Data processing method and task processing method

Abstract

Embodiments of the present specification provide a data processing method and a task processing method, the data processing method comprising: determining execution trajectory data of a target task, the execution trajectory data being generated by a first agent depending on a first model; generating an intelligent component library by using the first intelligent agent according to the execution trajectory data and a system constraint, the system constraint being determined according to a model configuration of a second model on which a second intelligent agent depends; generating a reference workflow by using the first agent according to the execution trajectory data and an intelligent component library; executing a test task by utilizing a second intelligent agent according to the reference workflow and the intelligent component library, and determining a target workflow set according to a task execution result; by collecting the execution track data generated when the first agent executes the target task, the efficient and stable target workflow suitable for the second agent is constructed, and the second agent can subsequently execute the same type of target task with low cost and high efficiency based on the target workflow.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of computer technology, and in particular to a data processing method and a task processing method. Background Art

[0002] Due to their powerful learning and reasoning capabilities, large models have been widely used in various tasks within the conversation analysis field. Conversation analysis can help customer service and sales industries achieve deep tasks such as improving customer service capabilities, optimizing systems, and increasing transaction conversion rates. For complex deep analysis tasks, agent systems based on strong models (such as LLMs) are currently generally used, achieving good results through dynamic execution path planning. However, strong models are very expensive, and the cost of invoking a single complex task is high. Furthermore, the flexibility of agents, which can automatically explore feasible paths, becomes a disadvantage when performing repetitive tasks. Each execution requires re-exploring feasible paths, resulting in unnecessary token consumption and redundant computation, prolonged inference time, and low efficiency. Furthermore, the uncertainty of dynamic paths significantly reduces the stability of agent systems. Summary of the Invention

[0003] In view of this, embodiments of this specification provide a data processing method. One or more embodiments of this specification also relate to a data processing device, a task processing method, a task processing device, a data processing method applied to a conversation analysis task, a data processing device applied to a conversation analysis task, a computing device, an electronic device, a computer-readable storage medium, and a computer program product to address the technical shortcomings of the prior art of using strong model-based agent systems to perform complex deep analysis tasks, such as high cost, low efficiency, and weak stability.

[0004] According to a first aspect of an embodiment of this specification, there is provided a data processing method, including: Determining execution trajectory data of a target task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target task based on a first model; generating, using the first agent, an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; Using the first agent, generating a reference workflow according to the execution trajectory data and the intelligent component library; The second agent is utilized to execute a test task according to the reference workflow and the intelligent component library, and a target workflow set is determined according to the task execution result, wherein the test task is determined according to the target task.

[0005] According to a second aspect of the embodiments of this specification, there is provided a data processing device, including: a trajectory determination module, configured to determine execution trajectory data of a target task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target task based on a first model; a component library generation module configured to generate an intelligent component library using the first agent based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; a reference workflow generation module, configured to generate a reference workflow using the first agent according to the execution trajectory data and the intelligent component library; The workflow set generation module is configured to use the second intelligent agent to execute a test task according to the reference workflow and the intelligent component library, and determine a target workflow set according to the task execution result, wherein the test task is determined according to the target task.

[0006] According to a third aspect of the embodiments of this specification, a task processing method is provided, including: Receiving a task to be executed, parsing the task to be executed using a second agent, and determining an execution workflow from a target workflow set based on the parsing result; The second agent is used to execute the task to be executed according to the execution workflow and the intelligent component library, wherein the target workflow set and the intelligent component library are generated according to the above-mentioned data processing method.

[0007] According to a fourth aspect of the embodiments of this specification, there is provided a task processing device, including: a task receiving module configured to receive a task to be executed, parse the task to be executed using a second agent, and determine an execution workflow from a target workflow set based on the parsing result; The task execution module is configured to utilize the second agent to execute the task to be executed according to the execution workflow and the intelligent component library, wherein the target workflow set and the intelligent component library are generated according to the above-mentioned data processing method.

[0008] According to a fifth aspect of the embodiments of this specification, a data processing method for a conversation analysis task is provided, comprising: Determining execution trajectory data of a target conversation analysis task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target conversation analysis task based on a first model; generating, using the first agent, an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; Using the first agent, generating a dialogue analysis reference workflow based on the execution trajectory data and the intelligent component library; Utilize the second intelligent agent to perform the conversation analysis test task according to the conversation analysis reference workflow and the intelligent component library, and determine the conversation analysis target workflow set according to the task execution result, wherein the conversation analysis test task is determined according to the target conversation analysis task.

[0009] According to a sixth aspect of the embodiments of this specification, there is provided a data processing device for a conversation analysis task, comprising: A first determining module is configured to determine execution trajectory data of a target conversation analysis task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target conversation analysis task based on a first model; a first generating module configured to generate, using the first agent, an intelligent component library according to the execution trajectory data and system constraints, wherein the system constraints are determined according to a model configuration of a second model that the second agent depends on, and the model configuration of the first model is greater than the model configuration of the second model; a second generating module configured to generate a conversation analysis reference workflow using the first agent according to the execution trajectory data and the intelligent component library; The second determination module is configured to utilize the second intelligent agent to perform the conversation analysis test task according to the conversation analysis reference workflow and the intelligent component library, and determine the conversation analysis target workflow set according to the task execution result, wherein the conversation analysis test task is determined according to the target conversation analysis task.

[0010] According to a seventh aspect of the embodiments of this specification, a computing device is provided, including: memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above-mentioned data processing method, task processing method or data processing method applied to dialogue analysis tasks are implemented.

[0011] According to an eighth aspect of the embodiments of this specification, an electronic device is provided, including: a memory and a processor, wherein the memory and the processor are connected via a bus; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above-mentioned data processing method, task processing method or data processing method applied to dialogue analysis tasks are implemented.

[0012] According to the ninth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned data processing method, task processing method, or data processing method applied to a conversation analysis task.

[0013] According to the tenth aspect of the embodiments of this specification, a computer program product is provided, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned data processing method, task processing method, or data processing method applied to a conversation analysis task.

[0014] One embodiment of the present specification implements a data processing method, including: determining execution trajectory data of a target task, wherein the execution trajectory data is trajectory data generated by a first agent relying on a first model to execute the target task; using the first agent, generating an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on the model configuration of a second model that a second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; using the first agent, generating a reference workflow based on the execution trajectory data and the intelligent component library; using the second agent, executing a test task based on the reference workflow and the intelligent component library, and determining a target workflow set based on the task execution result, wherein the test task is determined based on the target task. This data processing method collects execution trajectory data of a first agent that relies on a first model (i.e., a strong model) to perform a target task. Using the strong model's first agent, the method generates a weak model-adapted intelligent component library based on this execution trajectory data and system constraints determined by the model configuration of a second model (i.e., a weak model) that a second agent relies on. The library contains independently reusable intelligent components that are adapted to the weak model, allowing the strong model's intelligent components to be directly reused by the weak model. This allows the strong model's task execution capabilities to be reused without additional training, ensuring that the second agent that relies on the weak model can still perform complex deep analysis tasks with high quality at low cost. Furthermore, the strong model's first agent generates a reference workflow based on this execution trajectory data and the intelligent component library, achieving standardized task execution paths. This allows subsequent weak model-based second agents to efficiently execute test tasks based on this reference workflow, eliminating dynamic search redundancy caused by the strong model's first agent's flexibility, reducing inference time, and improving efficiency and stability. Finally, the second agent of the weak model is used to execute the test task according to the reference workflow and the intelligent component library. The final target workflow set is precipitated according to the task execution results. Through the closed-loop test verification mechanism of the second agent of the weak model, the boundary risks are intercepted to determine the output consistency, thereby ensuring the improvement of the stability of the second agent of the weak model in task execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of an application scenario of a data processing method provided by an embodiment of this specification; Figure 2 is a flow chart of a data processing method provided by one embodiment of this specification; Figure 3 This is a flowchart of a data processing method provided by one embodiment of this specification; Figure 4 This is a schematic diagram of the structure of a data processing device provided by one embodiment of this specification; Figure 5 This is a structural block diagram of a computing device provided by one embodiment of this specification; Figure 6 This is a structural block diagram of an electronic device provided by an embodiment of this specification. DETAILED DESCRIPTION

[0016] The following description sets forth many specific details to facilitate a thorough understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0017] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a," "the," and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0018] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0019] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0020] In one or more embodiments of this specification, a large model refers to a deep learning model with large-scale model parameters, typically containing hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than ten trillion model parameters. A large model can also be called a foundation model. It is pre-trained on a large amount of unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks and has good generalization capabilities, such as a large language model (LLM) and a multi-modal pre-training model.

