Task processing method and device, intelligent terminal and storage medium

Abstract

The invention relates to the technical field of information processing, in particular to a task processing method and device, an intelligent terminal and a storage medium. The method comprises the steps of obtaining a target task of a user; the target task is input into a target task processing model, the target task processing model is used for processing the target task, the target task processing model is a pre-trained large language model, and the processing step of the target task processing model on the target task comprises the steps that the target task is disassembled, and the disassembled target task is obtained; determining a subtask corresponding to the target task and subtask information of the subtask; determining an intelligent agent matched with the subtask according to the subtask information; and calling an interface of the matched intelligent agent to process the subtask. By adopting the method, the AI agent meeting the requirement can be intelligently and quickly selected for the user, and the task processing efficiency of the user is improved.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a task processing method, device, intelligent terminal and storage medium. Background Art

[0002] AI agent is an intelligent agent that can perceive the environment, understand autonomously, make decisions and perform actions. It can perceive the environment, collect information, reason, plan and execute tasks to achieve preset goals or instructions.

[0003] At present, various enterprises and institutions focus on the application of AI agents in closed scenarios and vertical fields. In fact, the application fields of AI agents in the existing market are extensive and diverse. At the same time, due to the differences in deployment methods, this creates obstacles for users to choose and use AI agents. It is difficult for users to quickly select AI agent products that meet their actual needs among the numerous AI agent products.

[0004] In view of this, how to quickly select AI agents that meet users' needs and improve the efficiency of user task processing is a problem that needs to be solved at present. Summary of the invention

[0005] In view of this, the embodiments of the present application provide a task processing method, device, intelligent terminal and storage medium, which can help users quickly select an AI agent that meets their needs, improve the efficiency of user task processing, and optimize the duration of task processing.

[0006] A first aspect of an embodiment of the present application provides a task processing method, the task processing method comprising:

[0007] Get the user's target task;

[0008] The target task is input into a target task processing model, and the target task is processed by the target task processing model, wherein the target task processing model is a large language model that has been pre-trained, and the target task processing model processes the target task in the following steps:

[0009] Decomposing the target task to determine the subtasks corresponding to the target task and the subtask information thereof;

[0010] Determine an agent matching the subtask according to the subtask information;

[0011] Call the interface of the matching agent to process the subtask.

[0012] In a possible implementation manner of the first aspect, the decomposing the target task to determine subtasks corresponding to the target task and the subtask information thereof includes:

[0013] Decomposing the target task, and constructing an event directed graph according to the decomposition results, wherein the event directed graph is composed of vertices and edges connecting the vertices, wherein the vertices of the event directed graph represent task descriptions of subtasks of the target task, and the edges of the event directed graph represent dependency relationships between two subtasks;

[0014] According to the vertex set and the edge set of the event directed graph, the task description of the subtask corresponding to the target task and the dependency relationship of the subtask are determined.

[0015] In a possible implementation manner of the first aspect, the subtask information includes a task description of the subtask, and determining an agent matching the subtask according to the subtask information includes:

[0016] According to the task description of the subtask, searching in the agent repository for an agent matching the subtask, wherein the agent repository includes a pre-organized and maintained set of agent information;

[0017] If no intelligent agent matching the subtask is found, the subtask is disassembled, and the intelligent agent matching the disassembly result is continuously searched in the intelligent agent repository until a matching intelligent agent is found, and the disassembly is completed.

[0018] In a possible implementation manner of the first aspect, the subtask information includes dependencies between subtasks, and the calling of the interface of the matched agent to process the subtasks includes:

[0019] According to the subtask information and the identifier of the matched agent, an agent directed graph is constructed, wherein the vertices of the agent directed graph represent the agent information of the subtask and the agent to be called, and the edges of the agent directed graph represent the dependency relationship between two agents, wherein the dependency relationship between the agents is determined based on the dependency relationship between the matched subtasks;

[0020] Based on the agent directed graph, the interfaces of the matching agents are called in sequence to process the subtasks.

[0021] In a possible implementation of the first aspect, the step of sequentially calling interfaces of matched agents to process the subtasks based on the agent directed graph includes:

[0022] When the agent to be called by the subtask has an associated agent that it depends on, construct a prompt based on the processing result of the associated agent on the matching subtask and the task description of the subtask corresponding to the agent to be called;

[0023] The prompt is used to guide the agent to process the subtask.

