Multi-application instruction registration and large model driven instruction scheduling method and device, equipment and medium

Abstract

The invention discloses a multi-application instruction registration and large model driven instruction scheduling method and device, equipment and a medium, and the method comprises the steps: carrying out the information binding operation of a target user in response to a first request for calling an instruction registration interface, and obtaining first information data; based on the first information data, in response to a second request for calling a semantic interaction interface, calling a large model to perform intention analysis on the target user to obtain a to-be-executed instruction list; and calling an instruction scheduling mechanism to realize instruction scheduling according to the to-be-executed instruction list. The multi-application instruction scheduling efficiency can be improved, and the method can be widely applied to the technical field of instruction scheduling.

Description

Technical Field

[0001] The present invention relates to the technical field of instruction scheduling, and in particular to a multi-application instruction registration and large model-driven instruction scheduling method, device, equipment and medium. Background Art

[0002] With the rapid development of Internet and Internet of Things technologies, smart devices and applications have become indispensable in life. In multiple fields such as smart home control, smart office, and game applications, users may need to manage and use multiple applications on the same end-side system or environment. Such applications usually have different functions and uses. Although they are independent of each other, they can share the same set of basic system and resource management mechanisms. However, in a multi-application environment, most of the existing user instruction recognition and scheduling execution solutions are limited to single applications, resulting in a poor user experience. In addition, traditional instruction scheduling strategies, such as those based on static priorities or polling, often face problems such as slow response speed and low resource utilization. Summary of the Invention

[0003] In view of this, the main purpose of the embodiments of the present invention is to provide a multi-application instruction registration and large model-driven instruction scheduling method, device, equipment and medium, in order to solve at least one of the problems in the prior art. The present invention can improve the efficiency of multi-application instruction scheduling.

[0004] To achieve the above object, on the one hand, an embodiment of the present invention provides a multi-application instruction registration and large model-driven instruction scheduling method, the method comprising:

[0005] In response to a first request for calling an instruction registration interface, perform an information binding operation on a target user to obtain first information data;

[0006] Based on the first information data, in response to a second request for calling a semantic interaction interface, call a large model to perform intent analysis on the target user to obtain a list of instructions to be executed;

[0007] According to the list of instructions to be executed, call an instruction scheduling mechanism to implement instruction scheduling.

[0008] In some embodiments, before the step of performing an information binding operation on a target user in response to a first request for calling an instruction registration interface to obtain first information data, the following steps are further included:

[0009] In response to a filing request of a target application in the user device of the target user, perform qualification review on the target application;

[0010] When the result of the qualification review is passed, assign a unique code and key to the target application;

[0011] Among them, the unique code and the key are used for authentication and data encryption.

[0012] In some embodiments, in response to a first request to call the instruction registration interface, an information binding operation is performed on the target user to obtain first information data, including the following steps:

[0013] In response to the first request, verify the legality and security of the first request to obtain a verification result;

[0014] Obtain second information data reported by the instruction registration interface;

[0015] When the verification result is legal and secure, bind the target user to the second information data to obtain the first information data.

[0016] In some embodiments, based on the first information data, in response to a second request to call the semantic interaction interface, call a large model to perform intent analysis on the target user to obtain a list of instructions to be executed, including the following steps:

[0017] According to the first information data, call a large model to generate a first instruction recognition prompt word, and bind the first instruction recognition prompt word to the user device of the target user;

[0018] In response to the second request, extract a second instruction recognition prompt word input by the user device;

[0019] Based on the first instruction recognition prompt word, call the large model to perform intent analysis on the second instruction recognition prompt word to obtain the list of instructions to be executed.

[0020] In some embodiments, according to the list of instructions to be executed, calling an instruction scheduling mechanism to implement instruction scheduling includes the following steps:

[0021] According to the list of instructions to be executed, call the instruction scheduling mechanism to send a list of high-priority instructions to obtain a response message;

[0022] According to the callback address in the response message, call the instruction scheduling mechanism, and orderly execute the list of low-priority instructions through an asynchronous thread mechanism to obtain a first operation result.

[0023] In some embodiments, after calling the instruction scheduling mechanism to implement instruction scheduling according to the list of instructions to be executed, the following steps are further included:

[0024] According to the list of instructions to be executed, call the user device of the target user to trigger an instruction execution process to obtain a second operation result.

[0025] To achieve the above object, another aspect of the embodiments of the present invention provides an instruction scheduling device for multi-application instruction registration and large model-driven, the device comprising:

[0026] A first module, configured to perform an information binding operation on a target user in response to a first request for invoking an instruction registration interface, and obtain first information data;

[0027] A second module, configured to, based on the first information data, in response to a second request for invoking a semantic interaction interface, call a large model to perform intention analysis on the target user, and obtain a list of instructions to be executed;

[0028] A third module, configured to call an instruction scheduling mechanism to implement instruction scheduling according to the list of instructions to be executed.

[0029] In some embodiments, an instruction scheduling device for multi-application instruction registration and large model-driven further comprises:

[0030] A fourth module, configured to perform qualification review on a target application in response to a filing request of the target application in the user device of the target user;

[0031] A fifth module, configured to, when the result of the qualification review is passed, assign a unique code and a key to the target application;

[0032] Wherein, the unique code and the key are used for authentication and data encryption.

[0033] To achieve the above object, another aspect of the embodiments of the present invention provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above-mentioned method for multi-application instruction registration and large model-driven instruction scheduling when executing the computer program.

[0034] To achieve the above object, another aspect of the embodiments of the present invention provides a computer-readable storage medium, the computer-readable storage medium storing a computer program, and the computer program implementing the above-mentioned method for multi-application instruction registration and large model-driven instruction scheduling when being executed by a processor.

[0035] To achieve the above object, another aspect of the embodiments of the present invention provides a computer program product or a computer program, the computer program product or the computer program comprising computer instructions, and the computer instructions being stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the above-mentioned method for multi-application instruction registration and large model-driven instruction scheduling.

