embodiment.

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

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

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

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

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

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

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

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

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

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

[0040] Hereinafter, an example of an implementation of the system according to the present disclosure will be described with reference to the accompanying drawings.

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

[0042] In the following embodiments, the processor (hereinafter referred to as "processor") with reference numerals may be a single computing device or a combination of multiple computing devices. Furthermore, the processor may be a single computing device or a combination of multiple computing devices. Examples of computing devices include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), etc.

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

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

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

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

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

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

[0049] The data processing apparatus 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" as understood in this disclosure. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0050] The smart device 14 includes a computer 36, a receiving device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. In addition, the receiving device 38, output device 40, camera 42, and communication I / F 44 are also connected to the bus 52.

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

[0052] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting data in a form perceptible to the user 20 (e.g., sound and / or text). The display 40A displays visual information such as text and images according to instructions from the processor 46. The speaker 40B outputs sound according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

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

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

[0055] like Figure 2 As shown, in the data processing apparatus 12, specific processing is performed by the processor 28. A specific processing program 56 is stored in the memory 32. The specific processing program 56 is an example of a "program" as understood in this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0056] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. In the emotion inference function (emotion-specific function) using the emotion-specific model 59, various inferences and predictions related to the user's emotions are performed, including inferences and predictions of the user's emotions, but this is not limited to this example. Furthermore, emotion inference and prediction may also include, for example, emotion analysis (parsing).

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

[0058] Alternatively, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. Furthermore, the data processing device 12 may be a server device or a user-held terminal device (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of the processing of the data processing system 10 of the first embodiment will be described.

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

[0060] Existing technologies that utilize generative artificial intelligence models to automatically generate concepts or designs and transact them in a network environment suffer from the following shortcomings at the computer technology level: First, servers typically forward user-input prompts as plain text to generative AI services, lacking a unified text preprocessing and encoding mechanism. This leads to unstable model input formats, inconsistent generation quality, and low processing efficiency, making it difficult to operate efficiently in e-market environments with multiple concurrent users.

[0061] Second, existing systems generally store the generated results simply as files or unstructured data, without forming standardized conceptual data or design data structures on the server side. They also lack systematic association with metadata such as user identifiers, generation conditions, and prompt statements, which requires a lot of additional application layer processing for subsequent retrieval, tracing, and transaction management, increasing server resource consumption.

[0062] Third, the binding process between generated content and non-fungible tokens is mostly implemented in a fragmented manner by application logic. When the server interacts with the distributed ledger management device, it lacks a unified mechanism for generating transaction data and managing state. It cannot reliably correspond the work identifier, token identifier, and transaction record in the same data structure, which can easily lead to technical problems such as asynchronous ownership information and inconsistent transaction states.

[0063] Fourth, when users trade in electronic marketplaces, traditional systems often separate transaction matching, ownership transfer, and fee settlement into different modules or external services. The server lacks integrated control logic for buying and selling requests, non-fungible token ownership transfer, and fee calculation, resulting in complex transaction processes, difficulty in error handling, and poor system scalability.

[0064] Therefore, there is a need for a mechanism on the server side that improves the processing efficiency, data consistency, and scalability of generative AI-supported electronic market systems through improved text preprocessing, model input construction, structured storage of generated results, integrated interaction with distributed ledgers, and transaction and fee management mechanisms. This would improve the implementation of generation, listing, trading, and ownership management at the computer technology level.

[0065] 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.

[0066] In this invention, the server includes: a device for receiving input prompts from a user terminal via a communication function and obtaining the prompts as text data; a device for performing preprocessing on the text data, including string normalization, length detection, and format detection, and converting the text data into an input format for a generative artificial intelligence model; a device for performing symbol sequence partitioning on the text data, encoding it into an identifier sequence, and converting it into a numerical vector sequence to generate a tensor for model input, and inputting the tensor into the generative artificial intelligence model to instruct the generation of concept data or design data; a device for enabling the generative artificial intelligence model to perform inference processing based on the numerical vector sequence, including multi-layer matrix operations and attention mechanism operations, thereby generating concept data or design data; and a device for storing the concept data or design data in a data storage device and integrating it with work identifiers, user identifiers, prompts, generation conditions, and generation... The device includes: an apparatus for recording time as metadata; an apparatus for converting the conceptual or design data into product information that can be publicly disclosed as electronic market information, and generating distribution data for list display and detailed display containing index information; an apparatus for generating transaction data for issuing non-fungible tokens on a distributed ledger based on the product information, sending the transaction data to a distributed ledger management device to enable the issuance of non-fungible tokens, and storing the identification information of the non-fungible tokens in correspondence with the work identification; and an apparatus for receiving buying and selling requests between users through the electronic market, generating control information for executing transaction processing including the transfer of ownership of the non-fungible tokens, initiating ownership transfer transactions to the distributed ledger management device based on the control information, automatically calculating the transaction fee based on the transaction amount, storing the transaction fee and transaction records, and managing them as revenue information. This allows for the formation of a unified data processing and transaction control architecture within the server, oriented towards generative artificial intelligence. This enables integrated computational processing across the entire process, from prompt input, model reasoning, structured storage of generated results, product information generation, non-fungible token issuance, ownership transfer, and fee settlement. While ensuring data consistency and traceability, it improves the efficiency and scalability of generation and transaction processing, thereby enhancing the overall performance of the electronic market system at the computer technology level.

[0067] A "system" refers to an overall technical solution consisting of one or more computing devices and the programs, data storage devices, and communication devices running on them, used to perform functions such as prompt statement processing, generative artificial intelligence model invocation, data storage, and transaction control.

[0068] A "server" refers to a computing device or node that provides computing resources, storage resources, and service interfaces in a network environment to receive requests from user terminals and perform generation, storage, and transaction processing.

[0069] "User terminal" refers to an information processing device operated by a user to send prompts to the server, receive generated results, and perform browsing, listing, and trading operations, including but not limited to mobile terminals, fixed terminals, or other electronic devices with communication functions.

[0070] "Prompt statements" refer to text data input by users in natural language or structured text, used to instruct generative artificial intelligence models to generate target content, including generation requirements, constraints, or descriptive information.

[0071] "Text data" refers to information data represented in the form of a sequence of characters or symbols, including prompt statements input by the user terminal and received and processed by the server.

[0072] "Preprocessing" refers to the formatting, normalization, and verification processes performed on the text data before it is input into the generative artificial intelligence model. This includes string normalization, length detection, format detection, and other transformation operations that are beneficial to the model's processing.

[0073] "String normalization" refers to the operation of standardizing characters in text data, including removing redundant whitespace, standardizing encoding, standardizing punctuation marks, and eliminating unnecessary control characters.

[0074] "Length detection" refers to the process of measuring and judging the number of characters or symbols in text data to confirm whether it meets the predetermined minimum and maximum length range.

[0075] "Format checking" refers to the process of checking the structure and content of text data to confirm whether it meets predetermined input specifications, such as character type constraints, prohibited symbol checks, or basic syntax checks.

[0076] "Generative artificial intelligence models" refer to artificial intelligence models built on machine learning and deep learning technologies that can automatically generate text, concepts, designs or other data content based on input prompts, including but not limited to language models or multimodal models based on neural networks.

[0077] "Input format" refers to the specific data structure or encoding form that generative artificial intelligence models can receive and process, including numerical data after word segmentation, encoding, and vectorization.

[0078] "Symbol sequence segmentation" refers to the process of dividing continuous text data into several basic symbol units (such as characters, words, or sub-words) according to predetermined rules.

[0079] A “sequence of identifiers” refers to an ordered sequence of one or more discrete identifiers, each of which corresponds to a symbolic unit in the text and is used to represent the discrete encoding form of the text in the model.

[0080] A “numerical vector sequence” refers to a sequence of multiple numerical vectors arranged in order, with each numerical vector corresponding to an identifier, used as an input feature representation for generative artificial intelligence models.

[0081] "Model input tensor" refers to a multidimensional numerical array obtained by arranging and expanding the dimensions of a sequence of numerical vectors, which is used as the input data structure when a generative artificial intelligence model performs inference operations.

[0082] "Multi-layer matrix operations" refers to multiplication, addition, and other linear or nonlinear operations performed on parameter matrices and input tensors in the various layers within a generative artificial intelligence model.

[0083] "Attention mechanism computation" refers to the computational process in generative artificial intelligence models that selectively aggregates information by calculating the relevance weights between different positions, thereby improving the model's ability to model contextual relationships.

[0084] "Inference processing" refers to the process in generative artificial intelligence models of performing forward propagation calculations on input data and outputting result data under fixed model parameters.

[0085] "Conceptual data" refers to structured or semi-structured data generated by generative artificial intelligence models to describe abstract ideas, functional concepts, or solution frameworks.

[0086] "Design data" refers to structured or semi-structured data generated by generative artificial intelligence models to describe appearance design, structural design, or other feasible solutions.

[0087] "Data storage device" refers to storage resources used to preserve data in a system for a long or medium term, including database systems, file storage systems or other non-volatile storage media.

[0088] "Work Identifier" refers to the identification information used to uniquely identify the generated conceptual or design data. It can be a string, a number, or other data that can uniquely locate the work.

[0089] "User ID" refers to the identification information used to uniquely identify the user who issues the prompt or owns the copyright to the work.

[0090] "Generation time" refers to the time information corresponding to when a generative artificial intelligence model outputs conceptual data or design data, used to record the point in time when generation occurs.

[0091] Metadata refers to descriptive information attached to conceptual or design data, including prompts, user identifiers, generation conditions, model type information, and time information, used for management and retrieval.

[0092] "Electronic marketplace" refers to an information processing platform provided through communication networks for displaying, browsing, and trading digital content such as concept data or design data.

[0093] "Product information" refers to publicly available information associated with conceptual or design data used for display and trading in electronic marketplaces, including titles, descriptions, prices, metadata references, etc.

[0094] "List information" refers to a collection of brief product information displayed simultaneously to multiple users in an online marketplace in list format, for browsing and filtering.

[0095] "Detailed information" refers to product information displayed individually for a specific product, including complete attributes, metadata, and transaction options.

[0096] "Distribution data" refers to data generated and output by a server for display or processing on terminal devices, including structured data used to present list information or detailed information.

[0097] "Distributed ledger technology" refers to a ledger management technology that uses multiple computing nodes to jointly maintain and update ledgers, and uses a consensus mechanism to ensure data consistency.

[0098] A "distributed ledger management device" refers to a computing node or set of nodes used to run distributed ledger technology, receive transaction data, and record it in a ledger.

[0099] "Non-fungible tokens" refer to digital tokens issued based on distributed ledger technology that have unique identifiers and are not interchangeable, used to represent the rights or ownership of specific digital works or data objects.

[0100] "Transaction data" refers to the data structure constructed to record token issuance or ownership changes on a distributed ledger, including the transaction initiator, recipient, and related identification information.

[0101] "Identification information" refers to the tag data used to uniquely identify non-fungible tokens, including contract address, token number or other unique identifiers.

[0102] A “purchase request” refers to a request sent by a user through an online marketplace to buy or sell a work corresponding to a specific product listing.

[0103] "Transaction processing" refers to a series of calculation and data update operations performed based on buy and sell requests, including price confirmation, authorization verification, ownership transfer, and record updates.

[0104] "Control information" refers to instructional data generated by a server to instruct a distributed ledger management device or other module to perform specific transaction processing steps.

[0105] A “transfer of ownership transaction” is a transaction recorded on a distributed ledger that transfers ownership of nonfungible tokens from one party to another.

[0106] "Transaction amount" refers to the total payment amount corresponding to a specific buy and sell request, used to settle the consideration for the work and calculate handling fees.

[0107] "Transaction fee amount" refers to the service fee calculated based on the transaction amount and used as revenue for the system or platform.

[0108] A “transaction record” is a collection of data used to represent the details of a specific transaction, including the transaction amount, transaction fee amount, identification of the parties involved in the transaction, time information, and related works or token identifiers.

[0109] "Revenue information" refers to the recorded data associated with the revenue obtained by the system through handling fees or other means, which is used to statistically analyze and manage the system's revenue.

[0110] The embodiments of this invention will combine hardware structure, software modules, data structure, and the internal processing of generative artificial intelligence models to illustrate how servers, terminals, and users can collaboratively achieve concept or design generation, listing, and transaction control based on non-fungible tokens, and explain the resulting improvements in computer technology.

[0111] I. Overall System Composition A server is implemented on one or more computing devices. The server can employ general-purpose server hardware, including: a multi-core central processing unit, a graphics processing unit, main memory, and data storage devices based on disk or solid-state storage. The underlying software running the server can include: an operating system (e.g., a Unix-like operating system), network server software (e.g., an HTTP server), a database management system (e.g., relational database management software), distributed ledger interaction middleware, and deep learning inference frameworks (e.g., frameworks for executing generative artificial intelligence models).

[0112] A terminal is implemented on a user-operated information processing device. The terminal can be a mobile computing terminal, a desktop information processing terminal, or other electronic device equipped with a display and input device. The basic software running on the terminal may include: a graphical user interface system, a web browser or client application, a communication module for sending and receiving data, and an interface rendering module for presenting data returned from the server.

[0113] Users input prompts into the server through the human-computer interaction interface on the terminal, browse the conceptual or design data generated by the server, and operate the listing and trading functions in the electronic marketplace. Users do not directly access the server's internal data structures and model calculations, but interact with the server through the terminal's interface.

[0114] II. Server-side functional modules and data structures 1. Prompt Statement Receiving and Preprocessing Module The server includes a prompt message receiving and preprocessing module. The server receives prompt messages from the terminal via a communication interface and parses them into uniformly encoded text data. Internally, this module uses a character stream buffer and parser to process the input data, converting network layer messages into string objects that can be processed by the application layer.

[0115] The server stores metadata related to the prompt statements in this module, including user identifier, session identifier, reception time, and terminal type identifier. The server stores this metadata in an in-memory request context structure for use by subsequent modules.

[0116] The server performs string normalization operations on the prompt statement, including: converting all text to a unified encoding format, removing leading and trailing whitespace characters, merging consecutive spaces, standardizing punctuation marks, and deleting invisible control characters. Through this standardization, the server reduces ambiguity in subsequent word segmentation and encoding stages, improving the consistency of input to generative artificial intelligence models.

[0117] The server performs length and format checks within the same module. Based on pre-configured maximum and minimum length thresholds, the server checks the number of characters or symbols in the prompt statement. If the number exceeds the limit, it truncates the message or returns an error message on the server side. The server also checks the text format, such as disabling specific control symbols or illegal encoding sequences, thus preventing abnormal input from crashing the model's inference process or wasting computational resources. This server-side pre-screening directly reduces the probability of invalid requests entering the deep learning inference module, thereby reducing computational load and improving overall processing speed.

[0118] 2. Encoding and Model Input Construction Module The server includes encoding and model input construction modules. Using a pre-configured symbol table and word segmentation rules, the server performs symbol sequence partitioning on the normalized text data, dividing the text into characters, words, or sub-word units. The server maps each symbol unit to a corresponding integer identifier through table lookup operations, forming an identifier sequence. This identifier sequence is a one-dimensional integer array, representing the discrete input of the generative artificial intelligence model.

[0119] The server further transforms the identifier sequence into a sequence of numerical vectors by embedding a parameter matrix and a positional encoding vector. The server uses each identifier as an index to retrieve a row in the embedding matrix and generates a corresponding fixed-dimensional real-valued vector. The server then adds or otherwise linearly combines the positional encoding with the embedding vector according to the sequence position, thereby numerically introducing positional information. The server sequentially arranges all vectors into a two-dimensional numerical array and stacks them in batch processing mode to form a tensor for model input.

[0120] Through the specific data structures and algorithm processes described above, the server can uniformly construct efficient model input tensors without relying on terminal computing power, which facilitates parallel matrix operations on the graphics processing unit and improves inference throughput.

[0121] 3. Generative Artificial Intelligence Model Inference Module The server includes a generative artificial intelligence model inference module. Within this module, the server deploys a neural network model with a multi-layered self-attention structure. The generative AI model can employ a multi-layered encoder-decoder structure or an autoregressive decoder structure, with each layer including a multi-head attention sublayer, a feedforward network sublayer, and a normalization sublayer. The server inputs the aforementioned model data into the model using tensors.

[0122] During inference, the server performs the following data operations: In each layer, the server performs matrix multiplication on the input tensor and weight matrix to generate query, key, and value vectors, respectively; the server calculates the attention weight matrix using dot product and normalization functions, and then multiplies it by the value vector matrix to obtain a weighted sum representation; the server concatenates the outputs of each attention head and transforms them through linear layers and nonlinear activation functions. The server continues to perform linear and nonlinear transformations, as well as residual connections and normalization operations, in the feedforward network sublayers. Following a sequence generation strategy, the server progressively outputs the probability distribution of the next identifier and selects output identifiers based on algorithms such as temperature parameters, top-k or top-p sampling, thus constructing an output identifier sequence.

[0123] The server, through multi-layered matrix operations and attention mechanisms, does not simply reproduce user input. Instead, it leverages statistical patterns and semantic relationships learned on large-scale training data to generate conceptual or design data that meets the constraints based on the context of the prompts. Unlike traditional rule-based automatic generation, the model in the server obtains weight parameters through end-to-end training, enabling it to model complex constraints in a high-dimensional vector space. This achieves a combination capability far exceeding that of manually written rules, thus technically improving the quality and diversity of generated data.

[0124] The server schedules the graphics processing unit to perform the aforementioned floating-point operations using a deep learning inference framework. Because the server performs batch processing and tensor quantization on the input, multiple requests can be merged into batches and subjected to matrix operations in a unified manner, thereby improving the utilization of the hardware computing unit and significantly reducing inference latency.

