Task execution method, task execution apparatus, electronic device and storage medium

Abstract

Disclosed in the embodiments of the present application are a task execution method, an electronic device and a storage medium, which are used for improving the efficiency of executing a task corresponding to a standard procedure. The task execution method comprises: receiving a task request; on the basis of the task request, acquiring a first standard procedure from a standard procedure information base; on the basis of the task request, acquiring a first sub-standard procedure from the first standard procedure, the first sub-standard procedure being a sub-standard procedure of the first standard procedure, the first sub-standard procedure being used for instructing a plurality of first sub-tasks to be successively executed on the basis of a jump relationship, the first sub-standard procedure comprising a first root node, and the first root node being used for instructing executing a start sub-task among the plurality of first sub-tasks; and, on the basis of the first sub-standard procedure, executing the plurality of first sub-tasks.

Description

Task execution method and task execution apparatus, electronic device, and storage medium

[0001] The present application claims priority from the Chinese patent application No. 202411333787.3 filed on September 23, 2024, and entitled "Task execution method and task execution apparatus, electronic device, and storage medium", the whole content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present application relates to the field of tasks in computer technology, and in particular to a task execution method and task execution apparatus, electronic device, and storage medium. BACKGROUND

[0003] The use scenarios of large models are constantly expanding, and large models can be used to solve complex problems. A general-purpose large model is integrated into an agent framework. Under the agent framework, the agent is endowed with the general capabilities of the large model to solve one or more specific problems and can solve more complex problems. Therefore, large models play an important role in task planning and task execution.

[0004] In providing solutions to tasks, introducing a standard process can handle a relatively fixed type of problem and achieve the completion of the entire task. In the agent framework, the standard process is introduced, and the large model can focus more on task selection and execution.

[0005] At present, the above-mentioned standard process is obtained by manually writing the standard process, and there are problems of low execution efficiency and low success rate of tasks. SUMMARY

[0006] The embodiments of the present application provide a task execution method and task execution apparatus, electronic device, and storage medium, which are used to improve the execution efficiency and success rate of tasks corresponding to the standard process.

[0007] To solve the above technical problems, the embodiments of the present application provide the following technical solutions:

[0008] In a first aspect, the embodiments of the present application provide a task execution method, comprising: first receiving a task request; then acquiring a first standard process from a standard process information base according to the task request; next acquiring a first sub-standard process from the first standard process according to the task request, the first sub-standard process being a sub-standard process of the first standard process, the first sub-standard process being used to indicate that a plurality of first sub-tasks are executed in sequence based on a jump relationship, the first sub-standard process comprising a first root node, the first root node being used to indicate that a starting sub-task in the plurality of first sub-tasks is executed; and finally executing the plurality of first sub-tasks according to the first sub-standard process.

[0009] In the above scheme, the standard process information base is configured with a standard process, and the first standard process corresponding to the task request can be provided through the standard process information base, so that the standard process does not need to be manually configured for the task request, and the way of providing the standard process for the task request is simplified. In addition, the first sub-standard process can be acquired from the first standard process according to the task request, the first sub-standard process indicating that the plurality of first sub-tasks are executed in sequence, and the first root node included in the first sub-standard process indicating the starting sub-task in the plurality of first sub-tasks, so that the plurality of first sub-tasks can be executed according to the first sub-standard process. Since the first sub-standard process can be acquired from the first standard process according to the task request, the corresponding first sub-standard process can be acquired for different task requests, which can be suitable for different task requests with complex dynamics, improve the flexibility of executing the plurality of first sub-tasks, and improve the execution efficiency and success rate of the plurality of sub-tasks.

[0010] In a possible implementation manner of the first aspect, the acquiring the first standard process from the standard process information base according to the task request comprises:

[0011] performing semantic analysis on the task request to obtain a target semantic representation;

[0012] performing semantic matching between the target semantic representation and the standard process information base to obtain a first semantic representation matched with the target semantic representation, the standard process information base comprising: a plurality of standard processes and a first semantic representation corresponding to each of the plurality of standard processes;

[0013] acquiring the first standard process from the standard process information base according to the first semantic representation matched with the target semantic representation.

[0014] In the above scheme, the standard process information library stores a plurality of standard processes and corresponding first semantic representations, and therefore the task execution device performs semantic matching between the target semantic representation and the standard process information library to obtain a first semantic representation matched with the target semantic representation, and finally, the task execution device obtains a first standard process from the standard process information library according to the first semantic representation matched with the target semantic representation. In the embodiment of the present application, the first standard process can be quickly obtained from the standard process information library through matching between the target semantic representation and the first semantic representation.

[0015] In a possible implementation form of the first aspect, the obtaining, according to the task request, of the first sub-standard process from the first standard process comprises:

[0016] obtaining a sub-standard process information library according to the first standard process, the sub-standard process information library including a plurality of sub-standard processes, and the plurality of sub-standard processes belonging to the first standard process;

[0017] obtaining the first sub-standard process from the sub-standard process information library according to the task request.

[0018] In the above scheme, the task execution device establishes a sub-standard process information library, and the plurality of sub-standard processes included in the sub-standard process information library belong to the first standard process. The task execution device can further obtain the first sub-standard process by searching the sub-standard process information library, and the searching of the sub-standard process information library improves the task execution efficiency, and the execution of the standard process starts from the starting sub-task indicated by the first root node included in the first sub-standard process, thereby reducing unnecessary calculation.

[0019] In a possible implementation form of the first aspect, the obtaining, according to the task request, of the first sub-standard process from the sub-standard process information library comprises:

[0020] performing semantic analysis on the task request to obtain a target semantic representation;

[0021] performing semantic matching between the target semantic representation and the sub-standard process information library to obtain a second semantic representation matched with the target semantic representation, the sub-standard process information library including a plurality of sub-standard processes of the first standard process and corresponding second semantic representations of the plurality of sub-standard processes of the first standard process;

[0022] obtaining the first sub-standard process from the sub-standard process information library according to the second semantic representation matched with the target semantic representation.

[0023] In the above scheme, the sub-standard process information library stores a plurality of sub-standard processes and corresponding second semantic representations, and therefore the task execution device performs semantic matching with the sub-standard process information library according to the target semantic representation, to obtain a second semantic representation matched with the target semantic representation. Finally, the task execution device obtains the first sub-standard process from the sub-standard process information library according to the second semantic representation matched with the target semantic representation. In the embodiment of the present application, the first sub-standard process can be quickly obtained from the sub-standard process information library through matching of the target semantic representation and the second semantic representation.

[0024] In a possible implementation of the first aspect, the method further includes: when the first standard process is not obtained from the standard process information library, obtaining a second standard process according to the task request, the second standard process being a standard process information generated according to the task request and an experience database; and storing the second standard process to the standard process information library and the sub-standard process information library.

[0025] In the above scheme, the task execution device and the terminal device interact, receive the task request from the terminal device, and analyze the task request to obtain the task request to be executed. The task execution device can obtain an experience database, which can provide expert experience information for generating a standard process. The second standard process can be generated through the experience database and the task request. The second standard process is stored to the standard process information library and the sub-standard process information library, to realize dynamic updating of the standard process information library and the sub-standard process information library. In some optional implementation, the second standard process can also be stored to the sub-standard process information library, to realize dynamic updating of the sub-standard process information library.

[0026] In a possible implementation of the first aspect, the method further includes: when the first standard process is not obtained from the standard process information library, obtaining first multi-modal information according to the task request, the first multi-modal information including a plurality of types of information required for generating standard process information; inputting the first multi-modal information into a large model, and outputting a third standard process through the large model, the large model being used for generating standard process information corresponding to a plurality of tasks respectively; and storing the third standard process to the standard process information library and the sub-standard process information library.

[0027] In the above scheme, the task execution apparatus can train the large model, use the trained large model to predict the first multi-modal information to output a third standard process, the large model is used to generate standard process information corresponding to a plurality of tasks respectively, when the first standard process is not obtained from the standard process information library, the third standard process is stored to the standard process information library, to realize dynamic updating of the standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request when searching next time. For example, the third standard process is stored to the standard process information library and the sub-standard process information library, to realize dynamic updating of the standard process information library and the sub-standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request when searching next time.

[0028] In a possible implementation of the first aspect, the method further includes: obtaining a fourth standard process, and storing the fourth standard process to the standard process information library, the fourth standard process being standard process information generated according to an experience database; or, obtaining second multi-modal information, the second multi-modal information including a plurality of types of information required for generating standard process information; inputting the second state information into a large model, and outputting a fifth standard process through the large model, the large model being used to generate standard process information corresponding to a plurality of tasks respectively; and storing the fifth standard process to the standard process information library and the sub-standard process information library.

[0029] In the above scheme, the experience database can provide expert experience information for generating a standard process, without the task request providing a task, the fourth standard process can be generated through the experience database, and the fourth standard process is stored to the standard process information library, to realize dynamic updating of the standard process information library, so that the standard process information library obtains the standard process and the sub-standard process matched with the task request when searching next time. For example, the fourth standard process is stored to the standard process information library and the sub-standard process information library, to realize dynamic updating of the standard process information library and the sub-standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request when searching next time.

[0030] In the foregoing solution, the task execution apparatus can train the large model, use the trained large model to predict the second multi-modal information to output a fifth standard process, the large model is used to generate standard process information corresponding to a plurality of tasks respectively, and the fifth standard process is stored into the standard process information library and the sub-standard process information library, so as to realize dynamic updating of the standard process information library, so that the standard process information library obtains the standard process and the sub-standard process matched with the task request when performing retrieval next time. For example, the fifth standard process is stored into the standard process information library and the sub-standard process information library, so as to realize dynamic updating of the standard process information library and the sub-standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request when performing retrieval next time.

[0031] In a possible implementation of the first aspect, the method further includes: in the process of executing the plurality of first sub-tasks through the action model, obtaining an action sequence corresponding to the first sub-standard process, a plurality of actions in the action sequence are used to execute the plurality of first sub-tasks or sub-task jumps, and the plurality of actions in the action sequence are implemented based on sub-task information or sub-task jump logic information of the plurality of sub-tasks; performing abnormal evaluation on the action sequence through a first evaluation model according to execution results of the plurality of actions in the action sequence, the sub-task information of the plurality of sub-tasks and the sub-task jump logic information, to obtain a suspicious action from the action sequence; and performing abnormal detection on the suspicious action through a second evaluation model according to an execution result of the suspicious action and sub-task information of a first sub-task corresponding to the suspicious action, or according to the execution result of the suspicious action and sub-task jump logic information of the first sub-task corresponding to the suspicious action, to determine that the suspicious action is an abnormal action.

[0032] In the foregoing solution, the task execution apparatus can train the second evaluation model, the task execution apparatus obtains an execution result of a suspicious action, and uses the trained second evaluation model to perform abnormal detection on the suspicious action according to the execution result of the suspicious action and sub-task information or sub-task jump logic information of a first sub-task corresponding to the suspicious action, wherein the abnormal detection means that whether the suspicious action is abnormal is evaluated locally through the second evaluation model, and when the suspicious action is abnormal, it is determined that the suspicious action is an abnormal action.

[0033] In a possible implementation manner of the first aspect, the method further includes: performing optimization on the abnormal action by an optimization model according to an execution result of the abnormal action and subtask information of a subtask corresponding to the abnormal action, or according to the execution result of the suspicious action and subtask jump logic information of a first subtask corresponding to the suspicious action, to obtain an execution result of an optimized action; and performing evaluation on the execution result of the optimized action by the second evaluation model to obtain a first evaluation result.

[0034] In the above solution, the task execution apparatus again invokes the second evaluation model to perform local evaluation, and performs evaluation on the execution result of the optimized action by the second evaluation model to obtain a first evaluation result. In the embodiment of the application, the abnormal action is optimized by the optimization model, and the execution result of the optimized action is evaluated by the second evaluation model, which can detect whether the optimized action still has a problem, to determine whether the optimization on the abnormal action is successful.

[0035] In a possible implementation manner of the first aspect, the method further includes: when the first evaluation result indicates that the execution result of the optimized action has a problem, performing backtracking by a backtracking model according to the execution result of the optimized action, subtask information and subtask jump logic information of a plurality of first subtasks, to obtain a first associated action that causes the problem from the action sequence and obtain historical backtracking information; performing optimization by the optimization model from the first associated action that causes the problem until the abnormal action, to obtain an execution result of an optimized first associated action; and performing evaluation on the execution result of the optimized first associated action by the second evaluation model to obtain a second evaluation result.

[0036] In the above solution, the task execution apparatus determines the first associated action that causes the problem from the action sequence by the backtracking model, again invokes the optimization model to perform optimization from the first associated action that causes the problem until the abnormal action, to obtain an execution result of an optimized first associated action. Then, the second evaluation model is again invoked to perform evaluation on the execution result of the optimized first associated action to obtain a second evaluation result. In the embodiment of the application, the first associated action is optimized by the optimization model from the first associated action until the abnormal action, and the execution result of the optimized first associated action is evaluated by the second evaluation model, which can detect whether the optimized first associated action still has a problem, to determine whether the optimization on the first associated action until the abnormal action is successful.

[0037] In a possible implementation manner of the first aspect, when the first evaluation result indicates that the execution result of the optimized action has no problem, the method further includes: updating the standard process information base by using the optimized action; and / or training at least one of the following models by using the optimized action: the action model, the first evaluation model, and the second evaluation model.

[0038] In the above solution, the second action is optimized by the optimization model until the abnormal action, and the execution result of the optimized second action is evaluated by the second evaluation model, so that whether the action after re-optimization still has a problem can be detected, and whether the optimization for the second action to the abnormal action is successful can be determined.

[0039] In a possible implementation manner of the first aspect, when the first evaluation result indicates that the execution result of the optimized action has no problem, the method further includes: updating the standard process information base by using the optimized action; and / or training at least one of the following models by using the optimized action: the action model, the first evaluation model, and the second evaluation model.

[0040] In the above solution, when the first evaluation result indicates that the execution result of the optimized action has no problem, the task execution apparatus can update the information base by using the optimized action, and / or optimize the model by using the optimized action. The information base can include the standard process information base, or the updated database includes the standard process information base and the sub-standard process information base. The model is trained: the action model, the first evaluation model, and the second evaluation model. By updating the information base and / or optimizing the model, the information base and the model can be used for subsequent task execution, and the efficiency and success rate of task execution can be improved.

[0041] In a possible implementation manner of the first aspect, when the second evaluation result indicates that the execution result of the optimized first associated action has no problem, the method further includes: updating the standard process information base by using the optimized first associated action; and / or training at least one of the following models by using the optimized first associated action: the action model, the first evaluation model, the second evaluation model, and the optimization model.

[0042] In the above scheme, when the second evaluation result indicates that the execution result of the optimized first associated action is not problematic, the task execution apparatus can update the information base by using the optimized action, and / or optimize the model, the information base can include the standard process information base, or the updated database includes the standard process information base and the sub-standard process information base, and the model is trained: action model, first evaluation model, second evaluation model, optimization model. By updating the information base and / or optimizing the model, the information base and the model can be used for subsequent task execution, improving the efficiency and success rate of task execution.