[0036] Embodiments of the present invention at least include the following beneficial effects: The present invention provides a multi-application instruction registration and large model-driven instruction scheduling method, device, equipment, and medium. This solution performs information binding operations on the target user in response to a first request for invoking the instruction registration interface to obtain first information data. Based on the first information data, in response to a second request for invoking the semantic interaction interface, a large model is called to perform intent analysis on the target user to obtain a list of instructions to be executed. According to the list of instructions to be executed, an instruction scheduling mechanism is called to implement instruction scheduling. By efficiently performing user semantic analysis and executable instruction deduction through the integration of a large model, and by introducing a multi-level instruction scheduling mechanism for orderly scheduling of instructions, the efficiency of multi-application instruction scheduling is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0038] Figure 1 is a flowchart of the multi-application instruction registration and large model-driven instruction scheduling method provided by the embodiments of the present invention;

[0039] Figure 2 is a schematic diagram of the core module of the multi-application instruction registration and large model-driven instruction scheduling provided by the embodiments of the present invention;

[0040] Figure 3 is a schematic diagram of the multi-application instruction registration and large model-driven instruction scheduling device or system module object provided by the embodiments of the present invention;

[0041] Figure 4 is a timing diagram of application filing and qualification review provided by the embodiments of the present invention;

[0042] Figure 5 is a timing diagram of user login and device information synchronization provided by the embodiments of the present invention;

[0043] Figure 6 is a timing diagram of generating instruction recognition prompt words provided by the embodiments of the present invention;

[0044] Figure 7 is a timing diagram of application instruction processing and data transmission provided by the embodiments of the present invention;

[0045] Figure 8 is a timing diagram of intent analysis and generation of a list of instructions to be executed provided by the embodiments of the present invention;

[0046] Figure 9It is a timing diagram of executing instructions on the cloud side and the terminal side provided by an embodiment of the present invention;

[0047] Figure 10 It is a timing diagram of the response result on the cloud side provided by an embodiment of the present invention;

[0048] Figure 11 It is a timing diagram of instructions to be executed on the terminal side provided by an embodiment of the present invention;

[0049] Figure 12 It is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0050] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the embodiments of the present invention. They are only examples of devices and methods consistent with some aspects of the embodiments of the present invention described in detail in the appended claims.

[0051] It should be noted that although functional module division is performed in the system schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order from the module division in the system or the order in the flowchart. The terms "first / S100" and "second / S200" in the specification, claims and the above-mentioned drawings may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present invention, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "when" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0052] The terms "at least one", "multiple", "each", "any one", etc. used in the present invention, at least one includes one, two or more than two, multiple includes two or more than two, each refers to each of the corresponding multiple, and any one refers to any one of the multiple.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used herein are only for the purpose of describing the embodiments of the present invention, and are not intended to limit the present invention.

[0054] Before elaborating on the embodiments of the present invention, some nouns and terms involved in the embodiments of the present invention are first explained, and the nouns and terms involved in the embodiments of the present invention are applicable to the following explanations.

[0055] The launcher application, abbreviated as launcher, is an important component in the Android system. It is usually referred to as the "desktop launcher" or "homescreen". Its main function is to provide a user interface for the user to display the device's homescreen, application icons, desktop widgets, and manage the launching of application programs.

[0056] Inter-Process Communication (IPC) is a technology in computer systems that allows different processes to exchange data and communicate. In the Android system, IPC is the core mechanism for cross-process communication.

[0057] Server-Sent Events (SSE) is a real-time communication technology based on the HTTP protocol that allows the server to actively push data to the client. It achieves one-way communication (from the server to the client) through a persistent HTTP connection and is suitable for scenarios of real-time data push.

[0058] RPC (Remote Procedure Call Protocol) is a remote procedure call protocol and a special IPC that allows a program to request services on different computers without having to understand the underlying network technology. The main role of RPC is to make method calls between different services as convenient as local calls.

[0059] Multi-application refers to running and managing multiple application programs in the same end-side system or environment. These application programs can have different functions and uses, but they share the same set of infrastructure and resource management mechanisms.

[0060] Large model refers to the Large Language Model technology, which is a deep learning model usually with billions or even more parameters and can exhibit excellent performance in natural language processing tasks.

[0061] Instruction registration refers to reporting the functional instructions that an application program can support for recognition and execution. These instructions will be stored and managed in this system for subsequent parsing and scheduling for execution.

[0062] Instruction scheduling refers to the process of managing and allocating the execution of instructions in a computer system or application. Its purpose is to ensure that multiple instructions can be executed efficiently and orderly according to certain rules and priorities, thereby optimizing system performance and resource utilization.

[0063] With the rapid development of Internet and Internet of Things technologies, smart devices and applications have become an indispensable part of modern life. In many fields such as smart home control, smart office, and gaming applications, users' demand for smart devices is increasing day by day. These devices and applications not only improve the efficiency of life and work but also bring users unprecedented convenience and comfortable experiences.

[0064] In a smart home environment, users can manage all smart devices in their homes through a single control interface, such as lighting, temperature control, security monitoring, etc. In the field of smart office, employees can use smart devices and applications for remote collaboration, video conferencing, and automated task management. In gaming applications, players can obtain a more immersive and interactive gaming experience through smart devices. Users may need to manage and use multiple applications on the same end-side system or environment. Such applications usually have different functions and uses. Although they are independent of each other, they can share the same set of basic system and resource management mechanisms. For example, the lighting control application and the security monitoring application in a smart home system can share the user's voice command recognition system, and the video conferencing application and the task management application in a smart office system can share the user's schedule management and notification system.

[0065] However, in a multi-application environment, most of the existing user instruction recognition and scheduling execution solutions are limited to a single application, resulting in a poor user experience. When using multiple applications, users often need to switch between different applications, which not only increases the complexity of operations but also reduces efficiency. In addition, traditional instruction scheduling strategies, such as those based on static priorities or polling, often face problems such as slow response speed and low resource utilization.

[0066] In view of this, as Figure 1 shown, the embodiments of the present invention provide a multi-application instruction registration and large model-driven instruction scheduling method, which may include but is not limited to steps S100 to S300:

[0067] Step S100, in response to a first request to call an instruction registration interface, perform an information binding operation on the target user to obtain first information data;

[0068] Step S200, based on the first information data, in response to a second request to call a semantic interaction interface, call a large model to perform intent analysis on the target user to obtain a list of instructions to be executed;

[0069] Step S300: According to the to-be-executed instruction list, call the instruction scheduling mechanism to implement instruction scheduling.