[0021] When large models are used in practice, only a small number of samples are needed to fine-tune the pre-trained model and it can be applied to different tasks. Large models can be widely used in natural language processing (NLP), computer vision and other fields. Specifically, they can be applied to computer vision tasks such as visual question answering (VQA), image description (IC, Image Caption), and image generation, as well as natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. The main application scenarios of large models include digital assistants, intelligent robots, search, online education, office software, e-commerce, and intelligent design.

[0022] First, the terms involved in one or more embodiments of this specification are explained.

[0023] MCP: ModelContextProtocol, Model Context Protocol, is an open standard protocol in the field of large models, designed to solve compatibility and interoperability issues between models and external tools / data sources.

[0024] Agent: In the fields of computer science and information technology, especially in artificial intelligence and software engineering, an agent generally refers to an entity or program that can perceive the environment, understand input information, and make decisions and actions according to preset goals and rules.

[0025] LLMAgent: LargeLanguageModel-basedAgent, Agent based on large model.

[0026] Strong and weak models: Strong models are typically more complex, capable of capturing subtle patterns in the data, have a larger hypothesis space, can represent relatively complex functional relationships, and have strong expressive and reasoning capabilities. Weak models are relatively simple and generally can only capture relatively obvious or basic patterns in the data. Compared with strong models, weak models have a smaller hypothesis space and can only represent relatively simple functional relationships. Their expressive and reasoning capabilities are weaker than those of strong models.

[0027] Token: The meaning of a token varies depending on the specific application scenario. For example, in text processing and natural language processing, tokenization is the process of segmenting text into words, punctuation marks, or other meaningful fragments. Each such fragment is a "token." For example, the sentence "I like cats" can be segmented into three tokens: "I," "like," and "cats."

[0028] In order to solve the above technical problems, a data processing method is provided in this specification. This specification also involves a data processing device, a computing device, an electronic device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0029] Considering the huge number of model parameters of large models and the limited computing resources of mobile terminals, the data processing method provided in the embodiment of the present application can be applied to Figure 1 The application scenarios shown are not limited to these. Figure 1 , Figure 1 This is a schematic diagram of an application scenario of a data processing method provided by an embodiment of this specification. Figure 1 In the illustrated application scenario, the large model is deployed on a server 10. Server 10 can be connected to one or more client devices 20 via a local area network (LAN), a wide area network (WAN), the Internet, or other types of data networks. Client devices 20 herein include, but are not limited to, smartphones, tablet computers, laptops, PDAs, personal computers, smart home devices, and in-vehicle devices. Client devices 20 can interact with users via a graphical user interface (GUI) to access the large model and implement the methods provided in the embodiments of this specification.

[0030] In the embodiments of this specification, the system composed of the client device and the server can perform the following steps: The client device performs the following steps: A target workflow set generation task sent by a user through a graphical user interface of a client device is received, and the target workflow set generation task is sent to a server.

[0031] The server executes the data processing method provided in one embodiment of this specification to complete the task of generating a target workflow set. The specific implementation steps are as follows: Determining execution trajectory data of a target task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target task based on a first model; generating, using the first agent, an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; Using the first agent, generating a reference workflow according to the execution trajectory data and the intelligent component library; The second agent is utilized to execute a test task according to the reference workflow and the intelligent component library, and a target workflow set is determined according to the task execution result, wherein the test task is determined according to the target task.

[0032] Specifically, the first agent that relies on the first model can be understood as an agent that relies on a strong model; the second agent that relies on the second model can be understood as an agent that relies on a weak model.

[0033] It should be noted that, when the operating resources of the client device can meet the deployment and operating conditions of the large model, the embodiments of the present application can be carried out in the client device.

[0034] One or more embodiments of this specification provide a data processing method that uses execution trajectory data of a strong model agent in a target task (such as a complex deep analysis task) combined with system constraints determined based on the model configuration of a weak model to generate an intelligent component library containing independent, reusable intelligent components that are compatible with the weak model agent. This allows the strong model agent to automatically generate a reference workflow corresponding to the target task based on the execution trajectory data and the intelligent components in the intelligent component library. At the same time, a closed-loop testing and verification mechanism based on the weak model agent precipitates a final target workflow set that includes the target workflow. This allows the weak model agent to subsequently complete tasks that are the same or similar to the target task efficiently and stably according to the corresponding target workflow, effectively reducing the cost of using large models and improving the stability and execution efficiency of the weak model-based agent.

[0035] See also Figure 2 , Figure 2 A flow chart of a data processing method provided according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0036] Step 202: Determine the execution trajectory data of the target task.

[0037] The execution trajectory data is trajectory data generated when the first agent relies on the first model to execute the target task.

[0038] The data processing methods provided in one or more embodiments of this specification can be applied to various in-depth analysis application scenarios, such as customer service and sales, such as customer service satisfaction analysis and customer churn analysis. Of course, this data processing method is not only applicable to customer service and sales, but can also be applied to finance and insurance (risk analysis scenarios), healthcare (treatment plan analysis scenarios), and other fields with in-depth analysis requirements. For ease of understanding, the data processing methods provided in one or more embodiments of this specification are described in detail using the in-depth analysis application scenario in the customer service field as an example.

[0039] Specifically, the target task varies depending on the application scenario of deep analysis. For example, if the deep analysis application scenario is customer service, the target task may be a customer service satisfaction analysis task; if the deep analysis application scenario is finance and insurance, the target task may be a risk analysis task for abnormal transactions, etc. For ease of understanding, one or more embodiments of this specification are described using the customer service satisfaction analysis task as an example.

[0040] Among them, the first model can be understood as a large model, such as the strong model mentioned above, and the first agent that relies on the first model can be understood as an agent of the strong model. Then, in the case where the first agent that relies on the first model is an agent of the strong model, the execution trajectory data of the target task can be understood as the complete trajectory data of the target task collected when the agent of the strong model completes the execution of the target task in the target task set. In specific implementation, the target task set will include multiple target tasks, for example, multiple target tasks include but are not limited to satisfaction analysis tasks, customer churn analysis tasks, etc. In one or more embodiments of this specification, only the target task is introduced as an example of the satisfaction analysis task. For the processing of other target tasks in the target task set, please refer to the target task, which will not be repeated here.

[0041] In practical applications, there are two ways to collect execution trajectory data for a target task: one is to use a strong model agent to actually execute the target task, and in real time, record each step it takes and the underlying execution details of each step during the execution of the target task by the strong model agent. Another approach is to obtain the target task's history log after confirming that the strong model agent has completed the task, and use the history log to replace the re-execution. From the history log, obtain the target task's history log, and in real time record each step it takes and the underlying execution details of each step during the execution of the target task by the strong model agent. The specific implementation methods are as follows: The determination of the execution trajectory data of the target task includes: Determining a target task set, wherein the target task is included in the target task set; Utilizing the first agent to execute the target task and obtain execution trajectory data of the target task; or Obtain execution trajectory data of the target task from the historical execution log of the first agent executing the target task.

[0042] Specifically, first determine the target task set. This target task set can include multiple target tasks, each covering the typical difficulties and diversity of the target application scenario. For example, in the customer service field, multiple target tasks could include satisfaction analysis, customer churn analysis, and customer complaint analysis. In practical applications, each target task can be traversed, executed by the strong model agent, and the execution trajectory data of each target task can be obtained. Alternatively, the execution trajectory data of each target task can be obtained from the historical execution log of each target task executed by the strong model agent.

[0043] Assume that one of the target tasks in the defined target task set is "Based on given data (satisfaction analysis.xlsx) and format requirements (report_format_requirements.txt), write a satisfaction analysis report that meets specific format requirements."

[0044] The target task is performed by the strong model agent. During the execution of the target task, the thinking and operation process of the strong model agent at each step is fully and detailedly recorded. This record can be understood as the execution trajectory data of the target task. The execution trajectory data of the target task includes the execution process (Traj0) of the strong model agent when actually performing the target task, as well as the decisions and operations made by the strong model agent when actually executing each step in the execution process. In other words, it includes the triple information of the strong model agent's internal detailed reasoning process (reasoning, rj), tool call (action, aj), and tool feedback (observation, oj) when executing each step.

[0045] For example, the execution process (Traj0) includes the following steps: (1) View and extract the data in the "Satisfaction Analysis.xlsx" file, including conversation ID, conversation content, project scenario, satisfaction, date ID, and flower name, and provide a data overview.

[0046] (2) Review and extract the report format requirements in the "report_format_requirements.txt" file to ensure that you understand the specific structure and content requirements of the report.

[0047] (3) Analyze the satisfaction data, calculate the overall satisfaction rate and the daily satisfaction rate change trend, and generate the corresponding trend chart.

[0048] (4) Identify and analyze project scenarios that require special attention, including calculating the satisfaction rate, dissatisfaction rate, and contribution of each project scenario, and generating detailed indicator data tables.

[0049] (5) Conduct in-depth analysis of key project scenarios, analyze the reasons for customer dissatisfaction (product dissatisfaction, service dissatisfaction), customer service solution types (unsatisfied, delayed gratification, satisfied), customer service satisfaction rates, and review invitations.