[0043] In a possible implementation of the first aspect, when the third evaluation result indicates that the execution result of the optimized second associated action is not problematic, the method further includes: updating the standard process information base by using the optimized second associated action; and / or, training at least one of the following models by using the optimized second associated action: action model, first evaluation model, second evaluation model, backtracking model, optimization model.

[0044] In the above scheme, when the third evaluation result indicates that the execution result of the optimized second associated action is not problematic, the task execution apparatus can update the information base by using the optimized action, and / or optimize the model, the information base can include the standard process information base, or the updated database includes the standard process information base and the sub-standard process information base, and the model is trained: action model, first evaluation model, second evaluation model, backtracking model, optimization model. By updating the information base and / or optimizing the model, the information base and the model can be used for subsequent task execution, improving the efficiency and success rate of task execution.

[0045] In a possible implementation of the first aspect, the task execution method provided by the above-mentioned first aspect can be implemented by a task execution apparatus, which can be a terminal device or a server, for example, the terminal device or the server includes a processor, and the processor is configured to execute the task execution method in the above-mentioned first aspect.

[0046] For example, the standard process can specifically include computer program code, and the task execution apparatus can be a code executor.

[0047] For another example, the task execution apparatus can be an agent, and the agent can be integrated with a large model. The agent can be used to execute the task execution method provided by the above-mentioned first aspect.

[0048] In the framework of the agent, the agent is endowed with the general ability of the large model, which is used to solve one or more specific problems, and different types of agents can be automatically connected to solve more complex problems.

[0049] In a possible implementation of the first aspect, the task request in the task execution method provided by the first aspect can be a task in various business scenarios, which is not limited here. Specifically, the task request can be in the fields of smart home, autonomous driving, gaming, intelligent question answering, communication, etc. For example, in the field of smart home, the task request can be a user-initiated fault detection of a smart home appliance, or a user-controlled customized route of a sweeping robot in the whole house, etc. In the field of autonomous driving, the task request can be a user-initiated automatic obstacle avoidance of a smart vehicle, or a control of the smart vehicle to update a map, etc. In the field of gaming, the task request can be a game player controlling a game character to automatically find a route, or controlling a game character to perform multi-task assignment with other game characters, etc. In the field of communication, the task request can be specifically an evaluation of the signal quality of a terminal attached in a cell of a base station, etc. According to the task execution method provided by the embodiments of the present application, the task execution device is configured with a standard process in the standard process information base, so as to be able to provide a first standard process corresponding to the task request in the task request, thereby not needing to manually configure a standard process for the task, and simplifying the way of providing a standard process for the task. In addition, the task execution device is configured with a sub-standard process in the sub-standard process information base, so as to be able to provide a starting sub-task of the first sub-standard process corresponding to the task request in the first standard process, thereby being able to flexibly select the starting sub-task in the first standard process according to the task request, and thus being able to be applicable to different tasks of complex dynamics, improve the flexibility of executing the task request, and improve the execution efficiency of the task corresponding to the standard process.

[0050] In a possible implementation of the first aspect, the standard process information base is constructed in the following manner:

[0051] obtaining sub-task information of a plurality of second sub-tasks and sub-task jump logic information between the plurality of second sub-tasks, the sub-task jump logic information between the plurality of second sub-tasks being used to indicate a sub-task jump manner between the plurality of second sub-tasks, and the plurality of second sub-tasks being used to sequentially execute the jump of the second sub-task according to a jump relationship indicated by a sixth standard process;

[0052] performing the jump of the second sub-task in a reverse sub-task jump manner between the plurality of second sub-tasks, and adding the sub-task information of the jumped second sub-task to a node of the jumped second sub-task, to obtain description information of the sixth standard process, the reverse sub-task jump manner being opposite to the sub-task jump manner indicated by the sub-task jump logic information;

[0053] According to the description information of the sixth standard flow, the standard flow information base is constructed, and the sixth standard flow is any one of the standard flows in the standard flow information base.

[0054] In the above scheme, the description information of the sixth standard flow is obtained according to the subtask jump logic information of the plurality of second subtasks, and the description information of the plurality of standard flows in the standard flow information base can be obtained in the same way, so that the standard flow information base can be constructed. The embodiments of the present application construct the standard flow information base through the description information of the plurality of standard flows, so that the standard flow information base can be provided for a plurality of different tasks, and the manner of providing the standard flow is simplified.

[0055] In a possible implementation manner of the first aspect, the jump of the second subtask is performed in the reverse jump manner between the plurality of second subtasks, and the subtask information of the jumped second subtask is added to the node of the jumped second subtask to obtain the description information of the sixth standard flow, including:

[0056] The jump is performed in the reverse jump manner between the plurality of second subtasks, and the subtask information of the jumped second subtask is added to the node of the jumped second subtask to obtain the description information of the plurality of substandard flows of the sixth standard flow.

[0057] The description information of the sixth standard flow is generated according to the description information of the plurality of substandard flows of the sixth standard flow.

[0058] In the above scheme, the jump of the second subtask is compacted in the reverse jump manner, and the subtask information of the jumped second subtask is added to the node of the jumped second subtask to obtain the description information of the plurality of substandard flows, so that the description information of the sixth standard flow can be generated based on the description information of the plurality of substandard flows, and the manner of generating the description information of the sixth standard flow is simplified.

[0059] In a possible implementation manner of the first aspect, the standard flow information base is constructed according to the description information of the sixth standard flow, including:

[0060] The first semantic representation corresponding to the sixth standard flow is generated according to the description information of the sixth standard flow.

[0061] The standard flow information base is constructed according to the first semantic representation corresponding to the sixth standard flow.

[0062] In the above scheme, the first semantic representation corresponding to the sixth standard process is one or more vector representations corresponding to the standard process, which can be sparse or dense semantic representation. The task execution device generates the first semantic representation corresponding to the sixth standard process, and after obtaining the first semantic representation corresponding to the plurality of standard processes respectively, the standard process information library is generated. In an embodiment of the application, the first semantic representation corresponding to each standard process can be used for retrieval to realize the matching of the task and the standard process.

[0063] In a possible implementation of the first aspect, the sub-standard process information library is constructed in the following manner:

[0064] obtaining subtask information of a plurality of first subtasks and subtask jump logic information between the plurality of first subtasks, the subtask jump logic information being used to indicate a subtask jump manner between the plurality of first subtasks;

[0065] performing jump of the first subtask according to the subtask reverse jump manner between the plurality of first subtasks, and adding subtask information of the jumped first subtask to a position node of the jumped first subtask to obtain description information of a plurality of sub-standard processes of the first standard process;

[0066] constructing the sub-standard process information library according to the description information of the plurality of sub-standard processes of the first standard process.

[0067] In the above scheme, the jump of the second subtask is compacted by using the subtask reverse jump manner, and the second subtask information after jump is added to the node of the second subtask after jump to obtain the description information of the plurality of sub-standard processes of the first standard process, and the sub-standard process information library is constructed based on the description information of the plurality of sub-standard processes of the first standard process. Embodiments of the application construct the sub-standard process information library through the description information of the plurality of sub-standard processes, so as to provide the sub-standard process information library for a plurality of different tasks, and simplify the manner of providing the sub-standard process.

[0068] In a possible implementation of the first aspect, the constructing the sub-standard process information library according to the description information of the plurality of sub-standard processes of the first standard process comprises:

[0069] generating second semantic representations respectively corresponding to the plurality of sub-standard processes of the first standard process according to the description information of the plurality of sub-standard processes of the first standard process;

[0070] constructing the sub-standard process information library according to the second semantic representations respectively corresponding to the plurality of sub-standard processes of the first standard process.

[0071] In the foregoing solution, the second semantic representation corresponding to each of the plurality of sub-standard processes of the first standard process is one or more vector representations corresponding to the plurality of sub-standard processes. After obtaining the second semantic representation corresponding to each of the plurality of standard processes by generating the second semantic representation corresponding to each of the plurality of sub-standard processes of the first standard process, the sub-standard process information library is generated. The second semantic representation corresponding to each sub-standard process in the embodiment of the application can be used for retrieval to achieve matching of the task and the sub-standard process.

[0072] In a second aspect, the embodiment of the application further provides a task execution apparatus, comprising:

[0073] a receiving module configured to receive a task request;

[0074] a standard process obtaining module configured to obtain a first standard process from a standard process information library according to the task request;

[0075] a sub-standard process obtaining module configured to obtain a first sub-standard process from the first standard process according to the task request, the first sub-standard process being a sub-standard process of the first standard process, the first sub-standard process being used to indicate that a plurality of first sub-tasks are executed in turn based on a jump relationship, the first sub-standard process comprising a first root node, the first root node being used to indicate that a starting sub-task in the plurality of first sub-tasks is executed;

[0076] a task execution module configured to execute the plurality of first sub-tasks according to the first sub-standard process.

[0077] In the foregoing solution, the standard process information library is configured with a standard process, so as to be able to provide a first standard process corresponding to a task request in a task request, thereby not needing to manually configure a standard process for a task, and simplifying the manner of providing a standard process for a task. In addition, the sub-standard process information library is configured with a sub-standard process, so as to be able to provide a first sub-standard process corresponding to a starting sub-task of a task request in a first standard process, thereby being able to flexibly select a starting sub-task in a first standard process according to a task request, and thus being able to be applied to different tasks of complex dynamics, improve flexibility of executing a task request, and improve execution efficiency of a task corresponding to a standard process.

[0078] In the second aspect of the application, the component modules of the task execution apparatus can also perform the steps described in the foregoing first aspect and various possible implementation manners, and details are described in the foregoing description of the first aspect and various possible implementation manners.

[0079] In a third aspect, the embodiment of the application provides a computer readable storage medium, wherein instructions are stored in the computer readable storage medium, and when the instructions are run on a computer, the computer executes the method in the foregoing first aspect.

[0080] In a fourth aspect, an embodiment of the present application provides a computer program product containing instructions which, when executed on a computer, cause the computer to perform the method of the first aspect.

[0081] In a fifth aspect, an embodiment of the present application provides a communication apparatus, which can include a terminal device or a chip or the like entity. The communication apparatus includes a processor, a memory. The memory is configured to store instructions. The processor is configured to execute the instructions in the memory, so that the communication apparatus performs the method in any one of the first aspect or the possible implementation manners of the first aspect.

[0082] In a sixth aspect, an embodiment of the present application provides a chip system, which includes a processor configured to support a task execution apparatus to implement the functions involved in the above aspects, for example, to send or process the data and / or information involved in the above method. In a possible design, the chip system further includes a memory configured to store the necessary program instructions and data of the task execution apparatus. The chip system can be composed of a chip, or can include a chip and other discrete devices.

[0083] In a seventh aspect, an embodiment of the present application provides a chip, which includes one or more interface circuits and one or more processors. The interface circuit is configured to receive a signal from a memory of an electronic device, and send a signal to the processor. The signal includes computer instructions stored in the memory. When the processor executes the computer instructions, the electronic device performs the task execution method in the first aspect or any possible implementation manner of the first aspect.

[0084] The seventh aspect and any possible implementation manner of the seventh aspect correspond to the first aspect and any possible implementation manner of the first aspect respectively. For details, refer to the technical effects of the first aspect and any possible implementation manner of the first aspect, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0085] FIG. 1 is a schematic diagram of an architecture of a data center according to an embodiment of the present application;

[0086] FIG. 2a is a schematic diagram of an exemplary system according to an embodiment of the present application;

[0087] FIG. 2b is a schematic diagram of another exemplary system according to an embodiment of the present application;

[0088] FIG. 3 is a schematic diagram of an embodiment of a task execution method according to an embodiment of the present application;

[0089] FIG. 4 is a schematic diagram of a standard flow according to an embodiment of the present application;

[0090] FIG. 5 is a schematic diagram of an application scenario of a task execution method according to an embodiment of the present application;

[0091] FIG. 6 is a schematic diagram of dynamic retrieval of SOP and task entry according to an embodiment of the present application;

[0092] FIG. 7 is a schematic diagram of improved SOP execution according to an embodiment of the present application;

[0093] FIG. 8 is a schematic diagram of an embodiment of a task execution apparatus according to an embodiment of the present application;

[0094] FIG. 9 is a schematic diagram of a structure of a computing device according to an embodiment of the present application;

[0095] FIG. 10 is a schematic diagram of a structure of a computing device cluster according to an embodiment of the present application;

[0096] FIG. 11 is a schematic diagram of a structure of a computing device cluster according to an embodiment of the present application. DETAILED DESCRIPTION

[0097] The embodiments of the present application provide a task execution method and a task execution apparatus, an electronic device and a storage medium, and are used for improving the execution efficiency of a standard process task.

[0098] The embodiments of the present application are described below in conjunction with the accompanying drawings. The terms used in the embodiment part of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0099] Those of ordinary skill in the art can know that, as technology develops and new scenarios appear, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0100] In the present application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. The terms "first", "second", and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged under appropriate circumstances, and this is only a way of distinguishing the objects with the same properties used in the description of the embodiments of the present application. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, so that the process, method, system, product or equipment containing a series of units does not have to be limited to those units, but can include other units not clearly listed or inherent to these processes, methods, products or equipment.

[0101] The embodiments of the present application provide a task execution method, which can be applied to a computing device cluster, which can include one or more computing devices.

[0102] Any computing device is not limited herein. For example, any computing device can be a terminal device, a cloud server, a container, a virtual machine, etc.

[0103] The computing device cluster can provide a data processing platform, so as to implement the task execution method of the embodiments of the present application through the data processing platform. Specifically, the data processing platform can be provided in the terminal device, and the data processing platform can be provided in the cloud server. Based on the respective data processing platforms of the terminal device and the cloud server, consistency and calculation accuracy of the terminal side and the cloud side can be achieved, and the difference in computing power of different devices can also be fully utilized. The specific form of the computing device cluster and the corresponding data processing platform is not limited herein.

[0104] In an example, the computing device cluster can be used to implement a cloud management platform, in other words, the data processing platform of the embodiments of the present application can be implemented through the cloud management platform.

[0105] The cloud management platform is used to manage the infrastructure providing cloud services, and can provide computing, network and storage capabilities based on the services of hardware resources and software resources. For example, the cloud management platform can include one or more data centers to provide cloud resources through one or more data centers.

[0106] The data center is introduced below in combination with an architecture diagram shown in FIG. 1.

[0107] In FIG. 1, in the data center, a cloud management platform interacts with one or more servers (such as server 1 and server 2 in FIG. 1) through an internal network of the data center. The server includes a hardware layer and a software layer. The hardware layer includes hardware configured for the server, wherein a peripheral component interconnect (PCI) device may, for example, be a network card, a graphics processing unit (GPU), an offload card, and the like, which is a peripheral device that can be plugged into a peripheral component interconnect (PCI) or peripheral component interconnect express (PCIe) slot of the server. The software layer includes an operating system installed and running on the server (the operating system of the virtual machine can be referred to as a host operating system), and a virtual machine manager (also referred to as a hypervisor) is set in the host operating system. The virtual machine manager is used to implement computing virtualization, network virtualization, and storage virtualization of the virtual machine, and is responsible for managing the virtual machine. The virtual machine (VM) refers to a complete computer system that is simulated by software and runs in a completely isolated environment and has complete hardware system functions. In the system architecture shown in FIG. 1, a plurality of servers are set in a data center, and the servers can be used to run virtual machines. The specifications of the virtual machines can be the same or different. The virtual machine can also be referred to as an elastic compute service (ECS), an elastic instance, and the like. Different cloud service providers may have different names.