[0070] Before step S100 in some embodiments, it further includes: in response to a filing request of a target application in the user device of the target user, performing qualification review on the target application; when the result of the qualification review is passed, assigning a unique code and a key to the target application; wherein, the unique code and the key are used for identity authentication and data encryption. Exemplarily, as Figure 4 shown, taking the Xiaoyi Butler APP as an example of the target application, file relevant application information. File the application information and instruction intent of the Xiaoyi Butler APP, and perform qualification review on it through the business system. After passing the review, assign an application unique code and an application key. The application unique code is the unique identifier of the application identity, and the application key is used for secure encryption of the reported data. By filing the application and having it reviewed by a commissioner, it can ensure the compliance of application instructions and the security of user data before the application accesses the system.

[0071] Before step S100 in some embodiments, it may further include that after the target user starts the Xiaoyi Butler APP, first perform user authentication and login. At this time, the Xiaoyi Butler APP notifies the launcher that it is in the activated state. After completing the authentication, the Xiaoyi Butler transmits relevant data (which may include but is not limited to user desensitized information, the list of user-controllable devices, the function descriptions supported by the controllable devices (including operation descriptions, support for end / cloud execution identifiers, etc.) and the instruction format requirements for function control, etc.) to the launcher through the IPC mechanism, so as to lay a foundation for subsequent completion of application instruction registration. Exemplarily, as Figure 5 shown, the target user can log in to the Xiaoyi Butler APP through the Tianyi account. The Xiaoyi Butler authenticates the user, returns the authentication result to the target user, and obtains the information of the devices bound to the target user. Then the Xiaoyi Butler transmits relevant data to the launcher through the RPC protocol.

[0072] In some embodiments, step S100 may include but is not limited to steps S110 to S130:

[0073] Step S110: In response to the first request, verify the legality and security of the first request to obtain a verification result;

[0074] Step S120: Obtain the second information data reported by the instruction registration interface;

[0075] Step S130: When the verification result is legal and secure, bind the target user to the second information data to obtain the first information data.

[0076] In step S110 of some embodiments, in response to the first request for the call instruction registration interface, the business system authenticates the first request for the call instruction registration interface to verify the legality and security of the request.

[0077] In steps S120 to S130 of some embodiments, when the authentication verification result is legal and secure, by obtaining the second information data reported by the instruction registration interface, the business system binds the user to the second information data reported by the instruction registration interface.

[0078] Optionally, as Figure 6 shown, Launcher integrates the received data and the basic information of the intelligent terminal, and calls the application instruction registration interface provided by the business system. In response to the first request for the call instruction registration interface, the business system authenticates the request for the call instruction registration interface to verify the legality and security of the request. When the authentication verification result is legal and secure, the business system binds the user to the second information data reported by the instruction registration interface to obtain the first information data. Among them, the first information data may include, but is not limited to, user information, device information, application instruction information, etc.

[0079] In some alternative embodiments, the business system also assigns a unique identification code for the target user for the user device, which is used to store all the information generated during the subsequent interactions of the user on this device. Exemplarily, taking the control of an intelligent air conditioner as an example, functions such as on / off status, mode adjustment, and wind speed gear adjustment are supported. Different functions may have different precondition requirements and control instruction formats. For example: There are no preconditions for the on / off status, the precondition for mode adjustment is to confirm the on / off status, and the precondition for wind speed gear adjustment is to confirm the mode adjustment. To ensure the accuracy of the subsequent execution instructions generated by the large model, the description and instruction format of the supported functions must be clear and definite. Exemplarily, taking the air conditioner switch as an example, the instruction format can be defined as JSON format, and fields such as device information, location information, execution parameters, and execution conditions can be defined in the format.

[0080] In some embodiments, step S200 may include, but is not limited to, steps S210 to S230:

[0081] Step S210, according to the first information data, call the large model to generate a first instruction recognition prompt word, and bind the first instruction recognition prompt word to the user device of the target user;

[0082] Step S220, in response to the second request, extract the second instruction recognition prompt word input by the user device;

[0083] Step S230: Based on the first instruction, identify the prompt words, and call the large model to perform intent analysis on the second instruction identification prompt words to obtain the list of instructions to be executed.

[0084] In step S210 of some embodiments, according to the first information data, the business system calls a large language model to generate instruction identification prompt words. The large language model returns the generated instruction identification prompt words to the business system. The business system binds the instruction identification prompt words to the user device of the target user and caches the instruction identification prompt words. Exemplarily, as Figure 6 shown, according to the reported information such as the description of the bound device, its function instruction description, and format requirements, etc., the semantic abstraction ability of the large language model is used to generate corresponding recognizable intent types and their application instructions respectively, and the complete instruction identification prompt words are integrated. The prompt words are the core tools for the interaction of the large language model. Through appropriately abstracted instruction identification prompt words, the large model can be guided to generate a high-precision list of instructions to be executed. The business system binds the generated instruction identification prompt words to the user device, so that when the user triggers semantic interaction subsequently, the instruction content can be quickly extracted, and the user instruction can be quickly recognized and responded to.

[0085] In some embodiments, as Figure 6 shown, after the business system performs prompt word binding and caching, it returns the registration result to Launcher. Launcher marks that the binding of the Xiaoyi Housekeeper application instructions is completed, supports semantic interaction, and starts the wake word listening mechanism to wait for the user to trigger semantic interaction through the wake word. As Figure 7 shown, when the user mentions the wake word by voice or text, the cooperative APP enables the IPC mechanism to prepare to receive the user's text / audio data. Then the user expresses the control intention of the device through natural language. After receiving the data of the control intention, the cooperative APP transmits it to Launcher through the IPC mechanism. After receiving the data of the control intention, Launcher will call the semantic interaction interface provided by the business system, and the interface response is returned in the form of a data stream to support real-time message passing.

[0086] In step S220 of some embodiments, as Figure 8 shown, in response to the second request for calling the semantic interaction interface, the business system extracts the list of second instruction identification prompt words related to the user according to information such as the user identification and the intelligent terminal identification.

[0087] In step S230 of some embodiments, as Figure 8As shown, the business system invokes a large language model for intent analysis. The large language model can identify prompt words based on instructions, understand the semantics of the user's second instruction recognition prompt words, generate a list of instructions to be executed, and at the same time determine whether the user's intent can be supported. If the user's intent is not supported, a response message will be generated and fed back to the user, such as: Your request is not currently supported. If the user's intent is supported, the operation to be executed will be deduced with reference to the instruction registration information, and then the operation to be executed will be converted into an instruction object to be executed that conforms to the execution specifications.