[0125] 4. Result Decoding and Structured Storage Module The server includes decoding and structured storage modules. From the identifier sequence output by the model, the server removes the start and end identifiers and converts the identifier sequence into natural language text through a reverse vocabulary mapping. If the model uses a specific format (e.g., key-value pairs or structured paragraphs), the server parses the text based on predefined delimiters and tagging rules, converting it into internal conceptual data objects or design data objects. These objects can contain multiple fields, such as functional description fields, appearance description fields, dimensional parameter fields, material suggestion fields, etc.

[0126] The server assigns a unique artwork identifier to each generated concept or design data, and stores it as metadata in the database along with the user identifier, prompts, generation conditions (such as the model version used and sampling parameters), and generation time. The server can use a relational database structure to define artwork tables, metadata tables, prompt log tables, etc., and link the above data through primary keys and foreign keys to support efficient subsequent queries and retrievals.

[0127] By structuring the generated results instead of simply storing them as unparsed text, the server can build index structures (such as full-text indexes and field indexes) at the database level, thereby improving the efficiency of search and filtering operations in the e-market and reducing the burden of manual parsing and traversal at the application layer, thus achieving technical improvements in data management.

[0128] 5. Product Information Generation and Index Management Module The server includes a product information generation and index management module. Based on conceptual or design data, the server generates a product information structure for display. The server combines the product title, summary description, price parameters, and metadata such as model type information, generation condition information, and prompts to form a two-tiered data structure for list display and detail display.

[0129] In this module, the server generates index information for product information. For example, the server creates inverted indexes for product titles and abstracts, composite indexes for model type information and generation conditions, and search keys for keywords in prompts, enabling online marketplace search requests to receive rapid responses at the database level. Because the server manages metadata and product information in an integrated manner, search requests do not require extensive string scanning at the application logic layer, reducing redundant computations in the central processing unit and improving the overall system response speed.

[0130] 6. Distributed Ledger Interaction and Non-Fungible Token Management Module The server includes a distributed ledger interaction and non-fungible token management module. In this module, the server generates token metadata descriptions based on the work identifier and metadata, constructs a metadata structure containing fields such as work link, work hash, and creator identifier, and generates non-fungible token issuance and transaction data accordingly.

[0131] The server encodes the aforementioned transaction data according to a predetermined protocol format by invoking the distributed ledger interaction middleware and sends it to the distributed ledger management device via a network interface. The server receives transaction receipts and block confirmation information from the distributed ledger and extracts the identification information of the non-fungible tokens (NFTs), such as token number and contract address, from the returned data. The server stores this identification information in a database, mapping it to artwork identifiers, thus forming a two-way mapping relationship between artworks and tokens.

[0132] In ownership transfer scenarios, the server generates ownership transfer transaction data based on the buy / sell request information and submits it to the distributed ledger in a manner similar to issuance. Through this modular transaction data construction and state management mechanism, the server ensures the consistency of the artwork data with the ledger records in terms of time sequence and holder information, avoiding ownership mismatch issues caused by improper multi-system collaboration, thereby improving the reliability of digital asset management at the technical level.

[0133] 7. Transaction and Fee Calculation Module The server includes a transaction and fee calculation module. Within this module, the server receives buy and sell request data from the electronic marketplace, verifies the artwork status, the legality of the buyer and seller identifiers, and confirms whether conflicting transactions already exist. Based on a preset fee rate and transaction amount, the server calculates the fee amount using floating-point arithmetic, combines the fee amount, transaction amount, buyer identifier, seller identifier, artwork identifier, and time information into a transaction record structure, and stores it in the database.

[0134] The server ensures consistency between transaction records and changes in the ownership status of non-fungible tokens through a unified data structure and atomic database operations. The server can simultaneously update the artwork holder field and insert transaction records within a single transaction, thus guaranteeing either complete success or complete rollback in the event of system crashes or network anomalies, reducing the risk of data inconsistency in intermediate states.

[0135] III. Terminal-side functions and user interaction The terminal provides a user-friendly interface. In this invention, the terminal does not perform complex model inference calculations, but is primarily responsible for collecting prompts, displaying generated results and product information, and sending listing and transaction operation instructions. The terminal displays the product list and details returned by the server in a graphical interface format using text and images. Users only need to edit prompts using their input devices to trigger the entire technical processing flow on the server side.

[0136] Terminals can perform basic input checks locally, such as checking if prompts are empty or contain obviously illegal characters, thus filtering out obviously invalid requests before they reach the server. This collaborative filtering mechanism between the terminal and the server helps reduce network communication and server processing burden.

[0137] IV. Training and Technical Details of Generative Artificial Intelligence Models The server trains the generative AI model offline. It loads a training corpus from a data storage device, which can include various natural language texts, design specifications, product manuals, etc. The server converts the training text into identifier sequences and tensors for model input, and performs supervised learning using a deep learning training framework.

[0138] The server defines a loss function, such as cross-entropy loss, to measure the difference between the identifier probability distribution output by the model and the target sequence. The server calculates the gradient of the loss function with respect to the model parameters using backpropagation and updates the weight matrix and bias vector at each training step using an optimization algorithm (e.g., adaptive learning rate optimization). During training, the server can use data augmentation methods, such as random truncation, word order perturbation, and masking, to improve the model's robustness to deformed inputs.

[0139] Through the training methods described above, the generative AI model on the server learns the statistical dependencies and semantic structures between texts in a high-dimensional parameter space, enabling it to generate reasonable concepts or design data for new prompts during the inference phase. This training and inference mechanism, based on specific mathematical models and optimization algorithms, differs from the traditional method of manually summarizing rules and boasts higher generalization ability and generation efficiency.

[0140] V. Specific Usage Examples 1. Conceptual Design Example The user enters a prompt message on the terminal interface: "Please design a futuristic smartwatch for me. The target users are urban white-collar workers aged 20-30. It needs to emphasize both fashion and practicality. Please provide detailed descriptions of the watch face design, materials, color scheme, and interaction methods." The terminal sends the prompt as text data to the server. Following the preprocessing, encoding, model reasoning, and structured storage process described above, the server generates design data containing multiple fields, such as: an oval dial, the use of a metal and glass composite material, a main color scheme of dark blue and silver, and the use of touch and gesture recognition interaction methods. The server stores this design data along with the prompt and metadata, and generates product information for display in the online marketplace.

[0141] 2. Logo Design Examples The user enters a prompt message on the terminal interface: "Please generate three minimalist tech company logo concepts in different styles. Each concept should be described in about 200 words, along with the appropriate application scenarios (official website, app icon, business card, etc.)." The server uses a generative artificial intelligence model to generate multiple logo concept description texts, storing them as different works and generating product information for each. Users can select one from the terminal interface to list, and the server will handle the issuance and subsequent trading of non-fungible tokens.

[0142] 3. Examples of Industrial Products The user enters a prompt message on the terminal interface: "Please design the overall appearance and interaction flow of a smart pillbox for elderly users, focusing on large font display, voice reminders, and mechanisms to prevent accidental medication use, and provide a structured list of design points." Based on the prompt, the server generates structured design data that includes various aspects such as appearance features (e.g., large buttons, high-contrast displays), interaction logic (e.g., voice prompts), and security mechanisms (e.g., secondary confirmation buttons, time lock logic). The server stores this design data and displays it in the e-marketplace for further reference in physical product development, thereby connecting the system of this invention with the real-world device design and manufacturing process.

[0143] VI. Technical Effects and Causal Relationships By employing a unified text preprocessing, encoding, and tensor construction process on the server side, this invention achieves standardization of format and quality before data enters the generative artificial intelligence model, thereby reducing the model's sensitivity to abnormal inputs and improving the stability and accuracy of inference results.

[0144] By storing the generated results in a structured manner and indexing the metadata and product information, the server can directly utilize the database's optimization mechanisms to complete queries when performing search and retrieval operations. This eliminates the need for repeated parsing and scanning of large amounts of text at the application layer, thereby significantly reducing the computational overhead of the central processing unit and improving system response speed.

[0145] By implementing a unified interaction module with the distributed ledger within the server, the work identifier, non-fungible token identifier, and transaction records are managed in a unified data structure. The server can complete the synchronization update of the ledger status and the internal database status in a single transaction, reducing the risk of inconsistent ownership information and improving the system's reliability and fault tolerance.

[0146] By integrating transaction processing and fee calculation into a unified module on the server, the server can automatically perform operations such as price verification, fee calculation, transaction record storage, and ownership transfer request construction after receiving buy and sell requests. This avoids complex interface calls between multiple systems, reduces network communication load and error propagation paths, thereby simplifying the transaction process and improving stability.

[0147] In summary, this invention provides a technical solution that not only automates business processes but also improves the data processing capabilities, resource utilization efficiency, and reliability of the computer system itself by establishing clear data flow and functional division among the server, terminal, and user, and by implementing dedicated modules within the server for prompt statement processing, generative artificial intelligence model inference, data structured storage, index management, and distributed ledger interaction.

[0148] use Figure 11 The processing flow is explained.

[0149] Step 1: Users input prompts on the interface using the terminal. Users type natural language text into the input box on the terminal as a prompt, such as "Please design a futuristic smartwatch for me".

[0150] Input: The sequence of characters entered by the user via keyboard or touch.

[0151] The terminal caches each key press event in a local string buffer, and after the user confirms the input, saves the complete prompt statement as a piece of text data into memory.

[0152] Output: A text message containing a prompt in the terminal memory, along with the user identifier and the current session identifier.

[0153] Step 2: The terminal performs basic validation on the prompt statement and generates request data. The terminal performs length and null checks on the locally cached prompt statements to determine whether the text is empty or whether the length exceeds a preset threshold.

[0154] Input: The prompt text data in the terminal memory and the preset length threshold parameter.

[0155] The terminal counts the number of characters in the input text and compares the resulting length with the minimum and maximum values. If the length does not meet the requirements, an error message is displayed on the interface. If the length meets the requirements, the terminal encapsulates the prompt, user identifier, and session identifier into a request data structure.

[0156] Output: The validated prompt and the request data structure containing user information are ready to be sent to the server.

[0157] Step 3: The terminal sends a request containing a prompt statement to the server via the communication module. The terminal uses the communication module to encode the request data into network messages, such as converting it into a JSON string, and then encapsulates it into an HTTP request according to the application layer protocol.

[0158] Input: The encapsulated request data structure and the server interface address.

[0159] The terminal serializes the requested data, converts the structured data into a byte stream, and sends the byte stream in fragments to the server's network address through the transmission control protocol stack.

[0160] Output: The network request message arriving at the server port, which includes the prompt text data and related metadata.

[0161] Step 4: The server receives the network request and parses the prompt statement. The server receives HTTP requests from the terminal through the network interface, and the application server module parses the request messages.

[0162] Input: An HTTP request message in byte stream form received from the network layer.

[0163] The server decodes the message, restoring the byte stream into text-based header and body data. It then parses the JSON data from the body, extracting fields such as the prompt text, user identifier, and session identifier.

[0164] Output: The prompt text data and associated user information object stored in the server's memory.

[0165] Step 5: The server performs string normalization and validation on the prompt statement. The server performs character-level processing on the parsed prompt, including removing leading and trailing whitespace, merging consecutive spaces, and replacing exception control characters.

[0166] Input: The original prompt text data in the server's memory.

[0167] The server iterates through the character array, performing conditional checks and replacement operations on each character. Simultaneously, the server recalculates the text length and checks if it exceeds the maximum allowed length on the server side. If it does, the server performs a truncation operation or generates an error response.

[0168] Output: Normalized prompt text data, and validity flags for subsequent processing.

[0169] Step 6: The server performs symbol sequence division and identifier encoding on the prompt statement. The server uses a pre-loaded tokenizer component to divide the normalized text into symbol units (such as characters, words, or subwords) according to rules.

[0170] Input: Normalized prompt text data and word segmenter vocabulary.

[0171] The server executes a word segmentation algorithm to divide the string into several symbol units according to the matching strategy; then, the server looks up the corresponding integer identifier for each symbol unit and arranges the results into an identifier sequence in order.

[0172] Output: A sequence of identifiers in the form of a one-dimensional integer array and the corresponding sequence length information.

[0173] Step 7: The server converts the identifier sequence into tensors, which are then used as model inputs. The server uses an embedding matrix and a positional encoding matrix to transform the sequence of identifiers into a sequence of numerical vectors and construct a multidimensional tensor.

[0174] Input: Identifier sequence, embedding matrix parameters, and position encoding parameters.

[0175] The server performs a matrix row indexing operation on each identifier to obtain the corresponding embedding vector; the server generates a position vector based on the sequence number and performs vector addition with the embedding vector to obtain an input vector sequence containing position information; the server arranges all vectors in order to form a two-dimensional array, and expands it into a three-dimensional or higher-dimensional tensor for model input in the batch dimension.

[0176] Output: Numerical tensors adapted to the structure of generative artificial intelligence models and corresponding attention mask data.

[0177] Step 8: The server invokes a generative artificial intelligence model to perform forward inference operations. The server inputs the input tensor and attention mask into the generative artificial intelligence model deployed on the graphics processing unit, performing multi-layer matrix operations and attention mechanism operations.

[0178] Input: The model input consists of tensors, attention masks, and model weight parameters.

[0179] In each layer, the server performs matrix multiplication on the input tensor and weight matrix to calculate the query, key, and value vectors. The server then calculates attention weights through dot products and normalization, performing a weighted summation of the value vectors. Next, the server performs linear transformations and nonlinear activations through feedforward network layers, and performs residual connections and normalization between layers. Following an autoregressive generation strategy, the server calculates the probability distribution of the next identifier at each time step and selects the output identifier according to a predefined sampling algorithm, until the stopping condition is met.

[0180] Output: The output identifier sequence representing the generated sequence, along with the corresponding probability or scoring information.

[0181] Step 9: The server decodes the output identifier sequence and constructs conceptual or design data. The server maps the output identifier sequence into natural language text using a reverse vocabulary and parses it into an internal data structure according to a predefined format.

[0182] Input: Output a sequence of identifiers and a reverse vocabulary.

[0183] The server iterates through the sequence of identifiers, maps each identifier to a corresponding string fragment, and concatenates them into complete text. If the text contains specific delimiters or tags, the server segments the fields accordingly and constructs a conceptual data object or design data object containing multiple attributes.

[0184] Output: Conceptual data or design data structures used to describe the generated results, and the corresponding original generated text.

[0185] Step 10: The server stores the generated results along with metadata into a data storage device. The server assigns a new work identifier to the current generated result and writes the user identifier, prompt statement, generation conditions, generation time, and concept data or design data into the database.

[0186] Input: Conceptual data or design data, user ID, prompt text, model version information, and current timestamp.

[0187] The server constructs a database insert statement or calls a data access interface to map the above information to the works table and metadata table according to predefined fields, and creates indexes on the necessary fields; after completing the write, the server returns the newly generated works identifier.

[0188] Output: Persistently stored work records and metadata records in the database, along with the corresponding work identifiers.

[0189] Step 11: The server constructs product information based on the generated results and generates data for distribution. The server generates a data structure containing brief and detailed information for electronic market presentations, based on stored conceptual or design data.

[0190] Input: artwork identifier, concept data or design data, metadata (prompt statements, model information, generation conditions, etc.).

[0191] The server extracts key fields from conceptual or design data to form a title and summary, and combines price, author information, and metadata into a complete product information object. The server further serializes the product information into structured response data suitable for network transmission for list display or detail display.

[0192] Output: Product information object and corresponding list display data and details display data.

[0193] Step 12: The terminal receives product information from the server and displays the generated results to the user. The terminal receives product information data returned by the server through the communication module and renders it on the interface.

[0194] Input: Structured product information data received from the server.

[0195] The terminal parses the data, mapping titles, summaries, detailed descriptions, etc., to interface components; the terminal calls the rendering engine to draw text and interface elements, displaying the concept or design content on the display device for the user to browse and confirm.

[0196] Output: The generated results interface displayed on the terminal screen, as well as the interface status that the user can interact with.

[0197] Step 13: Users confirm the generated results via the terminal and issue an uploading command. After viewing the generated results, users can select the listing operation through the terminal interface and enter parameters such as price and authorization method.

[0198] Input: User clicks on interface controls and input text or numerical data such as price and authorization type.

[0199] The terminal combines this data with the work identifier to form an upload request data structure, performs basic format verification locally (such as whether the price is positive), and then prepares to send it to the server.

[0200] Output: Packaged upload request data, including artwork identifier, price information, and authorization information.

[0201] Step 14: The terminal sends the racking request to the server and waits for a response. The terminal uses the communication module to encode the listing request into a network message and sends it to the server's listing interface via HTTP or other application protocols.

[0202] Input: The validated upload request data structure and the server upload interface address.

[0203] The terminal serializes and encrypts the data (e.g., using TLS), sends it to the server via the network protocol stack, and maintains a pending request state locally to receive the listing result.

[0204] Output: The rack request message arrives at the server, and the terminal is in a state of waiting for the server's response.

[0205] Step 15: The server processes the upload request and updates the work's status. The server receives and parses the upload request, extracts information such as the work identifier, price parameters, and authorization type, and verifies whether the requester is a legitimate user of the work.

[0206] Input: Structured data parsed from the upload request message and existing work records stored in the database.

[0207] The server queries the database for work records and compares the user identifier in the records with the user identifier in the request. If they match, the work status field is updated to "listed" and the price and licensing method fields are populated. The server uses a transaction mechanism to ensure the atomicity of the update operation.