[0024] In a possible implementation manner of the first aspect, the task processing method further includes:

[0025] Acquire a task decomposition diagram data set, wherein the task decomposition diagram data set includes sample annotation data obtained by decomposing a sample task, and the sample annotation data includes task descriptions of subtasks in the sample task and dependencies between subtasks;

[0026] Using the task decomposition graph dataset to train the large language model's task decomposition capabilities, and the ability to determine and call agents that match subtasks;

[0027] When the training error of the large language model meets the preset requirements, the target task processing model is obtained.

[0028] In a possible implementation manner of the first aspect, after calling the interface of the matched agent to process the subtask, the method further includes:

[0029] Monitoring the processing status and progress of calling the interface of the matched agent to process the subtask;

[0030] When an abnormality is detected in the processing status and progress, the abnormality is processed according to a preset processing mechanism.

[0031] A second aspect of an embodiment of the present application provides a task processing device, the task processing device comprising:

[0032] A task acquisition unit, used to acquire the user's target task;

[0033] A task processing unit is used to input the target task into a target task processing model, and use the target task processing model to process the target task, wherein the target task processing model is a pre-trained large language model, and the target task processing model processes the target task in the following steps: decomposing the target task to determine the subtasks corresponding to the target task and their subtask information; determining an intelligent agent matching the subtask based on the subtask information; and calling the interface of the matching intelligent agent to process the subtask.

[0034] The third aspect of an embodiment of the present application provides an intelligent terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the task processing method provided in the first aspect of the embodiment of the present application are implemented.

[0035] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the task processing method provided in the first aspect of the embodiment of the present application.

[0036] A fifth aspect of the embodiments of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device executes the steps of the task processing method described in the first aspect of the embodiments of the present application.

[0037] In the embodiment of the present application, the user's target task is obtained; the target task is input into the target task processing model, and the target task is processed by the target task processing model, the target task processing model disassembles the target task, determines the subtasks corresponding to the target task and its subtask information; then, according to the subtask information, determines the agent matching the subtask, and then calls the interface of the matching agent to process the subtask. The present application can intelligently screen agents that meet the needs of users, without the need for users to fully understand the agents released on the market, greatly improving the efficiency of user task processing and enhancing user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 is a flowchart of the implementation of the task processing method provided in the embodiment of the present application;

[0040] Figure 2 It is a specific implementation flow chart of obtaining a target task processing model in the task processing method provided in the embodiment of the present application;

[0041] Figure 3 It is a specific implementation flow chart of target task processing by a target task processing model in the task processing method provided in an embodiment of the present application;

[0042] Figure 3.1This is a schematic diagram of the target task decomposition in the task processing method provided in the embodiment of the present application;

[0043] Figure 3.2 This is a schematic diagram of subtask decomposition in the task processing method provided in an embodiment of the present application;

[0044] Figure 4 It is a specific implementation flow chart of determining subtasks and their subtask information in the task processing method provided in the embodiment of the present application;

[0045] Figure 5 It is a specific implementation flow chart of processing subtasks in the task processing method provided in the embodiment of the present application;

[0046] Figure 6 is a structural block diagram of a task processing device provided in an embodiment of the present application;

[0047] Figure 7 It is a schematic diagram of a smart terminal provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] In the following description, specific details such as specific device structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0049] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0050] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0051] It should be understood that each method embodiment of the present application provides a task processing method applicable to various types of smart terminals or servers that need to perform task processing, which may specifically include smart terminals such as mobile phones, tablet computers, wearable devices, laptops, and desktop computers. The present application embodiment does not impose any restrictions on the specific types of smart terminals and servers.

[0052] The task processing method provided by the present application is exemplarily described below in conjunction with specific embodiments.

[0053] Figure 1 The implementation process of the task processing method provided in the embodiment of the present application is shown, and the method process may include the following steps S101 to S102.

[0054] Step S101: Obtain the user's target task.

[0055] The target task is the task that the user currently wants to complete. The target task includes the task description of the task.

[0056] The form of the acquired user target task is not limited, that is, the target task can be in text form or in audio form. The user can propose the target task by inputting text or voice.