[0088] Refer to Figure 7 、 Figure 8 , in some embodiments, when the user uses a cooperative APP (such as a game application), the launcher's instruction listening can be activated through a wake-up word, and the user's natural language expression can be transmitted through the IPC mechanism. The launcher will request the business system through the semantic interaction interface. The business system extracts relevant information registered for application instructions based on the user identification, then determines the intent expressed by the user based on the semantic understanding ability of the large language model, and abstracts relevant control information from it as much as possible, and then generates instruction objects with different priorities. At the same time, the information deduced from the user portrait can be used to assign different weights to instructions with the same priority. Exemplarily, taking the user's expression "Set the air conditioner to cooling, adjust the wind speed to gear 2, then turn on the fan and turn on the light" as an example, through the large model, it can be determined that the user's expression intent is a smart device control function, and this function is known to be reported by the Xiaoyi Butler APP for application instructions. Then, referring to the instruction registration information, the operations to be executed are deduced to include: turning on the air conditioner, setting the air conditioner mode to cooling and adjusting the wind speed to gear 2; turning on the fan; turning on the light. At the same time, the large model can refer to information such as the time of the user's expression or historical behavior habits, etc., to make decisions on different priorities and weights for the operations to be executed. Suppose the operations to be executed after decision are adjusted to turn on the light; turn on the fan; turn on the air conditioner, set the air conditioner mode to cooling and adjust the wind speed to gear 2. Among them, the light operation is the highest priority instruction, and the fan and air conditioner operations are secondary priority instructions. Among the air conditioner operations with the same priority, the operation of turning on the air conditioner has the highest weight and the operation of adjusting the wind speed gear has the lowest weight. The large model generates a corresponding list of instructions to be executed according to the operations to be executed after decision, and these instruction lists will be delegated to the instruction scheduling mechanism for processing.

[0089] In step S300 of some embodiments, as Figure 9 shown, the list of instructions to be executed is delegated by the business system to the instruction scheduling mechanism for processing. The instruction scheduling mechanism includes two different execution methods: cloud-side execution and end-side execution, and the execution method is determined during the application instruction registration. Optionally, cloud-side execution can be a process in which the instruction scheduling mechanism triggers the instruction to be executed on the cloud side and responds with the operation result to the end side. End-side execution can be a process in which the instruction scheduling mechanism issues the instruction list for execution on the end side and the end side triggers the instruction execution.

[0090] In some embodiments, step S300 may include but is not limited to steps S310 to S320:

[0091] Step S310, according to the to-be-executed instruction list, call the instruction scheduling mechanism to send a high-priority instruction list and obtain a response message;

[0092] Step S320, according to the callback address in the response message, call the instruction scheduling mechanism, and orderly execute the low-priority instruction list through the asynchronous thread mechanism to obtain a first operation result.

[0093] In steps S310 to S320 of some embodiments, refer to Figure 9 、 Figure 10 , according to the obtained to-be-executed instruction list, the business system calls the instruction scheduling mechanism to send and execute the high-priority instruction list to the intelligent device. After the intelligent device executes the high-priority instruction, it returns the execution result. The instruction scheduling mechanism creates an instruction operation result callback address and returns a response message such as the execution result, callback address, and activated application id to the business system. The business system timely feeds back the operation to the user. When there are multiple instruction objects to be executed, after creating the instruction operation result callback address, the low-priority instructions are orderly executed through the asynchronous thread mechanism. After the instructions are executed, the first operation result is cached, and the Launcher broadcasts / displays the first operation result. Optionally, the terminal side establishes an SSE request connection with the callback address through an asynchronous mechanism, and when the execution operation result is updated, it will be returned in the form of streaming data. The Launcher broadcasts / displays the operation result returned in a streaming manner.

[0094] In some embodiments, step S300 may further include step S330, according to the to-be-executed instruction list, call the user device of the target user to trigger the instruction execution process to obtain a second operation result. In step S330 of some embodiments, according to the to-be-executed instruction list obtained by the business system, the instruction list is sent to the terminal side for execution through the instruction scheduling mechanism and the instruction execution is triggered by the terminal side to obtain a second operation result, and the Launcher broadcasts / displays the second operation result.

[0095] Refer to Figure 2 , in the embodiments of the present invention, there are four core modules: intelligent terminal, business system, large model service, and instruction scheduling mechanism. The following is an introduction to each core module:

[0096] (1) Intelligent terminal module: It refers to a device with computing capabilities and a rich user interface, capable of running multiple applications and providing a rich user interaction experience. The intelligent terminal is equipped with a powerful processor, large-capacity storage, and a high-resolution screen, and can perform complex data processing and graphics display tasks. The intelligent terminal usually runs an operating system (such as Android, iOS, Windows, etc.) and supports the installation and operation of third-party applications. Common intelligent terminals include: smartphones, tablets, smart TVs, smart speakers, etc. In the intelligent terminal, a launcher application needs to be installed and continuously kept active. The launcher is responsible for interacting with the business system and also serves as a control end to coordinate the execution of instructions and the delivery of asynchronous messages for each application. The launcher application is an application developed and maintained by this system. Different applications can communicate through the IPC mechanism to share data, send signals, or cooperate to complete tasks. In order to register and execute application instructions, other cooperative applications (referred to as cooperative APPs) need to perform cross-process communication with the launcher application through the inter-process communication protocol (IPC), report the application status, and support the activation of the instruction listening mechanism of the launcher by the wake-up word and transmit the user's text or audio data to the launcher when the application status is active. To simplify the access of cooperative APPs, the system of the present invention provides a MiniSDK, which integrates operation codes such as waking up the instruction listening and performing semantic interaction.

