Intelligent agent capability description method, task execution method and device and intelligent agent

Abstract

The invention provides an intelligent agent capability description method and device, a task execution method and device and an intelligent agent, and relates to the technical field of artificial intelligence, in particular to the technical fields of deep learning, large models, human-computer interaction, intelligent education, video generation and the like. The capability description method of the intelligent agent comprises the following steps: acquiring a task example for a to-be-tested intelligent agent; the main agent is used for detecting the example execution result, an example detection result is obtained, the example execution result is determined based on the task example executed by the to-be-detected agent, and the example detection result represents the matching degree between the example execution result and a task demand condition for the task example; and based on the example detection result and the task example, performing task execution capability description on the to-be-detected agent by using the main agent to obtain capability description information for the to-be-detected agent.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to technical fields such as deep learning, large models, human-computer interaction, smart education, and video generation. Background Art

[0002] Artificial Intelligence Agents (AI Agents) are intelligent entities designed to perform specific tasks. With the rapid development of AI technology, tasks such as video generation, image editing, and text generation can now be performed using agents. Summary of the Invention

[0003] The present disclosure provides a method for describing the capabilities of an intelligent agent, a method for executing a task, a device and an intelligent agent, a device, an intelligent agent, and a storage medium.

[0004] According to one aspect of the present disclosure, a method for describing the capabilities of an intelligent agent is provided, comprising: obtaining a task example for an intelligent agent to be tested; using a main intelligent agent to detect the example execution result to obtain an example detection result, wherein the example execution result is determined based on the task example executed by the intelligent agent to be tested, and the example detection result represents the degree of match between the example execution result and the task requirement conditions for the task example; based on the example detection result and the task example, using the main intelligent agent to describe the task execution capability of the intelligent agent to be tested to obtain capability description information for the intelligent agent to be tested.

[0005] According to another aspect of the present disclosure, an agent-based task execution method is provided, including: obtaining task requirement information; performing task intent detection on the task requirement information to obtain task description information; determining a target agent from the execution agents based on the task description information and capability description information for a preset execution agent, wherein the capability description information is determined according to the capability description method of the agent provided in an embodiment of the present disclosure; and controlling the target agent to execute a target task related to the task requirement information to obtain a task execution result.

[0006] According to another aspect of the present disclosure, a device for describing the capabilities of an intelligent agent is provided, comprising: a first acquisition module for acquiring a task example for an intelligent agent to be tested; an example detection result acquisition module for detecting the example execution result using a main intelligent agent to obtain an example detection result, wherein the example execution result is determined based on the task example executed by the intelligent agent to be tested, and the example detection result represents the degree of matching between the example execution result and the task requirement conditions for the task example; and a capability description information acquisition module for describing the task execution capability of the intelligent agent to be tested using the main intelligent agent based on the example detection result and the task example, thereby obtaining capability description information for the intelligent agent to be tested.

[0007] According to another aspect of the present disclosure, an agent-based task execution device is provided, including: a second acquisition module for acquiring task requirement information; a task description information acquisition module for performing task intent detection on the task requirement information to obtain task description information; a target agent determination module for determining a target agent from among the execution agents based on the task description information and capability description information for a preset execution agent, wherein the capability description information is determined according to the capability description method of the agent provided in an embodiment of the present disclosure; and a task execution result acquisition module for controlling the target agent to execute a target task related to the task requirement information to obtain a task execution result.

[0008] According to another aspect of the present disclosure, an artificial intelligence agent is provided, configured to execute the method provided according to an embodiment of the present disclosure.

[0009] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method provided according to an embodiment of the present disclosure.

[0010] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method provided according to an embodiment of the present disclosure.

[0011] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, which implements the method provided according to the embodiment of the present disclosure when executed by a processor.

[0012] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings are used to better understand the present invention and do not constitute a limitation of the present invention.

[0014] Figure 1 Schematically illustrates an exemplary system architecture to which the method and apparatus for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure can be applied;

[0015] Figure 2 The flowchart of the method for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure is schematically shown;

[0016] Figure 3A diagram schematically illustrating an application scenario of the method for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure;

[0017] Figure 4 A diagram schematically illustrating an application scenario of the method for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure;

[0018] Figure 5 A flowchart schematically illustrating an agent-based task execution method according to an embodiment of the present disclosure;

[0019] Figure 6 A diagram schematically illustrating an application scenario of an agent-based task execution method according to an embodiment of the present disclosure;

[0020] Figure 7 Schematically shows a block diagram of a device for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure;

[0021] Figure 8 Schematically shows a block diagram of an agent-based task execution device according to an embodiment of the present disclosure;

[0022] Figure 9 A block diagram schematically illustrates a structure of an artificial intelligence agent according to an embodiment of the present disclosure; and

[0023] Figure 10 A schematic block diagram of an example electronic device 1000 is shown, which can be used to implement the method for describing the capabilities of an agent and the method for executing a task based on an agent, which can be used to implement the embodiments of the present disclosure. DETAILED DESCRIPTION

[0024] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0025] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved comply with the provisions of relevant laws and regulations, take necessary confidentiality measures, and do not violate public order and good morals.

[0026] The inventors discovered that with the rapid development of artificial intelligence technology, intelligent agents built based on large models such as large language models are able to perform more complex tasks. However, for personalized or diversified task requirements, intelligent agents with task execution capabilities have difficulty in accurately generating task execution results, which reduces the efficiency of task execution.

[0027] The embodiments of the present disclosure provide a method for describing the capabilities of an intelligent agent, a method for executing a task, an apparatus, an intelligent agent, an electronic device, and a storage medium. The method for describing the capabilities of an intelligent agent includes: obtaining a task example for the intelligent agent to be tested; using a master intelligent agent to detect the execution result of the example to obtain an example detection result, wherein the example execution result is determined based on the task example executed by the intelligent agent to be tested, and the example detection result represents the degree of match between the example execution result and the task requirement conditions for the task example; based on the example detection result and the task example, using the master intelligent agent to describe the task execution capability of the intelligent agent to be tested, thereby obtaining capability description information for the intelligent agent to be tested.

[0028] According to an embodiment of the present disclosure, by obtaining a task example and using a master agent to detect the example execution result obtained by the agent to be tested when executing the task example, an example detection result that characterizes the degree of matching between the example execution result and the task requirement conditions for the task example is determined, thereby enabling the execution capability of the agent to be tested for the task example to be more accurately represented based on the example detection result. By using the master agent to process the task example and the example detection result to describe the task execution capability of the agent to be tested, the capability description information can more accurately characterize the degree of matching between the execution result of the agent to be tested in executing the target task and the task requirement conditions, thereby enabling the agent to be tested that matches the capability requirements of the target task to be more accurately called through the capability description information to accurately execute the target task and obtain an execution result that matches the task requirement conditions of the target task, thereby improving the scheduling accuracy and efficiency for the agent, and improving the execution efficiency of the target task and the accuracy of the task execution results.

[0029] Figure 1 An exemplary system architecture to which the method and apparatus for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure can be applied is schematically shown.

[0030] It should be noted that Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present disclosure may be applied, to help those skilled in the art understand the technical content of the present disclosure, but do not imply that the embodiments of the present disclosure may not be applied to other devices, systems, environments, or scenarios. For example, in another embodiment, an exemplary system architecture to which the method and apparatus for describing the capabilities of an agent may be applied may include a terminal device, but the terminal device may implement the method and apparatus for describing the capabilities of an agent provided by the embodiments of the present disclosure without interacting with a server.

[0031] like Figure 1As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0032] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software (for example only).

[0033] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0034] Server 105 may be a server that provides various services, such as a background management server (for example only) that supports content browsed by users using terminal devices 101, 102, and 103. The background management server may analyze and process received data such as user requests, and feed back processing results (e.g., web pages, information, or data obtained or generated based on user requests) to the terminal device.

[0035] Server 105 can be a cloud server, also known as a cloud computing server or cloud host. This is a host product within a cloud computing service system that addresses the management difficulties and limited scalability of traditional physical hosts and VPS (Virtual Private Server) services. Server 105 can also be a server for a distributed system or a server integrated with blockchain.