[0050] (6) Based on the previous analysis results and the format requirements in report_format_requirements.txt, write a complete satisfaction analysis report, including the achievement of satisfaction rate targets, key project scenarios and their in-depth interpretation.

[0051] Then, for step (3), analyze the satisfaction data, calculate the overall satisfaction rate and the daily satisfaction rate change trend, and generate the corresponding trend chart. The corresponding triplet information may include the following: Reasoning process: "The user requires to calculate the overall satisfaction rate. I need to read the 'Satisfaction Analysis.xlsx' file first. The file path is known to be '. / data / Satisfaction Analysis.xlsx'. I will use the aa tool to read it.

[0052] Tool call: call_tool(name="aa", arguments={"file_path": ". / data / Satisfaction Analysis.xlsx"}).

[0053] Tool feedback: {"status": "success", "content": {"columns": ["Dialogue ID", "Dialogue Content", "Project Scenario", "Satisfaction", "Date ID", "Flower Name"], "data": [[1, "Dialogue Text 1...", "Scenario A", "Satisfied", 20240101, "Customer Service 1"], [2, "Dialogue Text 2...", "Scenario B", "Unsatisfied", 20240101, "Customer Service 2"], ... ]}}.

[0054] The data processing method provided in one or more embodiments of this specification can be implemented in at least two ways to collect the execution trajectory data of the target task. One method is to use a strong model agent to re-execute the target task to collect the execution trajectory data of the target task. When re-executing the target task, the strong model agent can generate appropriate execution trajectory data reflecting the current environment and model state, and may explore different execution trajectory data for the target task based on changes in the model or environment. Another method is to obtain the execution trajectory data of the target task from the historical execution log if the strong model agent has already completed the target task. This method eliminates the need for additional computing resources and time costs to re-execute the target task, improves the collection time of the execution trajectory data of the target task, and enhances the overall execution efficiency of the data processing method.

[0055] Step 204: Utilize the first intelligent agent to generate an intelligent component library according to the execution trajectory data and system constraints.

[0056] The system constraint is determined according to the model configuration of the second model on which the second agent depends, and the model configuration of the first model is greater than the model configuration of the second model.

[0057] Specifically, after using the strong model agent to collect the target task's execution trajectory data, the strong model agent can then be used to construct a smart component library containing multiple smart components based on the target task's execution trajectory data and system constraints. The purpose of this step is to transform the collected target task's execution trajectory data into multiple modular, reusable smart components that can be understood and executed by the weak model, within the constraints of the system, to generate the smart component library.

[0058] In specific implementations, the second model can also be understood as a large model, such as the weak model mentioned above. In this case, the second agent that relies on the second model can be understood as the agent of the weak model. System constraints are determined based on the model configuration of the weak model. For example, system constraints can be defined based on the maximum context length, maximum output token limit, available tool list, or memory limit of the weak model. If the model configuration of the first model is greater than that of the second model, it can be understood that the model configuration of the strong model is greater than that of the weak model. The model configuration here includes but is not limited to model parameters, maximum context length of processing, maximum output token limit, etc.

[0059] In practical applications, the smart component library can include multiple smart components. Each smart component can be understood as an MCP, that is, an independent functional module that encapsulates specific logic or operations. It encapsulates the complex reasoning and operation capabilities demonstrated by strong model agents in execution trajectory data into standardized components that weak model agents can understand and reliably execute. In other words, for each execution trajectory data, each subtask (such as specific tool calls, information extraction, etc.) is logically encapsulated as an independent MCP.

[0060] Specifically, the generation of the smart component library includes two implementation steps: first, using the strong model agent to generate a smart component set, then using the strong model agent to structure the multiple smart components in the smart component set, and then generating the smart component library based on the multiple structured smart components. The specific implementation method is as follows: The step of generating an intelligent component library using the first intelligent agent according to the execution trajectory data and system constraints includes: Using the first intelligent agent, generating an intelligent component set according to the execution trajectory data and system constraints, wherein the intelligent component set includes multiple intelligent components; The plurality of intelligent components are subjected to structured processing by utilizing the first intelligent agent to obtain the processed plurality of intelligent components, and the intelligent component library is generated based on the processed plurality of intelligent components.

[0061] In specific implementation, the generation of the intelligent component library is achieved through two steps. First, the strong model agent is used to design the generation process and overall control of the intelligent component based on the execution trajectory data of the target task and system constraints; multiple intelligent components are generated according to the generation process and overall control of the intelligent component; second, the strong model agent is used to perform structured processing on the generated multiple intelligent components, and the processed multiple intelligent components are placed in the intelligent component library to realize the generation of the intelligent component library.

[0062] Of course, in a fixed single scenario (such as the same report every day), you can also use the strong model agent to design the generation process and overall control of the intelligent component based on the execution trajectory data of the target task and system constraints; after generating multiple intelligent components according to the generation process and overall control of the intelligent component, no longer structure the multiple intelligent components, and directly put the multiple intelligent components into the intelligent component library to realize the generation of the intelligent component library.

[0063] The data processing method provided in one or more embodiments of this specification uses a strong model agent to generate a smart component set based on the execution trajectory data of the target task and system constraints, and then uses the strong model agent to structure the multiple smart components in the smart component set, and generates a smart component library based on the multiple smart components after structured processing. In this way, a scattered smart component set that may contain redundant or specific task details is refined and integrated into a universal, parameterized smart component library, so that the smart component library can be flexibly applied to different tasks in the same field (such as generating monthly report tasks, satisfaction report tasks for different products, etc.), and is not limited to generating target tasks (such as writing a satisfaction analysis report that meets specific format requirements based on given data (satisfaction analysis.xlsx) and format requirements (report_format_requirements.txt).

[0064] The generation of smart component sets is also achieved through two execution steps, as described below: The step of utilizing the first intelligent agent to generate an intelligent component set according to the execution trajectory data and system constraints includes: generating, using the first agent, a plurality of intelligent component generation tasks based on the execution trajectory data and system constraints; The first intelligent agent is utilized to generate tasks according to the plurality of intelligent components, and a plurality of intelligent components are constructed to generate the intelligent component set.

[0065] Among them, each intelligent component generation task can be understood as using the strong model agent to generate a detailed plan for the intelligent component for the next strong model agent based on the execution trajectory data of the target task and system constraints. It at least includes the analysis of the execution trajectory data of the target task, the identification of logically independent or reusable sub-task units that appear repeatedly in the trajectory, the input and output interface specifications defined for each intelligent component, and the generation strategy and constraints of each intelligent component to ensure that each generated intelligent component meets the system constraints.

[0066] Specifically, the planner first utilizes a strong model agent to design the smart component generation process (i.e., how to analyze execution trajectory data and identify encapsulated subtasks), overall control (e.g., guiding the smart component generator, receiving environmental feedback, and dynamically adjusting the generation strategy based on feedback), and constraint consideration (e.g., ensuring that the generated smart components and the workflows they form can effectively operate within system constraints) based on the target task's execution trajectory data and system constraints (e.g., the maximum context length of the weak model, the maximum output token limit, the list of available tools, and memory limits). Furthermore, the planner generates multiple executable smart component generation plans, i.e., multiple smart component generation tasks, based on the execution trajectory data of the target task and system constraints (e.g., the maximum context length of the weak model, the maximum output token limit, the list of available tools, and memory limits). The planner then utilizes the strong model agent to generate smart components corresponding to the tasks and places them into a smart component collection, thereby achieving the generation of the smart component collection.

[0067] Continuing with the above example, using the first agent, based on the execution trajectory data and system constraints, multiple intelligent component generation tasks may be generated, including but not limited to: 1. Obtain MCP of overall satisfaction rate; 2. Obtain the MCP of daily dimension satisfaction; 3. Calculate the MCP of each project scenario satisfaction data indicator; 4. Identify the MCPs that require special attention. 5. MCP who provides in-depth interpretation of key project scenarios; 6. When writing MCP reports, it is necessary to consider the output length limit of weak models.

[0068] Of course, in actual applications, each smart component generation task contains more than just the information in the example. For details, please refer to the above detailed explanation of the smart component generation task. This is just an illustrative example.

[0069] The data processing method provided in one or more embodiments of this specification first utilizes a strong model agent to generate multiple smart component generation tasks based on execution trajectory data and system constraints; then utilizes the strong model agent to construct multiple smart components based on the multiple smart component generation tasks to generate a smart component set. By splitting the generation of the smart component set into two implementation steps, each step is implemented by a different module, such as implementing the generation of the smart component set through the division of labor and collaboration between a planner and a smart component generator, the planner can focus on the reusability design of the smart components, while the smart component generator can focus on the robustness implementation of the smart components, thereby generating a highly reusable and robust smart component set.