[0057] When the target task is in audio form, the intelligent terminal performs speech recognition on the audio and converts the target task in audio form into text form.

[0058] Step S102: input the target task into a target task processing model, and use the target task processing model to process the target task.

[0059] The target task processing model is a pre-trained large language model for task processing. Large language models (LLMs) are a type of natural language processing model based on deep learning. They learn language patterns, grammar and semantics by processing a large amount of text data, so that they can understand and generate human language. In this embodiment, the target task processing model is obtained by training the basic large language model as required.

[0060] As a possible implementation of this application, Figure 2 A specific implementation process of obtaining a target task processing model in the method embodiment provided in the embodiment of the present application is shown, and is described in detail as follows:

[0061] Step S201: Acquire a task decomposition diagram data set, wherein the task decomposition diagram data set includes sample annotation data obtained by decomposing a sample task, and the sample annotation data includes task descriptions of subtasks in the sample task and dependencies between subtasks.

[0062] The task decomposition diagram dataset is a sample dataset used to train a large language model in the embodiments of this application. In some implementations, data containing complex tasks and their decomposition steps are collected from various sources (such as user logs, online forums, task management platforms, etc.), and the collected complex tasks and their decomposition steps are annotated to obtain sample annotated data, wherein each decomposition step is a subtask of the sample task, and the task description of the subtask in the sample task and the dependency relationship between the subtasks are annotated. A task decomposition diagram dataset is constructed based on the sample annotated data.

[0063] In a possible implementation, the sample annotation data in the task decomposition diagram data set is preprocessed to process the sample annotation data into a format suitable for model training. Preprocessing includes data cleaning, deduplication, and format unification. By preprocessing the sample annotation data, low-quality, repeated, or irrelevant content is removed, the quality and diversity of the data are ensured, which is conducive to improving the efficiency of model training.

[0064] Step S202: Use the task decomposition graph dataset to train the large language model's task decomposition capability, and the capability to determine and call an intelligent agent that matches the subtask.

[0065] The task decomposition capability of the large language model is trained through supervised learning. The task decomposition capability specifically includes the identification of key information in the task, the division of subtasks, and the determination of the dependencies between subtasks. In the process of decomposing tasks, a directed graph between subtasks is constructed to reflect the dependencies between subtasks, which helps the model understand the order and conditions of task execution.

[0066] In some implementations, a large language model can be guided to decompose a task into a series of subtasks by constructing prompts. For example, a prompt is constructed: "Please decompose this task into executable steps, each of which is independent and executable, and at the same time, construct a directed graph to reflect the dependency relationship between the steps."

[0067] The ability of the large language model to determine and call the agent matching the subtask is trained by supervised learning or unsupervised learning. The ability of the large language model to determine the agent matching the subtask is specifically to retrieve and find the agent matching the subtask.

[0068] In this embodiment, an agent repository is constructed, in which the identifications and labeled descriptions of currently published agents are integrated, and the agents in the agent repository have completed registration. The labeled description includes information such as application areas, functions, prices, interfaces, etc. The agents in the agent repository are able to handle various types of subtasks. By integrating the interfaces of each agent in the agent repository in the large language model, the large language model can call the agent interface that matches the subtask to process the subtask. Before calling the agent interface to process the subtask, an agent directed graph is constructed, and the calling order of each agent interface is determined based on the agent directed graph to optimize the execution process.

[0069] In one possible implementation, in order to further improve the task decomposition capability of the large language model, a reinforcement learning mechanism is introduced to optimize the task decomposition capability of the large language model by simulating the task decomposition process, determining the intelligent agent that matches the subtask, and calling the intelligent agent interface to give rewards or penalties.

[0070] In a possible implementation, during the training phase, the error handling of the large language model is also trained, specifically training the errors and abnormalities that may occur during the large language model's interface calls to the intelligent agent, such as interface unavailable, data format error, etc.

[0071] Step S203: When the training error of the large language model meets the preset requirement, a target task processing model is obtained.

[0072] When the training error of the large language model meets the preset requirements, the processing results of all subtasks are integrated, and the integrated result is used as the final output result of the large language model. In some implementations, the training error can be calculated and determined according to a preset error function.

[0073] In this embodiment, the large language model selected for training may be an open source large language model, and supervised fine-tuning training is performed based on the open source large language model to obtain a target task processing model.