[0036] It should be noted that the capability description method of the intelligent agent provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the capability description device of the intelligent agent provided in the embodiment of the present disclosure can generally be set in the server 105. The capability description method of the intelligent agent provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the capability description device of the intelligent agent provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0037] For example, when a user is reading an e-book online, the terminal devices 101, 102, and 103 can obtain the target content in the e-book that the user is looking at, and then send the obtained target content to the server 105. The server 105 analyzes the target content to determine the characteristic information of the target content, predicts the content that the user is interested in based on the characteristic information of the target content, and extracts the content that the user is interested in. Alternatively, a server or server cluster that can communicate with the terminal devices 101, 102, and 103 and / or the server 105 can analyze the target content and ultimately extract the content that the user is interested in.

[0038] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0039] Figure 2 The flowchart of the method for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure is schematically shown.

[0040] like Figure 2 As shown, the capability description method of the intelligent agent in this embodiment includes operations S210 to S230.

[0041] In operation S210 , a task example for the agent to be tested is obtained.

[0042] In operation S220, the master agent is used to detect the example execution result to obtain an example detection result.

[0043] In operation S230 , based on the example detection results and the task example, the master agent is used to describe the task execution capability of the agent to be tested, and capability description information for the agent to be tested is obtained.

[0044] According to an embodiment of the present disclosure, the execution subject of operation S210 may be a master agent, or the execution subject of operation S210 may also be a terminal or other server that is in communication with the server that sets the master agent.

[0045] According to an embodiment of the present disclosure, the execution subject of operations S220 and S230 can be a master agent, which is configured on a server or terminal device. The agent to be tested can be configured on the same or different electronic device as the master agent, such as a server or terminal device, as long as communication between the master agent and the agent to be tested is possible.

[0046] Both the main agent and the target agent can be capable of perceiving the environment and taking actions to achieve specific goals. Both the main agent and the target agent can be designed to be highly scalable, for example, they can be designed to be agents configured with large models. The type of large model is not limited. For example, a large model can include a combination of one or more of a large language model (LLM), a large vision model (LVM), or a multimodal large model (MLM). The main agent and the target agent differ from each other in that the configured large models have different large model knowledge and can perform different types of tasks.

[0047] According to embodiments of the present disclosure, task examples can be executed by the agent under test. Task examples can include any number or type. For example, a task example can include a video generation task example or a text generation task example. However, this is not limited to this. Task examples can also include multiple tasks with dependencies. The agent under test can execute multiple task examples based on the dependencies to obtain example execution results for each of the multiple task examples.

[0048] According to embodiments of the present disclosure, an example execution result is determined based on an example task performed by a tested agent. For example, the example execution result may be obtained by controlling a tested agent to perform the example task via a master agent. The example execution result may include any type of information output by the tested agent, such as text, images, or videos. The embodiments of the present disclosure do not limit the specific type of the example execution result.

[0049] According to an embodiment of the present disclosure, the example detection result characterizes the degree of match between the example execution result and the task requirement conditions for the task example. The task requirement conditions for the task example may indicate the attribute requirements for the example execution result, such as the type and quality of the example execution result. For example, the task requirement conditions may characterize the text word count requirements, the text logical coherence requirements, the text structure requirements, etc. for the copy generation example. For another example, the task requirement conditions may also characterize the correctness requirements, the number of problem-solving steps requirements, the problem-solving knowledge point requirements, etc. for the problem-solving example task. The embodiment of the present disclosure does not limit the type of requirements for the example execution result represented by the task requirement conditions.

[0050] The example detection results can represent the degree of match based on non-quantitative information such as character identifiers or text. For example, the example detection results can represent the degree of match between the example execution result and the task requirement conditions for the task example based on text such as "match" or "mismatch." However, the example detection results are not limited to this. The example detection results can also represent the degree of match based on quantitative data. For example, the degree of match between the example execution result and the task requirement conditions for the task example can be represented based on information such as numerical values ​​or matrices. The embodiments of the present disclosure do not limit the information type of the example detection results.

[0051] In some embodiments, the task examples for the agent to be tested may include multiple task examples with dependencies, and the main agent may control the agent to be tested to execute the multiple task examples according to the dependencies to obtain multiple example execution results.

[0052] For example, multiple task examples with dependencies can be multiple task examples for generating a problem-solving video. These multiple task examples can include a problem-solving task example, a problem-writing generation example, a problem-illustration generation example, and a video synthesis task example. Based on the dependencies, the agent under test executes the multiple task examples, and the example execution results obtained for the multiple task examples can be a problem-solving result, a problem-writing, a problem-illustration data, and a problem-writing video, respectively.

[0053] In some embodiments, a master agent can be used to test example execution results, suitable for performing generalized detection tasks. However, this is not a limitation. Multiple sub-expert agents can also be configured for the master agent, each of which is suitable for performing personalized detection tasks on example execution results of a specific type of task example, thereby improving the detection accuracy of the example detection results.

[0054] The model structure configured in the general agent or sub-expert agent is not limited. For example, it can include one or more deep learning models such as attention mechanisms, convolutional networks, recurrent networks, and long short-term memory networks, but it is not limited to these. It can also be a large language model with more powerful functions. As long as it can be used to perform the detection task, it will be sufficient. The difference is that by utilizing multiple sub-expert agents, a hybrid expert mechanism can be formed (Mixture of Experts), combining multiple targeted and personalized sub-expert agents to detect different example execution results based on different task requirements.

[0055] In some embodiments, using a master agent to detect example execution results may include using the master agent to process the example execution results and task requirements. This allows the master agent to use the task requirements as prompt information, enabling it to utilize its enhanced understanding and analytical capabilities to detect the degree of match between the example execution results and the task requirements, thereby improving the accuracy of the example detection results relative to the example execution results.

[0056] According to an embodiment of the present disclosure, based on example detection results and task examples, the main intelligent agent is used to describe the task execution capability of the intelligent agent to be tested, which may include using the main intelligent agent to process example detection results and task examples, so that the main intelligent agent can more accurately describe the execution capability and execution effect of the intelligent agent to be tested for the task example by understanding the degree of match between the example execution results represented by the example detection results and the corresponding task requirement conditions, so that the execution capability and execution effect of the intelligent agent to be tested for the target task to be executed can be more accurately represented by the capability description information, and then based on the capability description information, the intelligent agent suitable for executing the target task to be processed can be selected to improve the execution efficiency of the target task and the accuracy of the task execution results.

[0057] In one example, multiple task examples can be task examples used to test the capabilities of a test agent in different dimensions or attributes. By utilizing a master agent to process a large number of task examples and example detection results, the task execution capabilities of the test agent are described, thereby describing the multi-dimensional capabilities of the test agent and outputting a multi-dimensional capability description of the test agent. It should be noted that the dimensions or attributes of the task examples can include attributes of requirement information representing task requirements, such as the number of words in the generated text or the complexity of the generated text.

[0058] In some embodiments, the task examples for the agent to be tested may include multiple task examples with dependencies, and the agent to be tested includes multiple sub-agents to be tested arranged according to execution logic, and the multiple sub-agents to be tested can be used to execute multiple task examples respectively to obtain multiple example execution results. The example detection result can represent the degree of matching between the example execution result obtained by the sub-agent to be tested when executing the task example and the task requirement conditions for the task example, so that the execution capability of the sub-agent to be tested for the task example can be represented based on the example detection result. By using the main agent to process multiple example detection results and multiple task examples, the obtained capability description information can more accurately describe the execution capability of each sub-agent to be tested in the agent to be tested for the corresponding target task to be processed, as well as the overall execution capability of the agent to be tested to execute multiple task examples with dependencies. Therefore, based on the method provided in the embodiments of the present disclosure, the task execution capabilities of multiple sub-agents to be tested, which are arranged through relatively complex dependency relationships, and the collaboration capabilities between multiple sub-agents to be tested can be more accurately evaluated, so as to select multiple matching sub-agents to be tested for collaboration for multiple target tasks with responsible dependencies, so as to improve the execution efficiency of complex target tasks.

[0059] It should be noted that in this embodiment, the multiple task examples with dependencies can be understood as a task example topology. The agent under test executes the multiple task examples based on the dependencies between the multiple task examples indicated by the task example topology, and can obtain the execution results of the task example topology. For example, the task example topology may include a problem-solving task example, a lecture text generation task example, a lecture audio synthesis task example, and a video generation task example, all with dependencies. The execution result of the task example topology can be a generated lecture video.

[0060] The following will explain and illustrate the capability description method of the intelligent agent provided by the embodiment of the present disclosure in combination with the drawings and specific embodiments.