[0208] Output: Records of works with updated status and response data used to provide feedback on the listing results.

[0209] Step 16: The server generates non-fungible token issuance and transaction data for the artwork and submits it to the distributed ledger. The server constructs token metadata, including the work link, summary hash, and author identifier, based on the work identifier and its metadata, and generates the token issuance transaction data structure.

[0210] Input: Work records, metadata records, and distributed ledger contract parameters.

[0211] The server performs a hash operation on the work's metadata to obtain a fixed-length digest value, and combines the digest with the work's identifier to form a token description; the server assembles the issuance transaction data according to the contract format, and encodes the transaction data into a binary format of a specified protocol through the distributed ledger interaction module, and sends it to the distributed ledger management device.

[0212] Output: Issuance transactions that have been submitted to the distributed ledger network, and transaction identifier information awaiting confirmation.

[0213] Step 17: The server receives the token identifiers returned by the distributed ledger and stores them accordingly. The server receives transaction confirmation information from the distributed ledger management device and parses out the unique identifier of the non-fungible token.

[0214] Input: Transaction receipt data from the distributed ledger.

[0215] The server parses the receipt data, extracts identification information such as token number and contract address, and stores this identification information in the database in association with the corresponding work identifier; the server updates the token field in the work record to make the work correspond one-to-one with the on-chain token.

[0216] Output: Updated records of the work-to-token correspondence in the database, and token identifier data available for subsequent queries.

[0217] Step 18: The server returns the listing and token generation results to the terminal. The server constructs a response object containing the artwork's status, price, token identifier, and notification information, and returns it to the terminal via the network.

[0218] Input: Latest work records and token identification information.

[0219] The server serializes the response object, generates JSON or other structured response text, and sends it to the terminal via the network interface.

[0220] Output: Successful listing notification data arriving at the terminal, including status information that the work has been listed and tokens have been generated.

[0221] Step 19: The terminal updates its interface and displays the listing and token status to users. The terminal receives the response data from the server, parses the fields such as the work status and token identifier, and updates the corresponding display content on the interface.

[0222] Input: Listing results and token identification data received from the server.

[0223] The terminal marks the work as "listed" in the "My Works" or similar interface, and displays information that the non-fungible token has been bound in the details; the terminal also displays a prompt on the display device to inform the user that the work can now be viewed and purchased by other users in the online marketplace.

[0224] Output: Updated user interface status and visual confirmation of successful listing and token binding results.

[0225] 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".

[0226] With the widespread application of generative artificial intelligence models in the creation of digital content such as text and images, a large amount of digital content automatically generated based on prompts has emerged online. However, existing technologies mainly focus on the generation algorithm itself, and systematic support for the entire technical chain surrounding prompt input, model selection, content generation, digital asset ownership confirmation, and cross-user transactions remains insufficient. Specifically, this manifests in the following technical issues: First, in existing content generation platforms, the prompts input from the terminal are usually passed directly to the model in the form of simple strings. There is a lack of a unified formatting and natural language preprocessing mechanism, making it difficult to detect abnormal inputs or non-compliant content in a timely manner. It is also impossible to automatically match appropriate generative artificial intelligence models and generation parameters based on the characteristics of the prompts, resulting in unstable generation quality and low resource utilization efficiency.

[0227] Second, existing server-side processing typically separates "content generation" from "digital asset management." The generated digital content is often stored as ordinary files or records, lacking an automated non-fungible token issuance and binding mechanism tightly integrated with distributed ledgers. This makes it difficult to reliably establish and track ownership information for each piece of generated content at the technical level, and thus hinders large-scale, automated ownership confirmation and cross-platform circulation.

[0228] Third, existing electronic trading systems are mostly designed for traditional commodities or pre-existing digital files, lacking integrated transaction management technology for "digital content dynamically generated and uploaded to the blockchain by generative artificial intelligence models." For example, they do not support automatically generating commodity information based on prompts and generated content features, do not support consistent association between token ownership changes on the distributed ledger and platform transaction records, and lack technical solutions for unified management of transaction fee revenue, ownership change history, and access control.

[0229] Fourth, in terms of user-side access control, traditional systems often rely on simple account verification at the application layer. The corresponding linkage between the server and the distributed ledger is weak, and it is impossible to make fine-grained judgments on access permissions for digital content based on the ownership status of on-chain non-fungible tokens. This can easily lead to coarse-grained access control, complex implementation, or insufficient security, making it difficult to meet the requirements for secure access and compliant distribution when generated content is used as a digital asset.

[0230] Therefore, a new computer system and its server-side processing mechanism are needed to organically combine the structured processing of prompt statements, the call control of generative artificial intelligence models, the automatic association of digital content and non-fungible tokens, and the transaction and access management on electronic trading platforms within the same technical system. This will improve the computer's ability to manage the entire lifecycle of generated content at the system architecture and processing flow levels, and enhance the automation, security, and scalability of the generation and transaction processes.

[0231] 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.

[0232] In this invention, the server includes: means for receiving a prompt statement input by a user from a terminal, and performing formatting and natural language preprocessing on the prompt statement to generate generation request information containing the prompt statement and generation conditions; means for selecting an appropriate generative artificial intelligence model based on the prompt statement and additional generation conditions contained in the generation request information, and inputting the prompt statement into the generative artificial intelligence model to perform inference processing on computing resources, thereby generating corresponding digital content; means for storing the generated digital content, as well as the prompt statement, generative artificial intelligence model information, and user identification information associated with the digital content, as content information in a storage device; and means for executing a non-fungible token management program on a distributed ledger based on the content information, automatically generating digital content on the distributed ledger. The device comprises: an apparatus for issuing corresponding non-fungible tokens and associating the non-fungible tokens with the content information to form ownership information; an apparatus for publishing the digital content as commodity information on an electronic trading platform and providing it to other users for browsing based on the content information and the ownership information; a transaction management apparatus for receiving purchase requests for the commodity information, updating the ownership of the non-fungible tokens on a distributed ledger to complete the ownership transfer process, calculating transaction fees based on the price information corresponding to the ownership transfer process, and storing the transaction fees and the purchase and sale price as transaction records; and an access control apparatus for determining access rights to the digital content based on the ownership information and the transaction records, and providing or allowing the download of the digital content to the corresponding user terminal when access rights are determined to be granted. This allows for the tight integration of prompt-driven generative AI reasoning, non-fungible token ownership confirmation on a distributed ledger, and transaction and access control on an electronic trading platform within the same server-side processing architecture. This enables fully automated management of the entire process of generating digital content, from input, generation, ownership confirmation, transaction to access, thereby improving the security, consistency, and scalability of generated content processing at the computer technology level.

[0233] "Terminal" refers to an electronic device operated by a user for inputting prompts and communicating data with a server, including portable information processing devices, display devices, head-mounted display devices, and other information processing devices with network communication and human-computer interaction functions.

[0234] "User" refers to the entity that uses a terminal to input prompts into the system, triggers a generative artificial intelligence model to generate digital content, and then browses or trades it through an electronic trading platform.

[0235] "Prompt statements" refer to natural language text or instruction messages containing natural language components that users input in the terminal to instruct generative artificial intelligence models to generate specific types of digital content.

[0236] "Generation request information" refers to a structured data set generated by the server based on the prompt statement and its related parameters, used to invoke the generative artificial intelligence model. It includes at least the prompt statement, generation conditions, and user-related identification information.

[0237] "Generation conditions" refers to the set of parameters used to control the generation process and characteristics of the generated results when calling a generative artificial intelligence model. These parameters include model type, output length, style parameters, randomness parameters, and other control parameters related to digital content generation.

[0238] "Generative artificial intelligence models" refer to data processing models trained using machine learning or deep learning techniques that can automatically generate text, images, or other digital content based on input prompts.

[0239] "Inference processing" refers to the process by which a generative artificial intelligence model, based on its trained parameters, calculates and outputs digital content from input prompts and generation conditions. This includes steps such as encoding the input, performing mathematical operations within the model, and decoding the output.

[0240] "Digital content" refers to data results generated by generative artificial intelligence models based on prompts, stored and processed electronically, including text data, image data, audio data, video data, 3D data, and other content that can be presented through electronic devices.

[0241] "Content information" refers to a set of structured information associated with a digital content and stored by a server, including at least the digital content itself or its storage location, the corresponding prompts, the generative artificial intelligence model information used, and user identification information.

[0242] "Storage device" refers to information storage resources used to electronically store content information, transaction records and other related data, including semiconductor memory, magnetic storage media, optical storage media and network-based data storage services.

[0243] A "distributed ledger" refers to a data structure and its operating system that maintains transaction or state records jointly by multiple nodes, reaches consensus through a consensus mechanism, and cannot be arbitrarily tampered with, including blockchain systems and other distributed ledger systems.

[0244] "Nonfungible tokens" are token units issued on a distributed ledger that are unique in their identification and non-fungible, used to identify and prove ownership or associated rights to specific digital content.

[0245] "Nonfungible token management program" refers to the program logic that runs in a distributed ledger environment and is used to create, manage and update nonfungible tokens and their association with content information, including smart contracts and their calling interfaces.

[0246] "Ownership information" refers to a collection of electronic data used to represent the ownership status of digital content, including at least a non-fungible token identifier associated with the digital content, the current rights holder identifier, and metadata related to ownership.

[0247] An “electronic trading platform” refers to an electronic trading system that operates in a network environment to display digital content product information and support transactions between different users, including its front-end interface, back-end services, and components that interact with distributed ledgers.

[0248] "Product information" refers to structured descriptive information related to digital content used for display and trading on electronic trading platforms, including at least digital content identifiers, ownership information, price information, title, description, and media data for display.

[0249] A "purchase request" refers to an instruction and related data sent by a user through a terminal to a server or electronic trading platform to acquire ownership or access rights to certain digital content.

[0250] "Price information" refers to data such as payment amount, payment medium, and settlement method related to digital content transactions, which is used for fee calculation and transaction recording during the transaction management process.

[0251] "Transaction management device" refers to the functional module in a server used to handle operations related to digital content transactions, including programs and hardware resources for receiving purchase requests, updating non-fungible token ownership, calculating transaction fees, and generating and storing transaction records.

[0252] "Transaction record" refers to a set of data that electronically records one or more digital content transaction processes, including at least the transaction time, participant identification, price information, transaction fee information, and information on changes in ownership of non-fungible tokens.

[0253] "Access permissions" refer to the technical restrictions on the scope and manner in which a user has access to certain digital content, including the rights to browse, copy, download, or otherwise use the digital content.

[0254] "Access control device" refers to a functional module in a server used to determine access permissions for digital content based on ownership information and transaction records, and to provide or allow the download of digital content to the user terminal when the authorization conditions are met.

[0255] In the following embodiments, the system of the present invention is based on collaborative processing among servers, terminals, and users. Through the integration of generative artificial intelligence models and distributed ledger technology, it achieves full lifecycle management of digital content generated based on prompt statements. The description of the present invention is specifically illustrated by one or more of the following embodiments, but is not limited to a specific hardware or software platform. Any solution that satisfies the same function and data processing flow can be considered a variation or equivalent of the present invention.

[0256] I. Overall System Structure In one implementation, servers can be deployed on computing nodes in a cloud computing environment, such as general-purpose cloud server instances, including computing devices with multi-core central processing units (CPUs), graphics processing units (GPUs), main memory, and non-volatile memory. Servers can run general-purpose operating systems, such as Unix-like operating systems or other server operating systems, and run backend application frameworks on them, such as Python-based, Java-based, or JavaScript-based application frameworks. Servers can also distribute production and storage modules across multiple nodes using container orchestration systems to improve scalability and fault tolerance.

[0257] In one implementation, the terminal can be a smartphone, tablet computer, desktop computer, laptop computer, head-mounted display device, smart glasses, or other information processing device with network communication and human-computer interaction capabilities. The terminal can run a mobile operating system or a desktop operating system and run native applications or browser applications to achieve interaction with the server.

[0258] In this invention, users interact with the server through a terminal, inputting prompts, browsing digital content, initiating or completing transactions, etc.

[0259] II. Handling Prompt Statements and Constructing Requests In one implementation, the server structures and standardizes the prompts received from the terminal to enable efficient reasoning by subsequent generative artificial intelligence models.

[0260] 1. After receiving the prompt statement sent by the terminal, the server organizes the prompt statement and its additional parameters (such as target content type, target language, expected length, style preference, etc.) into an internal generation request data structure. In one implementation, this data structure can take the form of a set of key-value pairs, containing fields such as: user identifier, original prompt statement, preprocessed prompt statement, generation conditions, timestamp, etc.

[0261] 2. In one implementation, the server uses a natural language processing component to preprocess the prompt statement. The server can use a text processing algorithm based on a combination of dictionary rules and statistical models to perform character normalization, illegal character removal, and whitespace compression; the server can also use a pre-trained word segmentation model to segment and tag the prompt statement for subsequent model selection and parameter setting.

[0262] 3. In another implementation, the server uses a sensitive content detection model to determine the security of prompts. This model can be a shallow neural network or a rule-based and machine learning hybrid decision module. It utilizes a predefined keyword list, regular expression templates, and a trained classifier to calculate risk scores at both the character and word levels. The server determines whether to reject the request or downgrade it based on whether the risk score exceeds a set threshold. By uniformly performing natural language preprocessing and security determination on the server side, the impact of non-compliant input on generative AI models can be reduced, and the waste of resources in invalid reasoning can be minimized, thereby improving overall computational efficiency.

[0263] For example, a user can enter the following prompt statement through the terminal: "Please design an architectural design for a future urban public library and explain the design concept in 300 words." or: "Please generate a composition description for a concept poster based on 'Cyberpunk Nighttime Elevated Road' for subsequent drawing." After receiving the message, the server performs character standardization, length checking, and sensitive word filtering on the above prompt statement to generate structured request information.

[0264] III. Generative Artificial Intelligence Model Structure and Inference Processing In one implementation, the server generates digital content using a generative artificial intelligence model. To avoid the "internal processing of the model" being too abstract, the server can employ a deep neural network model with a clearly defined network structure and training method.

[0265] 1. Model Structure The server employs a generative artificial intelligence model based on a transformer architecture in one implementation. This model comprises multiple stacked encoder-decoder modules or decoder-only modules, each layer including a self-attention sublayer and a feedforward network sublayer. The self-attention sublayer uses a multi-head attention mechanism to map the input sequence into multiple distinct subspace representations, thereby capturing long-range dependencies; the feedforward network sublayer includes at least two layers of linear transformations and non-linear activation functions for non-linear mapping of intermediate features.

[0266] In text generation scenarios, the server can use a decoder-only language model. This model takes word embedding vectors as input and outputs subsequent words step by step through a multi-layer self-attention decoder based on autoregression principles. In image generation scenarios, the server can use a model structure that combines a text encoder and an image decoder. The text encoder is also based on a transformer architecture to encode the prompts. The image decoder generates image features based on the encoded text features and random noise vectors and finally decodes them into a pixel matrix.

[0267] 2. Model Training Methods In one embodiment of this invention, the server employs a training method combining supervised and self-supervised learning for the generative artificial intelligence model. During training, the server uses a large-scale text or image-text pairing dataset and updates the network parameters by maximizing the likelihood of the training data under the model's conditions. The server can use cross-entropy as a loss function to calculate the error between the predicted probability and the actual word label for each word in the generated sequence; the server iteratively updates the network weights using stochastic gradient descent or its variants, such as adaptive moment estimation. The server can also use adversarial loss in the image generation model, combining the generative network with a discriminative network. The discriminative network distinguishes between real and generated images to further improve the generation quality.

[0268] During the training phase, the server can employ data augmentation methods, such as randomly masking some words in text training, or randomly cropping, flipping, or perturbing colors in image training, to improve the model's generalization ability. Through the explicit setting of the aforementioned network structure and training methods, the generative AI model can be guaranteed to have predictable performance and controllable generative behavior during the inference phase.

[0269] 3. Inference Parameters and Judgment Criteria During the inference phase, the server sets the model's inference parameters based on the generation conditions in the generation request information, including maximum output length, sampling temperature, and top-k or top-p sampling thresholds. Temperature controls the smoothness of the output distribution; lower temperatures help improve output stability and repeatability, while higher temperatures increase diversity. The server can dynamically adjust the maximum output length based on the length and complexity of the prompt statements to avoid resource waste or insufficient information caused by excessively long or short outputs.

[0270] In one implementation, the server can also assess the quality of the generated results. The server can use a set of rules, such as detecting whether the generated content contains a large number of repetitive segments, exceeds a set length limit, or contains high-frequency words unrelated to the prompt, or use an auxiliary discriminant model to score the generation quality. When the server determines that the quality of the generated result does not meet the preset standards, it can automatically trigger regeneration or adjust the generation parameters and perform another inference, thereby improving the overall quality of the generated results without increasing user interaction.

[0271] For example, the user inputs a prompt statement: "Please design a brand story and advertising copy for an environmentally friendly smart electric vehicle targeting 2035." After selecting the text generation model, the server sets the maximum output length to a moderate value and the temperature parameter to a medium value, so as to generate brand stories and advertising copy that are both creative and not overly divergent.

[0272] IV. Digital Content Storage and Data Structure Design In one implementation, the server uses a structured data structure to manage the generated results and their associated information to support subsequent non-fungible token generation and transaction management.

[0273] 1. After generating digital content, the server assigns a unique content identifier to that content. This identifier can be a randomly generated globally unique identifier used to uniquely reference the content throughout the system.

[0274] 2. The server maintains content information records in the storage device. Each record may include: content identifier, user identifier, original prompt statement, preprocessed prompt statement, generative artificial intelligence model type and version, generation conditions, generation timestamp, content type (text, image, etc.), storage path or access address, current ownership status, etc. The server can use relational data tables or document data storage structures to store this information.