[0097] (2) Business System Module: It includes, but is not limited to, units such as intelligent terminal registration, application instruction registration, user semantic interaction, large semantic model configuration, user information, and historical behavior portraits. The business system is an indispensable part of the system. It is not only a registration management center and a configuration management center but also responsible for providing interfaces for semantic interaction. The definitions and functions of each unit are briefly explained as follows: The intelligent terminal registration unit refers to reporting information related to intelligent terminals installed with the launcher application (such as device model, device serial number, IP address, etc.) to the business system, so as to ensure that the business system can identify and manage the device and provide personalized services (such as message push, weather forecast, etc.); The application instruction registration unit refers to reporting the list of executable instructions supported by each application program and preset conditions, etc. to the business system for storage, so that subsequent large model services can match relevant manipulation intentions and generate corresponding instructions; The user semantic interaction unit is an instant response stream data interface developed based on Server-Sent Events (abbreviation: SSE, an HTTP technology that allows the server to push data to the client). All text or audio interactions of users will trigger a request for the semantic interaction interface through the launcher for processing, and the response results will be quickly returned in the form of a data stream to enhance the user's experience perception of operations; User information and historical behavior portraits refer to the system collecting and analyzing users' basic information and historical behavior data, analyzing users' behavior habits and preferences, generating user behavior portraits, which can be used to specify personalized instruction processing and instruction scheduling strategies, thereby meeting the needs and preferences of different users.

[0098] (3) Large Model Service Module: It refers to large language model (abbreviation: LLM) technology, usually natural language processing models with large-scale parameters and computing capabilities. These models are usually constructed by deep neural networks and have billions or even hundreds of billions of parameters. They can show excellent performance in natural language processing tasks. In the embodiments of the present invention, it supports scheduling and using large models with different parameter levels and different categories, and can be configured and switched according to the differences in the recognition degrees of different application business instructions, ensuring the effective utilization of computing power resources. At the same time, when the application instruction list is reported to the business system, it will generate an initial version of the corresponding prompt words with the help of the capabilities of the large model for business and developers to review, further improving work efficiency. When the launcher conducts semantic interaction, the large model service is responsible for extracting the possible expected operation descriptions in the user's natural language and accurately determining the user's intention, and then generating corresponding instructions or reply words according to the context information of short-term interaction. In addition, it can combine long-term memory to evaluate the priority order of instruction scheduling.

[0099] (4) Instruction Scheduling Mechanism Module: The core responsibility of this module is to schedule instructions with different priorities in an orderly manner and ensure that the asynchronous notifications of instruction results can be accurately delivered. This module supports two instruction execution modes: edge-side execution and cloud-side execution. When a user issues an instruction through natural language and it is parsed by the large model service, a list of instructions to be executed is generated. The identification of instructions is uniformly configured and managed by the business system to ensure the consistency and accuracy of the identification. The format of the instruction to be executed is in the form of a JSON object. The mandatory field "payload" is the payload of the information, where the "action" field is the instruction identification. The format of the "payload" instruction is also different, and business processing needs to be carried out according to different instruction identifications. In addition, in the instruction format, the mandatory field "type" represents three different instruction priorities: immediate, interactive, and active. Among them, the immediate instruction has the highest priority due to its urgency, while the active instruction has a relatively lower priority. For example, the warning detected by the intelligent camera belongs to the immediate type; the content of the follow-up question when there are multiple instruction manipulation objects that cannot be determined belongs to the interactive type; the operations under the currently active application belong to the active type. To further improve the flexibility and intelligence of scheduling, the instruction scheduling mechanism module adopts a multi-level feedback queue mechanism, which can schedule and assign according to different instruction priorities. At the same time, according to the user's historical behavior habits, this module can assign different weights to the instructions in the same priority queue, so as to achieve more refined instruction scheduling. For example, taking the user's question "Turn on the light" but there are multiple devices as an example, the instruction format for generating the follow-up question content is: {"type": "interactive", "payload": {"action": "speak", "payload": {"content": "You have two devices. Do you want to turn on the light in the bedroom or the living room?", "listen": true, "ttsFinish": true, "ttsPlay": true}}}.

[0100] Reference Figure 3 , the device or system module object descriptions involved in the embodiments of the present invention may include: smart speaker, Launch APP, XiaoYi Butler APP, chess and card APP, business system, large model service, instruction scheduling mechanism module, smart devices (such as smart air conditioner, smart gateway, etc.). It is known that Launcher, XiaoYi Butler (APP), and chess and card (APP) are application programs installed on the smart speaker. Through the XiaoYi Butler APP, smart home devices can be bound and device control operations can be provided. Exemplarily, taking the registration of XiaoYi Butler application instructions and the process of the user using other applications (taking the chess and card APP as an example) as an example, the relevant processes such as semantic interaction through the launcher and completing instruction scheduling execution are as follows:

[0101] Step S1: Filing of relevant information. Before accessing this system, filing is required to ensure the compliance of application instructions and the security of user data, which is reviewed by a dedicated staff. After passing the review, a unique application code and an application key are assigned. The unique application code is the sole identifier of the application's identity, and the application key is used for the secure encryption of reported data. In this implementation example, the Xiaoyi Butler APP is used as an example for application filing.

[0102] Step S2: The end - side application starts and completes the process of registering application instructions. After the user starts the Xiaoyi Butler APP, user authentication and login are first performed. At this time, the Xiaoyi Butler APP is in an active state and notifies the launcher. After completing the authentication, the Xiaoyi Butler uses the IPC mechanism to transfer relevant data with the launcher to complete the registration of application instructions, and the business system will bind user information, device information, and application instruction information. At the same time, a unique identification code for the device is assigned to the user, which is used to store all the information generated during the subsequent interactions of the user on this device. For example, when controlling a smart air conditioner, functions such as on - off status, mode adjustment, and wind speed gear adjustment are supported. Different functions can have different pre - condition requirements and control instruction formats. For example: there is no pre - condition for the on - off status, the pre - condition for mode adjustment is to confirm the on - off status, and the pre - condition for wind speed gear adjustment is to confirm the mode adjustment. To ensure the accuracy of the execution instructions generated by the subsequent large - model, the description and instruction format of the supported functions must be clear and definite. Taking the air conditioner on - off as an example, the instruction format can be defined as JSON format, and fields such as device information, location information, execution parameters, and execution conditions can be defined in the format. Taking "turn on the air conditioner in the bedroom on the 2nd floor in one hour" as an example, the instruction is:

[0103] {"deviceType":"switch","deviceName":"air conditioner","location":{"zone":"2nd floor","room":"bedroom"},"cmdParams":[{"paramErrCode":"SET_POWER","paramValue":"1"}],"timeTrigger":{"afterTime":"one hour"}}.