[0070] In addition, to ensure the availability and stability of the generated multiple smart components, when using a strong model agent to generate multiple smart components based on multiple smart component generation tasks, the corresponding component script can be generated according to each smart component generation task, and then each component script can be input into the virtual environment for execution, and the smart component can be constructed based on the script execution results. The specific implementation method is as follows: The step of utilizing the first intelligent agent to generate tasks according to the plurality of intelligent components and constructing the plurality of intelligent components includes: Using the first agent, based on each component generation task in the plurality of intelligent component generation tasks, generate a component script corresponding to each component generation task; The component scripts corresponding to the component generation tasks are input into the virtual environment for execution, the script execution results of the component scripts are obtained, and multiple smart components are constructed according to the script execution results of the component scripts.

[0071] Specifically, for each of the multiple smart component generation tasks, a strong model agent is used to analyze the reasoning process and tool calls, extract the tool call pattern and data processing logic, and then generate a basic component script corresponding to each smart component generation task. After generating the basic component script corresponding to each smart component generation task, a virtual environment is created, and the basic component script corresponding to each smart component generation task is input into the virtual environment for verification. The script execution results of the basic component script corresponding to each smart component generation task are obtained, and multiple smart components are constructed based on the script execution results. Furthermore, following the above example, the specific implementation of this step can be executed by the smart component generator.

[0072] The data processing method provided in one or more embodiments of this specification utilizes a strong model agent to generate tasks based on multiple intelligent components and construct multiple intelligent components. The method splits the generation process of each intelligent component into two steps: first generating the component script, then entering it into a virtual environment for execution, and then constructing the intelligent component based on the execution results. During the component script generation phase, the core logic can be centrally implemented, improving the code quality and reusability of the component script. Secondly, testing the component script in an isolated virtual environment can effectively verify the feasibility and robustness of the component script, identify potential problems in advance, and reduce the risk of subsequent use. This separation of development and verification facilitates the debugging and optimization of intelligent components, improving the controllability and maintainability of the overall intelligent component construction.

[0073] Specifically, the specific implementation method of executing the component script in the virtual environment and constructing multiple smart components according to the script execution results of the component script is as follows: The component scripts corresponding to the component generation tasks are input into the virtual environment for execution, the script execution results of the component scripts are obtained, and multiple smart components are constructed according to the script execution results of the component scripts, including: Inputting a target component script corresponding to a target component generation task into a virtual environment for execution, and obtaining a script execution result of the target component script, wherein the target component generation task is any one of the multiple smart component generation tasks; If it is determined that the target component script is successfully executed according to the script execution result of the target component script, encapsulating the target component script to construct a target smart component, wherein the target smart component is any one of the multiple smart components; When it is determined that the execution of the target component script fails based on the script execution result of the target component script, the target component script is updated using the script execution result of the target component script, and the step of inputting the target component script corresponding to the target component generation task into the virtual environment for execution to obtain the script execution result of the target component script is continued.

[0074] During implementation, each component script generated for a component generation task is entered into a virtual environment for execution, and its execution result is obtained. Each component script entered into the virtual environment can be considered a target component script. If the target component script's execution result confirms successful execution, the target component script is encapsulated to generate a target smart component, which can be considered any of the multiple smart components subsequently constructed.

[0075] If the target component script execution result indicates that the target component script has failed, the cause of the failure needs to be analyzed based on the script execution result, and the error classification (such as logical errors, dependency issues, performance limits, etc.) is performed. The target component script is then updated based on the error (such as by correcting the script using a strong model agent). After the update is successful, the updated target component script is executed again in the virtual environment until the updated target component script is successfully executed in the virtual environment. Of course, in rare cases, after a target component script has been iteratively executed multiple times, in order to ensure the stability of the production environment, the target component script that meets the number of iterations can be encapsulated and built into a target smart component.

[0076] The data processing method provided in one or more embodiments of the present specification, in order to ensure the reliability of the constructed intelligent components, after generating the component script corresponding to each component generation task, it is necessary to input the component script into the virtual environment for execution. If the script execution result of the component script is successful, it is packaged to realize the construction of the intelligent component. If the script execution result of the component script fails, the script is corrected by using the strong model agent and then re-executed in the virtual environment. Through such verification and multiple iterative corrections, until the corrected component script can be successfully executed, it is packaged to realize the construction of the intelligent component, thereby ensuring the quality of the constructed intelligent component and improving the availability and reliability of the intelligent component.

[0077] After generating an intelligent component set based on multiple intelligent components, in order to ensure the universality and flexibility of the intelligent component set, it is necessary to structure the multiple intelligent components in the intelligent component set based on the intelligent component set, thereby obtaining multiple intelligent components after structured processing, and then generate a highly reusable, universal, and easy-to-manage intelligent component library based on the multiple intelligent components after structured processing. The specific implementation method is as follows: The step of utilizing the first intelligent agent to perform structured processing on the plurality of intelligent components to obtain the processed plurality of intelligent components includes: Using the first intelligent agent, abstracting the plurality of intelligent components to obtain a plurality of intelligent components after abstraction processing; Using the first intelligent agent, clustering the plurality of intelligent components after the abstraction process to obtain a plurality of intelligent components after the clustering process; The first intelligent agent is used to merge the plurality of intelligent components after the clustering process to obtain the plurality of intelligent components after the merge process, thereby obtaining the plurality of intelligent components after the process process.

[0078] Specifically, the first intelligent agent is used to abstract the multiple intelligent components to obtain multiple intelligent components after abstraction. This can be understood as using a strong model agent to analyze the specific implementation code of each intelligent component, parameterize each intelligent component, remove task-specific dependencies, retain general strategies, improve reusability and other abstract processing to obtain multiple intelligent components after abstraction, so that each intelligent component becomes a callable module with clearly defined parameters.

[0079] Continuing with the above example, the smart component MCP1 (calculating the overall satisfaction rate): its underlying implementation code may directly fix the file path ". / data / Satisfaction Analysis.xlsx".

[0080] Abstracted MCP1: The strong model LLM rewrites it, replacing the hard-coded file path with an input parameter. This allows MCP1 to calculate the overall satisfaction rate for any specified file. Specifically, the underlying code implementation of each intelligent component is extracted and the LLM is used to rewrite the implementation into a parameterized, reusable form, removing task-specific dependencies and retaining only the general strategy.

[0081] Using the first intelligent agent, clustering is performed on the multiple intelligent components after the abstraction process to obtain multiple intelligent components after clustering process. This can be understood as using the strong model agent to analyze the core functions of each intelligent component after the abstraction process, and grouping the multiple intelligent components after the abstraction process according to the similarity or correlation of the functions. For example, the strong model agent identifies the intelligent components with the same or relatively similar functions or those that are often used together among the multiple intelligent components after the abstraction process, and clusters these intelligent components together to prepare for the next step of merging. For example, if there are two intelligent components after the abstraction process that both calculate the satisfaction rate, they will be clustered together. That is, all the abstracted MCPs are clustered by function to obtain a group set.

[0082] Using the first intelligent agent, the multiple intelligent components after the clustering process are merged to obtain multiple intelligent components after the merger process. This can be understood as using the strong model agent to analyze all the intelligent components in each group and merge multiple similar intelligent components in the group into a single intelligent component with a universal interface. For intelligent components that need to be merged, such as two intelligent components whose core functions are both to calculate the satisfaction rate, the file parameters cannot be simply merged, because the two intelligent components may have different calculation dimensions in addition to the file path. In this case, the strong model needs to fuse and expand the parameters when merging the intelligent components, not only to cover the functions of the two intelligent components, but also to enable the merged intelligent component to have the expansion capability of calculating the satisfaction rate according to any specified dimension. That is, for each group of MCPs after clustering, the MCPs with similar kinetic energy are merged into a parameterized universal interface to generate the final intelligent component library.

[0083] The data processing method provided in one or more embodiments of this specification utilizes a strong-model agent to perform abstraction, merging, clustering, and other processing on multiple intelligent components in sequence, converting and merging them into more powerful and configurable general intelligent components; by merging intelligent components with similar functions, the number of intelligent components in the subsequent intelligent component library is reduced, making the intelligent component library easier to manage and use; through parametric design, an intelligent component can be adapted to a variety of different specific needs, thereby improving the reusability and flexibility of the intelligent component. The weak-model agent does not need to call multiple intelligent components to complete a logical function, thereby reducing the complexity of task execution and the probability of error. Subsequently, based on the intelligent components in the intelligent component library that have undergone structured processing, a workflow for solving complex tasks can be more efficiently and clearly combined.

[0084] Step 206: Utilize the first agent to generate a reference workflow based on the execution trajectory data and the intelligent component library.