[0074] In a possible implementation, a performance evaluation is performed on the obtained target task processing model to evaluate the performance of the model in terms of task decomposition, dependency modeling, and agent interface calling. The training data set and training strategy of the target task processing model are adjusted according to the results of the performance evaluation, and the target task processing model is trained and optimized again based on the adjusted training data set and training strategy.

[0075] In the embodiments of the present application, during the model training and use process, relevant laws and regulations are strictly observed to ensure the security and privacy of data.

[0076] like Figure 3As shown, the target task processing model processes the target task in the following steps: Step S301 to Step S303:

[0077] Step S301: Decompose the target task to determine the subtasks corresponding to the target task and their subtask information.

[0078] Subtasks are execution steps in the target task. In this embodiment, the large language model breaks down the user's target task, which helps the model capture key information in the task and reduce misunderstandings and deviations. By breaking down complex tasks into simple subtasks, each step of the target task can be made more specific and actionable, which not only helps the model itself to better handle the user's target task, but also facilitates possible subsequent automated processing. When a subtask includes more than one execution step, the subtask can be further broken down.

[0079] For example, taking an application scenario as an example, Figure 3.1 A schematic diagram of task decomposition is shown, where the user's target task is decomposed into subtask 1, subtask 2, subtask 3, subtask 4, and subtask 5. Figure 3.2 As shown, subtask 2 can be further decomposed.

[0080] In one possible implementation, the subtasks obtained after the target task is decomposed can be reasonably allocated resources and scheduled according to factors such as their nature, difficulty and time requirements, which helps to ensure that key tasks are processed in a timely manner while avoiding resource waste and conflicts.

[0081] As a possible implementation of the present application, the subtask information includes descriptions of the subtasks and dependencies between the subtasks. Figure 4 A specific implementation process of decomposing the target task and determining the subtasks corresponding to the target task and their subtask information in the method provided in the embodiment of the present application is shown, and is described in detail as follows:

[0082] A1: Decompose the target task, and construct an event directed graph according to the decomposition results. The event directed graph is composed of vertices and edges connecting the vertices.

[0083] The vertices of the event directed graph represent the task descriptions of the subtasks of the target task, and the edges of the event directed graph represent the dependency relationship between two subtasks.

[0084] A2: Determine the task description of the subtask corresponding to the target task and the dependency relationship of the subtask according to the vertex set and the edge set of the event directed graph.

[0085] The vertices in the event directed graph represent the task descriptions of the subtasks, and the edges in the event directed graph are directed edges, which are used to represent the logical relationships between subtasks, such as timing, cause and effect, conditions, and hierarchical relationships. The event directed graph can be used to show the evolutionary laws and patterns of the disassembled subtasks that occur successively in time and space. The event directed graph converts discrete subtasks and relationships into a structured knowledge representation form, making knowledge easier to understand and use. In some implementations, by introducing graph neural networks, the potential relationships between subtasks can be further explored to improve the richness and accuracy of knowledge representation.

[0086] This embodiment constructs an event directed graph according to the decomposition result of the target task, and determines the task descriptions of all subtasks of the target task and the dependencies of the subtasks based on the event directed graph, thereby effectively ensuring the accuracy and comprehensiveness of the subtask information determination.

[0087] Step S302: Determine an agent matching the subtask according to the subtask information.

[0088] The subtask information includes the task description of the subtask. In this embodiment, according to the task description of the subtask, the agent matching the subtask is traversed in the agent repository, wherein the agent repository includes a pre-organized and maintained agent information set. The agent information set includes a labeled description of the agent, and the labeled description includes application field, function, price, and interface. If an agent matching the subtask is not found, the subtask is disassembled, and the agent matching the disassembly result is continued to be searched in the agent repository until a matching agent is found, and the disassembly ends. In this embodiment, when an agent matching the subtask is not found, the subtask is disassembled. If an agent matching the disassembly result is not found, the disassembly continues. Task disassembly is not necessarily completed in one step, but is dynamically updated through interaction with the agent repository.

[0089] In this embodiment, based on the task description of the subtask and the labeled description of the agent, the matching degree between the agent and the task description of the subtask is fully considered from the aspects of application field, function, price, etc., so that the user can achieve the target task at the lowest cost. Based on the matching degree, an agent list is constructed, and the agent list is used to summarize the agent information of the agents matching the subtask.