[0061] In some embodiments, the task examples include at least one of: a video generation task example, an image editing task example, and a text generation task example.

[0062] The agent under test or its sub-agents can generate any type of video by executing the video generation task example, such as advertisements, animations, lectures, etc. The agent under test or its sub-agents can generate videos by processing video material data such as spoken text and avatar images.

[0063] The image editing task example may represent an editing task of performing color editing on an image, adding screen objects, etc. An example execution result of the image editing task example may be an edited image.

[0064] The copy generation task example can represent a task example for generating copy such as novels, press releases, and speech copy. The intelligent agent to be tested or the sub-intelligent agent to be tested can generate copy by executing the copy generation task example according to the required conditions.

[0065] It should be noted that task examples can also include examples for performing other tasks. For example, task examples can also be image detection task examples, problem-solving task examples for solving problems, code generation task examples for generating code, etc. The embodiments of the present disclosure do not limit the specific task types of task examples.

[0066] In some embodiments, the main agent may include any one or more of a first sub-agent, a second sub-agent, a third sub-agent, and a fourth sub-agent. The first sub-agent, the second sub-agent, the third sub-agent, and the fourth sub-agent may be expert agents for performing different specified tasks, respectively. Thus, the capability description method of the agent provided in the embodiment of the present disclosure may be executed through collaboration between multiple sub-agents in the main agent.

[0067] In some embodiments, the main agent is used to detect the example execution results, and obtaining the example detection results includes: using the fourth sub-agent of the main agent to process multiple example execution results, and the task requirement conditions for each of the multiple example execution results, to obtain multiple example detection results.

[0068] According to the embodiments of the present disclosure, multiple example detection results can characterize the execution capabilities of the intelligent agent to be tested or the sub-intelligent agent to be tested to respectively execute different task examples. Therefore, by obtaining multiple different task examples with different task attributes such as different task types and different task parameters, the fourth sub-intelligent agent can be used to process multiple task examples and multiple example detection results, so that the capability description information can represent the execution capabilities of the intelligent agent to be tested or the sub-intelligent agent to be tested for the target task with multi-dimensional specific task attributes, so as to facilitate accurate detection and evaluation of the execution capabilities of the intelligent agent to be tested, and improve the efficiency and scheduling accuracy of subsequent intelligent agent scheduling for the target task.

[0069] In some embodiments, among multiple dependent task examples, the execution results of the preceding task example can serve as task element data for the subsequent task example. For example, if the preceding task example is a lecture copy generation task example, the execution result of the lecture copy generation task example is the lecture copy. If the subsequent task example is a speech synthesis task example, the test agent can execute the speech synthesis task example by processing the lecture copy, and the generated execution result is spoken speech data representing the lecture copy.

[0070] By utilizing the main agent to process multiple task examples with dependencies and the example detection results of each of the multiple task examples, it is possible to detect based on the main agent's relatively strong understanding ability that the example execution result is an abnormality in the process of executing a specific task example among multiple task examples, so as to more accurately analyze the defects in the execution ability of the agent to be tested for the specific task example, so that the capability description information can accurately describe the execution ability of the agent to be tested to perform complex multi-step tasks.

[0071] In some embodiments, the agent to be tested includes multiple sub-agents to be tested arranged according to the execution logic. The multiple sub-agents to be tested can be used to execute multiple task examples with dependencies respectively. By using the main agent to process multiple task examples with dependencies and the example detection results of the multiple task examples, the capability description information can more accurately describe the collaboration capabilities between the multiple sub-agents to be tested, so as to improve the accuracy of the capability boundary description of the agent to be tested as a workflow execution component for the collaboration of multiple sub-agents.

[0072] In some embodiments, obtaining task examples for the intelligent agent to be tested includes: obtaining multiple task examples for describing the capabilities of the intelligent agent from a task example library, wherein the multiple task examples may have different task attributes and different task requirements, so as to generate capability description information that accurately describes the capabilities of the intelligent agent to be tested through the example execution results corresponding to each of the multiple task examples.

[0073] In some embodiments, obtaining a task example for the agent to be tested includes: using a master agent to process agent attributes for the agent to be tested to obtain a task example.

[0074] According to an embodiment of the present disclosure, agent attributes represent attribute information used to describe the specific functions of the agent under test or the sub-agent under test. Agent attributes may be acquired during a preliminary testing phase, or may be determined based on the model performance or model capabilities of a large model used to construct the agent. Agent attributes may include, for example, information such as the task types and task attribute parameters that the agent under test or the sub-agent under test can perform.

[0075] In some examples, a master agent can be used to process agent task attributes to generate task examples for at least one sub-agent in the agent under test. This allows the execution of multiple task examples generated by the master agent based on the orchestrated dependencies between the multiple agents under test, thereby accurately describing the execution capability boundaries of each sub-agent under test for different tasks. This improves the accuracy of the capability description information for each sub-agent under test in the workflow component under test, and further improves the accuracy of the capability description for complex multi-tasks of the agent under test.

[0076] In some embodiments, using the main agent to process the agent attributes for the agent to be tested to obtain a task example includes: using the first sub-agent of the main agent to process the agent attributes for the agent to be tested to obtain initial capability description information related to the agent to be tested; using the third sub-agent of the main agent to process the initial capability description information to obtain a task example.

[0077] According to an embodiment of the present disclosure, using the first sub-agent of the main agent to process the agent attributes for the agent to be tested can include using the first sub-agent to process the agent attributes for the agent to be tested to obtain the initial capability description information of the agent to be tested, as well as the initial capability description information of each of the multiple sub-agents to be tested in the agent to be tested. The initial capability description information can indicate the capability boundary range of the agent to be tested or the sub-agent to be tested to perform the task. For example, the initial capability description information can indicate the word count range, logical coherence score, etc. of the agent to be tested or the sub-agent to be tested to perform the text generation task, or the initial capability description information can indicate the scope of knowledge points involved in the problem-solving task performed by the agent to be tested or the sub-agent to be tested. The embodiment of the present disclosure does not limit the specific type of the capability boundary range, as long as it is compatible with the task requirement conditions for generating task examples.

[0078] The initial capability description information can be used as an initial prompt word for the capability boundary range, and combined with the example prompt word indicating the intention of the requirement for the task example, to control the third sub-agent to generate a variety of task examples. The diverse task examples can correspond to different task requirement conditions, so as to facilitate the control of the tested agent or the sub-agent to be tested to execute task examples with different task requirement conditions. For example, the initial capability description information and the example prompt word can be used to prompt the third sub-agent to generate an out-of-bounds task example that exceeds the execution capability boundary range of the sub-agent to be tested, or the initial capability description information and the example prompt word can be used to prompt the third sub-agent to generate an in-bounds task example that falls within the execution capability boundary range of the sub-agent to be tested. For example, the out-of-bounds task example can be a 100,000-word explanation copy for the representation, and the in-bounds task example can be a 500-word explanation copy for the representation.

[0079] Since the initial capability description information is usually not accurate in describing the execution capability boundary of the agent to be tested or the sub-agent to be tested for the task, and the quality of the task execution results of the agent may fluctuate greatly, the initial capability description information and example prompt words can be processed by a third sub-agent to generate multiple different within-boundary task examples and multiple identical outside-boundary task examples to perform multi-dimensional testing on the agent to be tested or the sub-agent to be tested, thereby improving the execution capability dimension represented by subsequent example detection results, and thereby improving the accuracy of the capability description information.

[0080] In one embodiment, example prompt words can be obtained by the first sub-agent by processing agent attributes to achieve control of the diversity of task examples through the understanding ability of the first sub-agent.

[0081] According to an embodiment of the present disclosure, the first sub-agent and the third sub-agent are agents with different functions and are suitable for performing different types of tasks. By generating task examples through the collaboration of the first sub-agent and the third sub-agent, the task examples can be combined with example prompt words and agent attributes to generate task examples for evaluating the task execution capabilities of the agent to be tested, so as to increase the data volume and diversity of the task examples, and then the example execution results can be detected by using the main agent, and the capability description information can be obtained to accurately represent the task execution capabilities of the agent to be tested. In this way, for the agent to be tested obtained by logically arranging multiple sub-agents to be tested, the task execution capabilities of the multiple sub-agents to be tested and the collaboration capabilities between the multiple sub-agents to be tested can be more accurately represented through the capability description information.