[0085] Specifically, after building the smart component library, we can use the strong model agent to generate a reference workflow for the target task based on the execution trajectory data and the smart component library. The specific implementation method is as follows: The step of utilizing the first intelligent agent to generate a reference workflow according to the execution trajectory data and the intelligent component library includes: The execution trajectory data is analyzed by using the first agent, and a reference workflow corresponding to the target task is generated according to the analysis results of the execution trajectory data and the component capabilities of the smart components in the smart component library.

[0086] During implementation, the execution trajectory data of the target task and the smart component library generated in the above steps are input into the strong model agent. The strong model agent will analyze each step of the execution process in the execution trajectory data and generate a corresponding reference workflow for the target task based on the capabilities of each smart component in the smart component library.

[0087] Continuing with the above example, the strong model agent generates a reference workflow based on the execution trajectory data of the target task and the intelligent component library generated in the above steps. 1. Obtain overall satisfaction rate; 2. Obtain the daily dimension satisfaction rate change trend; 3. Calculate the satisfaction data indicators for each project scenario; 4. Identify scenarios that require focus; 5. Provide in-depth interpretation of key project scenarios; 6. Report writing.

[0088] Of course, the reference workflow during specific implementation will also include the calling information of the smart components in the smart component library in each step, etc. The above reference workflow is only an illustrative example.

[0089] One or more embodiments of this specification provide a data processing method that utilizes a strong-model agent to analyze the execution trajectory data of a target task. Combined with the capabilities of intelligent components in an intelligent component library, the method converts the unstructured reasoning chain of the strong-model agent for the target task into a structured sequence of operational instructions (i.e., a reference workflow) that can be implemented by a weak-model agent. This method converts the dynamic search capability of the strong-model agent into a standardized solution that can be executed by the weak-model agent. This allows the reference workflow for the target task to achieve the scale benefits of the weak-model agent while retaining the breakthrough capabilities of the strong-model agent.

[0090] Step 208: Utilize the second agent to execute the test task according to the reference workflow and the intelligent component library, and determine the target workflow set according to the task execution result.

[0091] Wherein, the test task is determined according to the target task.

[0092] During implementation, the test task can be the target task or a test task generated by sampling the target task. To test the robustness of the reference workflow and intelligent component library under different data distributions and task requirements, expose potential boundary conditions, missing error handling, model understanding biases, and other issues, and provide a rich source of feedback data for closed-loop optimization, the target task can be sampled in a diversified manner to generate multiple variants of smaller-scale test tasks. By simulating the diversity and boundary conditions of real-world scenarios, a better set of target workflows can be identified. The specific implementation is as follows: The utilizing the second agent to execute the test task according to the reference workflow and the intelligent component library, and determining the target workflow set according to the task execution result, includes: Sampling the target task to generate the test task; Inputting the reference workflow and the test task into the second agent, obtaining a task execution result of the second agent executing the test task by calling the smart component in the smart component library based on the reference workflow; A target workflow set is determined according to the task execution result and the reference workflow.

[0093] Sampling the target task includes, but is not limited to, random sampling of the original data file corresponding to the target task, such as sampling only 50% of the data in the original data file; performing time slicing, such as analyzing only a certain week's data; constructing specific scenarios, such as including only data from a certain project scenario, facilitating data loss / abnormality, etc.; and modifying the requirements of the target task, such as requiring the report to include specific indicators. Multiple test tasks are generated through diversified sampling of the target task.

[0094] For each test task, the weak model agent is fed with the test task, its corresponding data, and a reference workflow. The weak model agent then combines components from the intelligent component library to execute each test task and generate a task execution result. Based on each test task's task execution result and the reference workflow, a suitable target workflow is identified and placed into a target workflow set to generate a target workflow set.

[0095] The data processing method provided in one or more embodiments of this specification generates a variety of test tasks, a reference workflow, executes each test task using a strong-model agent in conjunction with an intelligent component library, and generates a target workflow set based on the execution results of each test task and the reference workflow. This method, through repeated trial and error and optimization on a variety of test tasks, ensures that the resulting target workflow set and intelligent component library can cover a variety of real-world scenarios and can subsequently be stably and reliably executed by a weak-model agent.

[0096] In addition, when the weak model agent can stably and successfully execute the corresponding reference workflow or optimized reference workflow on all test tasks, the reference workflow or optimized reference workflow can be considered mature and can be added to the target workflow set for subsequent task execution. The specific implementation method is as follows: The determining of a target workflow set according to the task execution result and the reference workflow includes: If it is determined that the test task is successfully executed according to the task execution result, the reference workflow is determined as the target workflow of the target task; If it is determined that the test task has failed to execute according to the task execution result, the smart component library and / or the reference workflow are updated, and the steps of inputting the reference workflow and the test task into the second agent are continued, and obtaining the task execution result of the second agent calling the smart components in the smart component library based on the reference workflow to execute the test task are obtained; The target workflow set is determined according to the target workflow.

[0097] In practical applications, each target task can be sampled as multiple test tasks. When each test task is executed, the corresponding reference workflow can be understood as a reference workflow generated by the agent using a strong model based on the execution trajectory data of the corresponding target task and the intelligent component library.

[0098] Specifically, if the test task execution results determine that the test task was successfully executed, the reference workflow can be determined as the target workflow of the target task. However, in actual implementation, a target task may sample multiple test tasks. There are two possibilities for determining the reference workflow corresponding to the target task as the target workflow of the target task. One is that the strong model agent can determine the reference workflow corresponding to the target task as the target workflow of the target task if the reference workflow is successfully executed on all test tasks corresponding to the target task. The other is that the strong model agent can determine the reference workflow corresponding to the target task as the target workflow of the target task if the reference workflow is successfully executed on more than 80% of the test tasks corresponding to the target task.

[0099] If the task execution results indicate that the test task has failed, the cause of the failure needs to be analyzed and the smart component library updated accordingly (e.g., MCP does not consider empty data, non-existent files, field name changes (e.g., "date ID" becomes "date"), numerical overflow, etc.). Modify the internal logic of the MCP based on the specific error, adding input validation, error handling branches, default values, and compatibility processing; or, if the weak model has difficulty understanding the complex parameters or internal logic of an MCP, simplify the MCP or split it into smaller MCPs) and / or reference workflows (e.g., modify the order of steps or dependencies: if it is found that some steps require unmet prerequisites, add necessary steps: if it is found that data cleaning is often required, add a "data pre-cleaning" step to the workflow and map it to the corresponding smart component library, etc.). Then, continue executing the test steps according to the updated smart component library and / or reference workflow until the two aforementioned successful execution conditions are met.

[0100] The data processing method provided in one or more embodiments of this specification uses the success or failure of test task execution results, combined with reference workflows, to precipitate a low-cost target workflow for each target task, which can be stably executed by a weak-model agent, either directly or by adjusting the intelligent component library or reference workflow based on the task execution results. Subsequently, when executing tasks of the same type as the target task, the weak-model agent can directly execute the target workflow in the target workflow set with high quality and completeness, ensuring low-cost and highly stable operation.

[0101] Specifically, this data processing method collects execution trajectory data of a first agent that relies on a first model (i.e., a strong model) to perform a target task. Using the strong model's first agent, the method generates a weak model-adapted intelligent component library based on the execution trajectory data and system constraints determined based on the model configuration of a second model (i.e., a weak model) that the second agent relies on. Through the independently reusable intelligent components in the intelligent component library that are adapted to the weak model, the strong model's intelligent components can be directly reused by the weak model, allowing the strong model's task execution capabilities to be reused without additional training. This ensures that the second agent that relies on the weak model can still perform complex deep analysis tasks with high quality at low cost. At the same time, the strong model's first agent generates a reference workflow based on the execution trajectory data and the intelligent component library, achieving standardization of task execution paths. This allows subsequent weak model-based second agents to efficiently execute tasks based on the reference workflow, eliminating dynamic search redundancy caused by the strong model's flexibility, reducing inference time, and improving efficiency and stability. Finally, the second agent of the weak model is used to execute the test task according to the reference workflow and the intelligent component library. The final target workflow set is precipitated according to the task execution results. Through the closed-loop test verification mechanism of the second agent of the weak model, the boundary risks are intercepted to determine the output consistency, thereby ensuring the improvement of the stability of the second agent of the weak model in task execution.

[0102] In addition, after the target workflow set is built, when a new task arrives, the weak model agent can be used in conjunction with the matching target workflow in the target workflow set to execute the new task at low cost and high efficiency. The specific implementation method is as follows: A task processing method, comprising: Receiving a task to be executed, parsing the task to be executed using a second agent, and determining an execution workflow from a target workflow set based on the parsing result; The second agent is used to execute the task to be executed according to the execution workflow and the smart component library, wherein the target workflow set and the smart component library are generated according to the above-mentioned data processing method.

[0103] The pending task can be understood as a task of the same type as the target task. For example, if the target task is "Based on the given data (Satisfaction Analysis.xlsx) and format requirements (report_format_requirements.txt), write a satisfaction analysis report that meets specific format requirements," the pending task can also be a satisfaction analysis report task based on other given data and format requirements. The execution workflow can be understood as a target workflow included in the target workflow set that matches the pending task.