[0104] Step S3: The process of the edge side triggering the semantic interaction and the fusion large model to determine the intention, parse the semantics, and generate the instruction list. When the user uses the cooperative APP (i.e., the chess and card APP), the user can activate the instruction listening of the launcher through the wake-up word and transmit the natural language expression of the user through the IPC mechanism. The launcher will request the business system through the semantic interaction interface. The business system extracts the relevant information registered for the application instructions based on the user identification, then determines the intention of the user's expression based on the semantic understanding ability of the large language model, and abstracts the relevant control information from it as much as possible, and then generates instruction objects with different priorities. At the same time, the information deduced from the user portrait can be used to assign different weights to the instructions with the same priority. Taking the user's expression "Turn on the air conditioner for cooling and adjust the wind speed to gear 2. Then turn on the fan and turn on the light" as an example, the large model can determine that the user's expression intention is the intelligent device control function, and this function is known to be reported by the XiaoYi Housekeeper APP for application instructions. Then, referring to the instruction registration information, the deduced operations to be executed include: turning on the air conditioner, adjusting the air conditioner mode to cooling and adjusting the wind speed to gear 2; turning on the fan; turning on the light. At the same time, the large model can refer to information such as the user's expression time or historical behavior habits, etc., to decide different priorities and weights for the operations to be executed. Suppose the operations to be executed after decision are adjusted to turn on the light; turn on the fan; turn on the air conditioner, adjust the air conditioner mode to cooling and adjust the wind speed to gear 2. Among them, the light operation is the highest priority instruction, and the fan and air conditioner operations are the secondary priority instructions. Among the air conditioner operations with the same priority, the operation of turning on the air conditioner has the highest weight and the operation of adjusting the wind speed gear has the lowest weight. The large model generates the corresponding instruction list to be executed according to the operations to be executed after decision, and these instruction lists will be delegated to the instruction scheduling mechanism for processing. The instruction scheduling mechanism includes two different ways of cloud-side execution and edge-side execution of instructions, and the execution method is determined when the application instructions are registered.

[0105] Step S4 (optional): The process of triggering the execution of instructions on the cloud side through the instruction scheduling mechanism and responding the operation result to the edge side. The instruction scheduling mechanism will schedule the instructions according to the priority and weight of the instructions. For instructions with the same priority, the higher the weight, the earlier the execution order. When executing instructions on the cloud side, first execute the high-priority instruction object (abbreviation: priority instruction). After the priority instruction is executed, create an instruction operation result callback address, and return information such as the execution result, callback address, and activated application id, etc., to immediately feedback the execution result to the user and reduce the user's perception delay. And the lower-priority instruction objects enter the instruction queue to be executed, and the execution results are cached asynchronously. After the edge side obtains the instruction operation result callback address, it will establish an SSE request connection through the asynchronous mechanism, and this interface supports returning the operation result of the instruction in a streaming manner.

[0106] Step S5 (Optional): The process of issuing an instruction list for execution on the terminal side and triggering the execution of instructions on the terminal side through an instruction scheduling mechanism. The instruction scheduling mechanism will schedule instructions according to the priority and weight of the instructions. For instructions with the same priority, the higher the weight, the earlier the issuing order. When executing instructions on the terminal side, the launcher needs to transmit the instruction list to be executed to the XiaoYi Housekeeper APP through the IPC mechanism, and the XiaoYi Housekeeper triggers the custom instruction execution process, such as calling the system infrared capability to perform operations or transmitting operations issued via the smart gateway device, etc.

[0107] As Figures 4 to 11 shown, the respective sub-steps in the above steps S1 to S5 are introduced as follows:

[0108] Step S1:

[0109] 1.1. Application information and instruction intent filing. Developers need to submit the basic information of the application and the instruction intent, including the application name, function description, supported instruction numbers and descriptions, etc., so that the model can understand the functions and instruction scope of the application. Assume that the function filed by the XiaoYi Housekeeper APP in this case is the control of smart home devices.

[0110] 1.2. Qualification review. The business system needs to conduct a qualification review on the submitted application information to ensure that the application complies with the platform's specifications and standards.

[0111] 1.3. Allocate a unique application code and key. After passing the review, a unique code and key will be allocated to the application for subsequent identity verification and data encryption.

[0112] Step S2:

[0113] 2.1. Login. The user logs in to the XiaoYi Housekeeper APP through the Tianyi account.

[0114] 2.2. Authentication. The XiaoYi Housekeeper authenticates the user.

[0115] 2.3. Obtain user-bound device information. After user authentication, asynchronously obtain the information of the bound smart devices, including device type, device ID, etc.

[0116] 2.4. Information reporting. The XiaoYi Housekeeper APP transmits data such as encrypted user information, descriptions of bound devices, their function instructions, and format requirements to the launcher APP through the IPC mechanism.

[0117] 2.5. Call the instruction registration interface. The Launcher integrates the received data and the basic information of the smart terminal, and calls the application instruction registration interface provided by the business system.

[0118] 2.6. Authentication. The business system authenticates the requests to the instruction registration interface, verifying the legitimacy and security of the requests.

[0119] 2.7. Information Binding. The business system binds the user with the information reported by the instruction registration interface.

[0120] 2.8. Invoking the Large Model. Based on the reported information such as the bound device description, its function instruction description, and format requirements, etc., with the help of the semantic abstraction ability of the large language model, corresponding recognizable intent types and their application instructions are generated respectively, and integrated into a complete instruction recognition prompt. The prompt is the core tool for the interaction of the large language model. Through appropriately abstracted instruction recognition prompts, the large model can be guided to generate a high-precision list of instructions to be executed.

[0121] 2.9. Prompt Binding and Caching. The business system binds the generated instruction recognition prompt with the user device. When the user triggers semantic interaction, it can be quickly retrieved.

[0122] 2.10. Marking Application Binding. Launcher marks the completion of the binding of the Xiaoyi Butler application instructions, supporting semantic interaction.

[0123] 2.11. Starting the Wake Word Listening Mechanism. Start the wake word listening mechanism and wait for the user to trigger the semantic interaction step S3 through the wake word:

[0124] 3.1. Mentioning the Wake Word. The user mentions the wake word through voice or text.