[0090] Step S303: Call the interface of the matched agent to process the subtask.

[0091] In this embodiment, generally, the user's target task is a complex task. When the user's target task is a simple task, the target task processing model directly determines the identification of the matching agent according to the task information of the target task, and calls the agent corresponding to the identification to process the target task.

[0092] As a possible implementation of the present application, the subtask information includes dependencies between subtasks. Figure 5 A specific implementation process of calling the interface of the matched agent to process the subtask in the task processing method provided in the embodiment of the present application is shown, and is described in detail as follows:

[0093] B1: Construct an agent directed graph according to the subtask information and the identifier of the matching agent.

[0094] The vertices of the agent directed graph represent the agent information of the subtask and the agent to be called, and the edges of the agent directed graph represent the dependency relationship between two agents. The dependency relationship between the agents is determined based on the dependency relationship between their matched subtasks.

[0095] In this embodiment, the agent information includes information such as application field, function, price, interface, etc. The vertices of the agent directed graph correspond to the vertices of the event directed graph one by one.

[0096] B2: Based on the directed graph of intelligent agents, the interfaces of the matching intelligent agents are called in sequence to process the subtasks.

[0097] According to the task description of the subtask in the directed graph of the intelligent agent, a prompt for the matching intelligent agent is constructed, and the prompt is used to guide the intelligent agent to process the subtask.

[0098] In one possible implementation, when the agent to be called for the subtask has an associated agent on which it depends, a prompt is constructed based on the processing result of the associated agent on the matching subtask and the task description of the subtask corresponding to the agent to be called; and the prompt is used to guide the agent to process the subtask.

[0099] In this embodiment, the execution result of the associated agent of the agent to be called affects the agent to be called. By dynamically updating the prompt, the agent is guided to process the subtask, which is beneficial to improving the effectiveness and accuracy of task processing.

[0100] In a possible implementation, after calling the interface of the matched agent to process the subtask, the processing status and progress of calling the interface of the matched agent to process the subtask is monitored; when an abnormality is detected in the processing status and progress, the abnormality is processed according to a preset processing mechanism. The preset processing mechanism includes retry, rollback, and user notification.

[0101] In this embodiment, by monitoring the processing status and progress of the target task processed by the large language model, the smooth execution of the target task can be effectively ensured.

[0102] In a possible implementation, user feedback data on target task processing results is collected regularly, and the target task processing model is continuously optimized and improved based on the feedback data.

[0103] As can be seen from the above, in the embodiment of the present application, by obtaining the user's target task; inputting the target task into the target task processing model, using the target task processing model to process the target task, the target task processing model disassembles the target task, determines the subtasks corresponding to the target task and its subtask information; then, based on the subtask information, determines the agent that matches the subtask, and then calls the interface of the matching agent to process the subtask. The present application can intelligently screen agents that meet the needs of users, without the need for users to fully understand the agents that have been released on the market, and assists users to complete the processing of target tasks at the lowest cost, saving costs for users while greatly improving the efficiency of user task processing and enhancing user experience.

[0104] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0105] Corresponding to the task processing method described in the above embodiment, Figure 6 A structural block diagram of a task processing device provided in an embodiment of the present application is shown. For ease of explanation, only the parts related to the embodiment of the present application are shown.

[0106] Reference Figure 6 The task processing device is applied to an intelligent terminal, and the device comprises: a task acquisition unit 61, a task processing unit 62, wherein:

[0107] A task acquisition unit 61 is used to acquire a user's target task;

[0108] The task processing unit 62 is used to input the target task into a target task processing model, and use the target task processing model to process the target task, wherein the target task processing model is a pre-trained large language model, and the target task processing model processes the target task in the following steps: decomposing the target task to determine the subtasks corresponding to the target task and their subtask information; determining an intelligent agent matching the subtask based on the subtask information; and calling the interface of the matching intelligent agent to process the subtask.

[0109] As a possible implementation of the present application, the task processing unit 62 includes:

[0110] A disassembly construction module, used to disassemble the target task and construct an event directed graph according to the disassembly result, wherein the event directed graph is composed of vertices and edges connecting the vertices, wherein the vertices of the event directed graph represent task descriptions of subtasks of the target task, and the edges of the event directed graph represent dependencies between two subtasks;

[0111] The task information determination module is used to determine the task description of the subtask corresponding to the target task and the dependency relationship of the subtask according to the vertex set and edge set of the event directed graph.