[0275] 3. When storing digital content, the server can directly save text content as a character sequence in the database, while large files such as images can be stored in an object storage system, with corresponding access links saved in the content information. The server can also generate thumbnails or preview clips suitable for display to reduce network bandwidth consumption when the terminal browses the list.

[0276] Through the above content information structure design, the server can quickly retrieve and manage a large amount of generated content without traversing a large amount of original content, thereby improving query efficiency and data management scalability.

[0277] V. Non-fungible token generation and distributed ledger management In one implementation, the server binds content information to non-fungible tokens on a distributed ledger to technically establish and track ownership of digital content.

[0278] 1. When generating a non-fungible token for digital content, the server reads the content identifier, storage location, digest information, and user identifier from the storage device. The server can generate metadata to identify the content in the distributed ledger environment, including fields such as content digest, storage address hash, and creation time.

[0279] 2. The server interacts with the distributed ledger through a non-fungible token management program. In one implementation, this program can be pre-deployed program logic on the distributed ledger. The server sends a token creation request to this program via a network interface, passing in parameters such as content identifier, user address, and metadata address. Distributed ledger nodes, under a consensus mechanism, write this request into the distributed data structure, generating a unique token identifier.

[0280] 3. After receiving the token identifier and transaction confirmation information from the distributed ledger, the server writes the token identifier and content identifier into a one-to-one correspondence into the storage device, forming ownership information. This ownership information may include the token identifier, the current holder identifier, a list of historical holders, and relevant transaction summaries.

[0281] In this way, the server establishes a two-way mapping between the content information on the server side and the token status on the distributed ledger. It can query the token status by content identifier and query the corresponding content information by token identifier, making the ownership information difficult to tamper with and traceable.

[0282] VI. Electronic Trading Platform and Transaction Management In one implementation, the server manages proprietary digital content as commodities on an electronic trading platform. This involves not only business processes but also how the server-side constructs and maintains an internal state consistent with the distributed ledger.

[0283] 1. When uploading digital content as a product, the server generates product information based on the content information. The server can extract titles and summaries from prompts and generated content using rules or auxiliary models. For example, it can extract key phrases from prompts as titles and select the first few characters or a portion of an image from the generated content as a display summary. The server records fields such as product identifier, content identifier, token identifier, pricing information, currency type, current ownership status, and uploading time in the product information data structure.

[0284] 2. When managing transactions, the server creates an order record internally for each purchase request. The server writes the user identifier, product identifier, expected quantity (generally one in non-fungible token scenarios), and expected price into the order data structure and caches it in the pending payment state. After detecting a successful token transfer transaction on the distributed ledger, the server updates the order status to completed and records the actual transaction price, transaction fee amount, and on-chain transaction identifier.

[0285] 3. In one implementation, the server automatically calculates transaction fees according to a pre-agreed percentage. For example, the server can multiply the transaction price by a preset percentage to obtain the fee amount, and record the platform's revenue and the seller's net revenue separately in the transaction log. The server can aggregate and statistically analyze the transaction fee revenue from multiple transactions for platform operation analysis and resource allocation optimization.

[0286] By maintaining order and transaction records corresponding to token ownership on the distributed ledger within the server, this invention ensures the consistency between the internal accounting state and the on-chain state, avoiding state asynchrony caused by network latency or abnormal situations, and technically improving the stability and reliability of the system in high-concurrency scenarios.

[0287] VII. Access Control and Content Distribution In one implementation, the server performs fine-grained access control based on ownership information and transaction records. When a user requests access to certain digital content, the server first queries the storage device for the token identifier and current ownership status corresponding to the content, and then determines whether to authorize access based on whether the requesting user's identifier matches the ownership information.

[0288] When authorizing access, the server can generate a one-time access link or a temporary download credential for users. This credential is valid for a short period, thus reducing the risk of link abuse. The server can also control the resolution or fragment length of content provided based on access levels; for example, providing only a low-resolution preview to users who haven't purchased the content, while providing the full original content to token holders. This access control method based on on-chain ownership state allows the server to directly utilize the state information provided by the distributed ledger for access decisions at the technical level, thereby reducing additional manual review or registration operations and improving the automation and security of access control.

[0289] VIII. Technical Effects and Improvements in Computer Technology The server, through the combination of the aforementioned modules and data structures, implements an integrated processing framework encompassing prompt statement preprocessing, model selection and inference, content information management, token ownership confirmation, and transaction management. This invention improves upon computer technology itself in the following aspects: 1. By performing natural language preprocessing on the prompt statements and automatically selecting model parameters on the server side, the server avoids a large number of invalid or low-quality model calls, thereby reducing redundant calculations in inference requests, improving the utilization of computing resources (especially GPU resources), and thus improving the overall processing speed.

[0290] 2. By designing a structured data structure for the content information and mapping it one-to-one with the token identifier of the distributed ledger, the server can complete fast retrieval and consistency verification without traversing large files, thereby improving data management efficiency and reducing the load on the storage system and network system.

[0291] 3. By introducing a quality judgment and regeneration mechanism for generated results within the server, the server automates some of the work that was traditionally done manually by users into the joint judgment of the model and rules. Furthermore, by adopting non-simple repetitive generation methods such as probability sampling parameter adjustment, the system can improve the overall quality and stability of the generated results with less interaction, thereby reducing the communication and computation waste caused by multiple rounds of trial and error by users.

[0292] 4. By tightly integrating token ownership confirmation with access control, the server can directly perform access authorization determination based on the ownership status on the blockchain, avoiding the inconsistencies and complexities caused by simply relying on application-layer account checks. This simplifies access control logic, reduces error rates, and improves the security and auditability of access control.

[0293] 5. By using explicit deep neural network structures and optimization algorithms in the training and inference phases, the server can deploy multiple types of generative artificial intelligence models on a unified hardware platform, thereby enabling the system to have high flexibility and scalability when facing different types of prompts and content requirements.

[0294] IX. Optional Implementation Forms and Variations In other implementations, the server can employ different types of generative artificial intelligence models, such as models based on a combination of convolutional neural networks and recurrent neural networks, image generation models based on diffusion processes, or sequence generation models based on autoencoder structures. As long as the server uses the above-described structured processing flow to manage prompts, generation conditions, content information, and token information, it can achieve the same or similar technical effects as this invention.

[0295] In another implementation, the server can extend the transaction management module to integrate multiple settlement methods into a unified billing framework, such as supporting different digital assets, different fee strategies, or different profit-sharing rules; while still maintaining consistency with the distributed ledger through its internal transaction record structure.

[0296] The terminal can employ different human-computer interaction methods in different implementations. For example, it can use a voice input module to transcribe the user's speech into prompts, or use an augmented reality interface to allow the user to select a prompt template and fill in parameters using gestures. As long as the terminal ultimately sends structured prompt data to the server, the server can process it according to the above process.

[0297] In summary, the collaborative work of the server, terminal, and user in the above-described embodiments enables the system of the present invention to not only realize generative artificial intelligence content generation and transactions based on prompt statements, but also bring substantial technical improvements to the computer system in terms of resource utilization, data management, and access control through specific data structures, network model structures, and distributed ledger integration methods.

[0298] use Figure 12 The processing flow is explained.

[0299] Step 1: The user enters a prompt statement using the terminal. Users launch an application or browser page on their terminal, type a prompt in a text input box, and select the content type (such as text or image) and desired effect (such as length or style).

[0300] Input: Natural language prompts entered by the user in the terminal interface and generation condition parameters checked or selected by the user in the interface.

[0301] Output: The text of the prompt statement as displayed internally in the terminal and the set of generation condition parameters.

[0302] The terminal caches each character or word in local memory based on the user's keyboard input or speech transcription results. When the user clicks buttons such as "Generate" or "Submit", the terminal organizes the current input box content and the selected option value into a key-value pair to prepare for subsequent network transmission.

[0303] Step 2: Terminal preprocesses prompts and packages requests The terminal performs preliminary formatting on the prompts entered by the user, such as removing leading and trailing spaces, merging extra line breaks, and limiting the maximum length. It then combines the prompts with user identifiers, device identifiers, timestamps, etc., to generate request data.

[0304] Input: The prompt text output in step 1 and the set of generated conditional parameters.

[0305] Output: Structured request data containing prompts, generation conditions, user ID, device information, and timestamps.

[0306] The terminal uses string processing functions to remove redundant spaces and control characters, truncates prompts exceeding a preset length, and then combines the processed text with information such as the user login token, terminal model, and operating system version into a structured object. This object is then serialized into a format suitable for network transmission.

[0307] Step 3: The terminal sends a generation request to the server via the network. The terminal establishes a secure network connection to the server and sends the structured request data generated in step 2 to the specified interface of the server in the form of network packets.

[0308] Input: The structured request data output from step 2.

[0309] Output: The network request message sent to the server and the session state used to receive the server's response.

[0310] The terminal constructs a data packet containing a request header and a request body through the network communication module. The serialized generated request data is placed in the request body. Then, the network stack is called to send the packet, and a session identifier corresponding to the request is maintained locally for subsequent reception of the results returned by the server.

[0311] Step 4: The server receives and parses the request. The server receives request messages from the terminal at the network interface, parses the requests, extracts the prompts and related parameters, and performs basic validity checks.

[0312] Input: A network request message sent by the terminal.

[0313] Output: An internal request object containing prompts, generation conditions, and user information, or error status information.

[0314] The server reads the message into memory through the network access module, parses the message header to check the protocol version and authentication information, and then parses the message body to deserialize fields such as the prompt statement text, generation conditions, and user identifier. The server checks whether the fields are complete and the format is correct according to predefined rules, and saves the results as an internal data structure. If the verification fails, an error object is generated.

[0315] Step 5: The server performs natural language preprocessing on the prompt statement. The server performs natural language preprocessing on the parsed prompts, including character standardization, word segmentation, length checking, and security determination, to obtain a standardized text representation suitable for the model input.

[0316] Input: The internal request object output from step 4 (which contains the original prompt statement and generation conditions).

[0317] Output: An enhanced request object containing normalized prompts, word segmentation results, language features, and security tags.

[0318] The server uses text processing algorithms to unify characters with different encodings into an internal encoding format and remove invisible or illegal characters; it uses word segmentation algorithms to split the prompt statement into a sequence of words, counts the number of words and compares it with a preset maximum length; it uses keyword matching and a trained classification model to score the security of the text, and if high-risk words are found, it adds a risk marker to the request object or stops subsequent processing.

[0319] Step 6: The server selects a generative artificial intelligence model based on the prompts and generation conditions. Based on the standardized prompt content, length, and generation conditions, the server determines the appropriate generative artificial intelligence model type and specific model instance, including the model structure, size, and computing resources used.

[0320] Input: The enhanced request object output from step 5 (containing normalized prompts and generation conditions).

[0321] Output: The selected generative artificial intelligence model identifier and its corresponding inference configuration parameters.

[0322] The server analyzes the keywords and content type fields in the prompt statement, such as determining whether it is text generation or image generation, and selects the appropriate model version from the model pool based on features such as length and target style. The server also determines inference parameters such as maximum output length, temperature, and sampling strategy, and combines the model identifier with the parameters into an inference configuration object.

[0323] Step 7: The server encodes the prompt statement into a model input vector. The server converts the normalized prompts into vector representations that the model can process, including mapping the text to a sequence of lexical numbers and further converting it into embedded vectors for the neural network to perform numerical calculations.

[0324] Input: The normalized prompt statement output in step 5 and the model identifier output in step 6.

[0325] Output: A sequence of vectors or tensor data structures that can be used as input to generative artificial intelligence models.

[0326] The server maps each word in the prompt statement to an integer index based on the vocabulary of the selected model, forming an integer sequence. Then, by looking up the embedding matrix, it obtains the corresponding floating-point vector for each index and arranges these vectors according to time steps to form a two-dimensional or three-dimensional tensor. If necessary, the server will also add positional encoding vectors and then add them to the embedding vectors element by element to obtain the final model input tensor.

[0327] Step 8: The server performs inference operations on the generative artificial intelligence model. On the configured computing resources (such as GPUs), the server feeds the input vector obtained in step 7 into the generative artificial intelligence model, performs multi-layer matrix multiplication, self-attention calculation and nonlinear transformation, and generates the corresponding digital content representation.

[0328] Input: The model input tensor output from step 7 and the inference configuration parameters output from step 6.

[0329] Output: The output tensor representing the generated result (such as a text word distribution sequence or an image feature tensor).

[0330] The server executes forward propagation operations sequentially according to the neural network structure. In each layer, it performs linear transformations, attention weight calculations, and activation function operations on the input tensor. In the autoregressive generation mode, it iteratively samples output words or updates image features at each time step. Based on the temperature and sampling strategy, the server selects the next word or feature value from the output probability distribution, repeating until a predetermined length is reached or a termination condition is met, ultimately obtaining a complete output tensor sequence.

[0331] Step 9: The server decodes the model output into usable digital content. The server decodes the model's output tensor into human-understandable text or renderable image data, including the conversion of word sequences to strings or feature tensors to pixel matrices.

[0332] Input: The output tensor from step 8.

[0333] Output: Application-layer digital content data (such as text strings or image file data).

[0334] In text scenarios, the server reverse maps the output lexical index sequence to words in the vocabulary, concatenates them into a string in sequence, and performs deduplication and formatting, such as removing extra spaces or repeated sentences. In image scenarios, the server feeds multidimensional feature tensors into the image decoding module or decoding network, maps continuous numerical values ​​to pixel color values, and generates a data buffer in standard image format.

[0335] Step 10: The server generates and stores content information records. The server creates a content identifier for the generated digital content, combines the content itself with prompts, model information, user information, etc., into a content information record, and writes it to a storage device.

[0336] Input: The digital content data output in step 9 and the request-related information retained in steps 4-6.

[0337] Output: A record of content information including content identifier, metadata, and storage location.

[0338] The server generates a unique content identifier, combines this identifier with the user identifier, the original prompt statement, the model type and version used, the generation time and generation conditions, etc., into a record, and writes the record to the database; at the same time, the digital content itself is saved to object storage or database fields, and its access path or index position is recorded in the content information.

[0339] Step 11: Servers create non-fungible tokens for digital content on a distributed ledger. The server constructs token metadata based on the content information, and calls the non-fungible token management program through the distributed ledger interface to issue a corresponding non-fungible token on the chain for the content.

[0340] Input: The output information record of step 10 and the user's address on the chain.

[0341] Output: Non-fungible token identifiers registered on the distributed ledger and on-chain transaction confirmation information.

[0342] The server extracts the content identifier, summary information, and storage address, packages them into metadata description, and submits a token creation request transaction by calling the on-chain program interface over the network. After consensus is reached, the distributed ledger node returns the transaction hash and token identifier, and the server receives the data and updates its local ownership information record.

[0343] Step 12: The server associates the token information with the content information and updates the ownership information. In the local storage system, the server associates the token identifier obtained in step 11 with the corresponding content identifier to form a complete ownership information record.

[0344] Input: Record of the information output from step 10 and the token identifier and transaction information output from step 11.

[0345] Output: A record of ownership information including token identifier, current owner, and on-chain transaction information.

[0346] The server updates the database with new fields for this content, writing the token identifier, initial holder identifier, on-chain transaction hash, and network parameters. It also updates the content status from "not on-chain" to "on-chain" for subsequent access control and transaction processing.

[0347] Step 13: The server generates and publishes product information to the electronic trading platform. Based on content and ownership information, the server automatically generates product information for display and registers this product information in the product list of the electronic trading platform.

[0348] Input: The content information record output in step 10 and the ownership information record output in step 12.

[0349] Output: A data structure for displaying and trading product information.

[0350] The server automatically generates a product title and description based on the prompt and content summary, fills in the initial sales parameters such as price and currency type into the product record, and associates the content identifier with the token identifier; the server writes the product record into the transaction platform database and marks it as "for sale", so that the front end can read and display it through the interface.

[0351] Step 14: The terminal acquires and displays product information for users to browse. The terminal requests a list of products or product details from the server and displays the received product information to the user in a list or detail format on the interface.

[0352] Input: The browsing request sent by the user on the terminal and the product information data returned by the server.

[0353] Output: A list of products or a details page displayed on the terminal interface.

[0354] The terminal sends a query request to the server via the network. After receiving data containing multiple product items, it parses fields such as title, price, and thumbnail links, and draws the corresponding list items or details page on the screen, allowing users to scroll through and select digital content of interest.

[0355] Step 15: The user initiates a purchase request through the terminal. When a user clicks the "Buy" button or a similar button on the product details page on the terminal, confirms the price and related information, the terminal is instructed to send a purchase request to the server and trigger the on-chain transaction process.

[0356] Input: User actions on the terminal interface and displayed product information.

[0357] Output: Purchase request data sent from the terminal to the server and signature information used for on-chain transactions.

[0358] After user confirmation, the terminal collects data such as product identifier, user identifier, and payment parameters, constructs a purchase request object, and, if necessary, invokes the local wallet module to guide the user to complete the transaction signature. Then, it sends the signed transaction summary and purchase request to the server.

[0359] Step 16: The server processes purchase requests and monitors on-chain ownership transfers. After receiving a purchase request, the server creates a pending order internally and monitors the ownership transfer transaction status of the corresponding non-fungible token on the distributed ledger based on the request information.

[0360] Input: Purchase request data sent by the terminal and on-chain transaction signature information.

[0361] Output: Internal records representing order status and on-chain transaction status.