[0082] In some embodiments, based on example detection results and task examples, using the main agent to describe the capabilities of the agent to be tested includes: based on the example detection results, using the first sub-agent of the main agent to update the task example to obtain an intermediate task example; and using the second sub-agent of the main agent to process the intermediate task example and the intermediate detection results to obtain capability description information.

[0083] According to an embodiment of the present disclosure, an intermediate detection result is determined based on an intermediate execution result. For example, the intermediate execution result can be obtained by a master agent detecting the intermediate execution result, which is determined by the tested agent executing the intermediate task example. The intermediate detection result can indicate the degree of match between the intermediate task requirement for the intermediate task example and the intermediate execution result.

[0084] According to an embodiment of the present disclosure, updating the task examples using the first sub-agent can include using an agent with a task example generation function to process example detection results and task examples to obtain updated intermediate task examples. In this way, the first sub-agent can understand the execution capability of the agent to be tested or the sub-agent to be tested for the task examples by processing the example detection results, and the intermediate task examples obtained can be updated to represent new in-boundary task examples and out-of-boundary task examples. In this way, the second sub-agent with a relatively strong data semantic understanding ability can be used to process intermediate task examples and intermediate detection results to further deeply understand the execution capability of the agent to be tested or the sub-agent to be tested for task examples of various task attributes, and accurately describe the execution capability boundary range of the agent to be tested or the sub-agent to be tested by generating capability description information, thereby improving the accuracy of capability description for the agent to be tested, and improving the scheduling efficiency, task execution efficiency and execution quality of calling agents to perform tasks for different types of tasks.

[0085] In one embodiment, the intermediate detection result is obtained by processing the intermediate execution result using the fourth sub-agent of the main agent and the intermediate task requirement conditions for the intermediate task example. The intermediate task requirement conditions may be output by the third sub-agent. The intermediate task requirement conditions can serve as execution conditions for the intermediate task example to control the agent under test or the sub-agent under test to execute the intermediate task example based on the intermediate task requirement conditions.

[0086] In one embodiment, the method for describing the capabilities of an agent may also include iteratively taking at least one currently generated intermediate task example and intermediate detection result as the currently generated task example, using the first sub-agent of the main agent to process the example detection result of the currently generated task example, to update the currently generated task example in each iteration round, and obtain a new intermediate task example, so that the agent to be tested or the sub-agent to be tested can be controlled by the collaboration of multiple sub-agents in the main agent to execute the intermediate task examples generated in multiple rounds, and determine the intermediate execution results and intermediate detection results of multiple rounds. The second sub-agent is used to process the intermediate task examples, intermediate execution results and intermediate detection results of multiple rounds as the currently generated task example, example execution result and example detection result to understand the task execution capability of the agent to be tested, and obtain capability description information that more accurately describes the capability boundary range of the agent to be tested.

[0087] In some embodiments, based on the example detection results, the first sub-agent of the main agent is used to update the task example to obtain the intermediate task example, including: based on the example detection results, the first sub-agent is used to update the initial capability description information for the agent to be tested to obtain the intermediate capability description information; and the third sub-agent of the main agent is used to process the intermediate capability description information to obtain the intermediate task example.

[0088] In one embodiment, based on the example detection results, the first sub-agent is used to update the initial capability description information for the agent to be tested, which can include using the first sub-agent to process the currently generated example detection results, so that the first sub-agent can understand the execution capability of the agent to be tested for a variety of task examples through the example detection results of the currently generated task examples, thereby enabling the intermediate capability description information to more accurately represent the execution capability boundary range of the agent to be tested.

[0089] In one embodiment, the agent capability description method may further include iteratively using at least one currently generated intermediate task example and intermediate detection result as the currently generated task example, processing the example detection result of the currently generated task example using the first sub-agent of the main agent to generate new intermediate capability description information in each iteration round. Processing the intermediate capability description information generated in each round using the third sub-agent to obtain the intermediate task example of the current round.

[0090] In one embodiment, the intermediate task examples generated by the main agent in at least one round may be multiple intermediate task examples applicable to different sub-agents to be tested in the agent to be tested, thereby detecting the execution capabilities of different sub-agents to be tested to collaboratively execute complex multi-tasks by generating multiple intermediate task examples with dependencies for multiple sub-agents to be tested that have been logically orchestrated, thereby improving the capability description information to accurately describe the multi-agent collaborative capabilities with more complex execution logic, and improving the scheduling efficiency and accuracy of the agents.

[0091] According to an embodiment of the present disclosure, the task examples may include multiple tasks, the agent to be tested includes multiple sub-agents to be tested with dependencies, and the multiple sub-agents to be tested execute the multiple task examples based on the orchestrated dependencies.

[0092] Using the third sub-agent of the main agent to process the intermediate capability description information to obtain the intermediate task example may include: using the third sub-agent to process the defect description information in the intermediate capability description information to obtain the first intermediate example.

[0093] According to an embodiment of the present disclosure, the defect description information represents a defect factor of a first example execution result, where the first example execution result is determined by a sub-agent under test executing a first task example among multiple task examples. The defect factor may represent defect attributes of the first example execution result, such as the defect type and defect score. For example, the defect factor may represent the type of code error in the generated code.

[0094] In one example, the defect description information indicates the degree of mismatch between the first example execution result and the corresponding task requirement. Alternatively, the defect description information may also describe the degree of difference between the first example execution result and a baseline execution result represented by the corresponding task requirement. For example, the defect factor may be represented by the difference between the logical coherence score of the generated text and the logical coherence score of the baseline text represented by the task requirement.

[0095] In one embodiment, the defect description information may describe that the sub-agent to be tested failed to successfully execute the corresponding task example or intermediate task example.

[0096] By utilizing the third sub-agent to process the defect description information in the intermediate capability description information, the defect factors indicated by the defect description information can be used as prompt words to control the third sub-agent to adjust the task requirement conditions for the currently generated task example or intermediate task example according to the defect description information, thereby adjusting the currently generated task example or intermediate task example so that the generated first intermediate example can be successfully executed by the sub-agent to be tested, or so that the example execution result obtained by the sub-agent to be tested executing the first intermediate example can be closer to the task requirement conditions for the first intermediate example. In this way, the first intermediate example can be iteratively generated through multiple rounds so that the sub-agent to be tested can eventually successfully execute the currently generated first intermediate example, thereby utilizing the second sub-agent to understand the execution capabilities of the sub-agent to be tested for multiple task examples corresponding to the changing task requirement conditions by understanding the intermediate execution results and intermediate detection results of multiple rounds, thereby more accurately describing the task execution capability boundary of the sub-agent to be tested and improving the accuracy of the description of the capability boundary of the agent.

[0097] In one embodiment, based on the example detection results, the first sub-agent is used to update the initial capability description information for the agent to be tested, and the intermediate capability description information is obtained, including: using the first sub-agent to process the defect detection results in the example detection results to obtain defect description information for the sub-agent to be tested.

[0098] The defect detection result indicates that the execution result of the first example does not match the task requirements for the first task example. For example, the defect detection result may indicate that the generated image does not match the vehicle type or color required by the task requirements. Alternatively, the defect detection result may indicate that one or more steps in the problem-solving step data contain a solution error.

[0099] By utilizing the defect detection results in the example detection results processed by the first sub-agent, it can be prompted that the first sub-agent and the sub-agent to be tested have difficulty in performing in accordance with the task requirements when executing a specific task example. The defect detection results can be used as a prompt to control the first sub-agent to adjust the capability description of the sub-agent to be tested. The defect description information that more accurately describes the defect factors of the sub-agent to be tested can be used to control the third sub-agent to generate a new intermediate task example, so as to facilitate the detection of the capability boundary range of the sub-agent to be tested based on the new intermediate task example.

[0100] In some embodiments, the task example also includes a second task example, and using the third sub-agent of the main agent to process the intermediate capability description information to obtain the intermediate task example can also include: using the three sub-agents to process the completion description information in the intermediate capability description information to obtain a second intermediate example.

[0101] The completion description information indicates that the execution result of the second example matches the task requirement conditions for the second task example, and based on the completion description information, it can be indicated that the sub-agent to be tested has successfully executed the second example. The intermediate task requirement conditions for the second intermediate example are different from the task requirement conditions of the second task example.