[0108] In an example of an embodiment of the present application, the cloud management platform can be a public cloud platform. At this time, a cloud service provider such as an individual or a software developer who has cloud resource development capabilities can provide cloud services for users, and the users can obtain cloud services through the Internet, but do not own cloud computing resources.

[0109] Specifically, in the example shown in FIG. 1, the cloud management platform can provide an access interface (such as an interface or an application programming interface (API)), and a user of the cloud management platform and a cloud service provider can operate a client to remotely access the access interface to register a cloud account and a password in the cloud management platform, and after the cloud management platform authenticates the cloud account and the password successfully, the user logs in the cloud management platform, thereby creating, managing, logging in, and operating a virtual machine in a cloud data center, and performing a corresponding task through a cloud resource of the cloud management platform.

[0110] For example, as shown in the example of FIG. 2a, the data processing platform implementing the task execution method in the embodiments of the present application can be provided to a user in the form of a cloud service.

[0111] When a data processing task of cross-device data storage needs to be performed, some enterprises, organizations, or individuals can purchase a cloud service including the data processing platform, thereby performing a related data processing task through a cloud resource of the cloud management platform, and obtaining a processing result of input data from the cloud management platform.

[0112] The data processing platform can be implemented based on one or more cloud data centers and the like in the cloud management platform.

[0113] The data processing platform can be provided to a user as an independent cloud service, and can also be compatible with a computing framework such as TensorFlow or PyTorch, that is, can be an operator in a TensorFlow or PyTorch computing framework, and in the computing framework, provide functions such as preprocessing, calculation, feature extraction, and / or other data processing of real-time data and offline data.

[0114] Of course, the cloud management platform can also be other types of cloud management platforms, which are not limited in the embodiments of the present application.

[0115] In another example, as shown in FIG. 2b, the data processing platform can provide the user in the form of a software product or a function service in the software product, the software product can be deployed in a terminal device of the user or a server cluster of an enterprise user, the user can purchase a software license to obtain a use right of the software product, and can implement the task execution method in the embodiments of the present application through the software product within a license period.

[0116] The following takes the task execution method field as an example to illustrate the actual application mode of the data processing platform implementing the task execution method. The data processing platform has the function of the task execution device described in the subsequent embodiments. The task execution device can be a terminal device or a server, for example, the terminal device or the server includes a processor, and the processor is configured to execute the task execution method in the first aspect.

[0117] For example, the standard flow can include computer program code, and the task execution device can be a code executor.

[0118] For another example, the task execution device can be an agent, and the agent can be integrated with a large model. The agent can be used to execute the task execution method provided in the first aspect.

[0119] In the framework of the agent, the agent is endowed with the general ability of the large model to solve one or more specific problems. Different types of agents can be automatically connected to solve more complex problems.

[0120] In an exemplary application scenario, a user can want to transmit the layout defect information of a circuit board to a cloud server for storage, sharing, or subsequent processing. The user can use an AI model to encode the layout defect information of the circuit board through a TensorFlow, PyTorch, or other computing framework, and then send the encoded code stream through a transmission network between the mobile phone and the cloud server. The computing framework can be deployed in a client device, and the cloud server can also deploy the computing framework. After receiving the compressed layout defect information of the circuit board, the cloud management platform provides processing of the layout defect information of the circuit board through cloud resources. For example, the cloud management platform is provided with a large language model, a detection tool library, and a circuit board layout engine. The cloud management platform performs semantic understanding on the layout defect information of the circuit board to obtain a circuit board layout defect description result. The target detection tool is used to detect the circuit board layout defect description result, and the target optimization instruction is obtained according to the information of the electronic components to be optimized on the target circuit board obtained by detection. Finally, the layout optimization of the electronic components is performed based on the target optimization instruction.

[0121] It can be understood that the above example is only an exemplary introduction to an application scenario, and the task execution method of the embodiments of the present application can be applied to multiple fields, which are not limited herein.

[0122] To solve the current need to rely on manual experience to write standard process way, the embodiment of the application provides a task execution method and a corresponding task execution device, and the task execution method can be a task execution method for process task automation. For example, the task execution method provided by the embodiment of the application can realize efficient execution of a plurality of tasks based on a standard process. The task execution method provided by the embodiment of the application can be continuously improved and evolved. Compared with the task execution scheme relying on manual experience, the embodiment of the application can greatly improve the efficiency of the task execution method and the result of the task execution method.

[0123] As shown in FIG. 3, the task execution method device and the terminal device can interact, and the task execution method can include steps 301-305.

[0124] 301. The terminal device sends a task request to the task execution device.

[0125] In the embodiment of the application, the terminal device is a terminal device operated by a user, and the terminal device obtains a task request from the user. For example, the task request includes a task to be executed. Different users have different needs, and the task can be determined in combination with the field. For example, the task is a problem expected to be solved by the user, and the task can belong to various fields. For example, the task in the embodiment of the application can be at least one of the following: analyzing the abnormal reason of order number abcde, helping the user plan a travel itinerary to city A, giving a brief introduction to the personal life of historical figure ABC, and giving dietary suggestions for student id 12345. The field to which the task belongs and the content of the task are not limited. For example, the task request can include description information of the task provided by the user, such as a text description or a voice description of the task to be executed by the user.

[0126] For example, as shown in FIG. 2a, the terminal device in the embodiment of the application can be a client, and the task execution device can be a data center. The client can interact with the data center, and the client can send a task request to the data center.

[0127] 302. The task execution device receives the task request.

[0128] The terminal device can generate a task request, the task execution device and the terminal device interact, the task execution device receives the task request from the terminal device, and analyzes the task request to determine the task to be executed according to the task request.

[0129] For example, a user can control a terminal device to send a task request through the terminal device, for example, the user needs to obtain a diet suggestion for a student with an id of 12345, and the terminal device sends a task request including a task of giving a diet suggestion for the student with the id of 12345. For another example, the task execution apparatus analyzes the task request sent by the terminal device and obtains a task of giving a diet suggestion for the student with the id of 12345.

[0130] The application embodiments do not limit the application scenarios of the task request. For example, the task request can be a root cause positioning task request in the operation and maintenance field, that is, the terminal device can request the task execution apparatus to execute the root cause positioning task. For another example, the task request can be a user-selected branch task request in the intelligent customer service field, that is, the terminal device can request the task execution apparatus to execute a branch task selected according to different tasks of the user.

[0131] 303、The task execution apparatus obtains a first standard process from a standard process information library according to the task request.

[0132] In the application embodiments, the standard process can be used to process various types of problems, which are solved step by step in the form of a flowchart and finally completed through the standard process. The standard process provided in the application embodiments can also be referred to as a standard operation procedure, a procedure, a standard operation process (SOP), a standardized work process, etc. In the subsequent examples, the standard process is taken as SOP for example. The standard process information library includes a plurality of standard processes, and the standard process information library can also be referred to as a SOP long-term memory.

[0133] In the application embodiments, different standard processes are used for different tasks to complete different tasks. In the application embodiments, the task execution apparatus can be configured with a standard process information library, which can include a plurality of different standard processes, and the plurality of different standard processes can be used to solve different tasks. The standard process information library stores a plurality of standard processes. For example, the standard process information library can include a standard process a, a standard process b, and a standard process c, wherein the standard process a can be used to execute a task 1, the standard process b can be used to execute a task 2, and the standard process c can be used to execute a task 3. The fields associated with the plurality of standard processes in the standard process information library and the contents included in the plurality of standard processes are not limited.

[0134] In the application embodiments, the standard process information library can be dynamically updated to continuously expand the standard process information library, so that the standard process information library can include more standard processes.

[0135] In the embodiments of the present application, after the task execution device obtains the task request, the task request is used to search the standard process information library to obtain a first standard process from the standard process information library. The first standard process is a standard process stored in the standard process information library. In the embodiments of the present application, the standard process information library is configured with a plurality of different standard processes to provide the first standard process corresponding to the task request, so that the standard process does not need to be manually configured for the task, and the way of providing the standard process for the task is simplified.

[0136] For example, the task request can include a task of giving a diet suggestion for a student with id 12345. According to the task request, the standard process information library is searched to obtain a first standard process from the standard process information library. The first standard process can provide a diet suggestion and a specified diet plan according to the student's student number, gender, height, or deviation from the standard body weight. For another example, the task request can include a task of providing a brief biography of a basketball star abcde in the sports field. The first standard process can query the team that the basketball star abcde has served and the main match that the basketball star has played according to the name of the basketball star to provide a brief biography of the basketball star abcde.

[0137] In the embodiments of the present application, the standard process library can be searched according to the task request to retrieve a standard process matched with the task request. Through the standard process, a type of task requested to be executed in the task request can be solved.

[0138] The standard process in the embodiments of the present application can include a plurality of sub-standard processes. In the embodiments of the present application, each standard process can include a plurality of sub-standard processes, and the plurality of sub-standard processes are used to generate a complete standard process. For example, the first standard process is a standard process matched with the task request found from the standard process information library. Through the first standard process, the task included in the task request can be disassembled into a series of a plurality of first sub-tasks. Each first sub-task has an explicit execution mode, and the jump logic between the plurality of first sub-tasks is associated through the execution result. The plurality of sub-standard processes included in the first standard process can be used to execute the plurality of sub-tasks. For example, the first standard process can include a sub-standard process a1, a sub-standard process a2, and a sub-standard process a3. The sub-standard process a1 can be used to execute a sub-task 1 and other sub-tasks jumped from the sub-task 1. The sub-standard process a2 can be used to execute a sub-task 2 and other sub-tasks jumped from the sub-task 2. The sub-standard process a3 can be used to execute a sub-task 3 and other sub-tasks jumped from the sub-task 3.

[0139] In some embodiments of the present application, for each standard process in the standard process information base, a sub-standard process information base can be configured, which can include a plurality of sub-standard processes corresponding to the standard process. For example, the first standard process can include: sub-standard process a1, sub-standard process a2 and sub-standard process a3, wherein sub-standard process a1, sub-standard process a2 and sub-standard process a3 belong to sub-standard process information base 1, and for other standard processes in the standard process library except the first standard process, other sub-standard process information bases can also be configured.

[0140] 304、The task execution apparatus acquires the first sub-standard process from the first standard process according to the task request.

[0141] The first sub-standard process is a sub-standard process of the first standard process, and the first sub-standard process is used to indicate that a plurality of first sub-tasks are executed in turn based on a jump relationship. The first sub-standard process includes a first root node, and the first root node is used to indicate execution of a starting sub-task in the plurality of first sub-tasks.

[0142] In the embodiments of the present application, the task execution apparatus acquires the first standard process, the first standard process includes a plurality of sub-standard processes, acquires the first sub-standard process from the first standard process according to the task request, the plurality of sub-standard processes include the first sub-standard process, the first sub-standard process is a sub-standard process matched with the task request, the first sub-standard process includes a first root node, and the first root node represents a first node in a data structure of the first sub-standard process. Therefore, the first root node has no parent node in the first sub-standard process. For example, the first sub-standard process can be a tree data structure. Each node in the first sub-standard process is used to execute a first sub-task, and the first root node is used to execute a starting sub-task in the plurality of first sub-tasks. The starting sub-task can also be referred to as the first sub-task, the first sub-task, or the root sub-task, which is not limited here.

[0143] For example, the first sub-standard process is used to execute the starting sub-task indicated by the first root node and other sub-tasks jumped to by the starting sub-task according to the jump relationship. The starting sub-task can also be regarded as a task entry of the first standard process.

[0144] In some embodiments of the present application, the task execution apparatus retrieves the sub-standard process information base corresponding to the first standard process according to the task request, determines the first sub-standard process, and the first sub-standard process corresponds to the task entry.

[0145] According to the first standard process, the first standard process corresponding sub-standard process information base is obtained, and based on the description of the sub-standard process information base in step 303, the sub-standard process information base can include a plurality of sub-standard processes. The sub-standard process information base stores a plurality of sub-standard processes included in the first standard process, and the plurality of sub-standard processes include a first sub-standard process. After the task execution device obtains the first standard process, the task request is used to retrieve the first standard process corresponding sub-standard process information base to obtain the first sub-standard process from the sub-standard process information base. The first sub-standard process is a sub-standard process stored in the sub-standard process information base, and the first root node included in the first sub-standard process indicates the execution of the starting sub-task in the plurality of first sub-tasks. The starting sub-task can also be considered as the task entry of the first standard process. For example, the task execution device retrieves the first standard process corresponding sub-standard process information base according to the task request, determines the first sub-standard process, and the first sub-standard process corresponds to the task entry.

[0146] In the embodiment of the application, the sub-standard process information base is configured with a plurality of different sub-standard processes to determine the task entry of the first standard process, so that the starting sub-task in the first standard process can be selected flexibly according to the task request, and therefore it can be applied to complex and dynamic different tasks, and the flexibility of executing tasks is improved.

[0147] 305, the task execution device executes the plurality of first sub-tasks according to the first sub-standard process.

[0148] In the embodiment of the application, after the task execution device obtains the first sub-standard process from the first standard process, the starting sub-task can be obtained according to the first sub-standard process, and other sub-tasks jumped to by the starting sub-task can be obtained according to the jump relationship between the first sub-tasks in the first sub-standard process. Then, according to the task content provided by the task request, the starting sub-task is executed using the first sub-standard process. After the execution of the starting sub-task is completed, the sub-task jump is performed according to the first sub-standard process, and the other first sub-tasks jumped to by the starting sub-task are executed using the first sub-standard process. It is not limited that the jump of the first sub-task in the embodiment of the application can include one jump or multiple jumps. When the end condition of the first sub-task is reached, the execution of the first sub-task is stopped, and the execution result for the task request is output.

[0149] As can be known from the foregoing examples, in the embodiments of the present application, the standard process information library is configured with a standard process, and the first standard process corresponding to the task request can be provided through the standard process information library, so that the standard process does not need to be manually configured for the task request, and the manner of providing the standard process for the task request is simplified. In addition, the first sub-standard process can be obtained from the first standard process according to the task request, the first sub-standard process indicates that the plurality of first sub-tasks are executed in sequence, and the first root node included in the first sub-standard process indicates a starting sub-task in the plurality of first sub-tasks, so that the plurality of first sub-tasks can be executed according to the first sub-standard process. Since the first sub-standard process can be obtained from the first standard process according to the task request, the corresponding first sub-standard process can be obtained for different task requests, which can be suitable for complex and dynamic different task requests, improve the flexibility of executing the plurality of first sub-tasks, and improve the execution efficiency and success rate of the plurality of sub-tasks.

[0150] As can be known from the foregoing steps 301 to 305, for the task request of the terminal device, only the terminal device needs to provide the task request to be executed, and the planning manner of obtaining the corresponding standard process on the terminal device side is not needed. The task execution device can configure the standard process information library, and the understanding degree of the standard process configuration information in the actual scene is reduced through the manner of retrieving the standard process information library, and the ease of use of the standard process is improved. At the same time, the retrieval of the task can also improve the coverage of the task.