[0362] The server creates new order entries in the transaction management module, records the product identifier, buyer user identifier, and expected price, and searches for or initiates token transfer transactions on the chain based on the signature information. It also periodically queries the distributed ledger nodes for the confirmation status of the transaction and updates the order status with the query results.

[0363] Step 17: The server updates ownership information and records transaction data. After detecting a successful transfer of ownership of the on-chain token, the server updates the local ownership information record and the status of the goods, and calculates and stores the transaction fee and revenue based on the transaction amount.

[0364] Input: On-chain transaction confirmation information and order records output from step 16.

[0365] Output: Updated ownership information record, product status, and transaction record.

[0366] The server reads the new token holder address and actual transaction amount from the on-chain transaction data, updates the current holder identifier in the ownership information to the buyer, and updates the product status to "sold". The server performs a multiplication operation on the transaction amount according to a preset ratio to obtain the transaction fee amount, writes the transaction fee and the seller's net income into the transaction record table, and saves it in association with the order record.

[0367] Step 18: The server determines access permissions based on ownership information and returns access credentials. After the transaction is completed, the server determines that the buyer has full access to the digital content based on the updated ownership information and transaction records, and generates credentials or links for downloading or online access and returns them to the terminal.

[0368] Input: Ownership information record and buyer access request output from step 17.

[0369] Output: Access response data containing access links or download authorization information.

[0370] The server compares the user identifier in the access request with the current holder identifier in the ownership information. If they match, it generates an access link or temporary token with validity period and signature parameters, packages the link or token into the response data and returns it to the terminal, and records the access authorization behavior for auditing purposes.

[0371] Step 19: The terminal uses access credentials to obtain and display digital content. After receiving the access credentials returned by the server, the terminal downloads digital content from the server or storage system based on the credentials and displays it locally through the corresponding application module.

[0372] Input: The access link or download authorization information output in step 18 and the terminal's existing network connection.

[0373] Output: A digital content interface presented on a terminal device or a digital content file stored locally.

[0374] The terminal parses the access link or token, initiates a download request to the specified address over the network, receives the data stream of digital content, writes it to local storage or loads it directly into the display component, and renders text, images or other forms of content on the screen, allowing users to view, read or use the purchased digital content.

[0375] 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.

[0376] 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."

[0377] With the widespread application of generative AI models in content creation and product design, users can automatically generate ideas or designs by inputting prompts. However, in existing technologies, generative models are typically loosely coupled with e-commerce platforms. (1) The server lacks a unified input processing mechanism for generative artificial intelligence models, and cannot perform structured parsing and parameter extraction of user prompts, which makes it difficult to control the quality, format and tradability of the generated results, which is not conducive to direct commercialization in the electronic market. (2) When the server manages the generated results as digital goods, it often stores them only in the form of ordinary files or records, without binding them in an integrated manner with the non-fungible tokens in the distributed ledger. The uniqueness of the goods, ownership and transaction traceability are insufficient, which can easily lead to problems such as plagiarism, tampering and unclear ownership. (3) When the server matches transactions and processes payments, it usually separates the transaction process from the management of generated content and the change of token ownership. It lacks a mechanism to link and update transactions, non-fungible tokens and generated content on the same information processing platform, resulting in a complex system architecture, difficulty in ensuring data consistency, and easy generation of ownership asynchrony and security risks. (4) In terms of user evaluation and credit management, most servers only record the evaluation score in a simple way, and do not manage the transaction fee data, evaluation data and credit data in a unified manner. They cannot conduct refined trust management and fee setting for users based on complete transaction history and income information, which limits the risk control capabilities and incentive mechanism design of the electronic market. (5) From the perspective of computer technology, existing systems often disperse natural language processing, generative artificial intelligence calls, e-commerce processing, and distributed ledger operations in different subsystems, lacking unified data flow and control flow management. This results in multiple redundant data conversions and network calls during the processes of generation, listing, transaction, settlement, and evaluation, which reduces the overall processing efficiency and scalability.

[0378] Therefore, an improved computer implementation is needed to enable servers to: perform structured parsing and parameterized processing of users' natural language prompts; collaboratively invoke generative artificial intelligence models within the same information processing platform to generate product information, mint and manage non-fungible tokens; automatically and synchronously update token ownership during transactions and settlements; and uniformly manage user credit and display strategies based on transaction records, fee information, and evaluation information. This would improve system architecture and data processing flow while enhancing the commodification of generated content, transaction security, and operational efficiency in the electronic marketplace.

[0379] 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.

[0380] In this invention, the server includes: a device for receiving prompt statements input by a user via a terminal and acquiring the prompt statements as natural language; a device for generating input data for a generative artificial intelligence model based on the prompt statements, providing the input data to the generative artificial intelligence model and acquiring generation results related to the concept or design; a device for generating product information based on the generation results, registering the product information and the generation results in a storage device and publishing the product information on an information processing platform for e-commerce; a device for generating non-fungible tokens corresponding to the product information in a distributed ledger and storing and managing the non-fungible tokens corresponding to the product information; a device for receiving purchase requests from a terminal, generating transaction information related to the product information, performing payment processing using a settlement information processing platform and updating the non-fungible token holders based on the payment processing results; and a device for acquiring evaluation information from the terminal after the transaction is completed, accumulating the evaluation information to update the credit information for each user, and controlling the display or retrieval results in the information processing platform for e-commerce based on the credit information. This allows for the integrated processing of natural language prompt parsing, generative artificial intelligence model invocation, structured registration of product data, non-fungible token management on a distributed ledger, and transaction settlement and reputation control within the same server-side computing architecture. This reduces data conversion and communication overhead between heterogeneous systems, improves the automation and data consistency of the entire process from content creation and listing to trading and evaluation, and thus enhances the efficiency, security, and scalability of digital product management and transaction processing in electronic marketplaces at the computer technology level.

[0381] A "system" refers to a computer implementation consisting of one or more servers, one or more terminals, and the communication network between them, used to perform functions such as prompt statement processing, generative artificial intelligence model invocation, product management, distributed ledger operations, and transaction and evaluation processing.

[0382] A "server" refers to an information processing device or cluster of devices equipped with a processor and memory, and running program code for receiving prompts, calling generative artificial intelligence models, generating product information, managing non-fungible tokens, and processing transaction and evaluation information.

[0383] "Terminal" refers to a client-side information processing device operated by a user and interacting with a server via a network, including but not limited to mobile terminals, fixed terminals, or other computing devices capable of running applications or browsers.

[0384] "User" refers to the entity that uses the terminal to input prompts into the system, generate ideas or designs, list products, purchase products, and provide evaluations. It can be a natural person or other legally qualified entity.

[0385] "Prompt statements" refer to text data input by users in natural language form, used to instruct generative artificial intelligence models to generate corresponding ideas or design content.

[0386] "Generative artificial intelligence models" refer to artificial intelligence models trained based on machine learning or deep learning techniques that can automatically generate text, images, or other data content related to ideas or designs based on input prompts.

[0387] "Input data" refers to structured or semi-structured data generated by the server based on prompts and provided to the generative artificial intelligence model to control the generation process, including but not limited to information such as the generation target category, generation parameters, and constraints.

[0388] "Idea or design" refers to creative data generated by a generative artificial intelligence model based on prompts, which can be used as product content, including but not limited to digital information such as product ideas, graphic designs, text schemes, and multimedia content.

[0389] "Generated results" refers to the specific data content related to the concept or design output by the generative artificial intelligence model after receiving input data, including data that can be used directly or after minimal processing as product content.

[0390] "Product information" refers to a set of data used to represent and manage a product in an e-commerce environment, including but not limited to product identifier, name, description, price, user identifier, associated generated result identifier, and display resource identifier.

[0391] "Storage device" refers to data storage resources used to persistently store data such as generation results, product information, transaction information, evaluation information, and reputation information, including database systems, file storage systems, or other non-temporary data storage media.

[0392] "Information processing platform for e-commerce" refers to an online transaction processing environment implemented by servers, used to display product information, receive transaction-related requests, provide search and browsing functions, and interact with settlement platforms and distributed ledgers.

[0393] A "distributed ledger" refers to a decentralized data storage infrastructure that is jointly maintained by multiple nodes and records and updates data through a consensus mechanism, including but not limited to ledger systems built on blockchain technology.

[0394] "Non-fungible tokens" refer to data records generated and registered on a distributed ledger that have unique identifiers and are not interchangeable. They are used to uniquely identify and prove the correspondence and ownership of specific product information or generation results.

[0395] A "purchase request" refers to request data sent by a terminal to a server to instruct the purchase of specific product information, including product identifier, buyer identifier, quantity, and payment method.

[0396] "Transaction information" refers to the set of transaction-related data generated and managed on the server to facilitate the buying and selling of specific goods, including transaction identifiers, product identifiers, buyer and seller identifiers, prices, time, and status.

[0397] "Settlement information processing platform" refers to a computer system or service used to perform payment processing, including but not limited to payment gateway systems, electronic payment services, or clearing and settlement systems.

[0398] "Payment processing" refers to the process of deducting funds from transaction amounts, transferring funds, and confirming payment results through a settlement information processing platform.

[0399] "Holder" refers to an entity recorded in the distributed ledger as the current owner of a non-fungible token, including but not limited to user accounts, addresses, or other identifiable entities.

[0400] "Review information" refers to the feedback data submitted by users regarding the traded object or product after the transaction is completed, including rating values, text comments, and time information.

[0401] "Credit information" refers to an indicator or set of data calculated or determined based on accumulated evaluation information, transaction records and other relevant data, used to reflect the credit status of each user in the system.

[0402] "Control of display or search results" refers to the server's adjustment and optimization of the product ranking, filtering, highlighting methods, or search result content presented on information processing platforms used for e-commerce, based on reputation information.

[0403] "Transaction fee information" refers to data calculated by the server based on completed transactions, representing the amount of fees charged by the platform for those transactions and related parameters.

[0404] "Revenue management data" refers to structured data used to record and manage the platform's revenue from transaction fees, etc., including revenue amount, source transaction identifier, time, and statistical classification information.

[0405] In an embodiment of this invention, the server is deployed in a cloud computing environment. The server includes a processor, memory, network interface, database system, file storage system, and distributed ledger access module. The server can use general-purpose computing hardware, such as multi-core central processing units and accelerated nodes with graphics processing units. The server runs application server software and middleware such as database management systems on top of the operating system to perform various data processing and computations required by this invention.

[0406] In one implementation, the server uses a relational database management system as storage to store structured data such as user information, prompts, generated results, product information, transaction information, evaluation information, and reputation information. The server further uses object storage services as file storage to store large object data such as images and documents, and associates these objects with database records using Uniform Resource Locators (URLs). The server communicates with the blockchain network through a distributed ledger access module to generate and update non-fungible tokens.

[0407] In one implementation, the server uses a generative artificial intelligence (AI) model as its content generation engine. In a preferred embodiment, this model is a deep neural network based on a self-attention mechanism. Its network structure includes multi-layer encoder and decoder modules, each with multi-head self-attention sublayers and feedforward neural network sublayers. The generative AI model is pre-trained on a large amount of labeled or unlabeled corpus and can be adapted to conceived or designed generation tasks in the e-marketplace through supervised or instructional fine-tuning. The server deploys the generative AI model on inference nodes equipped with graphics processing units (GPUs) to perform high-dimensional matrix multiplication and nonlinear transformation operations.

[0408] In one implementation, the server performs natural language processing on user prompts to convert unstructured text into structured input data that the model can directly use. The server can use techniques such as word segmentation, part-of-speech tagging, syntactic analysis, and named entity recognition to parse the prompts into word sequences and corresponding linguistic labels. The server further extracts the target category (e.g., "text idea," "image design," "logo design"), generation parameters (e.g., style, length, resolution), and usage conditions (e.g., commercial use, non-exclusive use) from the prompts using sequence labeling or classification models. This parsing process enables the server to automatically construct input vectors containing multi-dimensional features, which serve as conditional inputs to generative AI models, thereby improving the relevance and structure of the generated results.

[0409] In one implementation, the server maps the parsed prompts into a continuous vector sequence through an embedding layer, and then inputs these vectors into a generative AI model. Internally, the generative AI model calculates the correlation between word vectors at different positions using a multi-head self-attention mechanism and fuses contextual information using a weighted summation method. During the inference phase, the server controls the diversity and determinism of the generated results by setting temperature parameters, truncation lengths, and sampling strategies (such as greedy search, bundle search, or kernel sampling). In text ideation generation tasks, the server outputs a serialized natural language description; in image design tasks, it outputs a latent spatial representation or image pixel matrix, encoding it into a standard image format for subsequent processing.

[0410] In one implementation, the server performs error backpropagation and parameter updates during the training process of a generative artificial intelligence model. During the training phase, the server uses prompts and their corresponding expected outputs as training samples, measuring the deviation between the model output and the expected output using a cross-entropy loss function or other appropriate objective function. The server updates the model weights using gradient descent and its variants, and can combine techniques such as regularization, learning rate scheduling, and gradient clipping to improve the stability and generalization ability of the model training. In scenarios with insufficient training data, the server can employ data augmentation techniques on the prompts and target content, such as synonym substitution, sentence transformation, or image geometric transformation, to improve the robustness of the model.

[0411] In one implementation, the server structurally binds the generated results with product information. The server assigns a unique identifier to each generated result and records its association with user ID, generation time, generation parameters, and parsing results of prompt statements in the database. When generating product information, the server automatically fills in some fields based on the type and content of the generated result; for example, it automatically generates a brief description based on the conceptual text and tags based on image features, reducing the burden of manual input for users and improving data consistency. Through this structured data management approach, the server enables the retrieval and display of product information to leverage index structures and query optimization, improving data access efficiency.

[0412] In one implementation, the server utilizes a distributed ledger to generate and manage non-fungible tokens for product information. The server calls a smart contract interface through the distributed ledger access module, submitting a transaction containing product identifiers, metadata hashes, and associated information. After consensus is reached on the blockchain network, the server receives the transaction hash and token identifier and writes this on-chain data into the database. In this way, the server ensures that each product has a unique and immutable identifier on the blockchain, thereby guaranteeing product uniqueness and ownership. When processing token ownership changes, the server executes the smart contract's transfer function to transfer the token from the current address to a new address and updates the corresponding record in the local database, maintaining consistency between on-chain and off-chain data.

[0413] In one implementation, the server centrally manages transaction information and payment processing. Upon receiving a purchase request from a terminal, the server generates a transaction record in its database and invokes the settlement information processing platform to execute the payment process. The server sends the order amount, currency type, and payment authentication information to the settlement platform via an interface. The settlement platform then performs fund deduction and risk control calculations within its internal system. Upon receiving a success or failure result from the settlement platform, the server updates the transaction record status and, in the case of success, triggers an ownership update process for non-fungible tokens. By serializing transaction record updates, payment processing, and token transfers within the same data stream, the server can reduce the number of cross-system synchronizations, lower network communication latency, and reduce errors caused by data inconsistencies between different systems.

[0414] In one implementation, the server calculates and stores evaluation and reputation information. After obtaining user rating scores and text reviews, the server writes them to the evaluation record table and triggers reputation update logic. The server updates the reputation of a single user or product using aggregation algorithms, such as weighted average, time decay models, or Bayesian inference-based rating models. The server can assign higher weights to recent transactions and robustly handle abnormal scores to reduce the impact of malicious ratings on the overall reputation. The server stores the calculated reputation values ​​in a user table or a product table, using them as one of the features during retrieval and ranking to optimize the display order and recommendation results in the information processing platform.

[0415] In one embodiment, the terminal is a mobile computing device or desktop terminal, running a client program or web browser. The terminal interacts with the server via a secure communication protocol. The terminal is responsible for collecting user input prompts, product information, payment information, and review information, and sending this data to the server in a structured format. After receiving the generated results, product list, and reputation information from the server, the terminal visualizes them in a graphical user interface. The terminal can perform partial input validation and front-end caching locally to reduce the load on the server from invalid requests.

[0416] In one implementation, users interact with the system via a terminal. Users can input prompts to trigger a generative artificial intelligence model to generate ideas or designs. Users can also create products based on the generated results and list them on an online marketplace for other users to purchase. After completing a transaction, users can rate the traded object and the product, and the system updates its reputation accordingly.

[0417] In a specific example, the user can enter the following prompt in the terminal: "Based on the theme of 'futuristic streetwear T-shirts,' please generate 5 creative descriptions of T-shirt designs suitable for sale on the online marketplace." "Please generate a detailed design description for a digital poster themed 'Cyberpunk City Night View', to be used as product description text in the market." "Please generate a logo design description suitable for commercial sales, based on the following keywords: 'minimalist style, blue tones, tech company logo'." "Please explain the complete steps a user takes to register their designed item as an NFT and list it for sale in this online marketplace." "Please explain how the system sends notifications to the buyer and seller after a transaction is successfully completed in the market, and briefly provide an example of the message content." After receiving the aforementioned prompts, the server uses a natural language processing module to identify keywords such as "topic," "style," and "purpose," encoding them into multiple numerical feature dimensions and inputting them into the generative AI model. This feature processing is not simply transcribing the user's text; rather, through specific feature extraction rules and model structure, the generative AI model can adjust its attention distribution across different dimensions during internal decision-making, thereby improving the controllability and relevance of the generated results at the computational level. Compared to manually writing explanatory text line by line, this process not only increases the generation speed but also enhances the diversity and consistency of the generated results because the model can integrate multiple features for optimized searching in a high-dimensional space.