[0125] 3.2. Enabling the IPC Mechanism. The cooperative APP enables the IPC (Inter-Process Communication) mechanism to prepare for receiving user text / audio data.

[0126] 3.3. Device Control Expression. Suppose the user expresses the control intention for the device through natural language, such as "Set the air conditioner to cooling and the wind speed to gear 2. Then turn on the fan and turn on the light", and after receiving the data, it is transmitted to the launcher through the IPC mechanism.

[0127] 3.4. Invoking the Semantic Interaction Interface (SSE). The launcher invokes the semantic interaction interface provided by the business system, and the interface response is returned in the form of a data stream, supporting real-time message passing.

[0128] 3.5. Extracting User Instruction Recognition Prompts. The business platform extracts the list of user-related instruction recognition prompts based on information such as the user identification and the intelligent terminal identification, etc.

[0129] 3.6. Call the big model. The business system calls the big language model to perform intent analysis. The big language model can recognize prompt words based on instructions, understand user semantics, and determine whether the user intent is supported. If the user intent is not supported, a reply is generated to the user, such as: Your request is not supported at the moment. If the user intent is supported, the pending operation is deduced based on the instruction registration information, and then the pending operation is converted into a pending instruction object that meets the execution specifications.

[0130] 3.7. Set the priority and weight of the pending instruction object. It supports setting different priorities and weights for pending instructions based on information such as user statement time, historical behavior habits or default configuration.

[0131] Step S4:

[0132] 4.1. Execute the high priority instruction list. High priority instructions are executed first when executing on the cloud side.

[0133] 4.1. Create a callback address for the result of the instruction operation. After the high-priority instruction is executed, create a callback address for the result of the instruction operation, and return the response message such as the execution result, callback address, and activated application ID, so that the user can get timely operation feedback. The semantic interaction interface is disconnected, and the user can perform other operations without waiting for all instructions to be executed, which improves the user experience.

[0134] 4.1. Orderly execution of low-priority instruction list. If there are multiple instruction objects to be executed, after creating the instruction operation result callback address, the low-priority instructions are executed in order through the asynchronous thread mechanism, and the operation results are cached after the instruction execution is completed.

[0135] 4.1. Report / display the first operation result. The Launcher reports / displays the first operation result.

[0136] 4.5 (optional) Maintain connection with the callback address. The client establishes an SSE request connection with the callback address through an asynchronous mechanism, and when the operation result is updated, it will be returned in the form of streaming data.

[0137] 4.6 (optional) Report / display the operation results. The Launcher reports / displays the operation results returned in streaming.

[0138] Step S5:

[0139] 5.1. List of instructions to be executed.

[0140] 5.2. Sending pending instructions (IPC protocol). The launcher needs to transmit the pending instruction list to the Xiaoyi Guanjia APP through the IPC mechanism.

[0141] 5.3. Obtain and execute instructions in order. After the Xiaoyi Butler APP obtains an instruction, it executes the instruction in order. After the operation is completed, the result is returned to the launcher.

[0142] 5.4. Announce / display the operation result. The launcher announces / displays the operation result.

[0143] The embodiment of the present invention also provides a multi-application instruction registration and large model-driven instruction scheduling device, which can implement the above-mentioned multi-application instruction registration and large model-driven instruction scheduling method. The device includes:

[0144] The first module is used to perform an information binding operation on the target user in response to a first request for calling the instruction registration interface, and obtain first information data;

[0145] The second module is used to, based on the first information data, in response to a second request for calling the semantic interaction interface, call a large model to perform intention analysis on the target user, and obtain a list of instructions to be executed;

[0146] The third module is used to call an instruction scheduling mechanism to implement instruction scheduling according to the list of instructions to be executed.

[0147] In some embodiments, a multi-application instruction registration and large model-driven instruction scheduling device may further include:

[0148] The fourth module is used to, in response to a filing request for a target application in the user device of the target user, perform qualification review on the target application;

[0149] The fifth module is used to, when the result of the qualification review is passed, assign a unique code and a key to the target application;

[0150] Wherein, the unique code and the key are used for identity authentication and data encryption.

[0151] It can be understood that the content in the above method embodiments is applicable to the device embodiments of the present invention. The functions specifically implemented by the device embodiments of the present invention are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0152] The embodiment of the present invention also provides an electronic device, which includes a processor and a memory. The memory stores a computer program, and when the processor executes the computer program, it implements the above-mentioned multi-application instruction registration and large model-driven instruction scheduling method. The electronic device can be any intelligent terminal including a tablet computer, a vehicle-mounted computer, etc.

[0153] It can be understood that the content in the above method embodiments is applicable to the device embodiments of the present invention. The functions specifically implemented by the device embodiments of the present invention are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0154] Reference Figure 12 , Figure 12 schematically shows the hardware structure of an electronic device according to another embodiment. The electronic device includes:

[0155] A processor 401, which can be implemented in ways such as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present invention;

[0156] A memory 402, which can be implemented in forms such as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 402 can store an operating system and other application programs. When implementing the technical solutions provided by the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 402 and are called by the processor 401 to execute a method for multi-application instruction registration and large model-driven instruction scheduling according to the embodiments of the present invention;

[0157] An input / output interface 403, which is used to implement information input and output;

[0158] A communication interface 404, which is used to implement communication and interaction between the device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WI FI, Bluetooth, etc.);

[0159] A bus 405, which transmits information between various components of the device (such as the processor 401, the memory 402, the input / output interface 403, and the communication interface 404);

[0160] Among them, the processor 401, the memory 402, the input / output interface 403, and the communication interface 404 are communicatively connected to each other inside the device through the bus 405.

[0161] An embodiment of the present invention also provides a computer-readable storage medium storing a computer program, which when executed by a processor implements the above-mentioned method for registering multi-application instructions and scheduling instructions driven by a large model.

[0162] It can be understood that the content in the above method embodiments is applicable to this storage medium embodiment. The functions specifically implemented by this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those in the above method embodiments.

[0163] An embodiment of the present invention also provides a computer program product or a computer program. The computer program product or the computer program includes computer instructions stored in a computer-readable storage medium. The processor of the computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to enable the computer device to execute the above-mentioned method for registering multi-application instructions and scheduling instructions driven by a large model.