[0151] In some embodiments of the present application, the standard process information library is constructed in the following manner:

[0152] A1, the task execution device obtains sub-task information of a plurality of second sub-tasks and sub-task jump logic information between the plurality of second sub-tasks, the sub-task jump logic information between the plurality of second sub-tasks is used to indicate a sub-task jump manner between the plurality of second sub-tasks, and the plurality of second sub-tasks are used to jump the second sub-tasks in sequence according to a jump relationship indicated by a sixth standard process;

[0153] A2, the second sub-task is jumped according to a reverse sub-task jump manner between the plurality of second sub-tasks, and sub-task information of the second sub-task after the jump is added to a node of the second sub-task after the jump to obtain description information of the sixth standard process, the reverse sub-task jump manner is opposite to the sub-task jump manner indicated by the sub-task jump logic information;

[0154] A3, the task execution device constructs a standard process information library according to the description information of the sixth standard process, and the sixth standard process is any one of the standard processes in the standard process information library.

[0155] The sixth standard process is any one of the standard processes in the standard process information library, and the construction of any one of the standard processes in the standard process information library is described in steps A1 to A3. Specifically, first, the task execution apparatus obtains the subtask information of the plurality of second subtasks and the subtask jump logic information between the plurality of second subtasks. The subtask information of the second subtask is used to describe the task content of the second subtask. The subtask jump logic information between the plurality of second subtasks includes information that the plurality of second subtasks jump in sequence according to the jump relationship indicated by the sixth standard process. The subtask jump logic information between the plurality of second subtasks indicates the subtask reverse jump manner between the plurality of second subtasks. The plurality of second subtasks can jump in sequence according to the jump relationship to implement a complete task. Then, the task execution apparatus obtains the subtask reverse jump manner opposite to the subtask jump manner between the plurality of second subtasks, jumps the second subtasks according to the subtask reverse jump manner, and adds the subtask information of the second subtask after the jump to the node of the second subtask after the jump. For example, the jump of the second subtask can be based on a graph structure. In the graph structure, the node of the second subtask after the jump adds the subtask information of the second subtask after the jump. That is, the task execution apparatus can obtain the description information of the sixth standard process by aggregating the subtask information of the plurality of second subtasks. For example, the subtask jump logic information indicates that the second subtask at the starting position jumps other second subtasks in sequence according to the jump relationship, and then aggregates the subtask information of the plurality of second subtasks to obtain the description information of the sixth standard process. The description information of the sixth standard process is obtained through the above steps A1-A2. The description information of the plurality of standard processes in the standard process information library can be obtained through the A1-A2 manner. Therefore, the standard process information library is constructed through the step A3. The embodiments of the present application construct the standard process information library through the description information of the plurality of standard processes, so as to provide the standard process information library for a plurality of different tasks and simplify the manner of providing the standard process.

[0156] Further, in some embodiments of the present application, the A2 task execution apparatus jumps the second subtasks according to the subtask reverse jump manner between the plurality of second subtasks, and adds the subtask information of the second subtask after the jump to the node of the second subtask after the jump to obtain the description information of the sixth standard process, including:

[0157] A21, the task execution apparatus jumps according to the subtask reverse jump manner between the plurality of second subtasks, and adds the subtask information of the second subtask after the jump to the node of the second subtask after the jump to obtain the description information of the plurality of sub-standard processes of the sixth standard process;

[0158] A22, the task execution apparatus generates description information of the sixth standard process according to description information of a plurality of sub-standard processes of the sixth standard process.

[0159] The task execution apparatus can obtain sub-task jump logic information between a plurality of second sub-tasks, the sub-task jump logic information indicating a second sub-task at a starting position, starting from the second sub-task at the starting position, jumping in a reverse sub-task jump manner, adding sub-task information of the jumped second sub-task to the node of the jumped second sub-task, thereby converging sub-task information of a plurality of second sub-tasks, and finally obtaining description information of a plurality of sub-standard processes of the sixth standard process. Finally, the task execution apparatus generates description information of the sixth standard process according to description information of a plurality of sub-standard processes of the sixth standard process. In the embodiment of the application, the reverse sub-task jump is used to converge a plurality of sub-task information to obtain description information of a plurality of sub-standard processes, so that the description information of the sixth standard process can be generated based on the description information of a plurality of sub-standard processes, and the generation of the description information of the sixth standard process is simplified.

[0160] Further, in some embodiments of the application, the A3 task execution apparatus constructs a standard process information library according to the description information of the sixth standard process, including:

[0161] A31, the task execution apparatus generates a first semantic representation corresponding to the sixth standard process according to the description information of the sixth standard process;

[0162] A32, the task execution apparatus constructs a standard process information library according to the first semantic representation corresponding to the sixth standard process.

[0163] The task execution apparatus can perform semantic analysis on the description information of the sixth standard process to obtain a first semantic representation corresponding to the sixth standard process. The semantic algorithm used for semantic analysis is not limited, for example, the description information of the sixth standard process is analyzed by a large language model (Large Language Model, LLM) to obtain a first semantic representation. In the embodiment of the application, the first semantic representation corresponding to the sixth standard process is one or more vector representations corresponding to the standard process, and the first semantic representation can be a sparse or dense semantic representation. The task execution apparatus generates a standard process information library according to the first semantic representation corresponding to the sixth standard process, and obtains a plurality of first semantic representations corresponding to a plurality of standard processes. In the embodiment of the application, the first semantic representation corresponding to each standard process can be used for retrieval to realize the matching between the task and the standard process.

[0164] It is not limited that the standard process information library in the embodiments of the present application stores a plurality of standard processes and corresponding first semantic representations, and in addition, an index value of the corresponding standard process can be generated according to the first semantic representation corresponding to the plurality of standard processes, that is, the index values of the plurality of standard processes are used as keys, and the first semantic representations corresponding to the plurality of standard processes are used as values, and the key-value pairs are stored in the standard process information library.

[0165] In some embodiments of the present application, the standard process information library comprises a plurality of standard processes and a plurality of first semantic representations corresponding to the plurality of standard processes respectively.

[0166] The acquisition method of the first semantic representation corresponding to each of the plurality of standard processes can refer to the description of step A31.

[0167] For example, as shown in FIG. 4, taking SOP as a standard process and sub-SOP as a sub-standard process to illustrate the construction process of the SOP information library. For example, the task request can include a task of providing dietary advice for student id 12345, and the task request can provide dietary advice and specify a dietary plan according to the student's student number, gender, height, or deviation from the standard body weight. The acquisition process of the sub-SOP description information and the SOP description information is as follows: the sub-SOP corresponding to the second sub-task 3 is the description information of the second sub-task, that is, "providing dietary advice and specifying a dietary plan according to the student's deviation from the standard body weight"; the sub-SOP corresponding to the second sub-task 2 is obtained by combining the downstream sub-SOP (i.e., sub-SOP 3) and the task information of the second sub-task 2, that is, "providing dietary advice and specifying a dietary plan according to the student's gender and height"; similarly, the sub-SOP corresponding to the second sub-task 1 is obtained by combining the downstream sub-SOP (i.e., sub-SOP 2) and the task information of the second sub-task 1, that is, "providing dietary advice and specifying a dietary plan according to the student's student number". Therefore, the description information of the overall SOP can be obtained by combining the above sub-SOP information, for example, the description information of the SOP can be "providing dietary advice and specifying a dietary plan according to the student's student number, gender, height, or deviation from the standard body weight".

[0168] The above FIG. 4 illustrates the construction of the SOP information library by taking one task as an example, and the embodiments of the present application can also extend other SOPs in the SOP information library in a similar manner, for example, constructing a SOP for a brief introduction of a historical figure ABC.

[0169] In some embodiments of the present application, the task execution device acquires a first standard process from the standard process information library according to a task request, comprising:

[0170] B1, the task execution device performs semantic analysis on the task request to obtain a target semantic representation;

[0171] B2, the task execution device performs semantic matching on the target semantic representation and the standard process information library to obtain a first semantic representation matched with the target semantic representation, the standard process information library including a plurality of standard processes and a plurality of first semantic representations corresponding to the plurality of standard processes respectively;

[0172] B3, the task execution device acquires a first standard process from the standard process information library according to the first semantic representation matched with the target semantic representation.

[0173] The task execution device can perform semantic analysis on the task request to obtain the target semantic representation. The semantic algorithm used for semantic analysis is not limited, for example, the task request is analyzed by a large language model (LLM) to obtain the target semantic representation. Based on the foregoing process of the standard process information library, it can be known that the standard process information library stores a plurality of standard processes and corresponding first semantic representations. Therefore, the task execution device performs semantic matching on the target semantic representation and the standard process information library to obtain a first semantic representation matched with the target semantic representation. Finally, the task execution device acquires a first standard process from the standard process information library according to the first semantic representation matched with the target semantic representation. In the embodiments of the present application, the matching of the target semantic representation and the first semantic representation can quickly acquire the first standard process from the standard process information library.

[0174] In some embodiments of the present application, the sub-standard process information library is constructed in the following manner:

[0175] C1, the task execution device acquires sub-task information of a plurality of first sub-tasks and sub-task jump logic information between the plurality of first sub-tasks, the sub-task jump logic information between the plurality of first sub-tasks being used to indicate a sub-task jump manner between the plurality of first sub-tasks;

[0176] C2, the task execution device performs jump of the first sub-task according to the sub-task reverse jump manner between the plurality of first sub-tasks, and adds sub-task information of the jumped first sub-task to a position node of the jumped first sub-task to obtain description information of a plurality of sub-standard processes of the first standard process;

[0177] C3, the task execution device constructs a sub-standard process information library according to the description information of the plurality of sub-standard processes of the first standard process.

[0178] The construction manner of the plurality of sub-standard processes in the sub-standard process information library is described in steps C1 to C3. Specifically, first, sub-task information of a plurality of first sub-tasks and sub-task jump logic information between the plurality of first sub-tasks are acquired. The sub-task information of the first sub-task is used to describe the task content of the first sub-task. The sub-task jump logic information between the plurality of first sub-tasks includes information that the plurality of first sub-tasks jump in turn according to a jump order. Then, the task execution apparatus acquires a sub-task reverse jump manner opposite to the sub-task jump manner between the plurality of first sub-tasks, jumps the first sub-tasks according to the sub-task reverse jump manner, and adds the sub-task information of the jumped first sub-task to the node of the jumped first sub-task. For example, the jump of the first sub-task can be based on a graph structure. In the graph structure, the node of the jumped first sub-task adds the sub-task information of the jumped first sub-task. That is, the task execution apparatus can gather the sub-task information of the plurality of first sub-tasks according to the sub-task jump logic information between the plurality of first sub-tasks to obtain the description information of the plurality of sub-standard processes in the first standard process. For example, the sub-task jump logic information indicates that the first sub-task located at a starting position sequentially finds other first sub-tasks according to a jump relationship, and gathers the sub-task information of the plurality of first sub-tasks to obtain the description information of the plurality of sub-standard processes in the first standard process. The description information of the plurality of sub-standard processes in the first standard process is obtained through the above steps C1-C2 manner, and therefore the sub-standard process information library is constructed through the step C3 manner. The sub-standard process information library is constructed by the description information of the plurality of sub-standard processes in the application embodiment, so that the sub-standard process information library can be provided for a plurality of different tasks, and the manner of providing the sub-standard process is simplified.

[0179] Further, in some embodiments of the application, the C3 task execution apparatus constructs the sub-standard process information library according to the description information of the plurality of sub-standard processes in the first standard process, including:

[0180] C31, the task execution apparatus generates a second semantic representation corresponding to the plurality of sub-standard processes in the first standard process according to the description information of the plurality of sub-standard processes in the first standard process.

[0181] C32, the task execution apparatus constructs the sub-standard process information library according to the second semantic representation corresponding to the plurality of sub-standard processes in the first standard process.

[0182] The task execution apparatus can perform semantic analysis on the plurality of sub-standard processes of the first standard process to obtain second semantic representations corresponding to the plurality of sub-standard processes. The semantic algorithm used for semantic analysis is not limited, for example, the description information of the plurality of sub-standard processes of the first standard process is analyzed by a large language model (LLM) to obtain the second semantic representation. In the embodiment of the application, the second semantic representation corresponding to the plurality of sub-standard processes of the first standard process is one or more vector representations corresponding to the plurality of sub-standard processes. The second semantic representation can be a sparse or dense semantic representation. The task execution apparatus generates a sub-standard process information library according to the second semantic representation corresponding to the plurality of sub-standard processes of the first standard process, and obtains the second semantic representation corresponding to the plurality of standard processes. In the embodiment of the application, the second semantic representation corresponding to each sub-standard process can be used for retrieval to realize the matching between the task and the sub-standard process.

[0183] In some embodiments of the application, the sub-standard process information library includes the plurality of sub-standard processes of the first standard process and the second semantic representation corresponding to the plurality of sub-standard processes of the first standard process.

[0184] The method for obtaining the second semantic representation corresponding to the plurality of standard processes can refer to the description of step C31.

[0185] In some embodiments of the application, the task execution apparatus obtains a first sub-standard process from the first standard process according to a task request, including:

[0186] D1, the task execution apparatus obtains a sub-standard process information library from the first standard process, the sub-standard process information library includes a plurality of sub-standard processes, and the plurality of sub-standard processes belong to the first standard process;

[0187] D2, the task execution apparatus obtains a first sub-standard process from the sub-standard process information library according to a task request.

[0188] The task execution apparatus establishes a sub-standard process information library, and the plurality of sub-standard processes included in the sub-standard process information library belong to the first standard process. The task execution apparatus can also obtain the first sub-standard process by retrieving the sub-standard process information library. The retrieval of the sub-standard process information library improves the task execution efficiency, and the execution of the standard process starts from the starting sub-task indicated by the first root node included in the first sub-standard process, reducing unnecessary calculation.

[0189] Further, in some embodiments of the application, step D2, the task execution apparatus obtains a first sub-standard process from the sub-standard process information library according to a task request, including:

[0190] D21, the task execution apparatus performs semantic analysis on the task request to obtain a target semantic representation;

[0191] D22, the task execution device performs semantic matching on the target semantic representation and the sub-standard process information library to obtain a second semantic representation matched with the target semantic representation, the sub-standard process information library including a plurality of sub-standard processes of the first standard process and second semantic representations corresponding to the plurality of sub-standard processes of the first standard process respectively;

[0192] D23, the task execution device obtains the first sub-standard process from the sub-standard process information library according to the second semantic representation matched with the target semantic representation.

[0193] The task execution device can perform semantic analysis on the task request to obtain the target semantic representation. The semantic algorithm used for semantic analysis is not limited, for example, the task request is analyzed by a large language model LLM to obtain the target semantic representation. Based on the foregoing process of the sub-standard process information library, it can be known that a plurality of sub-standard processes and corresponding second semantic representations are stored in the sub-standard process information library. Therefore, the task execution device performs semantic matching on the target semantic representation and the sub-standard process information library to obtain a second semantic representation matched with the target semantic representation. Finally, the task execution device obtains the first sub-standard process from the sub-standard process information library according to the second semantic representation matched with the target semantic representation. In the embodiment of the application, the matching of the target semantic representation and the second semantic representation can quickly obtain the first sub-standard process from the sub-standard process information library.