[0104] Specifically, upon receiving a pending task, the agent using the weak model first parses the pending task to obtain a parsed result, specifically determining the task description, the task data it carries, and the requirements. This parsed result is then matched against all target workflows in the target workflow set, typically based on parameters such as task type, input data structure, and output requirements. If a matching target workflow is identified in the target workflow set, that target workflow is designated as the execution workflow corresponding to the pending task. The agent using the weak model then executes the pending task based on the execution workflow and the smart components in the mounted smart component library.

[0105] In practical applications, generally, when the task to be executed has a high degree of match with a target workflow in the target workflow set, the target workflow can be directly selected as the execution workflow corresponding to the task to be executed. In another case, when the task to be executed does not match all target workflows in the target workflow set, the target workflow with a high degree of match can be fine-tuned. For example, if the core logic is the same and only the parameters are different, the parameter configuration of the target workflow with a high degree of match can be expanded or modified. If the task to be executed is a combination of multiple existing target tasks, a new target workflow can be constructed by combining the steps of multiple existing target workflows or calling smart components from multiple smart component libraries. However, if the construction of a new target workflow is triggered, it needs to be treated as a new target task. According to the specific implementation described above, the complete process of "trajectory collection, smart component library construction, reference workflow construction, and target workflow construction" must be executed.

[0106] One or more embodiments of the present specification provide a data processing method that, after receiving a task to be executed and determining the execution workflow corresponding to the task to be executed, inputs the task to be executed, the execution workflow, and the smart component library into a weak model agent. The weak model agent understands the structure of the target workflow corresponding to the task to be executed (such as the sequence of steps, dependencies, etc.). For each step in the execution workflow, the weak model agent determines the smart component in the smart component library that needs to be called, provides accurate input parameters for the determined smart component based on the execution workflow and the current state, executes the call of these smart components, receives the results of the execution of these smart components, and executes the task to be executed step by step according to the logic of the execution workflow. Therefore, during the execution process of the entire task to be executed, the weak model agent does not need to perform complex task planning and multi-step reasoning. It only needs to understand and execute the relatively simple instructions defined in each step of the execution workflow corresponding to the task to be executed (such as calling the smart component) and handle the data transfer between steps, thereby achieving low-cost and high-efficiency task execution.

[0107] At the same time, in order to achieve closed-loop optimization and continuous evolution, when using the weak model agent to execute the task to be executed, the target workflow in the smart component library and / or target workflow set will be tuned based on the task execution results of the task to be executed. The specific implementation method is as follows: After executing the task to be executed, the method further includes: When it is determined based on the execution result of the task to be executed that the task to be executed has failed, the smart component library and / or the execution workflow are updated.

[0108] Specifically, when using a weak model agent to execute a task to be executed using an execution workflow, the execution results of the task to be executed are collected, and its execution trajectory data can also be collected; if the execution of the task to be executed fails (unable to complete the execution, the result is biased or the efficiency is low, etc.), the smart components and / or execution workflow in the smart component library can be updated according to the above implementation by processing the task to be executed and collecting feedback.

[0109] The data processing method provided in one or more embodiments of this specification determines the cause of the execution failure by analyzing the execution results of the task to be executed, and updates the smart components and / or execution workflows in the smart component library based on the cause of the execution failure, so that the smart component library and the target workflow set continue to evolve, with a wider coverage and stronger robustness, thereby achieving closed-loop optimization and continuous evolution.

[0110] The following combined Figure 3, taking the data processing method provided in this specification as an example of an in-depth analysis application scenario in the customer service field, the data processing method is further explained. Figure 3 A flowchart of a data processing method provided in one embodiment of this specification is shown, which specifically includes the following steps.

[0111] Step 302: define a target task set, input each target task in the target task set into the strong model agent, use the strong model agent to execute each target task, and obtain execution trajectory data of each target task.

[0112] The target task set includes multiple target tasks of the same type, for example, all of which are satisfaction analysis generation tasks.

[0113] Step 304: Construct an MCP set.

[0114] Specifically, the specific implementation steps for constructing the MCP set are as follows: 1. Set system constraints based on the weak model agent used later, and input the system constraints and each target task into the planner. The planner uses the strong model agent to generate multiple MCP generation tasks based on the system constraints and each target task, and sends the multiple MCP generation tasks to the MCP generator.

[0115] 2. The MCP generator generates an MCP script corresponding to each MCP generation task according to each MCP generation task in the multiple MCP generation tasks, and then executes each MCP script in a virtual environment, and optimizes the MCP script according to the execution result to generate multiple MCPs.

[0116] 3. Construct an MCP set based on multiple MCPs.

[0117] Step 306: Use the strong model agent to abstract, cluster, and merge multiple MCPs in the MCP set, and construct the processed multiple MCPs into an MCP toolbox.

[0118] Step 308: Build a target workflow set.

[0119] Specifically, the specific implementation steps for building the target workflow set are as follows: 1. Sample each target task to obtain multiple test tasks.

[0120] 2. Using the strong model agent, a reference workflow is generated based on the execution trajectory data of each target task and the MCP toolbox.

[0121] 3. Using the strong model agent, according to the reference workflow of each target task and the MCP toolbox, the test task corresponding to each target task is executed. If the execution is successful, the reference workflow is deposited as the target workflow to construct the target workflow set.

[0122] Step 310: In case of execution failure, the MCP in the MCP toolbox is updated according to the execution result.

[0123] Specifically, in the event of an execution failure, the system will feed back the execution results to the planner, and the MCP and / or reference workflow in the MCP toolbox will be tuned and retested, and the reference workflow that is finally successfully executed will be stored in the target workflow set.

[0124] Step 312: Run online.

[0125] Specifically, the specific implementation steps of online operation are as follows: when there is a new task, the weak model agent can be used to match the new task with the target workflow in the target workflow set. After finding the matching target workflow, the weak model agent can refer to the target workflow to execute the new task based on the weak model agent and the MCP toolbox.

[0126] One or more embodiments of the present specification provide a data processing method that, first, automatically generates an independent, reusable MCP for a target task type using the execution trajectory data and system constraints of a strong model agent in a target task, eliminating the need for manually designing complex processes and significantly reducing development costs. Secondly, the MCP generated by the strong model agent can be directly used by the weak model agent, reusing the problem-solving capabilities of the strong model without additional training, ensuring that the weak model agent can still complete tasks with high quality while operating at a low cost, thus resolving the high dependence on model performance. At the same time, by consolidating the target workflow to support task requirements, the uncontrollable problems caused by the highly flexible strong model agent are avoided, effectively ensuring the stability and reliability of the weak model agent's execution of tasks based on the target workflow, further reducing costs. Specifically, the data processing method collects the execution trajectory data generated by the strong model agent when executing the target task, and automatically constructs an efficient and stable target workflow suitable for the weak model agent. This allows the weak model agent to subsequently execute the same type of target task based on the target workflow at low cost and high efficiency, greatly reducing inference costs and improving execution stability.

[0127] Corresponding to the above method embodiment, this specification also provides a data processing device embodiment, Figure 4 FIG1 shows a schematic diagram of the structure of a data processing device provided by an embodiment of this specification. Figure 4As shown, the device includes: a trajectory determination module 402 configured to determine execution trajectory data of a target task, wherein the execution trajectory data is generated by a first agent relying on a first model to execute the target task in a target task set; a component library generation module 404 configured to generate an intelligent component library using the first agent based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; A reference workflow generation module 406 is configured to generate a reference workflow using the first agent according to the execution trajectory data and the intelligent component library; The workflow set generation module 408 is configured to utilize the second agent to execute a test task according to the reference workflow and the intelligent component library, and determine a target workflow set according to the task execution result, wherein the test task is determined according to the target task.

[0128] Optionally, the trajectory determination module 402 is further configured to: Determining a target task set, wherein the target task is included in the target task set; Utilizing the first agent to execute the target task and obtain execution trajectory data of the target task; or Obtain execution trajectory data of the target task from the historical execution log of the first agent executing the target task.

[0129] Optionally, the component library generating module 404 is further configured to: Using the first intelligent agent, generating an intelligent component set according to the execution trajectory data and system constraints, wherein the intelligent component set includes multiple intelligent components; The plurality of intelligent components are subjected to structured processing by utilizing the first intelligent agent to obtain the processed plurality of intelligent components, and the intelligent component library is generated based on the processed plurality of intelligent components.

[0130] Optionally, the component library generating module 404 is further configured to: generating, using the first agent, a plurality of intelligent component generation tasks based on the execution trajectory data and system constraints; The first intelligent agent is used to generate tasks according to the plurality of intelligent components, to construct a plurality of intelligent components, and to generate the intelligent component set according to the plurality of intelligent components.