[0102] In one embodiment, the requirement type used in the intermediate task requirement condition representation of the second intermediate example can be completely different from the requirement type used in the task requirement condition representation of the second task example. For example, the task requirement condition used in the second task example is the word count requirement condition of "at least 500 words", while the intermediate task requirement condition used in the second intermediate example can be expressed as "the copy has a summary section, an execution step section, and a summary section." By generating second intermediate examples with different requirement types, multi-dimensional capability boundary detection is performed on the sub-agents to be tested, thereby improving the accuracy of the capability description information for the multi-dimensional capability descriptions of multiple sub-agents to be tested in the tested agent.

[0103] In one embodiment, the requirement type represented by the intermediate task requirement condition for the second intermediate example can be the same as the requirement type represented by the task requirement condition for the second task example, but the requirement attribute information of the requirement type is different. For example, the task requirement condition for the second task example is the word requirement condition "at least 500 words", and the intermediate task requirement condition for the second intermediate example can be expressed as "at least 5000 words". By adjusting the requirement attributes of the task requirement conditions of the same requirement type to generate a new intermediate task example, the collaboration of multiple sub-agents in the main agent can be utilized to explore the execution capabilities of the sub-agents to be tested for task examples with the same requirement type, and the capability description information output by the main agent can be used to more accurately characterize the task execution capability boundaries and capability range of the agent to be tested, thereby improving the scheduling accuracy of the agent.

[0104] Figure 3 A diagram schematically illustrates an application scenario of the method for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure.

[0105] like Figure 3 As shown, the main agent may include a first sub-agent 311, a second sub-agent 312, a third sub-agent 313, and a fourth sub-agent 314. The agent under test 320 may include multiple sub-agents under test based on logical arrangement. The sub-agents under test may be represented by solid circle elements, and dependencies may be represented as edges between different solid circle elements.

[0106] In round 0 (or the initial round), agent attributes are input to the first sub-agent 311. This first sub-agent 311 processes the agent attributes of the agent under test 320 to obtain initial capability description information. The third sub-agent 313 processes this initial capability description information to generate task examples for each of the multiple sub-agents under test. By calling an interface, the multiple sub-agents under test 320 are controlled to execute multiple task examples and transmit the example execution results to the fourth sub-agent 314. The fourth sub-agent 314 processes the multiple task examples, the task requirements for the task examples, and the multiple example execution results to obtain multiple example detection results.

[0107] In the first round, the first sub-agent 311 is used to process multiple example detection results to obtain first intermediate capability description information that characterizes the collaboration capabilities of the multiple sub-agents to be tested and the task execution capabilities of each of the multiple sub-agents to be tested. The third sub-agent 313 is used to process the first intermediate capability description information to generate first intermediate task examples for each of the multiple sub-agents to be tested. By calling the interface, the multiple sub-agents to be tested of the agent to be tested 320 are controlled to execute the first intermediate task examples and send the first intermediate execution results to the fourth sub-agent 314. The fourth sub-agent 314 is used to process the multiple first intermediate task examples, the first intermediate task requirement conditions for the first intermediate task examples, and the multiple first intermediate execution results to obtain multiple first intermediate detection results.

[0108] In the second round, the first intermediate detection result and the example detection result of round 0 are used as the currently generated example detection result. The first sub-agent 311 is used to process the currently generated example detection result to obtain the second intermediate capability description information for the agent to be tested. The third sub-agent 313 is used to process the second intermediate capability description information to generate second intermediate task examples for each of the multiple sub-agents to be tested. By calling the interface, the multiple sub-agents to be tested of the agent to be tested 320 are controlled to execute the second intermediate task examples and send the second intermediate execution results to the fourth sub-agent 314. The fourth sub-agent 314 is used to process the multiple second intermediate task examples, the second intermediate task requirement conditions for the second intermediate task examples, and the multiple second intermediate execution results to obtain multiple second intermediate detection results.

[0109] It should be understood that in the nth round, the example detection results from the n-1th intermediate detection result to the 0th round are used as the currently generated example detection results, and the first sub-agent 311 is used to process the currently generated example detection results to obtain the nth intermediate capability description information for the agent to be tested. The third sub-agent 313 is used to process the nth intermediate capability description information to generate the nth intermediate task examples for each of the multiple sub-agents to be tested. The multiple sub-agents to be tested of the agent to be tested 320 are controlled by calling the interface to execute the nth intermediate task example, and the nth intermediate execution result is sent to the fourth sub-agent 314. The fourth sub-agent 314 is used to process the multiple nth intermediate task examples, the nth intermediate task requirement conditions for the nth intermediate task example, and the multiple nth intermediate execution results to obtain multiple nth intermediate detection results. Wherein, n is an integer greater than 1.

[0110] In the case of N=n, the intermediate task examples, intermediate execution results, and intermediate detection results of rounds n, as well as the task examples, example execution results, and example detection results of round 0, can be input as the currently generated task examples, example detection results, and example execution results to the second sub-agent 312. The second sub-agent 312 is used to describe the task execution capability of the agent to be tested 320 to obtain capability description information.

[0111] In some embodiments, the capability description information includes at least one of the following: dependency relationships between multiple sub-agents in the agent to be tested; data attribute conditions for input data and output data of the sub-agents to be tested, as well as defect examples and completion examples.

[0112] In one embodiment, the data attribute conditions for the input data and output data of the sub-agent to be tested may represent constraints on data attribute information such as the data format, data type, and numerical range of the input data or output data.

[0113] A defective example indicates that the first example execution result obtained by the sub-agent under test when executing the first task example in the task examples does not match the task requirement conditions for the first task example. A defective example may represent an out-of-bounds task example associated with the sub-agent under test, where the sub-agent under test has difficulty generating an execution result that matches the task requirement conditions according to the task requirement conditions of the defective example.

[0114] A completed example represents a second example execution result obtained by a sub-agent under test when performing a second task example in a task example, and matches the task requirements for the second task example. A completed task example may represent an in-boundary task example associated with the sub-agent under test. The sub-agent under test is able to efficiently execute the completed task example and obtain an example execution result that matches the task requirements.

[0115] It should be noted that the completed examples or defect examples can be task examples in the above embodiments, or can also be intermediate task examples. For example, the completed examples or defect examples are task examples and intermediate task examples generated by the master agent in each round based on multiple rounds of collaboration.

[0116] In some embodiments, the capability description information may also include boundary task examples. These boundary task examples can indicate that the sub-agent under test can successfully execute under specific conditions and obtain example execution results that meet the task requirements. However, the quality of these example execution results is not stable, or there may be errors in the results. Boundary task examples can more accurately represent the boundaries of the sub-agent's task execution capabilities. This allows for a more accurate characterization of the task execution capabilities of the agent under test through boundary task examples and the descriptive text used to describe them.

[0117] In some embodiments, the capability description information may also include example execution results of task examples and intermediate task examples, as well as the dependencies between multiple sub-agents within the test agent. These dependencies may include information such as communication sequences, data interfaces for input and output data, and logical control units for controlling the collaboration of multiple sub-agents. This capability description information can be used to accurately call an agent to execute a target task.

[0118] Figure 4 A diagram schematically illustrates an application scenario of the method for describing the capabilities of an intelligent agent according to an embodiment of the present disclosure.

[0119] like Figure 4As shown, the agent attributes for the agent to be tested 410 can be input into the agent capability evaluation system 401. The agent capability evaluation system 401 can be constructed based on multiple sub-agents in the main agent provided in the embodiment of the present disclosure, and the agent capability evaluation system 401 can generate capability description information 420 by executing the agent description method provided in the embodiment of the present disclosure. The capability description information 420 may include data attribute conditions for input data and output data of the sub-agent to be tested 410, dependency description text for describing the dependency between multiple sub-agents to be tested in the agent to be tested 410, defect examples, completion examples, defect example execution results, completion example execution results and capability description text. The capability description information can be stored based on structured information to facilitate calling the agent to be tested 410 through structured capability description information.

[0120] By executing the capability description method of the intelligent agent of the embodiment of the present disclosure, the capability description information for the intelligent agent to be tested is determined, and the task execution capability of the complex intelligent agent to be tested, which is constructed based on multiple sub-agent collaboration processes, can be accurately described. The sub-agent collaboration process that matches the target task can be accurately and flexibly scheduled based on the capability description information, so as to improve the scheduling accuracy and deployment flexibility of the intelligent agent collaboration mechanism in various application scenarios, and improve the execution efficiency and execution accuracy of complex tasks.