[0194] In the embodiment of the application, in addition to the task execution method shown in FIG. 3, the embodiment of the application also provides another task execution method, which can include the following steps:

[0195] E1, the task execution device obtains a second standard process according to a task request, the second standard process being a standard process generated according to the task request and an experience database;

[0196] E2, the task execution device stores the second standard process into a standard process information library.

[0197] The task execution device and the terminal device interact, receive a task request from the terminal device, and analyze the task request. The task execution device can obtain an experience database, and the experience database can provide expert experience information for generating a standard process. The second standard process can be generated through the experience database and the task request. The second standard process is stored into the standard process information library to realize dynamic updating of the standard process information library. In some optional implementation manners, the second standard process can also be stored into a sub-standard process information library to realize dynamic updating of the sub-standard process information library.

[0198] Further, in some embodiments of the present application, for example, when the first standard process is not obtained from the standard process information library, the standard process information library can be dynamically updated through steps E1 and E2, so that the standard process information library obtains the standard process and the sub-standard process matched with the task request next time. For another example, the standard process information library and the sub-standard process information library can be dynamically updated through steps E1 and E2, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request next time.

[0199] In some embodiments of the present application, the task execution method executed by the task execution device can include the following steps:

[0200] F1, the task execution device obtains first multi-modal information according to the task request, the first multi-modal information including multiple types of information required for generating standard process information;

[0201] F2, the task execution device inputs the first multi-modal information into a large model, and outputs a third standard process through the large model, the large model being used for generating standard process information corresponding to multiple tasks respectively;

[0202] F3, the task execution device stores the third standard process into a standard process information library and a sub-standard process information library.

[0203] Wherein, the task execution device obtains the first multi-modal information according to the task request, for example, the first multi-modal information includes multiple types of information required for generating standard process information; the multi-modal type of the first multi-modal information is not limited, for example, the first multi-modal information can be voice data, or image data, or text data, etc.

[0204] For example, the first multi-modal information can include process graph, language description, structured data and other multi-modal information, the process graph can be the process graph shown in FIG. 4; the language description information can be: obtaining detailed information such as gender, height, weight according to the student ID, if there is, then according to the gender and height of the student, the corresponding standard weight is searched, and in the case of searching the standard weight, the diet suggestion is provided according to the deviation of the student's weight from the standard weight, and the diet plan is specified. The structured information can include structured configuration and other information obtained by other fields or other structures.

[0205] The task execution device in the embodiments of the present application can train a large model, use the trained large model to predict the first multi-modal information, and output a third standard process. The large model is used to generate standard process information corresponding to a plurality of tasks respectively. When the first standard process is not obtained from the standard process information library, the third standard process is stored in the standard process information library to realize dynamic updating of the standard process information library, so that the standard process information library obtains the standard process and the sub-standard process matched with the task request during the next retrieval. For example, the third standard process is stored in the standard process information library and the sub-standard process information library to realize dynamic updating of the standard process information library and the sub-standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request during the next retrieval.

[0206] In some embodiments of the present application, the task execution method performed by the task execution device can include the following steps:

[0207] G1, the task execution device obtains a fourth standard process, stores the fourth standard process in the standard process information library, and the fourth standard process is a standard process information generated according to an experience database;

[0208] Or,

[0209] H1, the task execution device obtains second multi-modal information, and the second multi-modal information includes a plurality of types of information required for generating standard process information;

[0210] H2, the task execution device inputs the second state information into a large model, and outputs a fifth standard process through the large model. The large model is used to generate standard process information corresponding to a plurality of tasks respectively;

[0211] H3, the task execution device stores the fifth standard process in the standard process information library and the sub-standard process information library.

[0212] The experience database can provide expert experience information for generating a standard process. In the case that the task request does not provide a task, the fourth standard process can be generated through the experience database, and the fourth standard process is stored in the standard process information library to realize dynamic updating of the standard process information library, so that the standard process information library obtains the standard process and the sub-standard process matched with the task request during the next retrieval. For example, the fourth standard process is stored in the standard process information library and the sub-standard process information library to realize dynamic updating of the standard process information library and the sub-standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request during the next retrieval.

[0213] The task execution apparatus in the embodiments of the present application can train a large model, use the trained large model to predict the second multi-modal information to output a fifth standard process, the large model is used to generate standard process information corresponding to a plurality of tasks respectively, store the fifth standard process to the standard process information library, so as to realize dynamic updating of the standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request when searching next time. For example, the fifth standard process is stored to the standard process information library and the sub-standard process information library, so as to realize dynamic updating of the standard process information library and the sub-standard process information library, so that the standard process information library and the sub-standard process information library obtain the standard process and the sub-standard process matched with the task request when searching next time.

[0214] In some embodiments of the present application, the task execution method executed by the task execution apparatus can also detect abnormalities for the starting subtask and the subtask jumped to by the starting subtask, to detect problems existing in the subtask execution process. Specifically, the task execution method executed by the task execution apparatus can include the following steps:

[0215] I1. The task execution apparatus obtains an action sequence corresponding to the first sub-standard process in the process of executing a plurality of first subtasks through the action model, the plurality of actions in the action sequence are used to execute the plurality of first subtasks or subtask jumping, and the plurality of actions in the action sequence are realized based on the subtask information or the subtask jumping logic information of the plurality of first subtasks;

[0216] I2. The task execution apparatus performs abnormal evaluation on the action sequence through the first evaluation model according to the execution results of the plurality of actions in the action sequence, the subtask information of the plurality of first subtasks and the subtask jumping logic information, to obtain suspicious actions from the action sequence;

[0217] I3. The task execution apparatus performs abnormal detection on the suspicious actions through the second evaluation model according to the execution results of the suspicious actions and the subtask information of the first subtask corresponding to the suspicious actions, or according to the execution results of the suspicious actions and the subtask jumping logic information of the first subtask corresponding to the suspicious actions, to determine whether the suspicious actions are abnormal actions.

[0218] The task execution device can detect a process of executing a plurality of first sub-tasks by an action model, the action model can be used to execute a plurality of sub-actions in a first sub-standard process, the task execution device obtains an action sequence corresponding to the first sub-standard process, a plurality of actions in the action sequence are used to execute corresponding sub-tasks or sub-task jumps, the plurality of actions in the action sequence are implemented based on sub-task information or sub-task jump logic information of the plurality of first sub-tasks, and execution results of the plurality of actions in the action sequence can be obtained by executing the plurality of actions in the action sequence, for example, the execution results of the plurality of actions in the action sequence can be stored in the SOP short-term memory. The task execution device can train a first evaluation model, and use the first evaluation model after training to perform abnormal evaluation according to the execution results of the plurality of actions in the action sequence and the sub-task information and the sub-task jump logic information of the plurality of first sub-tasks, and obtain a suspicious action from the action sequence. The abnormal evaluation refers to evaluating whether the action sequence is abnormal as a whole by the first evaluation model, and the suspicious action is obtained from the action sequence when the action sequence is abnormal. The task execution device can train a second evaluation model, the task execution device obtains an execution result of the suspicious action, and uses the second evaluation model after training to perform abnormal detection on the suspicious action according to the execution result of the suspicious action and the sub-task information of the first sub-task corresponding to the suspicious action, or according to the execution result of the suspicious action and the sub-task jump logic information of the first sub-task corresponding to the suspicious action, wherein the abnormal detection refers to evaluating whether the suspicious action is abnormal locally by the second evaluation model, and the suspicious action is determined as an abnormal action when the suspicious action is abnormal.

[0219] In the embodiments of the present application, the first evaluation model and the second evaluation model can be large models, the first evaluation model can perform overall abnormal evaluation, and the second evaluation model can perform local abnormal detection. Through the coarse-to-fine two-stage evaluation method, problems in the process can be efficiently found, and the efficiency and effect of process execution are considered.

[0220] Further, in some embodiments of the application, the task execution method executed by the task execution device in addition to the foregoing steps I1 to I3 can further include the following steps:

[0221] J1, the task execution device performs optimization on the abnormal action according to the execution result of the abnormal action and the sub-task information of the sub-task corresponding to the abnormal action, or according to the execution result of the abnormal action and the sub-task jump logic information of the first sub-task corresponding to the abnormal action, and obtains an execution result of the optimized action;

[0222] J2, the task execution device evaluates the execution result of the optimized action by the second evaluation model to obtain a first evaluation result.

[0223] The task execution apparatus can train the optimization model, use the trained optimization model to perform action optimization according to the execution result of the abnormal action and the subtask information or subtask jump logic information of the subtask corresponding to the abnormal action, and obtain the execution result of the optimized action. For example, the optimization model can be a large model.

[0224] The task execution apparatus calls the second evaluation model again to perform local evaluation, evaluates the execution result of the optimized action through the second evaluation model, and obtains the first evaluation result. In the embodiments of the present application, the abnormal action is optimized through the optimization model, and the execution result of the optimized action is evaluated through the second evaluation model, so that whether the optimized action still has a problem can be detected to determine whether the optimization of the abnormal action is successful.

[0225] In some other embodiments of the present application, the task execution method performed by the task execution apparatus, in addition to performing the foregoing steps I1 to I3 and J1 to J2, can further include the following steps:

[0226] K1, when the first evaluation result indicates that the execution result of the optimized action has a problem, the task execution apparatus performs backtracking through a backtracking model according to the execution result of the optimized action and the subtask information and subtask jump logic information of the plurality of first subtasks to obtain the first associated action causing the problem from the action sequence and obtain historical backtracking information;

[0227] K2, the task execution apparatus performs optimization through the optimization model from the first associated action causing the problem until the abnormal action is optimized, and obtains the execution result of the optimized first associated action;

[0228] K3, the task execution apparatus evaluates the execution result of the optimized first associated action through the second evaluation model and obtains the second evaluation result.

[0229] The task execution apparatus can train the backtracking model according to the execution result of the optimized action and the subtask information and subtask jump logic information of the plurality of first subtasks, use the trained backtracking model to perform action backtracking, obtain the first associated action causing the problem from the action sequence, and obtain the historical backtracking information. For example, the backtracking model can be a large model.

[0230] The task execution apparatus determines a first associated action causing the problem from the action sequence through the backtracking model, and calls the optimization model again to optimize from the first associated action causing the problem until an abnormal action is optimized, to obtain an execution result of the optimized first associated action. The second evaluation model is called again to evaluate the execution result of the optimized first associated action, to obtain a second evaluation result. In the embodiments of the present application, the first associated action is optimized through the optimization model until the abnormal action is optimized, and the execution result of the optimized first associated action is evaluated through the second evaluation model, which can detect whether the action after the second optimization still has a problem, to determine whether the optimization from the first associated action to the abnormal action is successful.

[0231] In some other embodiments of the present application, the task execution method performed by the task execution apparatus, in addition to performing the foregoing steps I1 to I3, J1 to J2, K1 to K3, can further include the following steps:

[0232] L1, when the second evaluation result indicates that the execution result of the optimized first associated action has a problem, the task execution apparatus performs backtracking according to the historical backtracking information, the execution result of the optimized first associated action, the subtask information and the subtask jump logic information of the plurality of first subtasks through the backtracking model, to obtain a second associated action causing the problem from the action sequence;

[0233] L2, the task execution apparatus performs optimization from the second associated action through the optimization model until an abnormal action is optimized, to obtain an execution result of the optimized second associated action;

[0234] L3, the task execution apparatus evaluates the execution result of the optimized second associated action through the second evaluation model, to obtain a third evaluation result.

[0235] When the second evaluation result indicates that the execution result of the optimized first associated action has a problem, the task execution apparatus can call the backtracking model again to perform backtracking according to the historical backtracking information, the execution result of the optimized first associated action, and the subtask information and the subtask jump logic information of the plurality of first subtasks, to obtain a second associated action causing the problem from the action sequence, and call the optimization model again to optimize from the second associated action causing the problem until an abnormal action is optimized, to obtain an execution result of the optimized second associated action. The second evaluation model is called again to evaluate the execution result of the optimized second associated action, to obtain a third evaluation result. In the embodiments of the present application, the second associated action is optimized through the optimization model until the abnormal action is optimized, and the execution result of the optimized second associated action is evaluated through the second evaluation model, which can detect whether the action after the second optimization still has a problem, to determine whether the optimization from the second associated action to the abnormal action is successful.

[0236] It can be understood that if the third evaluation result indicates that there is a problem with the execution result of the optimized second associated action, steps L1 to L3 can still be executed again until the evaluation result output by the second evaluation model indicates that there is no problem with the execution result of the optimized associated action. In some other embodiments, when the foregoing steps L1 to L3 are executed multiple times, an end condition can also be set, and the optimized action is output when the end condition is met.

[0237] Further, in some embodiments of the application, when the first evaluation result indicates that there is no problem with the execution result of the optimized action, it means that the optimization of the abnormal action is successful, and the optimization model is used to optimize the abnormal action to overcome the problem of the action.

[0238] M1, when the first evaluation result indicates that there is no problem with the execution result of the optimized action, the method provided by the embodiments of the application further comprises:

[0239] The task execution apparatus updates the standard process information base by using the optimized action.

[0240] and / or,

[0241] M2, the task execution apparatus trains at least one of the following models by using the optimized action: the action model, the first evaluation model, and the second evaluation model.

[0242] In the embodiments of the application, when the first evaluation result indicates that there is no problem with the execution result of the optimized action, the task execution apparatus can update the information base by using the optimized action, and / or optimize the model, the information base can include the standard process information base, or the updated database includes the standard process information base and the sub-standard process information base, and the model is trained: the action model, the first evaluation model, and the second evaluation model. By updating the information base and / or optimizing the model, the information base and the model can be used for subsequent task execution, and the efficiency and success rate of task execution can be improved.

[0243] Further, in some embodiments of the application, when the second evaluation result indicates that there is no problem with the execution result of the optimized first associated action, it means that the optimization of the first associated action to the abnormal action is successful, and the optimization model is used to optimize the first associated action to the abnormal action to overcome the problem of the action.

[0244] When the second evaluation result indicates that there is no problem with the execution result of the optimized first associated action, the method provided by the embodiments of the application further comprises:

[0245] N1, the task execution apparatus updates the standard process information base by using the optimized first associated action.

[0246] And / or,

[0247] N2. The task execution device trains at least one of the following models using the optimized first associated action: action model, first evaluation model, second evaluation model, and optimization model.

[0248] In this embodiment, when the second evaluation result indicates that the execution result of the optimized first associated action has no problems, the task execution device can update the information base and / or optimize the model using the optimized action. The information base may include a standard process information base, or the updated database may include a standard process information base and a sub-standard process information base. The model is then trained: an action model, a first evaluation model, a second evaluation model, and an optimization model. By updating the information base and / or optimizing the model, the information base and model can be used for subsequent task execution, improving the efficiency and success rate of task execution.

[0249] Furthermore, in some embodiments of the application, when the third evaluation result indicates that there are no problems with the execution result of the optimized second associated action, it means that the optimization of the second associated action up to the abnormal action is successful. The optimization of the second associated action up to the abnormal action through the optimization model overcomes the problems existing in the action.