[0112] As a possible implementation of the present application, the subtask information includes a task description of the subtask, and the task processing unit 62 further includes:

[0113] An agent search module is used to traverse and determine an agent matching the subtask in an agent repository according to the task description of the subtask, wherein the agent repository includes a pre-organized and maintained set of agent information; if an agent matching the subtask is not found, the subtask is disassembled, and the agent repository is continued to be searched for an agent matching the disassembly result, until a matching agent is found, at which time the disassembly is completed.

[0114] As a possible implementation of the present application, the subtask information includes dependencies between subtasks, and the task processing unit 62 further includes:

[0115] An agent graph construction module, used to construct an agent directed graph according to the subtask information and the identifier of the matched agent, wherein the vertices of the agent directed graph represent the agent information of the subtask and the agent to be called, and the edges of the agent directed graph represent the dependency relationship between two agents, and the dependency relationship between the agents is determined based on the dependency relationship between the matched subtasks;

[0116] The task processing module is used to call the interfaces of the matching agents in sequence to process the subtasks based on the agent directed graph.

[0117] As a possible implementation of the present application, the task processing module is specifically used for:

[0118] When the agent to be called by the subtask has an associated agent that it depends on, construct a prompt based on the processing result of the associated agent on the matching subtask and the task description of the subtask corresponding to the agent to be called;

[0119] The prompt is used to guide the agent to process the subtask.

[0120] As a possible implementation of the present application, the task processing device further includes:

[0121] A data set acquisition unit, used to acquire a task decomposition diagram data set, wherein the task decomposition diagram data set includes sample annotation data obtained by decomposing a sample task, and the sample annotation data includes task descriptions of subtasks in the sample task and dependencies between subtasks;

[0122] The model training unit is used to use the task decomposition graph data set to train the task decomposition capability of the large language model and the capability of determining and calling the intelligent agent matching the subtask; when the training error of the large language model meets the preset requirements, the target task processing model is obtained.

[0123] As a possible implementation of the present application, the task processing device further includes:

[0124] A status and progress monitoring unit, used to monitor the processing status and progress of calling the interface of the matched agent to process the subtask;

[0125] The exception handling unit is used to handle the exception according to a preset processing mechanism when an exception is detected in the processing state and progress.

[0126] In the embodiment of the present application, the user's target task is obtained; the target task is input into the target task processing model, and the target task is processed by the target task processing model, the target task processing model disassembles the target task, determines the subtasks corresponding to the target task and its subtask information; then, according to the subtask information, determines the agent matching the subtask, and then calls the interface of the matching agent to process the subtask. The present application can intelligently screen agents that meet the needs of users, without the need for users to fully understand the agents released on the market, greatly improving the efficiency of user task processing and enhancing user experience.

[0127] The present application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, Figures 1 to 5 Represents the steps of any task processing method.

[0128] The embodiment of the present application also provides an intelligent terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, Figures 1 to 5 Represents the steps of any task processing method.

[0129] The embodiment of the present application also provides a computer program product, when the computer program product is run on a terminal device, the terminal device executes the following Figures 1 to 5 Represents the steps of any task processing method.

[0130] Figure 7 is a schematic diagram of a smart terminal provided by an embodiment of the present application. Figure 7 As shown, the intelligent terminal 7 of this embodiment includes: a processor 70, a memory 71, and a computer program 72 stored in the memory 71 and executable on the processor 70. When the processor 70 executes the computer program 72, the steps in the above-mentioned embodiments of the task processing method are implemented, for example Figure 1 Steps S101 to S102 shown, and Figure 2 Alternatively, when the processor 70 executes the computer program 72, the functions of each module / unit in the above-mentioned device embodiments are realized, for example Figure 6 The functions of the units 61 to 62 are shown.

[0131] The computer program 72 may be divided into one or more modules / units, which are stored in the memory 71 and executed by the processor 70 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 72 in the smart terminal 7.