[0121] Figure 5 A flowchart schematically illustrates an agent-based task execution method according to an embodiment of the present disclosure.

[0122] like Figure 5 As shown, the agent-based task execution method includes operations S510 to S540.

[0123] In operation S510 , task requirement information is obtained;

[0124] In operation S520, task intention detection is performed on the task requirement information to obtain task description information;

[0125] In operation S530, a target agent is determined from the execution agents based on the task description information and the capability description information for the preset execution agents.

[0126] According to an embodiment of the present disclosure, the capability description information is determined according to the capability description method of the intelligent agent provided by the implementation of the present disclosure.

[0127] In operation S540 , the target agent is controlled to execute the target task related to the task requirement information, and a task execution result is obtained.

[0128] According to an embodiment of the present disclosure, the task requirement information may be used to indicate the requirements of a target task to be performed. The task requirement information may be represented based on data types in any format, such as voice and text.

[0129] In some embodiments, performing task intent detection on task requirement information may include processing the task requirement information using a trained large language model to obtain task description information. The task description information may represent text, characters, or other data describing the target task requirements, target task attribute parameters, and other task attributes of the target task to be performed.

[0130] In some embodiments, the task description information can be matched with the capability description information of the execution agent to obtain a matching result, and a target agent suitable for executing the target task can be determined from a plurality of execution agents based on the matching result.

[0131] It should be noted that the above operations S510 to S540 can be performed based on electronic devices such as servers.

[0132] In some embodiments, operations S510 to S540 may be performed based on a scheduling agent located in a server or server cluster. The scheduling agent may be used to generate a target task based on user demand information and invoke a target agent to execute the target task, thereby obtaining a high-precision task execution result for the target task.

[0133] In some embodiments, the execution agent includes multiple sub-execution agents with dependency relationships. The multiple sub-execution agents can be used to execute target tasks according to the execution order or execution logic of the multiple target tasks to obtain the execution results of the target tasks.

[0134] In some embodiments, the scheduling agent may include multiple sub-scheduling agents, which may be used to perform different operation steps in the above operations S510 to S540.

[0135] In some embodiments, determining a target agent from among the executing agents includes: determining a target agent from among a plurality of executing agents using a second sub-scheduling agent based on capability description information and task description information.

[0136] For example, the second sub-scheduling agent can be used to process the capability description information and task description information of multiple execution agents to determine the target capability description information that matches the task description information, and then the execution agent corresponding to the target capability description information can be used as the target agent. In this way, when the execution agent includes multiple sub-execution agents, the capability description information and task description information can be used to quickly determine the target agent suitable for executing the target task, thereby improving the execution efficiency of the target task.

[0137] In one embodiment, determining the target agent from the execution agent includes: using the first sub-scheduling agent to process task description information and capability description information for the sub-execution agent to obtain a sub-target agent among multiple sub-execution agents; and using the first sub-scheduling agent to logically arrange multiple sub-target agents to obtain a target agent.

[0138] According to an embodiment of the present disclosure, a sub-target agent is used to execute related target tasks among multiple target tasks. By utilizing a first sub-scheduling agent to logically orchestrate multiple sub-target agents within a target agent, dependencies between the multiple sub-target agents are obtained, thereby enabling the fine-grained construction of a target agent capable of executing complex execution processes to execute multiple target tasks. By executing the corresponding related target tasks through the dependencies, task execution results for the multiple related target tasks are obtained, thereby improving the execution efficiency and accuracy of the target tasks.

[0139] In one embodiment, the execution agent includes multiple sub-target agents, at least two of which are determined from different execution agents. By determining sub-execution agents that match each of the multiple target tasks from the multiple execution agents and logically arranging the multiple sub-execution agents to obtain the target agent, when the current execution agent is unable to meet the task requirements, a complex execution process for executing the target task can be obtained by fine-grained disassembly and arrangement of the sub-execution agents, thereby improving the execution efficiency and accuracy of the target task.

[0140] Figure 6 A diagram schematically illustrates an application scenario of an agent-based task execution method according to an embodiment of the present disclosure.

[0141] like Figure 6 As shown, the scheduling agent can include multiple sub-scheduling agents, including a conversation understanding sub-agent, a demand understanding sub-agent, a logic orchestration sub-agent, and an interaction sub-agent. The scheduling agent can be deployed in a distributed server cluster, and information can be transmitted between the multiple sub-scheduling agents via communication links within the distributed server cluster. The scheduling agent makes decisions and controls user task execution requirements through the collaboration of multiple sub-scheduling agents.

[0142] Specifically, the user sends unstructured conversation information to the scheduling agent via a smart terminal such as a smartphone. The conversation understanding sub-agent processes the conversation information to detect task intent. For example, the conversation understanding sub-agent can perform operation S601 to determine whether the user intends to chat by processing the conversation information. If the result of operation S601 is yes, it can be determined that the user's conversation information can be general conversation content or simple question and answer content, and there is no need to call an agent. The interactive sub-agent can be used to process the conversation information to generate reply content, and the reply content can be sent to the user to meet the user's real-time conversation needs.

[0143] If the judgment result of operation S601 is negative, it can be determined that the user needs to call an agent to perform the task and obtain the task execution result. By using the dialogue understanding sub-agent to process the demand information and detect task intent, structured task description information is obtained. Structured task description information facilitates the demand understanding sub-agent to conduct in-depth analysis and accurate understanding of the task semantics and task execution logic represented by the task description information. The demand understanding sub-agent determines whether there is a matching target agent in the agent library by performing operation S602. Specifically, the demand understanding sub-agent processes the task description information and the capability description information for the executing agent to determine whether the capability description information matches the task description information. If the judgment result of operation S602 is positive, the demand understanding sub-agent can determine a target agent that matches the task description information from at least one executing agent. The target agent is suitable for performing the target task corresponding to the task description information. The target task can be obtained by updating the task attribute parameters of the task example for the target agent using the task description information, or it can be determined by the demand understanding sub-agent by processing the task description information. Then, the demand understanding sub-agent is used to perform operation S603, control the target agent to perform the target task, and obtain the task execution result of the target task.

[0144] In the case where the judgment result of operation S602 is yes, the logic arrangement sub-agent is used to assume the responsibility of dynamically constructing the collaboration of multiple sub-agents. The logic arrangement sub-agent is used to process the task description information and the capability description information for the sub-execution agent to determine multiple sub-target agents that match the task description information from the sub-execution agents of the multiple execution agents. The task description information is processed by the logic arrangement sub-agent to logically arrange the multiple sub-target agents, and the dependency relationship between the multiple sub-target agents is obtained, thereby constructing a target agent that can meet the new target task requirements of the current user, and realizing the construction of the collaboration process and method of multiple sub-target agents. The logic arrangement sub-agent uses the task requirement information and the target task to infer and filter out multiple sub-target agents that are suitable for executing the target task, and constructs a multi-agent collaboration method for the target task by logically arranging the multiple sub-target agents, thereby improving the scheduling flexibility and accuracy for the target agent. Therefore, the execution agent reuse and dynamic agent collaboration can be realized by scheduling the agent based on the collaboration of multiple sub-scheduling agents, so that efficient and robust task response capabilities can still be achieved in the face of task diversity, strong uncertainty in intelligent collaboration, and heterogeneity in the distribution of agent capabilities.

[0145] The interactive sub-agent processes the target task's execution results and converts one or more of them into user-understandable, natural language responses. For example, the interactive sub-agent can summarize and refine the task execution results to generate user-understandable responses.

[0146] Alternatively, a natural language response can be generated based on the user's configured response method. For example, voice data expressing the task response based on the timbre of a specified object can be fed back to the user. The interactive sub-agent can also proactively provide relevant task response information to the user based on the context of multiple rounds of dialogue.

[0147] In addition, the scheduling agent can also use adjustment mechanisms such as target task execution status tracking and target task execution failure rollback mechanism to improve the stability and robustness of target task execution.

[0148] It should be understood that Figure 6 In the illustrated embodiment, the first scheduling sub-agent is a logic orchestration sub-agent, and the second scheduling sub-agent is a demand understanding sub-agent.

[0149] Figure 7 The block diagram of the capability description device of an intelligent agent according to an embodiment of the present disclosure is schematically shown.

[0150] like Figure 7As shown, the capability description device 700 of an intelligent agent includes: a first acquisition module 710 , an example detection result acquisition module 720 and a capability description information acquisition module 730 .