[0250] When the third evaluation result indicates that there are no problems with the execution result of the optimized second associated action, the method provided in this application embodiment further includes:

[0251] P1. The task execution device updates the standard process information database using the optimized second associated action;

[0252] And / or,

[0253] P2. The task execution device trains at least one of the following models using the optimized second associated action: action model, first evaluation model, second evaluation model, backtracking model, and optimization model.

[0254] In this embodiment, when the third evaluation result indicates that there are no problems with the execution result of the optimized second associated action, the task execution device can update the information base and / or optimize the model using the optimized action. The information base may include a standard process information base, or the updated database may include a standard process information base and a sub-standard process information base. The model is then trained as follows: an action model, a first evaluation model, a second evaluation model, a backtracking model, and an optimization model. By updating the information base and / or optimizing the model, the information base and model can be used for subsequent task execution, improving the efficiency and success rate of task execution.

[0255] In the embodiments of the present application, a task execution method implemented by an agent is also provided. The agent refers to an agent that can perceive the environment and take actions to achieve a specific goal. The agent can be software, hardware or a system, and has autonomy, adaptability and interaction ability. The agent can perceive changes in the environment, make judgments and decisions according to the knowledge and algorithms learned by itself, and then perform actions to affect the environment or achieve the predetermined goal. The agent is widely used in the field of artificial intelligence. The agent can autonomously learn and continuously evolve to better complete tasks.

[0256] Specifically, the task execution method implemented by the agent can specifically include the following steps:

[0257] S1, the agent receives a task request from a terminal device.

[0258] The agent and the terminal device interact, receive the task request from the terminal device, and analyze the task request.

[0259] In the embodiments of the present application, the task request can be a task in various business scenarios, which is not limited here. Specifically, the task request can be in the fields of smart home, autonomous driving, games, intelligent question and answer, communication, etc. For example, in the field of smart home, the task request can be that the user initiates fault detection of smart home appliances, or the user controls the customized route of the sweeping robot in the whole house. In the field of autonomous driving, the task request can be that the user initiates automatic obstacle avoidance of the intelligent vehicle, or controls the intelligent vehicle to update the map. In the field of games, the task request can be that the game player controls the game character to automatically find the way, or controls a game character to perform multi-task allocation with other game characters. In the field of communication, the task request can be specifically to evaluate the signal quality of the attached terminal in the cell of the base station, etc.

[0260] The application scenarios of the task request in the embodiments of the present application are not limited, and can be determined in combination with the business scenarios involved in the task sent by the terminal device and the field to which the framework of the agent is applied.

[0261] S2, the agent acquires a first standard process from a standard process information library according to the task request.

[0262] In the embodiments of the present application, the agent can be configured with a standard process information library. The standard process information library can include a plurality of different standard processes, and the plurality of different standard processes can be used to solve different tasks. The standard process information library stores a plurality of standard processes, and the plurality of standard processes include the first standard process.

[0263] In the embodiments of the present application, the intelligent agent can dynamically update the standard process information base to continuously expand the standard process information base, so that the standard process information base can include more standard processes.

[0264] In the embodiments of the present application, after the intelligent agent obtains the task request, the intelligent agent uses the task request to search the standard process information base to obtain a first standard process from the standard process information base, and the first standard process is a standard process stored in the standard process information base. In the embodiments of the present application, the standard process information base is configured with a plurality of different standard processes to be able to provide a first standard process corresponding to the task request in the task request, so that it is not necessary to manually configure a standard process for the task, and the way of providing a standard process for the task is simplified.

[0265] S3, the intelligent agent obtains a first sub-standard process from the first standard process according to the task request, and the first sub-standard process is a sub-standard process of the first standard process.

[0266] The first sub-standard process is used to indicate that a plurality of first sub-tasks are executed in turn based on a jump relationship, and the first sub-standard process includes a first root node, and the first root node is used to indicate execution of a starting sub-task in the plurality of first sub-tasks.

[0267] The standard process in the embodiments of the present application can include a plurality of sub-standard processes, and for each standard process including a plurality of sub-standard processes, the plurality of sub-standard processes are used to generate a complete standard process. For example, the first standard process is used to execute the task request, and the task request can be disassembled into a series of a plurality of sub-tasks through the first standard process, each sub-task has an explicit execution manner, and the jump logic between the plurality of sub-tasks is associated through the execution result.

[0268] For example, the intelligent agent can configure a sub-standard process information base for each standard process in the standard process information base, and the sub-standard process information base can include a plurality of sub-standard processes corresponding to the standard process, and searching the sub-standard process information base can find the first sub-standard process.

[0269] In the embodiment of the present application, after the agent obtains the first standard process for executing the task request, the first standard process corresponding sub-standard process information base is obtained according to the first standard process. The sub-standard process information base can include a plurality of sub-standard processes. The sub-standard process information base stores a plurality of sub-standard processes included in the first standard process. After the agent obtains the first standard process, the task request is used to search the sub-standard process information base corresponding to the first standard process to obtain a first sub-standard process from the sub-standard process information base. The first sub-standard process is a sub-standard process stored in the sub-standard process information base. The first root node included in the first sub-standard process indicates a starting sub-task in a plurality of first sub-tasks. The starting sub-task can also be considered as the task entry of the first standard process. For example, the agent searches the sub-standard process information base corresponding to the first standard process according to the task request to determine the first sub-standard process. The first sub-standard process corresponds to the task entry.

[0270] In the embodiment of the present application, the sub-standard process information base is configured with a plurality of different sub-standard processes to determine the task entry of the first standard process, so that the starting sub-task in the first standard process can be flexibly selected according to the task request. Therefore, it can be applied to complex and dynamic different tasks, and improve the flexibility of executing the task request.

[0271] S4, the agent executes a plurality of first sub-tasks according to the first sub-standard process.

[0272] After the agent obtains the first sub-standard process, the starting sub-task and other first sub-tasks jumped by the starting sub-task can be obtained according to the first sub-standard process. Then, according to the task content provided by the task request, the starting sub-task is executed using the first sub-standard process. After the execution of the starting sub-task is completed, the sub-task jump is performed according to the first sub-standard process, and the other first sub-tasks jumped by the starting sub-task are executed using the first sub-standard process. It is not limited that the jump of the sub-task in the embodiment of the present application can be one jump or multiple jumps. When the end condition of the first sub-task is reached, the execution of the first sub-task is stopped, and the execution result for the task request is output.

[0273] According to the task execution method provided in the embodiments of the present application, the agent is configured with a standard process in the standard process information base, and the standard process information base can provide a first standard process corresponding to the task request, so that the standard process does not need to be manually configured for the task request, and the way of providing the standard process for the task request is simplified. In addition, the agent can obtain a first sub-standard process from the first standard process according to the task request, the first sub-standard process indicates that a plurality of first sub-tasks are executed in turn, and a first root node included in the first sub-standard process indicates a starting sub-task in the plurality of first sub-tasks, so that the plurality of first sub-tasks can be executed according to the first sub-standard process. Since the first sub-standard process can be obtained from the first standard process according to the task request, the corresponding first sub-standard process can be obtained for different task requests, which can be suitable for different task requests with complex dynamics, improve the flexibility of executing the plurality of first sub-tasks, and improve the execution efficiency and success rate of the plurality of sub-tasks.

[0274] For the method steps performed by the agent in the embodiments of S1 to S4 described above, the contents in the embodiment shown in the foregoing FIG. 3 can be referred to, and the examples in the foregoing steps 302 to S305 are described in detail.

[0275] In order to better understand and implement the above scheme of the embodiments of the present application, the following examples of corresponding application scenarios are used for specific description.

[0276] Next, taking the standard process as SOP and the sub-standard process as sub-SOP as an example.

[0277] The real business scenarios provided in the embodiments of the present application are exemplified. The large model agent (Agent) framework for process task automation can be used in various real business scenarios. For example, in the root cause positioning scenario in the operation and maintenance field, a defined fault tree or process diagram is needed to gradually locate the underlying root cause, and the selection and execution of the sub-tasks involved in the middle process can improve the accuracy and generalization ability of the task through the large model. For another example, in the intelligent customer service assistant scenario, different branches need to be selected according to different user tasks to achieve a constantly refined solution, so the agent (Agent) framework based on the large model is also needed to assist users in making decisions in the middle process.

[0278] Next, five embodiments will be exemplified respectively, specifically:

[0279] The embodiment of the application can improve the process execution efficiency of the SOP by illustrating the process of dynamic retrieval based on the SOP information base and the sub-SOP information base through subsequent embodiment one. Specifically, the SOP description information and the sub-task description information are constructed by using the SOP information, and the corresponding index is generated; and the adaptive SOP and task entry are dynamically retrieved according to the user task request. Through the dynamic retrieval of the SOP and the task entry based on the user task request, the coverage of the user task request and the execution efficiency of the process are improved.

[0280] The embodiment of the application can improve the task coverage of the SOP by illustrating the construction process of the SOP information base and the sub-SOP information base through subsequent embodiment two, and the application scenarios of the SOP planning are expanded through the automatic production method of the SOP. Specifically, the SOP information is generated by using the multi-modal data, and the SOP information base is formed. The SOP information base can be used for retrieval for the task request of the user.

[0281] The embodiment of the application can improve the SOP by illustrating the process of SOP evaluation feedback through subsequent embodiment three, and the coarse-to-fine process evaluation is fused. The coarse-to-fine abnormal positioning method is adopted, and the efficiency and effect of the execution are taken into account. Specifically, the coarse-to-fine evaluation method is adopted to locate the abnormal action.

[0282] The embodiment of the application can improve the SOP by illustrating the process of SOP execution improvement through subsequent embodiment four. The reflection and adaptive problem backtracking mechanism is fused, the complexity of the action dependence in the process is met, and the problem improvement ability is improved.

[0283] The embodiment of the application can realize the long-term optimization of the SOP by illustrating the process of SOP iterative optimization through subsequent embodiment five. Specifically, through the evaluation, backtracking and reflection improvement of the problem, the experience data can be accumulated, and the execution of the SOP planning process is optimized. The utilization method of the multiple experience data is fused, and the evaluation, improvement and execution ability of the SOP process is improved respectively, and the self-evolution of the method is realized.

[0284] As shown in FIG. 5, a flowchart of a task execution method applied by the embodiment of the application mainly includes:

[0285] S1, receiving a task request sent by a terminal device.

[0286] S2, performing SOP retrieval from the SOP information base according to the task request.

[0287] For details of the generation process of the SOP in the SOP information base, see subsequent embodiment one.

[0288] S3, determining whether there is an SOP for executing the task.

[0289] S4, if no SOP for performing the task is retrieved from the SOP information base, generating a SOP according to the task and the expert experience base, and storing the generated SOP into the SOP information base, which can also be referred to as a SOP long-term memory. For example, the SOP can be a code in particular.

[0290] For details, see the example of generating a SOP in Embodiment II below.

[0291] S5, if the SOP for performing the task is retrieved from the SOP information base, determining the sub-SOP in the sub-SOP information base corresponding to the SOP according to the task, and performing SOP planning and execution. The execution result of the action sequence generated in the process of SOP planning and execution is stored into the SOP short-term memory.

[0292] S6, performing SOP evaluation feedback for the process of SOP planning and execution.

[0293] For details, see the example of SOP evaluation feedback in Embodiment II below.

[0294] S7, determining whether a problem is found in the SOP evaluation feedback.

[0295] For details, see the example of finding a problem from the SOP evaluation feedback in Embodiment III below.

[0296] S8, if a problem is found, performing SOP improvement, and triggering S5 to be re-executed after the improvement is completed.

[0297] For details, see the examples of SOP improvement in Embodiments IV and V below.

[0298] If no problem is found, the execution is ended.

[0299] The embodiment of the present application aims at real user task requests online. The SOP and the starting sub-task suitable for solving the task are located through SOP retrieval and task entry retrieval. If no suitable SOP is found, the corresponding SOP information is generated in an automatic manner and stored into the SOP information base for later retrieval. After the SOP information base is formed, the SOP retrieved or generated is executed step by step according to the flow logic, and the execution process and the historical execution result are interacted to ensure that the context information is accurately utilized. After the execution is ended, the evaluation feedback mechanism combining coarse and fine granularity is used to locate potential abnormal problems. If a problem is found, adaptive problem improvement is automatically performed to optimize the execution of the flow. Finally, when the execution data of multiple flows are accumulated, the ability of the entire framework to solve the SOP planning task can be further improved through the method of optimization configuration or model.

[0300] Next, the task execution method provided by the embodiment of the present application is described in detail.

[0301] Embodiment one of the present application:

[0302] As shown in FIG. 6, the present embodiment includes the following steps:

[0303] Step S01, aggregate SOP information to obtain complete SOP description information.

[0304] In step S01, for the aggregation of SOP information, the method of text summary and graph learning can be used to aggregate the subtask description information organized in the form of flowchart to obtain the description information of the entire SOP. For example, the node aggregation method can be used to continuously aggregate the information of downstream subtasks from bottom to top, and the aggregation method can use a large model to summarize to obtain the aggregation of the entire SOP information.

[0305] For example, SOP information at least includes: subtask information of each subtask, and jump information between subtasks. For example, the following is the information configuration file corresponding to the diet suggestion scene:

[0306] Step S02, for each subtask in the SOP, aggregate the subSOP information corresponding to each subtask to obtain the description information required when the subtask is used as a task entry.

[0307] In step S02, for each subtask corresponding to the subSOP, the method similar to S01 can be used to aggregate the subSOP information as the information of the task entry. Specifically, a dynamic programming method can be used to construct the subSOP description information from bottom to top, thereby improving the efficiency of generating the SOP description information. Among them, dynamic programming refers to continuously aggregating subSOP from bottom to top, and gradually constructing subSOP description information.

[0308] Step S03, respectively based on the SOP description information and the subSOP description information of the task entry, construct the index of the SOP and the index of the task entry.

[0309] In step S03, based on the description information obtained in steps S01 and S02, sparse or dense semantic representation can be used to correspond the information to one or more vector representations, based on which key-value pairs are constructed to form the SOP information library and the subSOP information library, respectively.

[0310] For example, as shown in FIG. 6, the SOP stored in the long-term memory of the SOP i includes a plurality of subSOPs, for example, the starting subtasks corresponding to the plurality of subSOPs are A, B, C, D, E, F, G, H and I, and each subSOP is used to execute different subtasks and subtasks jumped to by the subtasks.

[0311] Step S04, for the user task request, the relevant SOP and task entry are respectively located based on the retrieved method.

[0312] In step S04, for the user task request, sparse or dense semantic representations are constructed by modal forms, and the most relevant SOP is located from the above-mentioned constructed SOP information base by using the semantic matching retrieval method for SOP configuration retrieval. i , and the task entry TASK is retrieved from the sub-SOP information base by using the semantic matching retrieval method for task entry retrieval. ij For example, if the starting sub-task of the retrieved sub-SOP is B, then B is used as the task entry to execute the user's task request.

[0313] Semantic parsing parses the task request into semantic representations, such as tokenizing the task by a model preset word table and mapping it to a corresponding identifier ID. The semantic representation generates a feature vector by using a model to process the semantic representation obtained by task parsing, such as generating a feature vector according to the ID sequence mapped by the task by using a model for generating a sentence vector.