[0418] In one implementation, the server optimizes end-to-end data flow through specific data structures and module partitioning. For example, the server designs structured records for the parsing results of prompt statements, the generated results, product information, and token information, allowing each step of data processing to be accessed in memory with fixed field offsets, thereby reducing conversion overhead and serialization counts. When calling a generative artificial intelligence model, the server merges multiple prompt statements into batch inputs to fully utilize the parallel computing capabilities of the graphics processing unit and reduce the average inference latency per request. During product retrieval and sorting, the server uses pre-built multi-column indexes and caching strategies to cache frequently accessed product lists, thereby reducing database disk accesses and improving overall response speed.

[0419] In one implementation, the server integrates generative AI model invocation, commodity data management, distributed ledger interaction, and settlement platform integration within the same system, reducing the number of network round trips caused by multi-system cascading in traditional architectures. The server can simultaneously update database records and trigger token transfer requests within a single transaction. This design reduces the probability of inconsistencies in intermediate states and minimizes the compensation computations required to restore consistency. Since the server can maintain the mapping between on-chain and off-chain data locally, querying commodity ownership and transaction history only requires accessing the local database, significantly reducing the number of real-time queries to distributed ledger nodes and thus lowering the distributed communication load.

[0420] In one implementation, the server employs specific rules and unconventional steps to coordinate result generation and transaction processing. Instead of simply fixing the generated result immediately after creation, the server performs secondary filtering based on parameters parsed from prompts and user-defined usage conditions. This includes sensitive content filtering, format constraint checks, and duplicate detection. The server can compare newly generated results with existing results using local sensitive hashing or vector similarity calculations to avoid listing highly duplicated content. This automated filtering completes risk control before products are publicly disclosed on the electronic marketplace, thus technically reducing the burden of subsequent disputes and deletion operations.

[0421] In another implementation, the server can choose different types of generative AI model architectures. For example, the server can employ a multimodal model combining convolutional neural networks and self-attention mechanisms to simultaneously process text prompts and reference images. This allows for full utilization of visual features extracted from the images and semantic features from the prompts during image generation design. In this architecture, the server performs feature extraction on the text and images separately, and then performs cross-modal attention computation in a unified multimodal fusion layer, thereby generating design results that better meet user needs. Under this architecture, the server can choose to use only a text model or use a multimodal model according to task requirements, achieving system scalability.

[0422] In another implementation, the server can use credit information to dynamically adjust transaction fees. Based on a user's historical transaction volume, complaint records, and average rating, the server can adjust fee rates using preset rules or learned functions. In this process, the server not only updates revenue management data but also incentivizes high-credit users and restricts potentially risky users by automatically adjusting fees. This data-driven dynamic adjustment method binds credit calculation and charging strategies through specific algorithms, achieving technical linkage between internal computer state adjustments and external business parameter settings, resulting in improved overall system efficiency and security.

[0423] Through the aforementioned implementations, the server, terminal, and user together constitute a complete technical chain, encompassing prompt message parsing, generative artificial intelligence model inference, structured commodity management, non-fungible token minting and transfer, transaction settlement, and reputation calculation. Because each stage is optimized through specific data structures, model structures, algorithmic processes, and module divisions, the system of this invention offers substantial improvements over traditional systems in processing speed, data consistency, ownership management, and risk control. This demonstrates an improvement in computer technology itself, rather than simply automating manual business processes.

[0424] use Figure 13 The processing flow is explained.

[0425] Step 1: Users input prompts via the terminal and send them to the server.

[0426] Users type natural language text as prompts in the terminal's input interface, such as "Please generate 5 creative descriptions of T-shirt designs suitable for sale in the online market, based on the theme 'Futuristic Streetwear T-shirt'," and then click send.

[0427] The terminal takes the prompt text and user identifier as input, encapsulates them into request data, and sends it to the server through an encrypted communication channel.

[0428] The terminal performs basic validation on the input locally (such as length and whether it is empty), and then outputs the valid text directly as string data to the server for subsequent natural language parsing and processing.

[0429] Step 2: The server parses the prompt statements and generates input data for the generative artificial intelligence model.

[0430] The server takes the prompt text sent by the terminal and the user identifier as input. First, it uses word segmentation and part-of-speech tagging algorithms to decompose the prompt text to obtain word sequences and part-of-speech tags. Then, it extracts sentence structure and dependency relations through a syntactic analysis module, and then identifies semantic slots such as "generated target category" (e.g., "text conception" or "image design"), "style features" (e.g., "futuristic" or "street style"), and "purpose" (e.g., "electronic market sales") through a classification or sequence labeling model.

[0431] The server processes the data based on these parsing results: it maps word sequences to embedding vectors, encodes extracted categories and parameters into discrete feature vectors, and concatenates the two into a unified model input tensor.

[0432] The server outputs the tensor along with generation parameters (such as maximum generation length, temperature, and sampling strategy) to form structured input data for generative artificial intelligence models to perform inference.

[0433] Step 3: The server calls a generative artificial intelligence model to generate the intended or designed result.

[0434] The server takes the input tensor and generation parameters output from step 2 as input and feeds them into a generative artificial intelligence model deployed on a graphics processing unit.

[0435] Generative artificial intelligence models internally perform multi-layer matrix multiplication, self-attention weight calculation, and feedforward network nonlinear transformation to iteratively update input features and gradually generate target sequences (text descriptions) or target matrices (image latent representations).

[0436] The server performs post-processing on the model's raw output: for text, the server restores the token sequence into a natural language sentence based on the word segmentation boundaries, and performs deduplication, sensitive word filtering, and formatting; for images, the server decodes the latent representation into a pixel matrix, and then encodes it into a standard image format file.

[0437] The server outputs the standardized generated text, image file data, and associated prompt message parsing information to the subsequent product information generation module.

[0438] Step 4: The server generates product information and registers it in the storage device.

[0439] The server constructs product records using the generated results (text descriptions and / or image files), user identifiers, and parsed category and usage information as input.

[0440] The server extracts the core content from the generated results into the product name using predefined rules, uses the detailed description as the product description, uses the parsed purpose and style as the tag field, and determines the initial price or suggested price according to the system policy.

[0441] The server creates a unique identifier for each product in the database, performs an insert operation, and writes fields such as product name, description, tags, price, user identifier, generated result identifier, and file storage address into the product table. For image files, the server uploads the file to the object storage service, obtains the file access address, and writes the address into the product record.

[0442] The server will output the product records (including product identifiers and their metadata) that have been successfully written to the database, which will be used for subsequent display on the e-commerce information processing platform and for binding non-fungible tokens.

[0443] Step 5: The server generates non-fungible tokens corresponding to the product information and completes on-chain registration.

[0444] The server uses the product identifier, product metadata (such as name, description, file hash) output from step 4, and the user's on-chain account address as input to construct token minting transaction data for the distributed ledger.

[0445] The server performs a hash operation on the product metadata to obtain a fixed-length hash value, which is used to record the product content fingerprint on the blockchain. Then, the server calls the smart contract interface through the distributed ledger access module to submit the product identifier, metadata hash, and owner address as transaction parameters to the blockchain network.

[0446] The distributed ledger executes smart contract logic on each node, updates token state, generates transaction records, and returns transaction hashes and token identifiers.

[0447] The server receives the transaction hash and token identifier, writes them together with the product identifier into the token table of the local database, forming a mapping relationship between on-chain tokens and off-chain product records, and provides this mapping relationship as output to the transaction management module for subsequent ownership changes and verification.

[0448] Step 6: The terminal displays product information and receives purchase requests.

[0449] The terminal takes the product list data returned by the server (including product name, preview image address, price, reputation, etc.) as input and renders product cards and detail pages in the graphical user interface.

[0450] When a user browses products on the terminal, the terminal retrieves the next batch of product data from the local cache or server based on the user's scrolling and clicking actions, and performs sorting or filtering display locally.

[0451] When a user selects a product on the terminal and clicks "Buy," the terminal encapsulates the product identifier, user identifier, and the quantity and payment method selected by the user as input into a purchase request. For scenarios requiring online payment, the terminal calls the payment component to convert the original payment information (such as a bank card number) into a payment token to avoid directly exposing sensitive data.

[0452] The terminal sends request data, including product identifier, quantity, payment method, and payment token, as output to the server's transaction processing interface.

[0453] Step 7: The server generates transaction information and executes payment processing.

[0454] The server takes the purchase request data sent by the terminal (including product identifier, user identifier, quantity and payment token) as input, first queries the database for the corresponding product record, verifies whether its status is available for sale, and calculates the total amount (unit price multiplied by quantity plus possible handling fees or taxes).

[0455] The server creates a transaction record in the transaction table, assigns a transaction identifier, and writes the buyer's identifier, seller's identifier, product identifier, amount, and initial state to the database. Then, using the transaction amount, currency type, and payment token as input, the server calls the settlement information processing platform interface to send the payment request to the payment system.

[0456] After performing risk control and deduction calculations internally, the payment system returns the payment result and transaction serial number. The server uses the payment result and transaction serial number as input to update the status of the local transaction record: if the payment is successful, it is marked as "paid"; if it fails, it is marked as "payment failed".

[0457] When the payment is successful, the server outputs the updated transaction record and the account information of the buyer and seller to the token ownership change module to trigger the on-chain transfer operation.

[0458] Step 8: The server updates the holders of non-fungible tokens and synchronizes on-chain and off-chain data.

[0459] The server takes transaction records (including product identifiers, buyer addresses, and seller addresses) and token mapping relationships (the correspondence between product identifiers and token identifiers) as input to determine the tokens to be transferred and their current holders.

[0460] The server calls the smart contract's transfer function through the distributed ledger access module, submitting a transfer transaction request to the blockchain network with the token identifier, the current holder's address, and the new holder's address as parameters.

[0461] The distributed ledger executes token transfer logic on network nodes, updates the on-chain token owner field, and returns the transfer transaction hash. Upon receiving this transaction hash, the server updates the current holder field of the token record in its local database and simultaneously writes the transfer transaction hash into the transaction record.

[0462] The server outputs the updated token records and transaction records to generate notification messages and update the front-end display, thereby technically ensuring the consistency between the on-chain ownership state and the off-chain database state.

[0463] Step 9: The server generates a transaction notification and sends it to the terminal.

[0464] The server uses the updated transaction records, product information, and token records as input to construct notification content for both buyers and sellers, including product name, transaction amount, transaction time, current ownership status, and download or access links.

[0465] The server encapsulates the notification content into message data or email body through the message push module or email sending module, attaches necessary identifiers (such as order number, token number), and calls the corresponding transmission service.

[0466] The push notification service delivers notifications to both the buyer's and seller's terminals, while the email service sends emails to the email addresses pre-registered by both parties.

[0467] The terminal uses the received notification or email content as output to display transaction completion information on the local interface, making it easy for users to confirm the transaction results and access the generated content.

[0468] Step 10: Users submit evaluation information through the terminal and trigger a reputation update.

[0469] Users fill in their rating (e.g., 1-5 stars) and text comments on the order details page of the terminal and submit them.

[0470] After performing basic checks on the range of rating values ​​and the length of the text, the terminal encapsulates the order identifier, rating value, comment text, and user identifier into evaluation submission data.

[0471] The terminal sends the evaluation submission data as output to the server's evaluation processing interface through a secure communication channel.

[0472] The server takes the submitted review data as input, verifies whether the order status allows for a review, inserts a new record in the review table, records the rating, comment, and timestamp, and triggers the reputation calculation module.

[0473] Step 11: The server calculates and updates the reputation information of users or products.

[0474] The server takes newly added rating records, historical rating records, and transaction records as input and calls the reputation calculation algorithm module. This module first aggregates historical ratings, calculates the average score and the number of ratings, and can use a time decay weighting function to give more weight to recent ratings than older ratings; the server can also weight ratings based on transaction amount or transaction frequency.

[0475] The server uses this data to calculate updated reputation metrics, such as overall score, confidence interval, or risk level, and can use smoothing techniques (such as Bayesian averaging) to reduce the impact of extreme scores.

[0476] The server writes the new reputation metrics into the user table or product table, replacing the old values, and updates the relevant index structure so that they can be used efficiently in subsequent retrieval and sorting operations.

[0477] The server outputs the updated reputation information to adjust the product sorting strategy and recommendation results in the e-commerce information processing platform, thereby improving the overall information display quality and risk control capabilities of the system through a data-driven approach.

[0478] 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".

[0479] With the widespread application of generative AI models in content creation, design generation, and other fields, prompt-driven generation results are gradually being commercialized and traded in the online environment. However, existing technologies suffer from the following problems: First, servers typically only forward user-input prompts to the generative AI model, lacking structured parsing and emotional contextual understanding of the prompts themselves. This results in the generated results failing to match the user's true intentions and emotional needs in a timely manner, leading to insufficient generation quality and personalization. Second, the registration and display process of generated results on electronic marketplaces or information exchange platforms is disconnected from the generation process. Servers cannot automatically manage the entire process from "prompts → generation → commercialization → listing → transaction → digital ownership establishment" within a unified software architecture. This results in multiple data format conversions and repetitive network interactions, increasing system resource consumption and reducing processing efficiency. Third, when processing transactions between users, servers generally only perform static order recording and amount settlement, lacking a mechanism for unified modeling of transaction data at the protocol level and automatic calculation and recording of platform fees. This makes it difficult to aggregate, analyze, and audit large amounts of transaction data in a timely manner, affecting the system's scalability and reliability. Fourth, existing solutions for linking non-fungible tokens (NFTs) often simply map generated content to identifiers on the blockchain without unifying the modeling and management of transaction data, product information, and NFT identification information on the server side. This makes it difficult to guarantee the consistency and traceability of digital asset ownership, circulation information, and the generation process. Fifth, most existing recommendation systems and pricing mechanisms are based on static attributes or historical behavior, lacking automatic analysis and utilization of users' real-time emotional states. Servers cannot incorporate emotional features into generation control, product recommendation, and price adjustment logic in the underlying data processing flow, resulting in the system's inability to dynamically adapt to user states at the algorithm level, thus limiting recommendation accuracy and transaction conversion rates. Sixth, existing systems rarely support the server automatically constructing new prompts based on user emotional states and existing prompts, and driving generative AI models to automatically generate new tradable content and list it. This fails to form a closed-loop computation process of "emotion-driven—prompt generation—content production—automatic listing" within the system.

[0480] Therefore, a new system architecture and data processing solution is needed to enable the server to: perform structured parsing of prompt statements within the same technical framework; dynamically construct generation parameters based on user emotions and attributes; automatically commercialize the generated results and register them on electronic markets or information exchange platforms; uniformly manage transaction data, fees, and non-fungible tokens; and improve the computer technology of generative AI content trading platforms through emotion-driven recommendation and price adjustment mechanisms. This will enhance the quality and personalization of the generated content while improving the overall processing efficiency, scalability, security, and maintainability of the system.

[0481] 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.

[0482] In this invention, the server includes: a device for receiving prompt statements input by a user via an information processing terminal; a device for parsing the prompt statements based on the prompt statements and user attribute information and / or user emotion information, and converting them into structured instruction data for use by a generative artificial intelligence model; a device for inputting the structured instruction data into the generative artificial intelligence model to generate abstract concept information and / or setting information; a device for registering the generated abstract concept information and / or setting information as product information and publishing it in an electronic market and / or information exchange platform; a device for receiving search conditions and / or recommendation conditions and filtering and outputting multiple product information items; and a device for receiving... The system includes: a device for processing purchase requests and coordinating with a payment processing device to execute transaction processing and calculate and record transaction fees; a device for generating non-fungible digital identification information based on transaction data and registering it as a non-fungible token in collaboration with a distributed ledger management device, and associating it with the buyer's digital asset management device; a device for parsing emotion-related data and / or behavioral history data to infer the user's emotional state and dynamically adjusting product recommendations and / or price information based on that emotional state; and a device for automatically generating new prompts based on the user's emotional state and existing prompts, driving a generative artificial intelligence model to add abstract concept information and / or setting information, and automatically publishing it on electronic marketplaces and / or information exchange platforms. This allows for a unified data processing flow on the server side, encompassing prompt parsing, generative artificial intelligence invocation, content commodification, digital ownership establishment, and emotion-driven recommendations and pricing. By reducing cross-system data copying and format conversion, decreasing the number of network interactions, and introducing dynamic control logic based on the emotional dimension, the system comprehensively improves the processing efficiency, resource utilization, personalization capabilities, and security traceability of the generative artificial intelligence content trading platform from the perspectives of computer system architecture and algorithm implementation.

[0483] "System" refers to a data processing whole consisting of at least one server, at least one information processing terminal, and optional distributed ledger management device, digital asset management device, etc., which are interconnected through a network to perform the various functions of this invention.

[0484] A "server" is an information processing device equipped with a processor and memory, used to execute program code to perform functions such as receiving and parsing prompts, calling generative artificial intelligence models, managing product information, processing transactions, registering non-fungible tokens, and analyzing emotions.

[0485] "Information processing terminal" refers to an electronic device operated by a user for inputting prompts, browsing product information, issuing transaction requests, and receiving and displaying the processing results from the server, including but not limited to mobile terminals, fixed terminals, or wearable terminals.

[0486] "User" refers to an individual or organizational entity that interacts with the server through an information processing terminal, inputs prompts, browses product information, and buys or sells information resources.

[0487] "Prompt statements" refer to natural language text or equivalent instruction data that is input by the user and used to drive generative artificial intelligence models to perform content generation processing, including requirements, constraints, or contextual descriptions of target concepts or setting information.

[0488] "Generative artificial intelligence models" refer to artificial intelligence models that are based on machine learning algorithms and / or deep learning algorithms and can automatically generate text information, image information, audio information or other data forms based on input prompts.

[0489] "Structured data" refers to instruction information in a predetermined data format obtained by the server after parsing the prompt statement, which includes at least a set of parameters, labels, or fields corresponding to the input of the generative artificial intelligence model.