[0131] Optionally, the component library generating module 404 is further configured to: Using the first intelligent agent, abstracting the plurality of intelligent components to obtain a plurality of intelligent components after abstraction processing; Using the first intelligent agent, clustering the plurality of intelligent components after the abstraction process to obtain a plurality of intelligent components after the clustering process; The first intelligent agent is used to merge the plurality of intelligent components after the clustering process to obtain a plurality of intelligent components after the merger process.

[0132] Optionally, the reference workflow generation module 406 is further configured to: The execution trajectory data is analyzed by using the first intelligent agent, and a reference workflow corresponding to the target task is generated according to the component capabilities of the intelligent components in the intelligent component library. Optionally, the workflow set generation module 408 is further configured to: Sampling the target task to generate the test task; Inputting the reference workflow and the test task into the second agent, obtaining a task execution result of the second agent executing the test task by calling the smart component in the smart component library based on the reference workflow; A target workflow set is determined according to the task execution result and the reference workflow.

[0133] Optionally, the workflow set generation module 408 is further configured to: If it is determined that the test task is successfully executed according to the task execution result, the reference workflow is determined as the target workflow of the target task; If it is determined that the test task has failed to execute according to the task execution result, the smart component library and / or the reference workflow are updated, and the steps of inputting the reference workflow and the test task into the second agent are continued, and obtaining the task execution result of the second agent calling the smart components in the smart component library based on the reference workflow to execute the test task are obtained; The target workflow set is determined according to the target workflow.

[0134] Optionally, the component library generating module 404 is further configured to: Using the first agent, based on each component generation task in the plurality of intelligent component generation tasks, generate a component script corresponding to each component generation task; The component scripts corresponding to the component generation tasks are input into the virtual environment for execution, the script execution results of the component scripts are obtained, and multiple smart components are constructed according to the script execution results of the component scripts.

[0135] Optionally, the component library generating module 404 is further configured to: Inputting a target component script corresponding to a target component generation task into a virtual environment for execution, and obtaining a script execution result of the target component script, wherein the target component generation task is any one of the multiple smart component generation tasks; If it is determined that the target component script is successfully executed according to the script execution result of the target component script, encapsulating the target component script to construct a target smart component, wherein the target smart component is any one of the multiple smart components; When it is determined that the execution of the target component script fails based on the script execution result of the target component script, the target component script is updated using the script execution result of the target component script, and the step of inputting the target component script corresponding to the target component generation task into the virtual environment for execution to obtain the script execution result of the target component script is continued.

[0136] Optionally, the device further includes: The task execution module is configured as follows: Receiving a task to be executed, parsing the task to be executed using the second agent, and determining an execution workflow from the target workflow set according to the parsing result; The second agent is utilized to execute the task to be executed according to the execution workflow and the intelligent component library.

[0137] Optionally, the device further includes: Update module, configured as follows: When it is determined based on the execution result of the task to be executed that the task to be executed has failed, the smart component library and / or the target workflow are updated.

[0138] The data processing device provided by one or more embodiments of this specification collects execution trajectory data of a first agent that relies on a first model (i.e., a strong model) to perform a target task. The first agent, based on the strong model, generates a library of intelligent components adapted to the weak model based on the execution trajectory data and system constraints determined based on the model configuration of the second model (i.e., a weak model) that the second agent relies on. The library contains independently reusable intelligent components adapted to the weak model, making the intelligent components of the strong model directly reusable by the weak model. The strong model's task execution capabilities can be reused without additional training, ensuring that the second agent, which relies on the weak model, can still perform complex deep analysis tasks with high quality at low cost. At the same time, the first agent, based on the strong model, generates a reference workflow based on the execution trajectory data and the intelligent component library to standardize the task execution path. This allows the subsequent second agent, based on the weak model, to efficiently execute tasks based on the reference workflow, eliminating dynamic search redundancy caused by the flexibility of the first agent, reducing inference time, and improving efficiency and stability. Finally, the second agent of the weak model is used to execute the test task according to the reference workflow and the intelligent component library. The final target workflow set is precipitated according to the task execution results. Through the closed-loop test verification mechanism of the second agent of the weak model, the boundary risks are intercepted to determine the output consistency, thereby ensuring the improvement of the stability of the second agent of the weak model in task execution.

[0139] The above is a schematic diagram of a data processing device according to this embodiment. It should be noted that the technical solution of the data processing device and the technical solution of the above-mentioned data processing method are based on the same concept. For details not described in detail in the technical solution of the data processing device, please refer to the description of the technical solution of the above-mentioned data processing method.

[0140] Corresponding to the above-mentioned task processing method embodiment, this specification also provides a task processing device embodiment, which includes: a task receiving module configured to receive a task to be executed, parse the task to be executed using a second agent, and determine an execution workflow from a target workflow set based on the parsing result; The task execution module is configured to utilize the second agent to execute the task to be executed according to the execution workflow and the intelligent component library, wherein the target workflow set and the intelligent component library are generated according to the above-mentioned data processing method.

[0141] The above is a schematic scheme of a task processing device of this embodiment. It should be noted that the technical scheme of the task processing device and the technical scheme of the task processing method described above are of the same concept. For details not described in detail in the technical scheme of the task processing device, please refer to the description of the technical scheme of the task processing method described above.

[0142] In addition, this specification also provides a data processing method for conversation analysis tasks, including: Determining execution trajectory data of a target conversation analysis task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target conversation analysis task based on a first model; generating, using the first agent, an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; Using the first agent, generating a dialogue analysis reference workflow based on the execution trajectory data and the intelligent component library; Utilize the second intelligent agent to perform the conversation analysis test task according to the conversation analysis reference workflow and the intelligent component library, and determine the conversation analysis target workflow set according to the task execution result, wherein the conversation analysis test task is determined according to the target conversation analysis task.

[0143] In specific implementation, the specific implementation of the data processing method applied to the conversation analysis task is the same as the specific implementation of the above-mentioned data processing method, and will not be repeated here.

[0144] Corresponding to the above-mentioned data processing method embodiment applied to a conversation analysis task, this specification also provides an embodiment of a data processing device applied to a conversation analysis task, the device comprising: A first determining module is configured to determine execution trajectory data of a target conversation analysis task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target conversation analysis task based on a first model; a first generating module configured to generate, using the first agent, an intelligent component library according to the execution trajectory data and system constraints, wherein the system constraints are determined according to a model configuration of a second model that the second agent depends on, and the model configuration of the first model is greater than the model configuration of the second model; a second generating module configured to generate a conversation analysis reference workflow using the first agent according to the execution trajectory data and the intelligent component library; The second determination module is configured to utilize the second intelligent agent to perform the conversation analysis test task according to the conversation analysis reference workflow and the intelligent component library, and determine the conversation analysis target workflow set according to the task execution result, wherein the conversation analysis test task is determined according to the target conversation analysis task.

[0145] The above is a schematic diagram of a data processing device for a conversation analysis task according to this embodiment. It should be noted that the technical solution of this data processing device for a conversation analysis task shares the same concept as the technical solution of the data processing method for a conversation analysis task described above. For details not described in detail in the technical solution of the data processing device for a conversation analysis task, please refer to the description of the technical solution of the data processing method for a conversation analysis task described above.

[0146] Figure 5 FIG. 5 shows a structural block diagram of a computing device 500 provided according to an embodiment of the present specification.

[0147] The computing device 500 includes: Memory 510 and processor 520; The memory 510 is used to store computer programs / instructions, and the processor 520 is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor 520, the steps of the data processing method, task processing method or data processing method applied to the dialogue analysis task are implemented.

[0148] In one or more embodiments of this specification, the computing device can be understood as an integrated intelligent terminal, including but not limited to a server, a desktop computer, a PC (Personal Computer), an all-in-one model machine, a mobile phone, a tablet computer or other portable intelligent terminal, etc., and the computing device can be pre-installed with the model described in the above embodiments of this application.

[0149] Specifically, the computing device can pre-install multiple types of models, including but not limited to models in the fields of natural language processing, visual processing, speech processing, code processing, and multimodal task processing, thereby providing a diverse selection of models. In different product forms, the computing device can support one or more model usage methods, including but not limited to model training, model calling, model fine-tuning, model deployment, model reasoning, and application. In some product forms, the computing device also supports model management, including but not limited to multi-type model management (supporting the management of multiple types of models such as discriminants and genesis), model version control (supporting the control of different model versions), and model evaluation (evaluating the performance and effectiveness of models based on model evaluation tools). In other product forms, the computing device can also create applications based on models and provide API (Application Programming Interface) calling capabilities, allowing models to be called into created applications through the API interface. Application management tools are also provided to enable management and monitoring of applications.