[0314] As can be known from the description of the first embodiment of the present application, for an actual task request, the user does not need to obtain the corresponding SOP planning details, and the user can obtain whether a specific task can be solved. By using the retrieval method, the understanding of the SOP configuration information in the actual scene is reduced, and the usability of the SOP is improved. At the same time, the task retrieval can also improve the coverage of the task, and the retrieval method is repeatedly called in the task solving process to ensure that various task requirements of the user are met. Finally, the retrieval of the task entry improves the execution efficiency, and the process execution only focuses on the task request of the user, reducing unnecessary calculation.

[0315] Embodiment two of the present application:

[0316] This embodiment includes the following steps:

[0317] Step S11, SOP planning generation based on expert experience.

[0318] Based on the task request, in the case where no relevant SOP is retrieved, the SOP information is generated using expert experience.

[0319] For example, for a task of providing dietary suggestions, the SOP corresponding to the task is not stored in the SOP information base, and the task "providing dietary suggestions for student id12345" is combined with the experience database, for example, the tools for executing sub-tasks 1, 2 and 3 are obtained from the experience database, and the field information required in the above SOP information configuration file is filled in.

[0320] Step S12, SOP planning automatic generation based on a large model.

[0321] According to the task request, combined with the flowchart, language description, structured data and other multi-modal information, the SOP configuration information is automatically generated by the large model.

[0322] In step S12, the business-related SOP information can be automatically mapped to the space of structured configuration or code by multi-modal generation technology, which involves techniques such as multi-modal data retrieval based on user query, example provision, and output format prompt word engineering. For example, multi-modal data retrieval refers to retrieving available multi-modal databases using user task requests, which is similar to SOP information library retrieval. Providing examples is a method in large model prompt word engineering, that is, providing similar input-output examples so that the large model can better understand the requirements of the output, such as providing multi-modal data that is not the same task and the configuration information generated under this multi-modal data as examples. Output format refers to the format information required for output. Examples of output format are as follows: You are an expert in generating json format data. Please refer to and understand the provided flowchart and generate the corresponding structured configuration information, where the dictionary key corresponds to the index of the subtask; the value must contain task and next, i.e. the definition of the subtask and the next subtask. Similar SOP configuration data is as follows: …, please generate configuration information that meets the requirements.

[0323] For example, a multi-modal large model is trained according to massive web page data and artificially collected question and answer data. Taking GPT-4V as an example, which is a GPT-4 with visual function, the multi-modal large model can automatically generate configuration information according to multi-modal information, such as the flowchart shown in FIG. 4, and the organization form of the information, such as the meanings of the subtask target (task) and next hop subtask (next) fields, to automatically generate the corresponding structured configuration. The language description information can be: according to the student ID, detailed information such as gender, height, and weight is obtained, if the student's gender and height exist, then according to the student's gender and height, the corresponding standard weight is retrieved, and in the case of retrieving the standard weight, according to the student's weight deviation from the standard weight, dietary suggestions and specified dietary plans are provided. The above multi-modal information is provided as input to the large model, and the output requirements are provided in the form of text, such as generating structured configuration, and the large model can generate the above structured configuration.

[0324] Step S13, constructing the SOP information library.

[0325] The generated SOP information can be used to construct the SOP information library; according to different domain tasks, the SOP generation method in steps S12 and S13 can also be used in the offline stage to expand the SOP information library.

[0326] In step S13, the SOP information generated by the expert database or the large model is used to expand the SOP information library, and an SOP index is constructed. In the offline stage, the SOP information can be generated through multi-modal data without receiving a user request, and the SOP information library is constructed offline. For example, without receiving a user task request, the SOP information can also be generated through the prompt word engineering technology by providing examples or requirements for the output format.

[0327] As can be known from the description of the second embodiment of the present application, the SOP generation can reduce or even eliminate the manual writing of related SOP information or code by experts, can greatly reduce the manual workload, and improve the ease of use of the framework. At the same time, the multi-modal information accumulated in the business can also make the generated SOP information more accurate, and make the problem solving process according to the business expectation.

[0328] The third embodiment of the present application:

[0329] The present embodiment is divided into the following steps:

[0330] In step S21, the whole rough evaluation is used to locate the problem action for the action sequence in the SOP execution process.

[0331] In step S21, the SOP execution process is combined with the subtask description information defined by the SOP and the jump logic relationship between the subtasks, and the large model is used to determine whether there is a problem action, including but not limited to: based on the prompt word engineering, the large model provides evaluation based on the SOP information and execution information, constructs positive and negative samples to learn the evaluator, and directly matches the action result and the expected information.

[0332] For example, the prompt word that can be used is: the action sequence of the execution is as follows: … Please judge whether there is a content error in the reply of the assistant in the above dialogue process, and give the earliest problem if there is. Through this prompt word and the provided action sequence, the rough evaluation of the whole action sequence can be realized.

[0333] For example, according to the prompt word input to the large model in step S21, the possible output of the model is as follows: there is a content error in the dialogue process, and an error occurs in the reply in the 4th round.

[0334] For example, the positive sample can be the action sequence corresponding to the normally completed task, and the negative sample is the action sequence corresponding to the task with problems in the execution. The positive and negative samples can be used to train a binary classifier and a sequence labeling model, and for any action sequence to be tested, the judgment of whether there is a problem and the location of the problem is given.

[0335] For example, the expected information mainly refers to that the execution of some actions in the action sequence does not meet the subtask information defined by the subtask, for example, the subtask is to search the life of the corresponding person through the Internet, and the search result is empty, which does not match the requirement of the subtask, and it can be considered that there is a problem.

[0336] In step S22, the problem of the located specific action is reviewed by using local fine evaluation.

[0337] In step S22, according to the action sequence in the SOP execution process and the specific subtask description information, it is judged by using the large model whether the execution of the current subtask has a problem, and the specific method is consistent with that in step S21.

[0338] For example, the local evaluation can be referred to as fine evaluation, and the fine evaluation is to evaluate whether the action has a problem according to the located subtask information and subtask jump logic information of the subtask and the action execution result, for example, the following prompt words can be input to the large model for action evaluation: for the current subtask information and subtask jump logic information: …, the preliminary action output is: …, please analyze whether the preliminary action output result meets the requirement.

[0339] For example, the task request includes a task of giving a diet suggestion for student id 12345. The action execution result of subtask 1 is that the height of student id 12343 is xxx, and the weight is xxx. The execution result of the action has a problem because the output id is 12343 and the id in the task request is 12345, which cannot be matched.

[0340] In step S23, if the fine positioning is successful, the abnormal action is returned, otherwise the feedback information is used to reposition the problem to determine whether there is other problem.

[0341] In step S23, if the result of the fine positioning is that there is no abnormality, the coarse positioning of step S21 is performed again. When the coarse positioning is performed again, the feedback information is additionally brought in, and the specific method still refers to the method of step S21. The other problem can be different from the previously located abnormal action, that is, other actions in the action sequence.

[0342] As known from the description of the third embodiment of the present application, if only the whole evaluation is used in the action evaluation method, the problem cannot be accurately located, and if only the local fine evaluation method is used, the problem locating method is relatively low in efficiency. For the SOP process execution, the coarse-to-fine two-stage evaluation method is used in the embodiment of the present application, which can efficiently find the problem in the process and balance the efficiency and effect of the process execution.

[0343] The fourth embodiment of the present application

[0344] As shown in FIG. 7, the embodiment is divided into the following steps:

[0345] Step S31, the execution result of the problem action is improved by reflection using the feedback information of the positioning process.

[0346] In step S31, according to the aforementioned abnormal positioning information and subtask information and subtask jump logic information, the execution optimization of the subtask is engineered by the large model based on the prompt word, so as to improve the problem.

[0347] For example, according to the aforementioned abnormal positioning information and subtask information and subtask jump logic information, the execution optimization of the subtask is engineered by the large model based on the prompt word, so as to improve the problem. For example, the prompt word can be constructed in the following way: for the feedback information of the evaluation: …, re-execute the action of … to avoid the problem indicated in the feedback information.

[0348] Step S32, if the improvement fails, a dynamic backtracking model is used to locate the related action for improvement.

[0349] In step S32, if the improved result still has a problem after evaluation, the action that may cause the current problem is backtracked based on the execution result of the previous action, subtask information and subtask jump logic information. A dynamic backtracking model is used to locate the related action for improvement, including but not limited to using a large model to give the related action of the action causing the problem according to the evaluation feedback, SOP information, execution result and other information. After the problem is backtracked, the backtracked action is re-executed, and the subsequent action of the problem action is reached.

[0350] For example, the following prompt word can be constructed: according to the execution sequence of the action: …, please analyze which round of the action abnormality is most likely to be caused by the result of the fourth round.

[0351] As shown in FIG. 7, after the optimization model improves the D action in the action sequence, the improved D action is input into the evaluation model, and the improved D is evaluated by the evaluation model. If the evaluation result indicates that the improvement is not successful, the backtracking model is used to backtrack, determine that the related action B of the problem action D has a problem, and then input the action B into the optimization model to improve the action B by the optimization model.

[0352] Among them, the optimization model, the evaluation model and the backtracking model shown in FIG. 7 can be a large model.

[0353] Step S33, if the problem is still not repaired, retrace.

[0354] In step S33, if the problem action is still not repaired, retrace is performed again using the historical trace information. The retracing method is consistent with that in step S32, and the historical trace information is added as additional information. For example, the historical trace information can be an analysis result output by the large model according to the constructed prompt word.

[0355] As can be known from the description of the fourth embodiment of the present application, the current improvement of the problem is implemented by fixed forms of reflection improvement, that is, the trigger condition and the dependent path of reflection are determined, and the improvement mode is too rigid. The reflection and adaptive problem retracing mechanism are combined in the embodiments of the present application, the complexity and flexibility of the action dependence in the process are met, and the problem improvement capability is improved.

[0356] The fifth embodiment of the present application is as follows:

[0357] The present embodiment is divided into the following steps:

[0358] In step S41, the data generated in the evaluation, retracing and improvement process is used to optimize the SOP information database, and the SOP information database stores the information of the SOP.

[0359] In step S41, the data of evaluation and retracing can be used to configure the key check point and the dependent path of retracing, and the data of SOP improvement can be adjusted by the prompt word fine tuning technology and the task description information, and the reference examples are provided.

[0360] In step S42, the data generated in the evaluation, retracing and improvement process is used to update the optimization model, the evaluation model and the retracing model shown in FIG. 7.

[0361] In step S42, the data of evaluation, retracing and improvement can be used to generate corresponding training samples, and the corresponding model capability is fine tuned.

[0362] The key check point is illustrated as follows. The historical execution result of the current SOP shows that the action problem always occurs at subtask 3, and the subsequent evaluation can focus on checking the action result of subtask 3.

[0363] The retracing path is illustrated as follows. The historical execution result of the current SOP shows that the problem of subtask 3 is always caused by subtask 1, so the retracing of the action problem of subtask 3 can focus on checking whether there is a problem in the action of subtask 1.

[0364] The prompt word fine tuning is illustrated as follows. The output of the large model is adjusted to meet the expectation by adjusting the prompt word, for example, the prompt word of the above evaluation is adjusted, so that the evaluation result is more accurate, and the specific adjustment mode of the prompt word is not limited.

[0365] An example of the tuning task description information is as follows, the tuning task information refers to the information of the subtask, the information of the subtask 1 in the SOP information "get detailed information such as gender, height, weight, etc. according to the student ID", and the tuning can get "get detailed information such as gender, height, weight, etc. according to the student ID", which can help the current subtask to be better executed, because the evaluation utilizes the information of the subtask, and the evaluation can also be affected.

[0366] An example of providing a reference is as follows, in order to improve the success rate of the subtask execution, the example of successful execution is provided as an additional input of the model.

[0367] As can be known from the description of the fifth embodiment of the present application, the historical experience data obtained in the cumulative improvement process can optimize the SOP information library and the model capability, so that subsequent user task requests can be more efficiently and accurately satisfied.

[0368] As can be known from the foregoing examples, the SOP and task entry dynamic retrieval are provided in the embodiments of the present application, the coverage of the user request is improved through the SOP retrieval, and the efficiency of the process execution is improved through the retrieval of the task entry. The SOP generation is provided in the embodiments of the present application, the application scenarios of the SOP planning are expanded in the manner of the SOP generation. The SOP evaluation feedback is provided in the embodiments of the present application, the coarse-to-fine abnormal positioning method is adopted to take into account the efficiency and effect of the execution. The SOP adaptive improvement is provided in the embodiments of the present application, the reflection and adaptive problem backtracking mechanism is fused, the complexity and flexibility of the action dependence in the process are satisfied, and the problem improvement capability is improved. The SOP iterative optimization is provided in the embodiments of the present application, the utilization methods of various experience data are fused, the evaluation, improvement and execution capabilities of the SOP process are respectively improved, and the self-evolution of the method is realized.

[0369] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited to the action sequence described, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0370] In order to better implement the foregoing scheme of the embodiments of the present application, the related device for implementing the foregoing scheme is also provided.

[0371] Please refer to FIG. 8, the task execution device 800 provided by the embodiments of the present application can include a receiving module 801, a standard process acquisition module 802, a substandard process acquisition module 803 and a task execution module 804, wherein,

[0372] receive a task request;

[0373] obtain a first standard process from a standard process information base according to the task request;

[0374] obtain a first sub-standard process from the first standard process according to the task request, the first sub-standard process being a sub-standard process of the first standard process, the first sub-standard process being used to indicate that a plurality of first sub-tasks are executed in turn based on a jump relationship, the first sub-standard process including a first root node, the first root node being used to indicate that a starting sub-task in the plurality of first sub-tasks is executed;

[0375] execute the plurality of first sub-tasks according to the first sub-standard process.

[0376] In the embodiments of the present application, the modules are an example of software functional units, and the data processing apparatus can include codes running on computing instances. The computing instances can be at least one of physical hosts (computing devices), virtual machines, containers, and the like. Further, the above computing devices can be one or more. For example, the data processing apparatus can include codes running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers used to run the application can be distributed in the same region, or can be distributed in different regions. The multiple hosts / virtual machines / containers used to run the codes can be distributed in the same available zone (AZ), or can be distributed in different AZs, each AZ including one data center or multiple data centers with similar geographical locations. Generally, one region can include multiple AZs.

[0377] Similarly, the multiple hosts / virtual machines / containers used to run the codes can be distributed in the same virtual private cloud (VPC), or can be distributed in multiple VPCs. Generally, one VPC is set in one region. Communication between two VPCs in the same region and between VPCs in different regions needs to be set in each VPC to set a communication gateway, and the interconnection between VPCs is realized through the communication gateway.

[0378] As an example of a hardware functional unit of a module, the data processing apparatus can include at least one computing device, such as a server or the like. Alternatively, the data processing apparatus can also be a device implemented by an application specific integrated circuit (ASIC) or a programmable logic device (PLD), and the like. The PLD can be implemented by a complex PLD (CPLD), a field programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0379] The plurality of computing devices included in the data processing apparatus can be distributed in the same region or in different regions. The plurality of computing devices included in the data processing apparatus can be distributed in the same AZ or in different AZs. Similarly, the plurality of computing devices included in the data processing apparatus can be distributed in the same VPC or in multiple VPCs. The plurality of computing devices can be any combination of servers, ASICs, PLDs, CPLDs, FPGAs, and GALs.