[0490] "Abstract conceptual information" refers to high-level information generated by generative artificial intelligence models based on prompts, used to represent non-concrete physical objects such as creative ideas, logical structures, worldview settings, and story frameworks.

[0491] "Setup information" refers to design information generated by generative artificial intelligence models based on prompts, which represents specific configurations, styles, appearances, or behavioral schemes, including but not limited to image design, interface layout schemes, or product structure schemes.

[0492] "Product information" refers to the recorded data used for transaction management in electronic markets and / or information exchange platforms, including at least identification information, price information, sales period information, public status information, and metadata associated with abstract concept information and / or setting information.

[0493] "Electronic marketplace" refers to an online trading environment that provides functions such as displaying, searching, and purchasing product information via the internet, and is a platform used for transaction management of abstract conceptual information and / or setting information.

[0494] An "information exchange platform" refers to a network platform that allows multiple users to publish, browse, purchase, or exchange information resources. Electronic marketplaces can be considered a form of information exchange platform.

[0495] "Search criteria" refers to a set of parameters input by the user or system to filter product information in electronic marketplaces and / or information exchange platforms, including keywords, price range, category, language, tags, etc.

[0496] "Recommendation criteria" refer to the parameters or rules used to generate a list of recommended products, including user attribute information, user emotional state, behavioral history data, and feature information related to product information.

[0497] "Transaction processing" refers to the process executed collaboratively by the server and payment processing device to enable users to purchase goods information, including operations such as order generation, payment status confirmation, and transaction data recording.

[0498] "Transaction data" refers to the collection of transaction-related information recorded by the server during or after the transaction process, including at least transaction identifier, product identifier, buyer identifier, amount information, time information, and status information.

[0499] "Transaction fees" refer to the fees calculated and recorded by the server based on the transaction amount and a preset fee rate, which are part of the revenue of the platform operator.

[0500] "Payment processing device" refers to a device or service system used to perform financial transaction-related operations such as fund deduction, payment verification, and refund processing, including but not limited to online payment service systems or settlement systems.

[0501] "Non-homogeneous digital identification information" refers to data identifiers generated by a server and used to uniquely identify a certain abstract concept, setting information, or product information, including but not limited to hash values, identifier strings, or combinations thereof.

[0502] "Non-fungible tokens" are cryptographic digital tokens registered on a distributed ledger management device and used to represent unique ownership of a specific digital asset. Each token corresponds to a unique digital identification information.

[0503] A "distributed ledger management device" refers to a network system or set of nodes that maintains ledger data based on a distributed consensus mechanism and is used to record the generation, transfer, and state changes of non-fungible tokens.

[0504] "Digital asset management device" refers to a combination of hardware and software used to store, display and manage digital assets such as non-fungible tokens, including but not limited to digital wallet applications, account management systems or security modules.

[0505] "Emotion-related data" refers to input data that can be used to infer a user's emotional state, including but not limited to the user's text input, voice features, image features, interactive behavior data, or physiological signal data.

[0506] "Behavioral history data" refers to time-series data generated and recorded during a user's use of the system, related to actions such as browsing, clicking, searching, purchasing, and adding to favorites.

[0507] "User emotional state" refers to the state information obtained by the server through parsing and inference of emotion-related data and / or behavioral history data, which represents the category and intensity of a user's emotions at a certain moment or period of time.

[0508] "Dynamic adjustment" refers to the process by which the server automatically updates the recommended order, displayed content, and / or price information of products based on real-time or near-real-time user sentiment and other parameters.

[0509] "Input parameters" refers to the set of parameters used to control the behavior of generative artificial intelligence models, including but not limited to temperature coefficient, maximum generation length, style tag, sentiment tag, or other control variables.

[0510] In the embodiments of the present invention, the server, terminal and user work collaboratively in a network environment. Through a generative artificial intelligence model, the prompt statements are structured, parsed, generated, commercialized and digitally owned. This enables a series of specific data structure transformations and algorithm processing flows within the computer, thereby improving the processing efficiency, generation quality and resource utilization of the content generation system.

[0511] In one implementation, a server includes a processor, main memory, a network interface, and non-volatile storage. Software-wise, the server can employ an operating system, a backend runtime environment, and middleware. For example, the server might use a Unix-like operating system, run a backend framework based on an interpreted language (e.g., a network framework based on a general-purpose scripting language), and achieve data persistence and on-chain interaction through a database management system and a distributed ledger access library. In generative artificial intelligence, the server can connect to external generative AI services (e.g., the Transformer-based language model GPT-4, or a diffusion-based image generation model) via remote calls, or it can deploy pre-trained models locally.

[0512] In one embodiment, a terminal is a user-held electronic device, such as a smartphone, tablet, or personal computing device. Hardware-wise, a terminal includes a display device, input devices (touchscreen, keyboard, microphone, camera), storage device, and communication module. Software-wise, the terminal can run mobile applications (e.g., applications developed based on a cross-platform framework) or browser programs. In this invention, the terminal is primarily used to collect user input, display prompts and editing interfaces, show product information and generated results, and communicate with a server via a secure transmission protocol.

[0513] In this invention, users can input prompts through a terminal, browse product information provided by the server, and initiate purchase requests. Users can also allow the terminal to access the camera and microphone so that the server can obtain emotion-related data such as text, voice, and images for the server to infer emotional states.

[0514] In one implementation, the server performs multi-level data processing on the prompt statement. The server first receives the prompt statement in natural language text form from the terminal, for example: "Please develop an outline for a novel set in a medieval fantasy world, featuring a young magician who attempts to rebuild a cursed, lost kingdom." Upon receiving the prompt, the server parses it using a natural language processing module. This module may include a Transformer-based sequence labeling sub-model to extract keywords, contextual tags, target genre (e.g., "outline of a novel"), stylistic features (e.g., "medieval fantasy"), and constraints (e.g., "the protagonist is a young magician") from the prompt. Internally, the server maps the parsed results to predefined data structures, such as storing them as a set of fields: topic tags, background settings, character settings, target output type, etc.

[0515] In one implementation, the server combines the parsed results into structured instruction data. The server maps these structured fields to input parameters for a generative artificial intelligence model. For example, in a language model scenario, the server sets temperature parameters, maximum generation length, discourse style control labels, and emotion-related control vectors for the model. When the server receives additional emotion-related data from the user, it embeds emotion labels (e.g., "sad," "joy," "surprised") into the model's control encoding to conditionally change the generation probability distribution.

[0516] When a server invokes a generative AI model in one implementation, it takes the prompts and structured control information as input. The generative AI model can employ a multi-layer Transformer encoder-decoder architecture, using a multi-head attention mechanism to represent the prompts and progressively generating text or image descriptions based on context and control vectors during the decoding phase. When invoking the model, the server can set different weight decay parameters and attention masks for each layer to control the length and diversity of the generated data while ensuring contextual integrity. During the generation process, the server controls the output distribution through sampling strategies (e.g., top-k sampling, kernel sampling), thereby achieving more stable and adjustable output quality at the computer level than traditional rule-based generation.

[0517] After generating abstract conceptual information or setting information, the server writes the results to the backend database. In one implementation, the server establishes a unified product data structure for each generated result. This data structure includes: a unique identifier, an association identifier with the prompt statement, a text body or image link, metadata information (such as style tags, mood tags, creation time), a price field, a sales period field, and a public status field. The server uses database indexing technology and a full-text search engine to index these fields to enable efficient queries during subsequent retrieval and recommendation.

[0518] In one implementation, the server performs unified modeling and calculation of transaction data. When the server receives a purchase request from a terminal, it reads the product price from the database and calculates the transaction fee based on a preset rate. During calculation, the server stores the amount and fee rate information as high-precision numerical types to avoid the accumulation of floating-point errors. After a successful transaction, the server writes the transaction data, including the transaction amount, fee amount, timestamp, and identifiers of both the buyer and seller, into a log table and a summary table for subsequent statistical analysis and auditing. This unified modeling approach enables the server to quickly calculate platform revenue, user revenue, and fee distribution through aggregate queries, thereby improving overall data management efficiency.

[0519] In one implementation, the server utilizes a distributed ledger management system to register goods using non-fungible tokens (NFTs). The server first performs a hash operation on the generated content, for example, using a one-way hash algorithm to calculate a digest value for text content or image files. This digest value is then used as part of the NFT identification information. The server then accesses the distributed ledger management system via a blockchain interface, writing this identification information, along with the goods' metadata and owner's identifier, into a smart contract storage structure to obtain a unique NFT identifier. Internally, the server maintains a mapping table between NFT identifiers and local goods identifiers, enabling consistency between on-chain and off-chain data when querying transaction history and verifying ownership.

[0520] In one implementation, the server estimates the user's emotional state. After receiving text, speech-to-text, or image features from the terminal, the server invokes an emotion recognition sub-model. This sub-model may include a convolutional neural network for image facial expression feature extraction, a spectral feature-based model for speech emotion recognition, and a Transformer-based text emotion classification model. The server merges the multimodal features through feature concatenation and attention fusion mechanisms and inputs them into the classification layer, outputting an emotion category and a confidence vector. The server records the dominant emotion with a confidence score higher than a threshold as the user's emotional state and stores this state in the user context data structure.

[0521] In one implementation, the server dynamically adjusts generation control, recommendations, and pricing based on the user's emotional state. When invoking the generative AI model for subsequent prompts, the server encodes the user's emotional state as a control vector, inputting it along with semantic control labels into the model. This causes the model to be more inclined to generate content that aligns with the current emotional state during decoding. For example, when the emotional state is "nervous," the server adjusts model parameters to reduce the frequency of emotionally charged words in the generated text and increase the weight of reassuring language. When the emotional state is "joyful," the server can appropriately increase the liveliness of the generated content. Through this parameter-level control, the server implements non-human rules at the computational level to adjust the output distribution based on emotion, thereby improving personalized matching.

[0522] In terms of recommendation and pricing, the server incorporates user emotional states into feature vectors, which are then input into the recommendation algorithm module. In one implementation, the server employs a vector retrieval and ranking model, generating feature vectors for each product (including content tags, historical sales data, average ratings, etc.), and concatenating user emotional features with historical preference features before inputting them into the ranking model. The ranking model can be a gradient boosting tree model or a neural ranking model. The server trains itself to finely adjust the weights of emotional states in the ranking results, ensuring that the recommendation results consider both traditional preference features and real-time emotional features. For price adjustments, the server performs algorithmic calculations on the price field according to predefined rules (e.g., "allowing a certain percentage price increase for some products during a joyful state"), and imposes upper and lower bound constraints on the calculation results. By performing these automated calculations on the server side, the system can achieve dynamic personalized pricing with lower computational costs, reducing errors and delays caused by repeated manual configuration.

[0523] In one implementation, the server can automatically construct new prompts. The server utilizes existing prompt text and the user's emotional state to invoke its internal prompt generation module. This module can be a relatively small language model or a hybrid system based on rules and templates. When generating new prompts, the server does not simply copy user input, but rather expands upon underutilized dimensions of the original prompts, such as adding plot layers, stylistic constraints, or introducing new character settings. For example, the server can generate the following new prompts: "The user is currently feeling tense. Please generate a story segment based on the original medieval magical world setting that can alleviate the tension. The story should include gentle character interactions and a positive ending." The server then uses the new prompt as input to invoke a generative artificial intelligence model, obtaining new abstract conceptual information or setting information, and automatically registers it as product information with the default or policy-specified price and public status. This chain-automated process of "emotion-prompt-content-listing" enables the server to form a repeatable and scalable processing pattern in its computing architecture, reducing the number of manual interventions and intermediate storage conversions.

[0524] In one implementation, the terminal is responsible for user interaction and local preprocessing. When prompting for input, the terminal can perform basic format checks and length limit checks on the text, reducing the load on the server from invalid requests. When capturing images and audio, the terminal can perform lightweight feature extraction, such as generating local feature vectors using a local model, and then sending these vectors to the server, thereby reducing network transmission load and improving response speed. When displaying generated content and product lists, the terminal maps the structured data returned by the server into interface elements, allowing users to filter, save, and partially modify the generated results.

[0525] In one implementation, users can directly input complex creative requirements and constraints in text form. For example, a user can input the following prompt: "Please design a limited-edition T-shirt design themed around 'Cyber ​​Dragon' and provide descriptive text suitable for NFT distribution." After receiving the prompt, the server analyzes the keywords and concepts such as "Cyber ​​Dragon," "limited edition," "T-shirt design," and "NFT issuance" to generate design theme, application scenario, product type, and digital ownership requirements fields. Based on this, the server calls the image generation model to generate the design and the language model to generate the explanatory text, and then combines the two into a setting information record for uploading.

[0526] This invention, through the aforementioned modular division and algorithmic processing, enables the server to form a tightly coupled computational process across multiple levels, including prompt statement parsing, generative AI invocation, product data structure construction, transaction data and non-fungible token management, and emotion-driven control. By introducing a structured data intermediary layer and emotion control vectors internally, the server reduces repeated parsing of raw natural language text and establishes a unified field mapping relationship between model invocation and data storage. This design improves the server's processing throughput and cache hit rate, thereby reducing latency in large-scale concurrent scenarios.

[0527] Furthermore, when registering non-fungible tokens using a distributed ledger management system, the server employs a dual mapping structure of content digest and local identifier to ensure consistency and traceability between on-chain and off-chain data. By maintaining a unified mapping table and transaction log in the server, the system can quickly locate the generation process and transaction history of a specific non-fungible token when needed, reducing the probability of erroneous and duplicate registrations and improving the accuracy and reliability of digital asset management.

[0528] Through the aforementioned implementation, the collaboration between the server, terminal, and user in this invention's system not only achieves automated production and trading of generated content, but more importantly, it establishes a complete technical processing mechanism within the computer, centered around prompt statements and generative artificial intelligence models. This includes specific data structure design, emotion control logic, non-fungible token mapping mechanism, and unified transaction modeling method. Therefore, this invention surpasses traditional solutions that simply embed the generation model into business processes in terms of processing speed, resource utilization, generation accuracy, and data management, achieving a substantial improvement in computer technology itself.

[0529] use Figure 14 The processing flow is explained.

[0530] Step 1: The user enters a prompt statement through the terminal. The user launches an application or browser page on the terminal and inputs a short text in natural language via keyboard or voice into the prompt input interface. The input is the user's natural language text, such as "Please develop an outline for a novel set in a medieval fantasy world, with a young magician attempting to rebuild a cursed, lost kingdom." The output is the original prompt text temporarily stored in the terminal's memory. During this step, the terminal can perform basic checks on the input, such as checking for emptiness or prohibited characters, and display the current input on the interface for user confirmation.

[0531] Step 2: The terminal packages the prompt message and sends it to the server. The terminal takes the original prompt text, user identifier, language settings, etc., as input, organizes them into a data structure with predefined fields, and encapsulates them into a request message. The input is the prompt text and user identifier data; the output is a request data packet sent to the server via the network interface. Before sending, the terminal encodes the data according to a preset protocol, such as adding message headers, length information, and signature information, so that the server can accurately parse and verify it, and then sends it to the server through a secure transmission channel.

[0532] Step 3: The server receives and parses the prompt request. The server takes request data packets sent by the terminal as input, receives them through the network interface, and passes them to the backend program for parsing. The program extracts information such as the prompt text, user identifier, and language type. The input is the data packet from the network layer, and the output is the original prompt text and associated user context data generated in the server's memory. After parsing, the server verifies the prompt length, character encoding, and illegal words, and records the verification results in a log. If an anomaly is detected, an error response is generated and returned to the terminal.

[0533] Step 4: The server performs natural language parsing on the prompt statement. The server takes the prompt text in memory as input, calls the natural language processing module to perform word segmentation, part-of-speech tagging, named entity recognition, and syntactic analysis on the prompt text. The input is the raw prompt text, and the output is an intermediate semantic representation structure containing multiple fields such as keywords, topic tags, target genres, and contextual information. During data processing, the server uses statistical models or Transformer-based sequence labeling models to calculate feature vectors for each word or phrase, and determines whether the word belongs to a topic word, style word, or constraint condition based on the model weights, thereby generating a set of structured semantic tags.

[0534] Step 5: The server combines user attributes and sentiment data to generate structured indicator data. The server takes the semantic representation of the prompt, user attribute information, and user emotional state as input, and converts this information into structured instruction data that can be directly used by the generative artificial intelligence model. The input consists of a parsed set of semantic tags, user age group, preference records, and emotional tags (e.g., "nervous" or "joyful"). The output is a structured data object containing fields such as control parameters, constraint labels, and emotion vectors. The data processing performed by the server in this step includes: mapping semantic tags to predefined control fields, encoding the user's emotional state into a numerical vector, and combining it with style control parameters to form a list of model input parameters, thereby subjecting the subsequent generation process to multi-dimensional control.

[0535] Step 6: The server inputs structured instruction data into the generative artificial intelligence model and generates content. The server takes structured instruction data and prompt text as input, establishes a call session with the generative AI model, and inputs control parameters, prompt text, and emotion vectors into the model. The input consists of encoded control parameters and original prompts; the output is abstract conceptual information or setting information generated by the model (such as a novel outline, design description, image description, etc.). During the call, the server performs vector encoding, multi-layer attention calculation, and decoding steps through an algorithm. The model adjusts the probability distribution based on the control parameters at each step when generating new tags. After receiving the streaming output, the server reassembles it into complete text or data fragments, ultimately forming a storable generated result.