[0132] The processor 70 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0133] The memory 71 may be an internal storage unit of the intelligent terminal 7, such as a hard disk or memory of the intelligent terminal 7. The memory 71 may also be an external storage device of the intelligent terminal 7, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the intelligent terminal 7. Further, the memory 71 may also include both an internal storage unit and an external storage device of the intelligent terminal 7. The memory 71 is used to store the computer program and other programs and data required by the intelligent terminal. The memory 71 may also be used to temporarily store data that has been output or is to be output.

[0134] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned device can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0135] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0136] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0137] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0138] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0139] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0140] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0141] If the integrated unit 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 present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0142] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A task processing method, characterized in that: The task processing method comprises: Get the user's target task; The target task is input into a target task processing model, and the target task is processed by the target task processing model, wherein the target task processing model is a large language model that has been pre-trained, and the target task processing model processes the target task in the following steps: Decomposing the target task to determine the subtasks corresponding to the target task and the subtask information thereof; Determine an agent matching the subtask according to the subtask information; Call the interface of the matching agent to process the subtask.

2. The task processing method according to claim 1, characterized in that: The step of decomposing the target task to determine subtasks corresponding to the target task and their subtask information includes: Decomposing the target task, and constructing an event directed graph according to the decomposition results, wherein the event directed graph is composed of vertices and edges connecting the vertices, wherein the vertices of the event directed graph represent task descriptions of subtasks of the target task, and the edges of the event directed graph represent dependency relationships between two subtasks; According to the vertex set and the edge set of the event directed graph, the task description of the subtask corresponding to the target task and the dependency relationship of the subtask are determined.

3. The task processing method according to claim 1, characterized in that: The subtask information includes a task description of the subtask, and determining an agent matching the subtask according to the subtask information includes: According to the task description of the subtask, searching in the agent repository for an agent matching the subtask, wherein the agent repository includes a pre-organized and maintained set of agent information; If no intelligent agent matching the subtask is found, the subtask is disassembled, and the intelligent agent matching the disassembly result is continuously searched in the intelligent agent repository until a matching intelligent agent is found, and the disassembly is completed.

4. The task processing method according to claim 1, characterized in that: The subtask information includes dependencies between subtasks, and the calling of the interface of the matched agent to process the subtasks includes: According to the subtask information and the identifier of the matched agent, an agent directed graph is constructed, wherein the vertices of the agent directed graph represent the agent information of the subtask and the agent to be called, and the edges of the agent directed graph represent the dependency relationship between two agents, wherein the dependency relationship between the agents is determined based on the dependency relationship between the matched subtasks; Based on the agent directed graph, the interfaces of the matching agents are called in sequence to process the subtasks.

5. The task processing method according to claim 4, characterized in that: The step of sequentially calling the interfaces of the matching agents to process the subtasks based on the agent directed graph includes: When the agent to be called by the subtask has an associated agent that it depends on, construct a prompt based on the processing result of the associated agent on the matching subtask and the task description of the subtask corresponding to the agent to be called; The prompt is used to guide the agent to process the subtask.

6. The task processing method according to claim 1, characterized in that: The task processing method further includes: Acquire a task decomposition diagram data set, wherein the task decomposition diagram data set includes sample annotation data obtained by decomposing a sample task, and the sample annotation data includes task descriptions of subtasks in the sample task and dependencies between subtasks; Using the task decomposition graph dataset to train the large language model's task decomposition capabilities, and the ability to determine and call agents that match subtasks; When the training error of the large language model meets the preset requirements, the target task processing model is obtained.

7. The task processing method according to claim 1, characterized in that: After calling the interface of the matched agent to process the subtask, the method further includes: Monitoring the processing status and progress of calling the interface of the matched agent to process the subtask; When an abnormality is detected in the processing status and progress, the abnormality is processed according to a preset processing mechanism.

8. A task processing device, characterized in that: The task processing device comprises: A task acquisition unit, used to acquire the user's target task; A task processing unit is used to input the target task into a target task processing model, and use the target task processing model to process the target task, wherein the target task processing model is a pre-trained large language model, and the target task processing model processes the target task in the following steps: decomposing the target task to determine the subtasks corresponding to the target task and their subtask information; determining an intelligent agent matching the subtask based on the subtask information; and calling the interface of the matching intelligent agent to process the subtask.

9. An intelligent terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the task processing method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the task processing method according to any one of claims 1 to 7 are implemented.

Images