[0164] In summary, the method, device, equipment, and medium for registering multi-application instructions and scheduling instructions driven by a large model according to the embodiments of the present invention first maintain a set of application registration and cross-application instruction interaction mechanisms. Other cooperative applications interact with the business system through the launcher application. After the cooperative application APP completes the filing of relevant information, the registration process of registering the application instruction information supported by the user reported actively by the cooperative application on the terminal side is more flexible. At the same time, it integrates a large language model, which can efficiently perform user semantic analysis and executable instruction deduction. In addition, a multi-level instruction scheduling mechanism is introduced. The instruction scheduling mechanism will perform an orderly scheduling of instructions according to the priority and weight ratio of the instructions, which can efficiently perform instruction scheduling among multiple application programs and enhance the intelligent interaction experience of users on intelligent terminals. Specifically, it has the following advantages:

[0165] 1. The embodiment of the present invention maintains a set of application registration and cross-application instruction interaction mechanisms, which are used to break through the limitations of the prior art in identifying user instructions in a multi-application program environment, and can significantly improve the response speed of requests, optimize resource allocation, and bring a more intelligent interaction experience to users.

[0166] 2. The embodiment of the present invention integrates advanced large language model technology, deeply analyzes by referring to the user's historical behavior and current context, so as to accurately understand and parse the user's complex natural language, determine the user's intention and deduce executable operations. Then, based on the intention analysis, it can intelligently generate application instructions to be executed and decide different priorities and weight ratios for them, enabling complex instruction execution tasks to be completed efficiently and orderly.

[0167] 3. The embodiments of the present invention also introduce a multi-level feedback instruction processing mechanism. This mechanism can ensure that high-priority instructions are executed promptly, while ensuring that low-priority instructions are completed within a reasonable time domain.

[0168] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order mentioned in the operation diagrams. For example, depending on the functions / operations involved, two consecutive blocks shown may actually be executed substantially simultaneously or the blocks can sometimes be executed in the reverse order. In addition, the embodiments presented and described in the flowcharts of the present invention are provided by way of example for the purpose of providing a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logical flows presented herein. Alternative embodiments are foreseeable, where the order of various operations is changed and the sub-operations described as part of a larger operation are executed independently.

[0169] In addition, although the present invention is described in the context of functional modules, it should be understood that unless otherwise stated to the contrary, one or more of the functions and / or features described may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It can also be understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present invention. More precisely, considering the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the module will be understood within the ordinary skills of an engineer. Therefore, those skilled in the art can implement the present invention as set forth in the claims without undue experimentation. It can also be understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0170] If the described function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0171] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0172] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.

[0173] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0174] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0175] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

[0176] The above has specifically described the preferred embodiments of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the present invention.

Claims

1. A multi-application instruction registration and large model-driven instruction scheduling method, characterized in that, It includes the following steps: In response to the first request for calling the instruction registration interface, perform an information binding operation on the target user to obtain the first information data; Based on the first information data, in response to the second request for calling the semantic interaction interface, call a large model to perform intent analysis on the target user to obtain a list of instructions to be executed; According to the list of instructions to be executed, call an instruction scheduling mechanism to implement instruction scheduling.

2. The multi-application instruction registration and large model-driven instruction scheduling method according to claim 1, wherein, Before the step of performing an information binding operation on the target user in response to the first request for calling the instruction registration interface to obtain the first information data, the following steps are further included: In response to the filing request of the target application in the user device of the target user, perform qualification review on the target application; When the result of the qualification review is passed, assign a unique code and a key to the target application; Wherein, the unique code and the key are used for identity authentication and data encryption.

3. A multi-application instruction registration and large model-driven instruction scheduling method according to claim 1, characterized in that The step of performing an information binding operation on the target user in response to the first request for calling the instruction registration interface to obtain the first information data includes the following steps: In response to the first request, verify the legality and security of the first request to obtain a verification result; Obtain the second information data reported by the instruction registration interface; When the verification result is legal and secure, bind the target user to the second information data to obtain the first information data.

4. A multi-application instruction registration and large model-driven instruction scheduling method according to claim 1, characterized in that The step of, based on the first information data, in response to the second request for calling the semantic interaction interface, calling a large model to perform intent analysis on the target user to obtain a list of instructions to be executed includes the following steps: According to the first information data, call a large model to generate a first instruction recognition prompt word and bind the first instruction recognition prompt word to the user device of the target user; In response to the second request, extract the second instruction recognition prompt word input by the user device; Based on the first instruction recognition prompt word, call the large model to perform intent analysis on the second instruction recognition prompt word to obtain the list of instructions to be executed.

5. The method for multi-application instruction registration and large model-driven instruction scheduling according to claim 1, characterized in that The step of calling an instruction scheduling mechanism to implement instruction scheduling according to the list of instructions to be executed includes the following steps: According to the list of instructions to be executed, call an instruction scheduling mechanism to send a list of high-priority instructions to obtain a response message; According to the callback address in the response message, call an instruction scheduling mechanism to orderly execute a list of low-priority instructions through an asynchronous thread mechanism to obtain a first operation result.

6. A method for multi-application instruction registration and large model-driven instruction scheduling according to claim 1, characterized in that After the step of calling an instruction scheduling mechanism to implement instruction scheduling according to the list of instructions to be executed, the following steps are further included: According to the list of instructions to be executed, call the user device of the target user to trigger an instruction execution process to obtain a second operation result.

7. A multi-application instruction registration and large model-driven instruction scheduling device, characterized in that, It includes: A first module for performing an information binding operation on the target user in response to the first request for calling the instruction registration interface to obtain the first information data; A second module, configured to perform intent analysis on the target user by invoking a large model based on the first information data in response to a second request for invoking a semantic interaction interface, so as to obtain a list of instructions to be executed; A third module, configured to implement instruction scheduling by invoking an instruction scheduling mechanism according to the list of instructions to be executed.

8. The instruction scheduling device for multi-application instruction registration and large model-driven according to claim 7, wherein, It further includes: A fourth module, configured to perform qualification review on the target application in response to a filing request of the target application in the user device of the target user; A fifth module, configured to assign a unique code and a key to the target application when the result of the qualification review is passed; Wherein, the unique code and the key are used for identity authentication and data encryption.

9. An electronic device, characterized in that, It includes a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The storage medium stores a program, and the program is executed by the processor to implement the method according to any one of claims 1 to 6.

Images