[0536] Step 7: The server converts the generated results into product information and writes it into the database. The server takes the generated abstract conceptual information or setting information as input to construct product information records, including fields such as unique identifier, generated content, title, tags, and creation time. The input is the model output content and related control parameters; the output is product data records persistently stored in the database. The data processing performed by the server during this process includes: generating a new product identifier, extracting a summary of the content for preview, writing sentiment tags, style tags, etc., into metadata fields, and indexing key fields in the database to support subsequent efficient searching and filtering.

[0537] Step 8: Users set prices and sales conditions through the terminal. In the terminal's product management interface, users use the product preview information returned by the server as input to edit sales conditions such as price, sales period, public access status, and whether to associate with non-fungible tokens. The input is the product title and content summary displayed on the terminal, and the output is settings including price, sales start and end times, public access status, and non-fungible token options. After user confirmation, the terminal packages these settings into a request message, ready to send it to the server to update the product information.

[0538] Step 9: The terminal sends product sales settings information to the server. The terminal takes the user-edited price and sales conditions as input, encapsulating them along with the product identifier into configuration request data. Inputs include fields such as price values, time ranges, and Boolean options; the output is an update request message sent to the server over the network. Before sending, the terminal performs basic validation on the values ​​(e.g., ensuring the price is non-negative and the time sequence is reasonable) and records the latest submission time locally to facilitate retrying or prompting the user in case of network errors.

[0539] Step 10: The server updates product information and builds a search index. The server takes the sales settings information sent by the terminal as input, searches the database for the corresponding product record, and updates the price, sales time, and public status fields. The input includes product identifiers and price settings; the output is the updated product record and corresponding index entries. During data processing, the server regenerates the retrieval index, including full-text indexes and multi-field composite indexes. For example, it creates an inverted index based on title, tags, and price range to enable fast filtering and sorting during subsequent queries, reducing query latency.

[0540] Step 11: Users browse and search for products in the online marketplace via their devices. Users open the online marketplace page on their devices, inputting their personal interests or keywords, and then enter search criteria or select category tags through the search box. The input includes keyword text, price range, and filter criteria; the output is a search request message submitted to the server. During this process, the device can perform simple normalization on the search criteria, such as converting keywords to lowercase, removing extra spaces, and displaying the currently active filters on the interface.

[0541] Step 12: The server filters products based on search criteria and returns a list of results. The server takes keywords and filter criteria from the search request as input, performs a query operation in the database and search index, and selects product records that meet the criteria. Inputs include keywords, price range, language type, and sentiment tags, while output is a list of brief product information sorted by relevance or sales volume. In its data processing, the server calculates a matching score for each candidate product, combining text matching degree, price proximity, and user sentiment preference weights into a ranking score. The top few items are then assembled into response data and returned to the terminal for display.

[0542] Step 13: Users view product details and decide to purchase on the terminal. After receiving the product list from the server, the user selects a product from the list as input and clicks to request detailed information about that product. The input is the selected product identifier, and the output is the details request message sent by the terminal to the server and the final details page displayed to the user. Upon receiving the details data, the terminal displays the product title, summary, price, sales status, and whether it includes non-fungible tokens on the interface, allowing the user to decide whether to continue purchasing.

[0543] Step 14: The terminal sends a purchase request to the server. When a user confirms a purchase on the product details page, using the product identifier and payment method as input, they trigger a purchase request by clicking the "Buy" button on the terminal. The input includes the product identifier, user account information, and the selected payment method; the output is a purchase request message containing order information. The terminal generates a temporary order number locally and sends it along with the user identifier to the server so that the server can create an order record in subsequent processing.

[0544] Step 15: The server creates the order and coordinates with the payment processing device to execute the transaction. The server takes the product identifier and user information from the purchase request as input, queries the database for the current price and inventory status, generates an order record, and calculates the amount due and expected transaction fees. Inputs include product price information and transaction fee rate configuration; outputs include transaction amount, transaction fee amount, and order status. The server then transmits the amount and payment method information to the payment processing device. After completing the deduction verification, the payment processing device returns the payment result, and the server updates the order status accordingly, changing it from "pending payment" to "paid" or "failed."

[0545] Step 16: The server generates non-fungible digital identification information based on transaction data and registers the non-fungible tokens. The server takes completed transaction order data and corresponding product details as input, executes a hash algorithm to calculate a product details summary, and combines this summary, order identifier, and user wallet identifier into a non-fungible digital identifier (NfGDI). The input consists of product details text or image data and an order identifier; the output is a unique NfGDI and a NfGDI identifier. The server then sends a registration request to the ledger management device via a distributed ledger interface, writes the identifier into the ledger, and obtains the on-chain transaction identifier and NfGDI identifier, storing them in a local mapping table for subsequent querying and ownership verification.

[0546] Step 17: The server returns the complete content and ownership information to the terminal. The server takes the order status, "paid" confirmation, and the generated non-fungible token identifier as input to construct response data. This data includes a complete prompt statement, generated abstract conceptual information or setting information, and non-fungible token information associated with the product. Inputs are order records and on-chain registration results; output is detailed delivery data returned to the terminal. The server filters sensitive fields when generating the response, providing the user only with the content they are authorized to view and proof of ownership.

[0547] Step 18: The terminal displays the purchase results and generated content to the user. The terminal takes the complete content and ownership information returned by the server as input, displays the generated abstract concept information or full text of the settings information in the user interface, and annotates the corresponding non-fungible token summary or identifier. The input consists of text content, a summary, and a non-fungible token identifier; the output is a visual interface presented on the display device. In this step, the terminal can also provide operation buttons such as "Copy Prompt Statement," "View On-Chain Records," and "Export to Digital Wallet," allowing users to continue using this content in other applications.

[0548] Step 19: Users access the purchased notification message via a generative artificial intelligence model through the terminal. The user takes the purchased notification text displayed on the terminal as input, selects the "Use in Generative AI Model" function, and specifies the generation goal (e.g., generating a complete novel or generating a variant design). The input is the purchased notification text and the user's goal setting; the output is a model invocation request initiated by the terminal. When organizing the request, the terminal packages the notification text and goal information and sends it to the server or directly to an external generation service to trigger secondary generation.

[0549] Step 20: The server updates the user's emotional state based on emotion-related data and adjusts recommendations and prices accordingly. The server takes the latest emotion-related data collected from the terminal (such as a user's comment "This design excites me") and behavioral records as input, infers the current user's emotional state through an emotion recognition module, and writes this state into the user's context data. The input includes text, timestamps, and interaction types; the output is a new emotion tag and confidence score. The server then adds the emotion tag to the recommendation feature vector, adjusts the ranking score of the recommendation algorithm, and calculates and updates the price fields of some products according to preset rules. For example, in the "joyful" state, the price of specific products is proportionally increased, and the adjusted results are applied to subsequent searches and displays for that user.

[0550] 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.

[0551] 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.

[0552] 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.

[0553] 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.

[0554] 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.

[0555] Second Implementation Method Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0556] 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.

[0557] 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).

[0558] 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.

[0559] 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.

[0560] 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).

[0561] 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.

[0562] 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.

[0563] 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.

[0564] 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).

[0565] 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.

[0566] 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".

[0567] 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.

[0568] Application Example 1 The process is the same as that of the specific processing described in Application Example 1 in the first embodiment above, so the description is omitted.

[0569] 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.

[0570] 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.

[0571] 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.

[0572] 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.

[0573] 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.

[0574] 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.

[0575] 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.

[0576] Third Implementation Method Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0577] 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.

[0578] 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).

[0579] 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.

[0580] 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.

[0581] 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).

[0582] 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.

[0583] 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.

[0584] 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.

[0585] 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.

[0586] 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.

[0587] 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".

[0588] 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.

[0589] Application Example 1 The process is the same as that of the specific processing described in Application Example 1 in the first embodiment above, so the description is omitted.

[0590] 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.

[0591] 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.

[0592] 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.

[0593] 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.

[0594] 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.

[0595] 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.

[0596] 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.

[0597] Fourth Implementation Method Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0598] 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.

[0599] 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).

[0600] 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.

[0601] 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.

[0602] 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).

[0603] 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.

[0604] 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.

[0605] 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.

[0606] 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.

[0607] 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.

[0608] 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.

[0609] 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".

[0610] 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.

[0611] Application Example 1 The process is the same as that of the specific processing described in Application Example 1 in the first embodiment above, so the description is omitted.

[0612] 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.

[0613] 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.

[0614] 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.

[0615] 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.

[0616] 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.

[0617] 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.

[0618] 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.

[0619] 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.

[0620] 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.

[0621] 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.

[0622] 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).

[0623] 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.

[0624] 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."

[0625] 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.

[0626] 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).

[0627] 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.

[0628] 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.

[0629] 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.

[0630] 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.

[0631] 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.

[0632] 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.

[0633] 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.

[0634] 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.

[0635] 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.

[0636] 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.

[0637] In addition, the following notes are provided in response to the above explanation.

[0638] Example 1 (Note 1) An information processing system, characterized in that it comprises: A device for receiving input prompt statements from a user terminal via a communication function and obtaining the prompt statements as text data; An apparatus for performing preprocessing on the text data, including string normalization, length detection and format detection, and converting the text data into an input format for a generative artificial intelligence model; An apparatus for inputting numerical data containing the transformed text data into the generative artificial intelligence model, and for causing the generative artificial intelligence model to perform inference processing including multi-layer matrix operations and attention mechanism operations based on the numerical data, thereby generating conceptual data or design data; A device for storing the generated concept data or design data into a data storage device, and for associating and recording the work identifier, user identifier, and generation time corresponding to the concept data or design data; An apparatus for converting the conceptual data or design data into product information that can be disclosed as electronic market information, and for generating distribution data as electronic market listing information and detailed information; An apparatus for generating transaction data based on the product information for issuing non-fungible tokens using distributed ledger technology, sending the transaction data to a distributed ledger management device to issue the non-fungible tokens, and storing the identification information of the non-fungible tokens in correspondence with the work identifier; An apparatus for receiving buy and sell requests between users through the electronic marketplace, generating control information for executing transaction processing including the transfer of ownership of the non-fungible tokens, and initiating ownership transfer transactions to the distributed ledger management device based on the control information; An apparatus for calculating a transaction fee based on the transaction amount corresponding to the buy / sell request, storing the transaction fee, the transaction amount, and the identifier of the transaction party as a transaction record, and managing the transaction fee as revenue information.

[0639] (Note 2) According to the information processing system described in Appendix 1, the device for converting the text data into an input format for a generative artificial intelligence model is configured to perform symbol sequence partitioning on the text data, encode the symbol sequence into an identifier sequence, convert the identifier sequence into a numerical vector sequence to generate a tensor for model input, and input the tensor into the generative artificial intelligence model to instruct the generation of the concept data or design data.

[0640] (Note 3) According to the information processing system described in Appendix 1, the device for converting the conceptual data or design data into commodity information that can be disclosed as electronic market information is configured to attach prompt statements corresponding to the conceptual data or design data, type information of generative artificial intelligence models, and generation condition information as metadata, and store the metadata in association with the commodity information, while generating index information that can be used as retrieval conditions for the commodity information.

[0641] Application Example 1 (Note 1) An information processing system, characterized in that it comprises: A device for receiving prompt statements input by a user from a terminal and formatting the prompt statements to generate generation request information containing the prompt statements; An apparatus for selecting a generative artificial intelligence model based on a prompt statement and additional generation conditions contained in the generation request information, and inputting the prompt statement into the generative artificial intelligence model to perform inference processing, thereby generating digital content; A device for storing the generated digital content, along with corresponding prompts, generative artificial intelligence model information, and user identification information, as content information into a storage device; An apparatus for performing a non-fungible token management procedure on a distributed ledger based on the content information, issuing a corresponding non-fungible token for the digital content on the distributed ledger, and associating the non-fungible token with the content information to establish ownership information; A device for publishing the digital content as product information to an electronic trading platform based on the content information and the ownership information, and providing the product information to other users for browsing; A transaction management device for receiving purchase requests for the product information, updating the ownership of the non-fungible token on the distributed ledger to complete the ownership transfer process, calculating transaction fees based on the price information corresponding to the ownership transfer process, and storing the transaction fees and the purchase and sale price as transaction records. An apparatus for determining access rights to the digital content based on the ownership information and the transaction records, and for providing or allowing the download of the digital content to the corresponding user terminal when access rights are determined to be granted.

[0642] (Note 2) The information processing system according to Appendix 1 is characterized in that, The device for receiving prompt statements input by the user from the terminal is configured to perform natural language processing on the prompt statements, including character normalization processing, length constraint checking, and content security determination, and to determine the type and generation conditions of the generative artificial intelligence model based on the results of the natural language processing.

[0643] (Note 3) The information processing system according to Appendix 1 is characterized in that, The transaction management device is configured to automatically generate product information based on the feature information extracted from the digital content and the prompt statements when displayed on an electronic trading platform, and to manage the sales status, ownership change history, and transaction fee revenue history of the digital content by updating the product information.

[0644] Example 2 (Note 1) An information processing system, characterized in that it comprises: A device for receiving prompt statements input by a user via a terminal and acquiring the prompt statements as natural language; An apparatus for generating input data for a generative artificial intelligence model based on the prompt statement, providing the input data to the generative artificial intelligence model, and obtaining the generated results related to the concept or design generated by the generative artificial intelligence model. An apparatus for generating product information based on the generation result, registering the product information and the generation result in a storage device, and disclosing the product information as product information on an information processing platform for e-commerce. An apparatus for generating non-fungible tokens corresponding to the product information in a distributed ledger, and for storing and managing the non-fungible tokens in correspondence with the product information; An apparatus for receiving purchase requests from terminals, generating transaction information related to the product information, performing payment processing using a settlement information processing platform, and updating the holder of the non-fungible token based on the result of the payment processing. An apparatus for obtaining evaluation information from a terminal after a transaction is completed, accumulating the evaluation information to update the credit information for each user, and controlling the display or retrieval results in the information processing platform for e-commerce based on the credit information.

[0645] (Note 2) According to the information processing system described in Appendix 1, the device for generating input data for a generative artificial intelligence model based on the prompt statement is further configured to: before inputting the prompt statement into the generative artificial intelligence model, parse the prompt statement using natural language processing technology, extract the generation target category, generation parameters, and usage conditions, generate generation instruction data corresponding to the extraction results, and use the generation instruction data to control the content or expression form of the conceived or designed generation result.

[0646] (Note 3) According to the information processing system described in Appendix 1, the system is further configured to: calculate transaction fee information for each transaction completed in the information processing platform for e-commerce, record the transaction fee information as revenue management data in the storage device, and save the transaction fee information in association with the evaluation information and the credit information, so as to perform trust management and fee setting for the user based on transaction history and revenue information.

[0647] Application Example 2 (Note 1) An information processing system, characterized in that it comprises: A device for receiving prompt statements input by a user through an information processing terminal; An apparatus for parsing the content of the prompt statement based on the prompt statement and user attribute information and / or user emotion information, and converting the prompt statement into structured data as input information for a generative artificial intelligence model; A device for inputting the structured data into the generative artificial intelligence model, so that the generative artificial intelligence model generates abstract conceptual information and / or setting information based on the prompt statement; A device for registering the generated abstract concept information and / or setting information as product information, and for listing the product information on an electronic marketplace and / or information exchange platform after associating price information, sales period information and public status information with the product information. An apparatus for receiving search criteria and / or recommendation criteria related to the product information, and based on the search criteria and / or the recommendation criteria, extracting target product information from multiple product information items published in the electronic market and / or the information exchange platform, and outputting the extraction result to the information processing terminal; An apparatus for receiving purchase requests for the product information, cooperating with a payment processing device to execute transaction processing, generating transaction data based on the transaction processing, calculating transaction fees, and recording them. An apparatus for generating non-fungible digital identification information based on the transaction data, the generated abstract concept information and / or setting information and / or the product information, cooperating with a distributed ledger management device to register the non-fungible digital identification information as a non-fungible token, and associating the non-fungible token with the purchaser's digital asset management device; An apparatus for analyzing emotion-related data and / or behavioral history data obtained from the user to infer the user's emotional state, and for dynamically adjusting the recommendation of product information and / or the price information of the product information based on the user's emotional state; An apparatus for automatically generating new prompts for the generative artificial intelligence model based on the user's emotional state and the prompts, and for automatically publishing abstract conceptual information and / or setting information on the new prompts in the electronic marketplace and / or the information exchange platform.

[0648] (Note 2) The information processing system according to Appendix 1 is characterized in that, The device for receiving prompt statements is configured to parse the prompt statements using natural language processing technology, extract keyword information, emotion tag information and / or usage category information from the prompt statements, and set input parameters to the generative artificial intelligence model based on the extraction results, so as to instruct the generation of the abstract concept information and / or the setting information.

[0649] (Note 3) The information processing system according to Appendix 1 is characterized in that, The device for performing transaction processing is configured to calculate the transaction fee based on the transaction amount information contained in the transaction data and the pre-stored fee rate information, record the transaction fee as revenue information of the platform operator, and notify the information processing terminal of the purchaser of the registration result and ownership information of the non-fungible token.

Claims

1. An information processing system, characterized in that, include: processor; The processor is configured to: receive prompt text input by the user; and input the prompt text into a generative artificial intelligence model. The generative artificial intelligence model generates a concept or design based on the prompt text; the generated concept or design is uploaded to an online marketplace; and a non-fungible token is assigned to the generated concept or design.

2. The information processing system according to claim 1, characterized in that, The processor is configured to: parse the prompt text using natural language processing technology and instruct the generation of the concept or design.

3. The information processing system according to claim 1, characterized in that, The processor is configured to: act as an intermediary in transactions between users; and calculate a transaction fee based on the transaction and record the transaction fee as revenue.

Images