[0151] The first acquisition module 710 is used to acquire task examples for the agent to be tested.

[0152] The example detection result acquisition module 720 is used to use the main intelligent agent to detect the example execution result to obtain the example detection result. The example execution result is determined based on the task example executed by the intelligent agent to be tested. The example detection result represents the degree of matching between the example execution result and the task requirement conditions used for the task example.

[0153] The capability description information acquisition module 730 is used to use the master agent to describe the task execution capability of the agent to be tested based on the example detection results and task examples, and obtain capability description information for the agent to be tested.

[0154] According to an embodiment of the present disclosure, the capability description information obtaining module 730 includes: a first obtaining submodule and a second obtaining submodule.

[0155] The first acquisition submodule is used to update the task example based on the example detection result using the first sub-agent of the main agent to obtain the intermediate task example.

[0156] The second acquisition submodule is used to use the second sub-agent of the main agent to process the intermediate task examples and intermediate detection results to obtain capability description information, wherein the intermediate detection results are determined based on the intermediate execution results, and the intermediate execution results are determined by the agent to be tested executing the intermediate task examples.

[0157] According to an embodiment of the present disclosure, the first obtaining submodule includes: an intermediate capability description information obtaining unit and an intermediate task example obtaining unit.

[0158] The intermediate capability description information obtaining unit is used to update the initial capability description information for the agent to be tested using the first sub-agent based on the sample detection result to obtain the intermediate capability description information.

[0159] The intermediate task example obtaining unit is used to use the third sub-agent of the main agent to process the intermediate capability description information to obtain the intermediate task example.

[0160] According to an embodiment of the present disclosure, the task examples include multiple ones, and the agent to be tested includes multiple sub-agents to be tested having a dependency relationship.

[0161] The intermediate task example obtaining unit includes a first obtaining sub-unit.

[0162] The first acquisition sub-unit is used to use the third sub-agent to process the defect description information in the intermediate capability description information to obtain a first intermediate example, wherein the defect description information represents the defect factor of the first example execution result, and the first example execution result is determined by the first task example among multiple task examples executed by the sub-agent to be tested.

[0163] According to an embodiment of the present disclosure, the intermediate capability description information obtaining unit includes a second obtaining sub-unit.

[0164] The second obtaining sub-unit is used to use the first sub-agent to process the defect detection results in the example detection results to obtain defect description information for the sub-agent to be tested, and the defect detection results represent that the first example execution results do not match the task requirement conditions for the first task example.

[0165] According to an embodiment of the present disclosure, the intermediate task example obtaining unit includes a third obtaining sub-unit.

[0166] The third acquisition sub-unit is used to use the three sub-agents to process the completion description information in the intermediate capability description information to obtain a second intermediate example, wherein the completion description information characterizes that the execution result of the second example matches the task requirement conditions for the second task example, and the task example also includes a second task example, and the intermediate task requirement conditions for the second intermediate example are different from the task requirement conditions of the second task example.

[0167] According to an embodiment of the present disclosure, the intermediate detection result is obtained by processing the intermediate execution result using the fourth sub-agent of the main agent and the intermediate task requirement conditions for the intermediate task example.

[0168] According to an embodiment of the present disclosure, the capability description information includes at least one of the following: dependency relationships between multiple sub-agents to be tested in the agent to be tested; data attribute conditions for input data and output data of the sub-agents to be tested; a defect example, which characterizes that the first example execution result obtained by the sub-agent to be tested when executing the first task example in the task example does not match the task requirement conditions for the first task example; a completion example, which characterizes that the second example execution result obtained by the sub-agent to be tested when executing the second task example in the task example matches the task requirement conditions for the second task example.

[0169] According to an embodiment of the present disclosure, the example detection result obtaining module 720 includes an example detection result obtaining submodule.

[0170] The example detection result acquisition submodule is used to use the fourth sub-agent of the main agent to process multiple example execution results, as well as the task requirement conditions for each of the multiple example execution results, to obtain multiple example detection results.

[0171] According to an embodiment of the present disclosure, the first acquisition module 710 includes a task example acquisition submodule.

[0172] The task example acquisition submodule is used to use the main agent to process the agent attributes for the agent to be tested to obtain task examples.

[0173] According to an embodiment of the present disclosure, the task example obtaining submodule includes: an initial capability description information obtaining unit and a task example obtaining unit.

[0174] The initial capability description information obtaining unit is used to process the agent attributes for the agent to be tested by using the first sub-agent of the main agent to obtain the initial capability description information related to the agent to be tested.

[0175] The task example obtaining unit is used to use the third sub-agent of the main agent to process the initial capability description information to obtain a task example.

[0176] According to an embodiment of the present disclosure, the task example includes at least one of the following: a video generation task example, an image editing task example, and a text generation task example.

[0177] Figure 8 A block diagram of an agent-based task execution device according to an embodiment of the present disclosure is schematically shown.

[0178] like Figure 8 As shown, the agent-based task execution device 800 includes: a second acquisition module 810, a task description information acquisition module 820, a target agent determination module 830 and a task execution result acquisition module 840.

[0179] The second acquisition module 810 is used to acquire task requirement information.

[0180] The task description information obtaining module 820 is used to perform task intention detection on the task requirement information to obtain task description information.

[0181] The target agent determination module 830 is used to determine the target agent from the execution agents based on the task description information and the capability description information of the preset execution agent, wherein the capability description information is determined according to the capability description method of the agent provided in the embodiment of the present disclosure.

[0182] The task execution result obtaining module 840 is used to control the target intelligent agent to execute the target task related to the task requirement information and obtain the task execution result.

[0183] According to an embodiment of the present disclosure, the execution agent includes a plurality of sub-execution agents having a dependency relationship; the target agent determination module 830 includes:

[0184] The sub-target agent acquisition submodule is used to utilize the first sub-scheduling agent to process the task description information and the capability description information for the sub-execution agent to obtain the sub-target agent among the multiple sub-execution agents, and the sub-target agent is used to execute the associated target tasks among the multiple target tasks.

[0185] The first obtaining submodule is used to use the first sub-scheduling agent to logically arrange multiple sub-target agents to obtain a target agent.

[0186] According to an embodiment of the present disclosure, the execution agent includes multiple, and at least two of the multiple sub-target agents are determined from different execution agents.

[0187] According to an embodiment of the present disclosure, the target agent determination module 830 includes a second obtaining submodule.

[0188] The second obtaining submodule is used to determine the target agent from multiple execution agents using the second sub-scheduling agent based on the capability description information and the task description information.

[0189] Figure 9 The structural block diagram of an artificial intelligence agent according to an embodiment of the present disclosure is schematically shown.

[0190] In the embodiments of the present disclosure, Figure 9 As shown, the AI ​​agent 900 may include an input module 910 , a processing module 920 and an output module 930 .

[0191] Input module 910, for receiving input information;

[0192] A processing module 920 is configured to execute the agent capability description method or the agent-based task execution method provided in accordance with an embodiment of the present disclosure based on the input information received by the input module to obtain output information;

[0193] The output module 930 is used to output the output information obtained by the processing module.

[0194] According to an embodiment of the present disclosure, the input module 910 is responsible for receiving or perceiving information such as queries, requests, instructions, signals, or data from the outside world (e.g., a user or the external environment) and converting it into a format that can be understood and processed by the AI ​​agent 900. The input module 910 is the primary link for the AI ​​agent 900 to interact with the outside world. It enables the AI ​​agent 900 to efficiently and accurately obtain the necessary "sensory" information from the outside world and respond to this information.

[0195] In an example, the input module 910 may input the task examples or requirement information described above.

[0196] In this example, the processing module 920 is the core support for the AI ​​agent 900 to process complex tasks. The processing module 920 can execute the agent capability description method or the agent-based task execution method described above.

[0197] In this example, the performance of processing module 920 may be closely related to the large model underlying AI agent 900. To fully leverage the capabilities of the large model, the internal structure of processing module 920 may be designed to be highly configurable and extensible to handle a variety of different tasks and requirements in real-world scenarios.

[0198] In the example, after the AI ​​agent 900 obtains the required voice, the processing module 920 can use the main agent to detect the example execution results to obtain the example detection results; based on the example detection results and task examples, the main agent is used to describe the task execution capabilities of the agent to be tested, and the capability description information for the agent to be tested is obtained, and the capability description information is passed to the output module 930.