[0150] Furthermore, the computing device can also include data management (supporting the creation and management of model tuning data sets), a training center (providing rich training resources to help users learn and master AI (Artificial Intelligence) technology), and basic management and control capabilities (providing enterprise-level basic management and control capabilities to ensure the security and efficient operation of the system). Through the above functions, a comprehensive, integrated AI development, training, deployment and application device is provided.

[0151] Figure 6 FIG. 6 is a block diagram of an electronic device 600 according to an embodiment of the present disclosure.

[0152] A memory 610 and a processor 620 , wherein the memory 610 and the processor 620 are connected via a bus 630 ; The memory 610 is used to store computer programs / instructions, and the processor 620 is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor 620, the steps of the data processing method, task processing method or data processing method applied to the dialogue analysis task are implemented.

[0153] Specifically, the components of the electronic device 600 include but are not limited to a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and the database 650 is used to store data.

[0154] The electronic device 600 also includes an access device 640 that enables the electronic device 600 to communicate via one or more networks 660. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 640 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0155] In one embodiment of the present specification, the above components of the electronic device 600 and Figure 6 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 6 The electronic device structure block diagram shown is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0156] Electronic device 600 may be any type of stationary or mobile electronic device, including a mobile computer or mobile electronic device (e.g., a tablet computer, personal digital assistant, laptop computer, notebook computer, netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable electronic device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary electronic device such as a desktop computer or a personal computer (PC). Electronic device 600 may also be a mobile or stationary server.

[0157] The above is a schematic diagram of an electronic device according to this embodiment. It should be noted that the technical solution of this electronic device is based on the same concept as the aforementioned data processing method, task processing method, or the technical solution of the data processing method applied to the conversation analysis task. For any details not described in detail in the technical solution of the electronic device, please refer to the description of the technical solution of the data processing method, task processing method, or the technical solution of the data processing method applied to the conversation analysis task.

[0158] An embodiment of the present specification also provides a computer-readable storage medium storing a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned data processing method, task processing method, or data processing method applied to a conversation analysis task.

[0159] The above is a schematic diagram of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium is based on the same concept as the technical solution of the aforementioned data processing method, task processing method, or data processing method applied to a conversation analysis task. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the aforementioned data processing method, task processing method, or data processing method applied to a conversation analysis task.

[0160] An embodiment of the present specification further provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned data processing method, task processing method, or data processing method applied to a conversation analysis task.

[0161] The above is a schematic diagram of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product shares the same concept as the aforementioned data processing method, task processing method, or the technical solution of the data processing method applied to a conversation analysis task. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the aforementioned data processing method, task processing method, or the technical solution of the data processing method applied to a conversation analysis task.

[0162] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0163] The computer program / instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium may be appropriately increased or decreased based on the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0164] It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0165] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0166] The preferred embodiments disclosed above are intended only to help illustrate this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made based on the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A data processing method, comprising: Determining execution trajectory data of a target task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target task based on a first model; generating, using the first agent, an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; Using the first agent, generating a reference workflow according to the execution trajectory data and the intelligent component library; The second agent is utilized to execute a test task according to the reference workflow and the intelligent component library, and a target workflow set is determined according to the task execution result, wherein the test task is determined according to the target task.

2. The data processing method according to claim 1, wherein determining the execution trajectory data of the target task comprises: Determining a target task set, wherein the target task is included in the target task set; Utilizing the first agent to execute the target task and obtain execution trajectory data of the target task; or Obtain execution trajectory data of the target task from the historical execution log of the first agent executing the target task.

3. The data processing method according to claim 1, wherein generating an intelligent component library using the first agent according to the execution trajectory data and system constraints comprises: Using the first intelligent agent, generating an intelligent component set according to the execution trajectory data and system constraints, wherein the intelligent component set includes multiple intelligent components; The plurality of intelligent components are subjected to structured processing by utilizing the first intelligent agent to obtain the processed plurality of intelligent components, and the intelligent component library is generated based on the processed plurality of intelligent components.

4. The data processing method according to claim 3, wherein the step of generating a set of intelligent components using the first agent according to the execution trajectory data and system constraints comprises: generating, using the first agent, a plurality of intelligent component generation tasks based on the execution trajectory data and system constraints; The first intelligent agent is utilized to generate tasks according to the plurality of intelligent components, and a plurality of intelligent components are constructed to generate the intelligent component set.

5. The data processing method according to claim 3, wherein the step of using the first agent to perform structured processing on the plurality of intelligent components to obtain the processed plurality of intelligent components comprises: Using the first intelligent agent, abstracting the plurality of intelligent components to obtain a plurality of intelligent components after abstraction processing; Using the first intelligent agent, clustering the plurality of intelligent components after the abstraction process to obtain a plurality of intelligent components after the clustering process; The first intelligent agent is used to merge the plurality of intelligent components after the clustering process to obtain the plurality of intelligent components after the merge process, thereby obtaining the plurality of intelligent components after the process process.

6. The data processing method according to claim 1, wherein the step of generating a reference workflow using the first agent according to the execution trajectory data and the intelligent component library comprises: The execution trajectory data is analyzed by using the first agent, and a reference workflow corresponding to the target task is generated according to the analysis results of the execution trajectory data and the component capabilities of the smart components in the smart component library.

7. The data processing method according to claim 1, wherein the step of utilizing the second agent to execute a test task according to the reference workflow and the intelligent component library, and determining a target workflow set according to the task execution result, comprises: Sampling the target task to generate the test task; Inputting the reference workflow and the test task into the second agent, obtaining a task execution result of the second agent executing the test task by calling the smart component in the smart component library based on the reference workflow; A target workflow set is determined according to the task execution result and the reference workflow.

8. The data processing method according to claim 7, wherein determining the target workflow set based on the task execution result and the reference workflow comprises: If it is determined that the test task is successfully executed according to the task execution result, the reference workflow is determined as the target workflow of the target task; If it is determined that the test task has failed to execute according to the task execution result, the smart component library and / or the reference workflow are updated, and the steps of inputting the reference workflow and the test task into the second agent are continued, and obtaining the task execution result of the second agent calling the smart components in the smart component library based on the reference workflow to execute the test task are obtained; The target workflow set is determined according to the target workflow.

9. The data processing method according to claim 4, wherein the step of utilizing the first agent to generate tasks based on the plurality of intelligent components and constructing the plurality of intelligent components comprises: Using the first agent, based on each component generation task in the plurality of intelligent component generation tasks, generate a component script corresponding to each component generation task; The component scripts corresponding to the component generation tasks are input into the virtual environment for execution, the script execution results of the component scripts are obtained, and multiple smart components are constructed according to the script execution results of the component scripts.

10. The data processing method according to claim 9, wherein the step of inputting the component scripts corresponding to the component generation tasks into a virtual environment for execution, obtaining the script execution results of the component scripts, and constructing multiple smart components based on the script execution results of the component scripts comprises: Inputting a target component script corresponding to a target component generation task into a virtual environment for execution, and obtaining a script execution result of the target component script, wherein the target component generation task is any one of the multiple smart component generation tasks; If it is determined that the target component script is successfully executed according to the script execution result of the target component script, encapsulating the target component script to construct a target smart component, wherein the target smart component is any one of the multiple smart components; When it is determined that the execution of the target component script fails based on the script execution result of the target component script, the target component script is updated using the script execution result of the target component script, and the step of inputting the target component script corresponding to the target component generation task into the virtual environment for execution to obtain the script execution result of the target component script is continued.

11. A task processing method, comprising: Receiving a task to be executed, parsing the task to be executed using a second agent, and determining an execution workflow from a target workflow set based on the parsing result; Utilize the second intelligent agent to execute the task to be executed according to the execution workflow and the intelligent component library, wherein the target workflow set and the intelligent component library are generated according to any one of the data processing methods according to claims 1-10.

12. A data processing method for a conversation analysis task, comprising: Determining execution trajectory data of a target conversation analysis task, wherein the execution trajectory data is trajectory data generated by a first agent performing the target conversation analysis task based on a first model; generating, using the first agent, an intelligent component library based on the execution trajectory data and system constraints, wherein the system constraints are determined based on a model configuration of a second model that the second agent relies on, and the model configuration of the first model is greater than the model configuration of the second model; Using the first agent, generating a dialogue analysis reference workflow based on the execution trajectory data and the intelligent component library; Utilize the second intelligent agent to perform the conversation analysis test task according to the conversation analysis reference workflow and the intelligent component library, and determine the conversation analysis target workflow set according to the task execution result, wherein the conversation analysis test task is determined according to the target conversation analysis task.

13. An electronic device comprising: a memory and a processor, wherein the memory and the processor are connected via a bus; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer program / instructions are executed by the processor, the steps of the method according to any one of claims 1 to 12 are implemented.

14. A computer-readable storage medium storing a computer program / instruction, wherein the computer program / instruction, when executed by a processor, implements the steps of the method according to any one of claims 1 to 12.

Images