[0380] Embodiments of the present application also provide a computing device 130. As shown in FIG. 9, the computing device 130 includes a bus 132, a processor 134, a memory 136, and a communication interface 138. The processor 134, the memory 136, and the communication interface 138 communicate with each other through the bus 132. The computing device 130 can be a server or a terminal device. It should be understood that the number of processors and memories in the computing device 130 is not limited by the present application.

[0381] The bus 132 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one line is shown in FIG. 9, but it does not mean that there is only one bus or only one type of bus. The bus 134 can include a path for transmitting information between various components (e.g., the memory 136, the processor 134, the communication interface 138) of the computing device 130.

[0382] The processor 134 can include any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP), among other processors.

[0383] The memory 136 can include volatile memory, such as random access memory (RAM), and non-volatile memory, such as read-only memory (ROM), a floppy disk drive, a hard disk drive (HDD), or a solid-state drive (SSD), among others.

[0384] The memory 136 stores executable program code that the processor 134 executes to respectively implement the functions of the aforementioned obtaining module and training module, thereby implementing the data processing method applied to the computing device cluster in the above embodiments. That is, the memory 136 stores instructions for executing the data processing method applied to the computing device cluster in the above embodiments.

[0385] The communication interface 138 uses a transceiver module, such as but not limited to a network interface card or a transceiver, to enable communication between the computing device 130 and other devices or communication networks.

[0386] Embodiments of the present application also provide a computing device cluster. The computing device cluster includes at least one computing device. The computing device can be a server, such as a central server, an edge server, or a local server in a local data center. In some embodiments, the computing device can also be a terminal device, such as a desktop computer, a notebook computer, or a smartphone.

[0387] As shown in FIG. 10, the computing device cluster includes at least one computing device 130. The memory 136 in one or more computing devices 130 in the computing device cluster can store the same instructions for executing the data processing method.

[0388] In some possible implementations, the memory 136 of one or more computing devices 130 in the computing device cluster can also respectively store partial instructions for executing the data processing method. In other words, the combination of one or more computing devices 130 can collectively execute the instructions for executing the data processing method.

[0389] It should be noted that the memories 136 in different computing devices 130 in the computing device cluster can store different instructions for performing part of the functions of the data processing method. That is, the instructions stored in the memories 136 in different computing devices 130 can implement one or more functions of the obtaining module.

[0390] In some possible implementation manners, one or more computing devices in the computing device cluster can be connected through a network. The network can be a wide area network, a local area network, or the like. FIG. 10 shows a possible implementation manner. As shown in FIG. 10, two computing devices 130A and 130B are connected through a network. Specifically, the computing devices are connected to the network through communication interfaces in the computing devices. In this type of possible implementation manner, the memory 136 in the computing device 130A can store instructions for performing the functions of the first processing module. Meanwhile, the memory 136 in the computing device 130B can store instructions for performing the functions of the second processing module. Alternatively, the memory 136 in the computing device 130A can store instructions for performing part of the functions of the second processing module. Meanwhile, the memory 136 in the computing device 130B can store instructions for performing another part of the functions of the second processing module, and the like.

[0391] It should be understood that the functions of the computing device 130A shown in FIG. 11 can also be completed by multiple computing devices 130. Similarly, the functions of the computing device 130B can also be completed by multiple computing devices 130.

[0392] Embodiments of the present application also provide another computing device cluster. The connection relationship between the computing devices in the computing device cluster can be similar to the connection manner of the computing device cluster in FIG. 10. The difference is that the memories 136 in one or more computing devices 130 in the computing device cluster can store the same instructions for performing the data processing method.

[0393] In some possible implementation manners, the memories 136 in one or more computing devices 130 in the computing device cluster can also respectively store part of the instructions for performing the data processing method. In other words, the combination of one or more computing devices 130 can collectively execute the instructions for performing the data processing method.

[0394] It should be noted that the memories 136 in different computing devices 130 in the computing device cluster can store different instructions for performing part of the functions of the data processing method. That is, the instructions stored in the memories 136 in different computing devices 130 can implement one or more functions of the processing module.

[0395] The embodiments of the present application further provide a computer program product containing instructions. The computer program product can be software or program product containing instructions, which can run on a computing device or be stored in any available medium. When the computer program product runs on at least one computing device, the at least one computing device is caused to perform the data processing method.

[0396] The embodiments of the present application further provide a computer readable storage medium. The computer readable storage medium can be any available medium or data storage device that can be accessed by a computing device or a data center containing one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk), etc. The computer readable storage medium contains instructions, which instruct the computing device to perform the data processing method.

[0397] The embodiments of the present application further provide a chip system, which includes a processor for implementing the steps performed by the computing device cluster described above. In a possible design, the chip system can further include a memory for storing necessary program instructions and data. The chip system can be composed of a chip, or can include a chip and other discrete devices.

[0398] 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 foregoing method embodiments, which will not be described here.

[0399] In the several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented by other means. For example, the device embodiments described above are only schematic, and the division of units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0400] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0401] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software function unit.

[0402] When the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application, essentially or in other words, the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, read-only memory), a random access memory (RAM, random access memory), a magnetic disk or an optical disk, and various media that can store program codes.

Claims

A task execution method characterized by, The method comprises: receiving a task request; obtaining a first standard process from a standard process information base according to the task request; obtaining a first sub-standard process from the first standard process according to the task request, the first sub-standard process being a sub-standard process of the first standard process, the first sub-standard process being used to indicate that a plurality of first sub-tasks are executed in turn based on a jump relationship, the first sub-standard process comprising a first root node, the first root node being used to indicate that a starting sub-task in the plurality of first sub-tasks is executed; executing the plurality of first sub-tasks according to the first sub-standard process. The method of claim 1, wherein The method further comprises: performing semantic analysis on the task request to obtain a target semantic representation; performing semantic matching between the target semantic representation and the standard process information base to obtain a first semantic representation matched with the target semantic representation, the standard process information base comprising a plurality of standard processes and first semantic representations corresponding to the plurality of standard processes respectively; obtaining the first standard process from the standard process information base according to the first semantic representation matched with the target semantic representation. The method according to claim 1 or 2, characterized in that The method further comprises: obtaining a sub-standard process information base according to the first standard process, the sub-standard process information base comprising a plurality of sub-standard processes, the plurality of sub-standard processes belonging to the first standard process; obtaining the first sub-standard process from the sub-standard process information base according to the task request. The method according to claim 3, characterized in that The method further comprises: performing semantic analysis on the task request to obtain a target semantic representation; performing semantic matching between the target semantic representation and the sub-standard process information base to obtain a second semantic representation matched with the target semantic representation, the sub-standard process information base comprising a plurality of sub-standard processes of the first standard process and second semantic representations corresponding to the plurality of sub-standard processes of the first standard process respectively; obtaining the first sub-standard process from the sub-standard process information base according to the second semantic representation matched with the target semantic representation. The method according to any one of claims 1 to 4, characterized in that The method further comprises: when the first standard process is not obtained from the standard process information base, obtaining a second standard process according to the task request, the second standard process being a standard process generated according to the task request and an experience database; storing the second standard process to the standard process information base. The method according to any one of claims 1 to 5, characterized in that The method further comprises: when the first standard process is not obtained from the standard process information base, obtaining first multi-modal information according to the task request, the first multi-modal information comprising a plurality of types of information required for generating a standard process; inputting the first multi-modal information into a large model, outputting a third standard process through the large model, the large model being used to generate a standard process corresponding to a plurality of tasks respectively; storing the third standard process to the standard process information base. The method according to any one of claims 1 to 6, characterized in that The method further comprises: obtaining a fourth standard procedure, storing the fourth standard procedure into the standard procedure information base, the fourth standard procedure being a standard procedure generated according to an experience database; or, obtaining second multi-modal information, the second multi-modal information including multiple types of information required for generating a standard procedure; inputting the second state information into a large model, outputting a fifth standard procedure through the large model, the large model being used for generating a standard procedure corresponding to each of multiple tasks; storing the fifth standard procedure into the standard procedure information base. The method according to any one of claims 1 to 7, characterized in that The method further includes: in a process of executing the multiple first sub-tasks through an action model, obtaining an action sequence corresponding to the first sub-standard procedure, multiple actions in the action sequence being used for executing the multiple first sub-tasks or sub-task jumping, the multiple actions in the action sequence being implemented based on sub-task information or sub-task jumping logic information of the multiple first sub-tasks; performing abnormal evaluation on the action sequence through a first evaluation model according to execution results of the multiple actions in the action sequence, the sub-task information of the multiple first sub-tasks and the sub-task jumping logic information, to obtain a suspicious action from the action sequence; performing abnormal detection on the suspicious action through a second evaluation model according to an execution result of the suspicious action and sub-task information of a first sub-task corresponding to the suspicious action, or according to the execution result of the suspicious action and sub-task jumping logic information of the first sub-task corresponding to the suspicious action, to determine the suspicious action as an abnormal action. The method of claim 8, wherein The method further includes: performing optimization on the abnormal action through an optimization model according to an execution result of the abnormal action and sub-task information of a first sub-task corresponding to the abnormal action, or according to the execution result of the abnormal action and sub-task jumping logic information of the first sub-task corresponding to the abnormal action, to obtain an execution result of an optimized action; performing evaluation on the execution result of the optimized action through the second evaluation model, to obtain a first evaluation result. The method of claim 9, wherein The method further includes: when the first evaluation result indicates that the execution result of the optimized action has a problem, performing backtracking through a backtracking model according to the execution result of the optimized action, sub-task information of the multiple first sub-tasks and sub-task jumping logic information, to obtain a first associated action causing the problem from the action sequence and to obtain historical backtracking information; performing optimization through the optimization model from the first associated action causing the problem to the abnormal action, to obtain an execution result of an optimized first associated action; performing evaluation on the execution result of the optimized first associated action through the second evaluation model, to obtain a second evaluation result. The method of claim 10, wherein The method further includes: when the second evaluation result indicates that the execution result of the optimized first associated action has a problem, backtracking is performed according to the historical backtracking information, the execution result of the optimized first associated action, the subtask information of the plurality of first subtasks, and the subtask jump logic information by using the backtracking model, so as to obtain a second associated action that causes the problem from the action sequence; optimization is performed from the second associated action to the abnormal action by using the optimization model, so as to obtain an execution result of the optimized second associated action; the execution result of the optimized second associated action is evaluated by using the second evaluation model, so as to obtain a third evaluation result. The method of claim 9, wherein when the first evaluation result indicates that the execution result of the optimized action has no problem, the method further comprises: updating the standard process information base by using the optimized action; and / or, training at least one of the following models by using the optimized action: the action model, the first evaluation model, and the second evaluation model. The method of claim 10, wherein when the second evaluation result indicates that the execution result of the optimized first associated action has no problem, the method further comprises: updating the standard process information base by using the optimized first associated action; and / or, training at least one of the following models by using the optimized first associated action: the action model, the first evaluation model, the second evaluation model, and the optimization model. The method of claim 11, wherein when the third evaluation result indicates that the execution result of the optimized second associated action has no problem, the method further comprises: updating the standard process information base by using the optimized second associated action; and / or, training at least one of the following models by using the optimized second associated action: the action model, the first evaluation model, the second evaluation model, the backtracking model, and the optimization model. The method according to any one of claims 1 to 14, characterized in that The standard process information base is constructed in the following manner: obtaining subtask information of a plurality of second subtasks and subtask jump logic information between the plurality of second subtasks, the subtask jump logic information being used to indicate a subtask jump manner between the plurality of second subtasks, and the plurality of second subtasks being used to perform second subtask jumps in sequence according to a jump relationship indicated by a sixth standard process; performing the second subtask jumps in a reverse subtask jump manner between the plurality of second subtasks, and adding subtask information of a jumped second subtask to a node of the jumped second subtask, so as to obtain description information of the sixth standard process, the reverse subtask jump manner being opposite to the subtask jump logic information; constructing the standard process information base according to the description information of the sixth standard process, the sixth standard process being any one of the standard processes in the standard process information base. The method of claim 15, wherein The performing the second subtask jumps in the reverse subtask jump manner between the plurality of second subtasks, and adding the subtask information of the jumped second subtask to the node of the jumped second subtask, so as to obtain the description information of the sixth standard process, comprises: According to the subtask reverse jump manner between the plurality of second subtasks, jump is performed, and subtask information of the jumped second subtask is added at a node of the jumped second subtask, to obtain description information of a plurality of substandard flows of the sixth standard flow; According to the description information of the plurality of substandard flows of the sixth standard flow, the description information of the sixth standard flow is generated. The method according to claim 15 or 16, characterized in that The construction of the standard flow information library according to the description information of the sixth standard flow comprises: According to the description information of the sixth standard flow, a first semantic representation corresponding to the sixth standard flow is generated; According to the first semantic representation corresponding to the sixth standard flow, the standard flow information library is constructed. The method according to any one of claims 3 to 17, characterized in that The substandard flow information library is constructed in the following manner: Obtaining subtask information of a plurality of first subtasks and subtask jump logic information between the plurality of first subtasks, the subtask jump logic information between the plurality of first subtasks being used to indicate a subtask jump manner between the plurality of first subtasks; According to the subtask reverse jump manner between the plurality of first subtasks, jump is performed, and subtask information of the jumped first subtask is added at a node of the jumped first subtask, to obtain description information of a plurality of substandard flows of the first standard flow; According to the description information of the plurality of substandard flows of the first standard flow, the substandard flow information library is constructed. The method of claim 18, wherein The construction of the substandard flow information library according to the description information of the plurality of substandard flows of the first standard flow comprises: According to the description information of the plurality of substandard flows of the first standard flow, a second semantic representation corresponding to each of the plurality of substandard flows of the first standard flow is generated; According to the second semantic representation corresponding to each of the plurality of substandard flows of the first standard flow, the substandard flow information library is constructed. A task execution device characterized by comprising: Comprise: The receiving module is used for receiving a task request; The standard flow acquisition module is used for acquiring a first standard flow from a standard flow information library according to the task request; The substandard flow acquisition module is used for acquiring a first substandard flow from the first standard flow according to the task request, the first substandard flow being a substandard flow of the first standard flow, the first substandard flow being used to indicate that a plurality of first subtasks are executed in sequence based on a jump relationship, the first substandard flow comprising a first root node, the first root node being used to indicate that a starting subtask in the plurality of first subtasks is executed; The task execution module is used for executing the plurality of first subtasks according to the first substandard flow. An electronic device comprising a processor and a memory, the memory and the processor being coupled, the processor being configured to execute the method of any one of claims 1 to 19. A computer-readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the method of any one of claims 1 to 19. A computer program product comprising instructions which, when executed on a computer, cause the computer to perform the method of any one of claims 1 to 19. A chip characterized by An electronic device comprising one or more interface circuits and one or more processors; the interface circuit is configured to receive a signal from a memory of the electronic device, and send the signal to the processor, the signal comprising computer instructions stored in the memory; when the processor executes the computer instructions, the electronic device is caused to perform the method of any one of claims 1-19.

Images