[0199] Understandably, while large language models possess excellent language understanding and generation capabilities, like humans, they are limited in the tasks they can perform without tools. However, when AI Agent 900 is empowered with tool-based capabilities, it can perform tasks such as mathematical calculations using a calculator, data analysis using Python, and weather forecasting using search engines.

[0200] In an example, the output module 930 may output the capability description information or task execution result described above.

[0201] The AI ​​agent 900 according to the embodiment of the present disclosure can simply and effectively improve the level of intelligence, and enhance flexibility and versatility.

[0202] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0203] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method described above.

[0204] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method described above.

[0205] According to an embodiment of the present disclosure, a computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the method described above.

[0206] Figure 10 A schematic block diagram of an example electronic device 1000 of an agent capability description method and an agent-based task execution method that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0207] like Figure 10 As shown, device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a random access memory (RAM) 1003. RAM 1003 may also store various programs and data required for the operation of device 1000. Computing unit 1001, ROM 1002, and RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to bus 1004.

[0208] Various components in device 1000 are connected to I / O interface 1005, including an input unit 1006, such as a keyboard, mouse, etc.; an output unit 1007, such as various types of displays, speakers, etc.; a storage unit 1008, such as a magnetic disk, optical disk, etc.; and a communication unit 1009, such as a network card, modem, wireless communication transceiver, etc. The communication unit 1009 allows device 1000 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0209] The computing unit 1001 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as the agent capability description method or the agent-based task execution method. For example, in some embodiments, the agent capability description method or the agent-based task execution method can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 1000 via the ROM 1002 and / or the communication unit 1009. When the computer program is loaded into the RAM 1003 and executed by the computing unit 1001, one or more steps of the agent capability description method or the agent-based task execution method described above can be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to execute the agent capability description method or the agent-based task execution method in any other appropriate manner (eg, by means of firmware).

[0210] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0211] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0212] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0213] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0214] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0215] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0216] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0217] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for describing the capabilities of an intelligent agent, comprising: Get examples of tasks for the agent to be tested; Using a master agent to test the example execution result to obtain an example test result, wherein the example execution result is determined based on the test agent performing the task example, and the example test result represents the degree of match between the example execution result and the task requirement conditions for the task example; Based on the example detection results and the task example, the master agent is used to describe the task execution capability of the agent to be tested, and capability description information for the agent to be tested is obtained.

2. The method according to claim 1, wherein The step of using the master agent to describe the task execution capability of the agent to be tested based on the example detection result and the task example includes: Based on the example detection result, using the first sub-agent of the main agent to update the task example to obtain an intermediate task example; and The second sub-agent of the main agent is used to process the intermediate task example and the intermediate detection result to obtain the capability description information, wherein the intermediate detection result is determined based on the intermediate execution result, and the intermediate execution result is determined by the agent to be tested executing the intermediate task example.

3. The method according to claim 2, wherein: The updating of the task example based on the example detection result using the first sub-agent of the main agent to obtain the intermediate task example comprises: Based on the example detection result, using the first sub-agent to update the initial capability description information for the agent to be tested to obtain intermediate capability description information; The intermediate capability description information is processed by the third sub-agent of the main agent to obtain the intermediate task example.

4. The method according to claim 3, wherein: The task examples include multiple, and the agent to be tested includes multiple sub-agents to be tested with dependency relationships; The step of using the third sub-agent of the main agent to process the intermediate capability description information to obtain the intermediate task example includes: The defect description information in the intermediate capability description information is processed by the third sub-agent to obtain a first intermediate example, wherein the defect description information characterizes the defect factors of the execution result of the first example, and the execution result of the first example is determined by the sub-agent to be tested executing the first task example among multiple task examples.

5. The method according to claim 4, wherein The updating of the initial capability description information for the agent to be tested using the first sub-agent based on the example detection result to obtain the intermediate capability description information includes: The first sub-agent is used to process the defect detection result in the example detection result to obtain defect description information for the sub-agent to be tested, and the defect detection result represents that the first example execution result does not match the task requirement conditions for the first task example.

6. The method according to claim 3 or 4, wherein: The using the third sub-agent of the main agent to process the intermediate capability description information to obtain the intermediate task example includes: The completion description information in the intermediate capability description information is processed by the three sub-agents to obtain a second intermediate example, wherein the completion description information characterizes that the execution result of the second example matches the task requirement conditions for the second task example, and the task example also includes the second task example, and the intermediate task requirement conditions for the second intermediate example are different from the task requirement conditions of the second task example.

7. The method according to claim 2, wherein: The intermediate detection result is obtained by processing the intermediate execution result and the intermediate task requirement condition for the intermediate task example using the fourth sub-agent of the main agent.

8. The method according to claim 1, wherein The capability description information includes at least one of the following: Dependency relationships between multiple sub-agents to be tested in the agent to be tested; Data attribute conditions for input data and output data of the sub-agent to be tested; A defective example, indicating that a first example execution result obtained by the sub-agent to be tested when executing the first task example in the task examples does not match a task requirement condition for the first task example; The completion example represents that the second example execution result obtained by the sub-agent to be tested executing the second task example in the task example matches the task requirement conditions for the second task example.

9. The method according to any one of claims 1 to 8, wherein The example detection results obtained by using the main agent to detect the example execution results include: The fourth sub-agent of the main agent is used to process the plurality of example execution results and the task requirement conditions for each of the plurality of example execution results to obtain the plurality of example detection results.

10. The method according to claim 1, wherein Examples of tasks for obtaining the agent to be tested include: The master agent is used to process the agent attributes for the agent to be tested to obtain the task example.

11. The method according to claim 10, wherein: The using the master agent to process the agent attributes for the agent to be tested to obtain the task example includes: Processing the agent attributes of the agent to be tested using the first sub-agent of the main agent to obtain initial capability description information related to the agent to be tested; The third sub-agent of the main agent is used to process the initial capability description information to obtain the task example.

12. The method according to any one of claims 1 to 8, wherein Examples of the tasks include at least one of the following: Examples of video generation tasks, image editing tasks, and text generation tasks.

13. An agent-based task execution method, comprising: Obtain task requirement information; Performing task intent detection on the task requirement information to obtain task description information; Determining a target agent from the executing agents based on the task description information and capability description information for a preset executing agent, wherein the capability description information is determined according to the method of any one of claims 1 to 12; as well as The target agent is controlled to execute a target task related to the task requirement information to obtain a task execution result.

14. The method according to claim 13, wherein The execution agent includes a plurality of sub-execution agents having a dependency relationship; and determining the target agent from the execution agents includes: Processing the task description information and the capability description information for the sub-execution agent using the first sub-scheduling agent to obtain a sub-target agent among the plurality of sub-execution agents, wherein the sub-target agent is used to execute an associated target task among the plurality of target tasks; and The first sub-scheduling agent is used to logically arrange multiple sub-target agents to obtain the target agent.

15. The method according to claim 14, wherein The execution agent includes multiple, and at least two of the multiple sub-goal agents are determined from different execution agents.

16. The method according to any one of claims 13 to 15, wherein The determining of the target agent from the executing agents comprises: Based on the capability description information and the task description information, the target agent is determined from the plurality of execution agents using a second sub-scheduling agent.

17. A device for describing the capabilities of an intelligent agent, comprising: A first acquisition module is used to obtain task examples for the agent to be tested; an example detection result obtaining module, configured to use a master agent to detect the example execution result to obtain an example detection result, wherein the example execution result is determined based on the execution of the task example by the agent to be tested, and the example detection result represents the degree of match between the example execution result and the task requirement conditions for the task example; The capability description information obtaining module is used to use the main agent to describe the task execution capability of the agent to be tested based on the example detection results and the task example, and obtain capability description information for the agent to be tested.

18. An agent-based task execution device, comprising: The second acquisition module is used to obtain task requirement information; A task description information acquisition module is used to perform task intention detection on the task requirement information to obtain task description information; a target agent determining module, configured to determine a target agent from the executing agents based on the task description information and capability description information for a preset executing agent, wherein the capability description information is determined according to the method of any one of claims 1 to 12; as well as The task execution result acquisition module is used to control the target agent to execute the target task related to the task requirement information and obtain the task execution result.

19. An artificial intelligence agent configured to perform the method according to any one of claims 1 to 16.

20. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 16.

21. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 16.

22. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 16.

Images